Oxford University Press, Walton Street, Oxford O X 2 6DP

Oxford New York

Athens Auckland Bangkok Bombay
Calcutta Cape Town Dares Salaam Delhi
Florence Hong Kong Istanbul Karachi
Kuala Lumpur Madras Madrid Melbourne
Mexico City Nairobi Paris Singapore
Taipei Tokyo Toronto PREFACE
and associated companies in
Berlin Ibadan
Oxford is a trade mark of Oxford University Press The material in this book represents the present state of evolution of the
Published in the United States course notes for a graduate level course in the dynamics and control of AC
by Oxford University Press Inc., New York drives. The period of evolution of this subject material reaches all the way
0 D. W. Novotny and T. A. Lipo, 19% back to the 1970's and is ongoing. Publication of the material at this time is not
All rights reserved. No part of th~publicationmay be
because the work is finished but rather a recognition that a formal, hard cover
reproduced, stored in a retrieval system, or transmitted, in any version of the material will be usehl to our students and will contribute to the
form or by any means, without thepriorpehnisswn in writing of Oxford
University Press. Within the UK, exceptions are allowed in respect of any present state of the art.
ffir dealing for the purpose o f research or private study, or criticbm or While the primary audience for the text material has been electrical engi-
revrw. as permitted under the Copyright, Designs nnd Patents Act, 1988, or
in I / I C <.useo f reprographic reproduction in accordance with the terms of neering graduate students, there were also many engineers in industrial posi-
lice~rcesi,ssrted by the Copyright Licensing Agency. Enquiries concerning tions and senior level students in elective courses who were exposed to various
reproduction outside rho.~et e r m and in other countries should be sent to
rllv right.^ Departmenr, Oxford University Press, at the address above. portions of the material. In particular, the concept of field orientation and the
This book is sold subject to the condition that it shall not,
resulting ability to directly control torque was, and is, of great interest to a
by way o f trade or otherwise, be lent, re-sold, hired our, or otherwise wide range of persons. This has influenced the means in which portions of the
circulated without the publisher's prior consent in any form of binding
o r cover ofher than that in which it is published and without a similar
text have been organized with the goal being to produce a flexible body of tuto-
condition including this condition being imposed rial material which can be used in a variety of ways.
on the subsequent purchaser.
Early on we adopted the concept of complex vector representation because
A curalogue record for this book is available from the British Library of its great value as a visualization tool in clarifying the dynamic interactions
Library o f Congress Cataloging in Publication Data between fluxes and currents in electric machines. This concept, despite having
(Data available) originated in the US, is widely used in Europe, Japan and elsewhere but is not
ISBN 0 19 856439 2 commonly used here. Our use of the concept in the development of the mate-
Typeset by the authors rial in this book grew until it is now central to almost the entire book. The
Printed in Great Britain by power of the complex vector as a mathematical tool for simplifying polyphase
Bookcraft Ltd, Midsomer Norton, Somerset
electric machine modeling as well as a conceptual method for visualizing and
interpreting results will become apparent to anyone reading Chapters 2,3 and 4
who has tried to do some of the same things using real variable matrix meth-
ods.
Although the primary purpose of the text material is to treat vector control
and field orientation, a number of other, more conventional topics are also
treated as illustrations of the use of the basic machine model. These include
the development of the standard steady state machine models, constant speed
switching transients, trapped flux transient models, inverter-machine models
 CONTENTS
1 Introduction to AC Drives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
and others. The material has been organized to allow a variety of ways it can
be used by readers with differing backgrounds and needs. To illustrate some of Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I
these possibilities: Review of DC Drives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1) As a graduate level textbook, working through the material in the Comparison of DC and AC Motor Drives . . . . . . . . . . . . . . . . . . . . 4
order presented. 1.3.1 Commutator and Brushes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2) For someone interested in the concept and steady state behavior of 1.3.2 Power vs. Speed Range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5
field oriented machines, skip directly to Chapter 5. The only prerequisite 1.3.3 Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
knowledge is steady state machine theory as presented in any introduc- 1.3.4 Power Factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5
1.3.5 Inenia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.
tory level book.
1.3.6 Protection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..( .I
3) For someone interested in the dynamics of field orientation, skip
1.3.7 Motor Maintcnance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.
directly to Chapter 6. The prerequisite knowledge is d,q machine model- 1 .3.8 Ruggedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .0
ling. This is available in Chapter 2 if needed. 1.3.9 Standstill Performance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7.
4) For an introduction to complex variable analysis, Chapter 2 followed 1.3.10 Size and Weight. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 7
by Chapter 4 provide the basic material plus enough illustrations to Comparison of Synchronous and lnduction Machines. . . . . . . . . . . 7
clearly show the power of the method.
Inverters for Adjustable Speed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
There are a number of example problems in the text and most chapters
have a set of problems at the end to allow self study. These problems are taken The Six Step Voltage Stiff Inverter (VSI). . . . . . . . . . . . . . . . . . . . . 9
from the course we have taught using the text material in note form. Answers Six Step Inverter Speed and Slip Frequency Control Techniques . . 15
are provided at the end of the book. The Pulse Width Modulated Inverter (PWM). . . . . . . . . . . . . . . . . . I 8
We dedicate this book to our many students who have helped shape the 1.8.1 Block Modulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19
material through their participation in our courses over the past years. 1.8.2 Sinusoidal PWM Modulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .20
1.8.3 Synchronous vs. Asynchronous Modulation . . . . . . . . . . . . . . . . . . . . . . 2 1
Madison, WI, U.S.A. 1.8.4 Pulse Dropping. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 2 3
D.W.N. and T.A.L.
1.8.5 Changeover to Six Step Mode. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 4
February, 1996
1.8.6 Losses Caused by Inverters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .24
1.8.7 Torque Pulsations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 2 5
1.8.8 Induction Motor Speed Control Using a PWM Inverter . . . . . . . . . . . . . 25
The Current Stiff Inverter (CSI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
High Performance Drives vs. General Purpose Drives. . . . . . . . . . . 3 1
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

2 d , q Modelling of Induction and Synchronoris Machines . . . . . . . . . . . . .35

2.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 5
2.2 Winding Inductances. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35

vii
 CONTENTS CONTENTS
.......

System Equations in the Stationary a.b. c Reference Frame . . . . . . . 43 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .146

Determination of Induction Machine Inductances .............. 45
Complex Vector Representation of Three Phase Variables ........ 48 4 Complex Vector Anufysis of Inducrion Machines . . . . . . . . . . . . . . . . . .I51

Complex Variable Model of Three Phase Induction Machine . . . . . 52 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .151

Turns Ratio Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . - 5 4 Complex Vector Equivalent Circuit . . . . . . . . . . . . . . . . . . . . . . . . . 152

Transformation to a Rotating Reference Frame . . . . . . . . . . . . . . . .56 Turns Ratio and Modified d.q Models . . . . . . . . . . . . . . . . . . . . . . . 153

Interpretation of Complex Vectors . . . . . . . . . . . . . . . . . . . . . . . . . .61 Steady State Equivalent Circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156

4.4.1 Sinusoidal Excitation in Complex Vector Form . . . . . . . . . . . . . . . . . . . . 156
Power Flow in the d.q Equivalent Circuit . . . . . . . . . . . . . . . . . . . . .68 4.4.2 Stator Referred Equivalent Circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . .158
Example - Stator 0.b.c and Rotor d.q Model (Stator Referred) . . . 70 Invariance of the Steady State Equivalent Circuit . . . . . . . . . . . . . . 164
The Electromagnetic Torque . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 Modified Equivalent Circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
Analysis of Induction Motor Starting Performance Using dq.0 Electrical Transients at Constant Speed . . . . . . . . . . . . . . . . . . . . . . 170
briables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .78 4.7.1 Complex Vector Operational Impedances . . . . . . . . . . . . . . . . . . . . . . . . 171
Extension of d.y. 0 Theory to Analysis of Salient Pole Synchronous 4.7.2 Example - Sudden Application of Stator Current Source Excitation . . . 172
Machines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .88 4.7.3 General Solution for Switching Trensicnts . . . . . . . . . . . . . . . . . . . . . . . 177

Extension of dq.0 Theory to Analysis of Permanent Magnet Motors The Constant Speed Eigenvalues . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
4.8.1 Example - Sudden Short Circuit of llnloudcd Induction Muchinc . . . . . 187
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
.
Transient Equivalent Circuits (Constant Rotor Flux Linkage) . . . . . 190
Rcfcrcnces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .103 4.9.1 Constant Rotor Flux Linkage Modcl . . . . . . . . . . . . . . . . . . . . . . . . . . . .190
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 4.9.2 Transient Equivalent Circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
4.9.3 Evaluation of Voltage Behind Tral~sicntReactance . . . . . . . . . . . . . . . . .192
4.9.4 Example - Sudden Short Circuit of Unloaded Induction Machine . . . . . 194
'\/ .3 ti.c/ Mocirl.sfi)rSolid Srclre Power Converters . . . . . . . . . . . . . . . . . . . . .109
4.9.5 Comments on the Transient Equivalcnr Circuit . . . . . . . . . . . . . . . . . . . . I96
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .109
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
ci.y Model for Voltage Source Inverter . . . . . . . . . . . . . . . . . . . . . . .I09
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
d.q Model for PWM Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . .115
(1.4 Model for CSI System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .116 5 Principles of Vector Control and Field Orirrztution . . . . . . . . . . . . . . . . 203
Inverter d.q Models in a Synchronous Reference Frame . . . . . . . . . 120 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .203
Examples of Inverter-Induction Motor Models . . . . . . . . . . . . . . . . 123 5.2 DC Machine Torque Control. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
Fundamental Component Approximation for Steady State Operation 5.3 Requirements for Torque Control . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 5.4 Synchronous Machine Vector Control . . . . . . . . . . . . . . . . . . . . . . . 208
of VSI and CSI Systems
Vt~i~lily . . . . . . . . . . . . . . . . . . . . . . . . . . .144 5.4.1 CSI -Synchronous Machine (Commu~utorlessDC Machinc) . . . . . . . . 208
5.4.2 Torque Control and Choice of y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .212
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .145
viii
 CONTENTS
Synchronous Machine Steady State d.q Model . . . . . . . . . . . . . . . . 213 6.2.4 Block Diagram of Field Oriented Induction Machine . . . . . . . . . . . . . . 263
5.5.1
5.5.2
Steady State Conditions in d.q Variables . . . . . . . . . . . . . . . . . . . . . . . . .213
d.q Variable Vector Diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
.
6.3.1
e* ..
Indirect Controllers for Induction Machine Field Orientation . . . . . 264
Indirect Controller with ids and iqs as Inputs . . . . . . . . . . . . . . . . . .265
Torque Control Implementations - Synchronous Machines . . . . . . . 217 e* e*
5.6.1 Torque Control Using Field Orientation with a CSI . . . . . . . . . . . . . . . . .217 6.3.2 Indirect Controller with hdr and iyS as Inputs . . . . . . . . . . . . . . . . . . -265
5.6.2 Torque Control Using a CRPWM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .218 6.3.3 Other Indirect Controllers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
5.6.3 Magnitudc -Angle Resolver and Use in CSI Torque Control . . . . . . . . .220 6.3.4 Indirect Control Using a CSI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .268
5.6.4 Torque C'ol~trolRequirements - Synclironous Machine. . . . . . . . . . . . . . 221 6.3.5 Indirect Field Orientation Start Up Transient . . . . . . . . . . . . . . . . . . . . . 269

Electrical Measurement of the Rotor Field Angle 8, . . . . . . . . . . . .222 Direct Controllers for Induction Machine Field Orientation . . . . . . 275
6.4.1 Direct Determination of Rotor Flux Angle . . . . . . . . . . . . . . . . . . . . . . .276
Brushless DC Machines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .222 6.4.2 Measurement of Air Gap Flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .276
5.8.1 Sine Wave Machines with Current Control . . . . . . . . . . . . . . . . . . . . . . .223 6.4.3 Voltage and Current Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
.
5.8.2 Trapezoidal Wave Machines with Currcnt Control . . . . . . . . . . . . . . . . .224 6.4.4 Implementation of Direct Field Orientation. . . . . . . . . . . . . . . . . . . . . . . 280
5.8.3 Other Brushless Machines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .225
.
Induction Machine Field Orientation Using Air Gap Flux . . . . . . . . 282
Induction Machine Vector Control - Steady State . . . . . . . . . . . . .226 6.5.1 Air Gap Flux Referred d.q Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
5.9.1 Conventional Equivalent Circuit Considerations . . . . . . . . . . . . . . . . . . .226 6.5.2 Dynamic Response of Air Gap Flux Controlled Induction Machine . . . .284
5.9.2 Modified Equivalent Circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
. 6.5.3 Steady State Interpretation of Air Gap Flux Control . . . . . . . . . . . . . . . . 286
5.9.3 Torque Control in Terms of and . . . . . . . . . . . . . . . . . . . . . . . . . . . 230 6.5.4 Indirect Field Orientation Controller for Air Gap Flux . . . . . . . . . . . . . .287
5.9.4 Terminal Behavior in Terms of I,,,, and . . . . . . . . . . . . . . . . . . . . . . .233 6.5.5 Direct Field Orientation Controller for Air Gap Flux . . . . . . . . . . . . . . .288
Induction Machine Steady State d,q Model . . . . . . . . . . . . . . . . . . .234 Dynamics of Synchronous Machine Vector Control and Field
5.10.1 Example - Induction Machine Field Orientation . . . . . . . . . . . . . . . . . . .235 Orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .290
Implementation of Field Orientation in Induction Machines . . . . . . 240 6.6.1 d.q Model of Synchronous Machine . . . . . . . . . . . . . . . . . . . . . . . . . . . .290
5.1 1.1 Indirect Field Orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
.. 6.6.2 Vector Control and Angle Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .291
5.1 1.2 Direct Field Orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
. 6.6.3 Dynamics of Synchronous Machine Fleld Orientation . . . . . . . . . . . . . . 293
5.1 1.3 Torque Control Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .247 6.6.3.1 Constant Field Current Operation (Constant Torque Region). . 293
6.6.3.2 Variable Field Excitation (Field Weakening Region) . . . . . . . .295
A Basic Difference Between Synchronous and Induction Machine
6.6.4 Dynamic Response With y+O (Angle Control) . . . . . . . . . . . . . . . . . . . .296
Field Orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
6.6.5 Example - Synchronous Machine Field Orientation . . . . . . . . . . . . . . . .298
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
Field Orientation Using Voltage as the Controlled Variable . . . . . . . 304
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252 6.7.1 Stator Voltage Equations in Terns of Rotor Flux - Induction Machine . 304
6.7.2 Decoupling Equations for Field Orientation . . . . . . . . . . . . . . . . . . . . . .305
6 Dynamics of Vector Control and Field Orientation . . . . . . . . . . . . . . . . 257 6.7.3 Examples of Field Orientation Using Voltage Controlled Invcrtcrs . . . . 308

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .257 Stator Flux Based Field Orientation . . . . . . . . . . . . . . . . . . . . . . . . .309

6.8.1 Mathematical Model of a Stator 1:lux Oriented Induction Machine . . . . 310
6.2 Dynamics of Induction Machine Field Orientation . . . . . . . . . . . . .257 6.8.2 Design of a Decoupler for Direct Stator Flux Oriented System . . . . . . .311
6.2.1 Induction Machine d.q Model with Axis Rotation at an Angular
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .313
Velocity 0, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .258
6.2.2 Rotor Flux Referred Equations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .258 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .315
.
6.2.3 Dynamic Response of Field Oriented Induction Machines ........... 260
 Monographs in Electrical and Electronlc Engineering

Series editors: P. Hammond, T. J. E. Miller, and T. Kenjo

 Monographs in Electrical and Electronic
Engineering
Vector Control and
2 1. Rrrtshlesspermanerl/-magnet and rchct(~r~ce mofor-drives (1989)
Dynamics of AC Drives
T. J. E. Miller
22. Vector control of a.c. machines (1990) Peter Vas D. W. Novotny and T. A. Lipo
23. Brushless servomotors: fundamentals and applications (1990) Y. Dote Department of Electrical and Computer Engineering
and S. Kinoshita University of Wisconsin- Madison
24. Semiconductor devices, circuits, and systems (1991) Albrecht Wisconsin, USA
Moichwitzer
25. Electrical machines and drives: a space-vector theory approach (1992)
Peter Vas
26. Spiral vector theory o f a.c. circuits and mactrines (1992) Sakae Yamamura
27. I'arameter estimatiotr, condirion moniroring, crnd rlirrgnosis of electrical
machines (1993) Pcter Vas I
28. An introduction to ultrasonic motors (1993) '1'. Sashida and T. Kenjo I
29. Ultrasonic motors: theory and applications (1993) S. Ueha and I
Y. Tomikawa I
30. Linear induction drives (1993) J . F. Gieras
31. Switched reluctance motors and their control (1993) T. J. E. Miller
32. Numerical modelling of eddy currents (1993) Andrzej Krawczyk and :
John A. Tegopoulos
33. Rectifiers, cycloconverters, and a x . controllers (1994) Thomas M. ~ a r t b n
34. Stepping motors and their microprocessor controls. Second edition (1994)
T. Kenjo and A. Sugawara
35. Inverse problems and optimal design in electricity and magnetism (1995)
P. Neittaanmaki, M. Rudnicki, and A. Savini
36. Electric drives and their control (1995) Richard M. Crowder
37. Design of brushless permanent magnet motors (1995) J . R. Hendershott
and T. J. E. Miller
I
38. Reluctance synchronous machines and drives (1995) I . Boldea
39. Geometry of electromagnetic systems (1996) D. Baldomir and
P. Hammond I

40. Permanent-magnet d.c. linear motors (1996) Amitava Basak

41. Vector control and dynamics of a.c. drives (1996) D.W. Novotny and 1
T. A. Lipo

C L A R E N I ~ O N1996
PRESS. OXFORD
 CONTENTS CONTENTS

8.4 Optimal Selection of Flux Level . . . . . . . . . . . . . . . . . . . . . . . . . . . .388

7 Current Regulation in Power Converters . . . . . . . . . . . . . . . . . . . . . . . .31 7 8.4.1 Maximum Torque Per Stator Ampere . . . . . . . . . . . . . . . . . . . . . . . . . . .389
8.4.2 Maximum Drive Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 390
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .317 8.4.3 Reduced Sensitivity to Controller Detuning . . . . . . . . . . . . . . . . . . . . . .393
Current Regulated Inverters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 17 8.4.4 Optimal Utilization of the Converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394
7.2.1 The Current Regulated CSI Inverter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318 8.4.4.1 Converter Current Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394
7.2.2 The Current Regulated PWM Inverter ........................... 32 1 8.4.4.2 Converter Current and Voltage Limits . . . . . . . . . . . . . . . . . . . 395
Hysteresis Regulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .397
Ramp-Comparison Controllers . . . . . . . . . . . . . . . . . . . . . . . . . . . .33 1 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .398
.
Stationary Frame Regulators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .335
9 Field Weakening Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
Synchronous Frame Regulators . . . . . . . . . . . . . . . . . . . . . . . . . . . .340
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
.
Feedforward Compensation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .344
Augmented Feedforward Compensation . . . . . . . . . . . . . . . . . . . . .347
Torque Demand and Availability of High Performance Drives . . . . 402
Induction Machine Control Strategy for Constant Horsepower
Augmented Feedforward Compensation with Decoupling. . . . . . . . 348
Operation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .405
Predictive Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
Inverter Imposed Voltage and Current Limits. . . . . . . . . . . . . . . . . .406
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
Maximum Torque Capability Curve Using the Conventional Method
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .351 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .409
.
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352 Scheme for Achieving Maximum Torque Capability . . . . . . . . . . . 410
9.6.1 Transition from Constant Torque to Constant Power Operation . . . . . . . 411
9.6.2 Region #2 -Field Weakening with Constant Output Power . . . . . . . . . . 412
8 Pururneter Sensitivity and Saturation Eflects in Indirect Field 9.6.3 Region #3 -Field Weakening with Constant Speed*Power . . . . . . . . . . 413
Orientation ................................................ 353 Control System Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . .414
8.1 Introduction............................................ 353 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 416
8.2 Phasor Diagram for Detuned Operation ...................... 353 Appendix .per Unit Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417
8.2.1 Field Orientation Torque Characteristics . . . . . . . . . . . . . . . . . . . . . . . . .355
8.2.2 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
Answers to Problen~s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419
8.2.3 Field Orientation Flux Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . .363
8.2.4 Influence of Saturation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .364 .
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .435
8.1.5 Power Loss Considerations ...................................365
. 8.2.6 Example - Detuned Induction Machine Field Orientation. . . . . . . . . . . 369
8.3 Transient Response ...................................... 377
8.3.1 Torque Response with Slip Gain Error ...........................377
8.3.2 Block Diagram Representation of Slip Gain Error. . . . . . . . . . . . . . . . . .383
8.3.3 Simulation of Transient Response with Slip @ i n Error . . . . . . . . . . . . .384

xii xiii
 1 Introduction to AC Drives
1.1 Introduction
An important factor in worldwide industrial progress during the past several
decades has been the increasing sophistication of factory automation. Manw
facturing lines in an industrial plant typically involve one or more variablc
speed motor drives which serve to power conveyor belts, robot arms, overhead
cranes, steel process lines, paper mills and plastic and fiber processing lines to
name a few. Prior to the 1950s all such applications required the use of a dc
motor drive since ac motors were not capable of true adjustable or smoothly
varying speed since they inherently operated synchronously or nearly synchro-
nously with the frqiency of electrical input. The inherent disadvantages of dc
drives, however, have prompted continual attempts to find better solutions to
the problem. To a large extent, applications which require only a gradual
change in speed are now being replaced by what can be called general purpose
ac drives. In general, such ac drives often feature a cost advantage over their dc
counterparts and, in addition, offer lower maintenance, smaller motor size, and
improved reliability. However, the control flexibility available with these
drives is very limited and their application is, in the main, restricted to fan,
pump and compressor types of application where the speed need be regulated
only roughly and where transient response and low speed performance are not
critical.
Drives used in machine tools, spindles, high speed elevators, dynamorne-
ters, mine winders, rolling mills, glass float lines and the like have much more
sophisticated requirements and must afford the flexibility to allow for regula-
tion of a number of variables, such as speed, position, acceleration and torque.
Such high performance applications typically require a high speed holding
accuracy better than 0.5%. a wide range speed control of at least 20: 1, and fast
transient response, typically better than 50 rads for the speed loop. Until
recently, such drives have almost exclusively been the domain of dc motors
combined with various configurations of acldc converters depending upon the
application. With suitable control, however, ac drives are more than a match
2 1 Inrroducrion to AC Drives Review of DC Drives 3

for dc drives in a high performance application. It will be shown in this book In the third type of motor drive, Figure I. l(c), the motor is supplied by a
that control of an ac machine is considerably more complicated than its dc double inverse parallel bridge type of arrangement. Both polarities of dc volt-
motor counterpart. However, with continual advancement of microelectronics, age and current are now possible if the field can again be reversed. In this case,
these control complexities are rapidly being overcome. The gradual replace- torque can be reversed simply by first zeroing the current in the bridge provid-
ment of dc drives has already begun, for example in machine tool drives, and ing motoring current and then increasing the current in the inverse parallel
ac drives can be expected to continue to overtake them over the next decade. It bridge. The response of the bridge is now limited only by the armature leakage
is still too early to determine if dc drives will eventually be relegated to the his- inductance, the sampling theorem limit of the converter, and the commutation
tory book as have nearly all other commutator machines such as rotary con- capability of the dc motor. Care must be taken upon reversing the dc current
verters, amplidynes and the like. However, the next decade will surely witness since a short circuit path exists if both the forward and reverse bridges are trig-
a marked increase in the use of ac drives. gered simultaneously. Hence, the currents in both bridges must remain at zero
for a few milliseconds to allow the forward conducting bridge to recover its
1.2 Review of DC Drives
Figure 1.1 shows three of the most common types of dc drives, consisting of a
one, two and four quadrant drive respectively. In Figure I.l(a) the one quad-
rant converter consists of a single bridge operating in a two pulse or three pulse
mode depending upon the type of supply. The term "one quadrant' is used to
denote that the bridge is capable of only one polarity of voltage and current.
Hence, only motoring operation or acceleration in the forward direction can be
obtained with a single quadrant drive. Adjustment of the dc m a t u r e voltage is
obtained by phase control of the bridge which supplies a prescribed current to
the machine and thus results in a prescribed torque. Braking torque is accom-
plished by the overhauling action of the load which drags the motor to a halt
when the torque from the dc motor is removed.
In F~gureI I(b) a more controlled form of braking is obtained by reversing
the field current in the machine, which, in turn, reverses the counter emf. In
this case the motor is supplied from a four or six pulse acldc converter. The
polanty ol [he converter output voltage can then be reversed by means of phase
control, thereby reversing the power flow so that operation is possible in two Figure 1.1 Major dc motor drive types with both single phase and three
quadrants. Since the field winding has an inherently large inductance, the phase supply. Additional components for three phase supply
response of this system during the torque reversal is limited to about a few shown shaded
tenths of a wcond. Speed reversal can also be obtained by reversing the field as b1'&king ability before turning on the reverse conducting bridge. Speed rever-
the motor reaches zero speed. However, since the torque becomes zero while sal is easily accomplished by reversing the roles of the forward and reverse
the field current is reversed, speed perforinance near zero speed is adversely bridges. Smooth torque reversal is now possible since the current continues to
affected and a smooth reversal through zero speed is not possible with this sys- flow in a single bridge as the motor reverses. Torque response of the system
tem. can now be improved to tens of milliseconds.
 4 1 Introduction to AC Drives 5
Comparison of DC and AC Motor Drives

'
The basic layout of an ac drive employing a converter with a dc link is per second. Although for modem laminatcd frame dc motors a rate of risc of
shown in Figure 1.2. In this case the line voltages and currents are first con- 200 per unit per second is possible, even with his improved cajnbilily this Sea-

- - ture can be the limiting factor affecting torque response in a dc motor. No such
inherent limitation exists in an induction motor and the rates of current rise
I or 3 Phase ' experienced in such a motor is limited only by the leakage inductance of the
AC Supply Converter Converter
Device
- - machine and the amount of voltage available to force the current from one
value to another.
t t Sensed Variables
I 1.3.2 Power us. Speed Range
Speed
Command * Controller I Because of the presence of the commutator the speed of the dc machine IS
inherently limited. In general, for large motors it is not possible to obtain dc
Figure 1.2 Basic ac motor drive employing a dc link motors with a speed*power product greater than 2.6 x 106 (kW)(rpm) For
example, it is difficult to locate a 1500 rpm, 1350 kW dc motor for pu~chase.
Induction motors of this rating can be constructed for speeds reaching several
times this rpm without great difficulty.
verted by a line side converter to dc quantities and subsequently by a machine
side converter to the ac voltages and curtents of variable frequency as required 1.3.3 EfJiciency
by the motor. Both converters consist of a bridge of solid state switching
devices in the case of an ac drive while the machine side conversion function is The efficiency of induction motors used for variable speed operation are gener-
accomplished by means of the commutator in the case of the dc drive. Note ally comparable and frequently better than the equivalent dc motor efficiency,
that the machine side converter together with the ac motor can be considered as even though the pnsencre of the rotor cage adds an additional loss component
the electromechanical equivalent of the dc motor. not encountered in a dc machine. Since the induction motor cage need not be
designed to allow for a direct on line start when driven from a converter, the
cage resistance can be selected solely to provide optimum running perfor-
1.3 Comparison of DC and AC Motor Drives mance and minimum loss. Also, dc motor losses such as brush drop can be
eliminated. + ,
I .3.1 Commutator and Brushes [ I ]
DC motor armatures have two major weak points, the mechanical commutator 1.3.4 Power Factor
and brush assembly. Neither of these components exist on squirrel cage induc- In general, the dc converter of a dc drive operates with an input fundamental
tion motors. The absence of the mechanical commutator means that higher component power factor which ranges from 0 to about 0.9 and increases
speeds are possible with the equivalent induction motor. Also, higher armature approximately in proportion to the motor speed. Unity power factor cannot be
voltages can be used with an induction motor due to limitations in the voltage achieved because the converter requires some voltage margin to allow for sup-
that can be supported between adjacent commutator segments in a dc machine ply voltage dips, dynamic requirements and to prevent loss of commutation
m a t u r e . Transient response is also limited by the rate of rise of armature cur- capability in the inversion mode. Although the induction motor always runs at
rent which is set by the ability of the brush to complete the reversal of the cur- lagging power factor, typically 0.95 at rated load for machines in the 300 kW
rent in the armature coil undergoing commutation. The maximum permissible range, the reactive power requirements of the motor are supplied by the dc link
rate of rise of current in modem solid frame dc motors is limited to 30 times filter capacitor and inverter. With modern pulse width modulated inverters act-
rated current per second. For older motors this limit can be as low as 5 per unit
6 I Introduction to AC Drives Comparison of Synchronous and Induction Machines 7

ing as the machine side converter, the input fundamental component power thermal cycling and creep is eliminated in a squirrel cage winding. Over the
factor remains high, typically above 0.95, regardless of the motor speed. years induction motors have been successfully used under conditions which
could not be sustained by other electrical machine structures. If deterioration of
the rotor does occur, such as a broken bar or end ring, the machine is often
An ac machine of the same power rating And speed will normally have a lower capable of continued operation, perhaps at reduced power, until maintenance
inertia than its dc motor equivalent. In a dc machine, a parameter that affects can be scheduled. Modem computer data acquisition and signal processing
comrnutat~onis the so called reactance voltage of the armature windings which techniques have made possible the early detection of broken bars and the like.
relates to the cornmutating capability of the motor. The need to keep this
parameter within certain bounds, constrains the length of core and hence 1.3.9 Standstill Pegormance
results in an increased diameter for a given power and speed. The lower inertia A normal dc machine cannot develop high torque for extended periods at
of an induction motor translates to faster speed response for a machine with standstill since the armature current then flows through a particular group of
equal torque producing capability. armature coils and commutator segments under this condition. Although spe-
cial mill duty motors can be purchased to provide rated torque over several
1.3.6 Protection I seconds they are essentially oversized relative to the induction motor which
Monitoring and protection of a squirrel cage induction motor is simpler than has no suct limitation.
for a dc motor. Without a direct on line s t k i n g capability and with the ability
of the squirrel cage rotor to withstand much higher temperatures than the sta- 1.3.10 Size and Weight
tor, simple stator temperature monitoring gives comprehensive thermal protec- Mainly because of the commutator assembly, a dc machine of the same torque
tion for the entire machine. Also, for high power applications it is usually capacity is significantly larger than the equivalent induction machine. As a
necessary to use a dc circuit breaker to provide protection for large dc result the 'footprint' of the motor is larger resulting in increased real estate to
machines. These breakers are very expensive and require regular maintenance. house the motor. Access doors must be larger and larger cranes need be used
However, DC circuit breakers need not be used for ac drive systems. for maintenance. All of these factors lead to a greater overall cost for the instal-
lation.
1.3.7 Motor Maintenance Table 1 . 1 shows a comparison of the two motors for three widely different
An essential part of any economic comparison is the cost of keeping spares as sizes corresponding to 6 MW, 75 kW and 1.5 kW [I]. The data for the two
well as the frequency of maintenance. A supply of brushes and brush holders smaller machines are for standard squirrel cage machines while the larger
are necessary stock items for dc motors. A dc motor must be regularly taken machine is specially designed for a variable speed application. Machines spe-
out of service to check or replace brushes and at less frequent intervals to cifically designed for variable speed would normally have a higher efficiency
resurface the commutator. Maintenance of these brushes is particularly burden- due to relaxation of the line starting requirement. Note the substantial irnprove-
some in harsh environmental conditions. Except for the bearings, an induction ment in efficiency and reduction in inertia available in the 6 MW cage induc-
motor is essentially maintenance free. tion machine
1

Being on the rotating member, the rotor windings are the 'Achilles heal' of any F.
!& &Y
1d Comparison of Synchronous and Induction Machines
The synchronous motor is another machine which possesses many of the
motor. In this regard the squirrel cage rotor is clearly far more robust than the
a@btages of the induction motor. However, the synchronous machine needs
dc machine. Progressive degradation of insulation due to electrical stressing,
&$fings, brush& and an insulated winding on the rotor except for small

,.
@%
8 1 Introduction to AC Drives Inverters for Adjustable Speed 9

motors where the field winding can be replaced by permanent rnagnets.The gate turn off thyristors (GTOs) continue to decrease in cost, such situations are
synchronous machine can be operated at unity or even leading power factor expected to diminish in the future.
resulting in a smaller kVA requirement for the machine side converter than for The future for variable speed permanent magnet motor drives is, on the
the induction motor. However, as we will see later in the book, operation at other hand, bright. However, issues concerning the cost of the magnets limit
unity power factor reduces the torque transient response of the machine. Also, their application to relatively small values of horsepower.
torque response is reduced due to the relatively high value of synchronous
Table 1.1 Comparison of Induction and DC Motors for Three Diverse Sizes 1.5 Inverters for Adjustable Speed
The process of converting dc to ac power is called inversion and it is the
inverter which creates the variable frequency from the dc source which is used
to drive an induction motQr at a variable speed. In general, two basic types of
inverters exist which are totally different in their behavior. The so called Volt-
age Stiff (or Source) Inverter abbreviated as VSI is more common and this type
Rating(kW) 6000 6000 79 75 1.5 1.5 of inverter creates a relatively well defined switched voltage waveform at the
terminals of the motor. The dc bus is maintained as voltage stiff by the use of a
Speed (rpm) 60 60 1500 1500 1500 1500
/ Efficiency
(%)
1 95.1 1 92 ( 93.5 1 88 1 75 1 83 1 large capacitor in the dc link. A voltage source inverter differs from the voltage
stiff inverter in that the dc link is established with a dc source (battery for
instance). The resulting motor current is then governed primarily by the motor
Inertia 30000 60000 0.75 0.673 0.0039 0.0098 load and the speed. The VSI is also termed a voltage source inverter. However,
it is clear that the dc link itself is truly not a power source if the link voltage is
(kg m2) supported only by an energy storage element such as a dc capacitor. The VSI is
Length (cm) 1500 1500 86.3 102.2 32.25 41.6 typically subdivided into two forms the so called six step inverter and the pulse
-

Width (cm) 410 473 45 39 17 width modulated inverter.

21
The second type of inverter, the Current Stiff (or Source) Inverter (CSI)
Weight (kg) - - 385 480 23 42.5 provides a switched current waveform at the motor terminals. The dc bus is
maintained as current sta#by use of a large inductor in the dc link. The voltage
reactance which must be compared with the armature leakage inductance of waveform is now governed primarily by the motor load and speed.
the dc motor. Application of wound field synchronous motor drives are there-
fore limited to very high horsepower applications in which their overall cost
1.6 The Six Step volkge Stiff Inverter (VSI) [2]
favors the synchronous motor over an induction motor. An important factor in
this cost trade off is the price advantage of a thyristor bridge which can be A simplified diagram of a basic three phase voltage stiff inverter bridge is
cornrnutated simply by means of the synchronous motor emf versus a force shown in Figure 1.3. Note that the structure of this circuit follows the general
commutated bridge. Since the switch itself must be relied upon to interrupt the depiction of an ac drive shown in Figure 1.2. The rectifier serves to establish a
current in an induction motor due to its inherent lagging power factor, expen- dc potential in much the same manner as in the dc motor drive. In addition, a
sive power switching devices such as bipolar transistors must be used which relatively large electrolytic capacitor is inserted to 'stiffen' the link voltage and
frequently gives a cost advantage in a high horsepower application to the syn- provide a path for the rapidly changing currents drawn by the inverter. It is the
chronous motor drive. However, as better power switching devices such as rms value of this ripple current which determines the value of the capacitor
used. The capacitor is of appreciable size and cost, frequently 2,000 to 20,000
10 I Introduction to AC Drives The Six Steo Voltage Stiff Inverter (VSI) 121 11

reversal inductors per phase. In addition, extra snubbing circuits are placed
across and in series with each thyristor in order to keep them within their spec-
ified tolerances with respect to their turn on and turn off capabilities. The result
is a complex circuit having numerous components and requiring a precisely
timed firing sequence to prevent accidental short circuits. Thyristors have one
important advantage over transistors which is their ability to withstand a sub-
Induction
Machine stantial fault current for a brief time before protective devices (fuses or circuit
breakers) operate. However, when used as an externally commutated switch
they are presently of more interest for historical than for practical reasons.
Figure 1.3 Basic three phase voltage stiff inverter (VSI)
Transistors have nearly completely replaced thyristors in inverter circuits
below 500 kW. They are available in ratings to 1800 V and several hundred
microfarads and is a major cost item in the system. It is also usual to interpose
amperes and increases in the voltage rating have been occurring almost yearly.
some inductive reactance between the rectifier and the ac supply to limit fault
They have the great advantage of being able to be turned off as well as to turn
current and to reduce the severity of the commutation dips produced by the
on the current in the device. Hence, auxiliary components to accomplish turn
rectifier. This series impedance is also helpful in attenuating the voltage spikes
off are not needed. Snubbing is still necessary if the transistors are to be used
which may enter through the rectifier bridge due to switching or lightning
I near their maximum ratings, which is generally an economic necessity. Of the
strikes out in the utility system.
transistor family, the bipolar junction transistor or BJT is the lowest cost and
The inverter acts somewhat as the equivalent of the commutator assembly
thus most widely used. The circuitry required to drive the transistor base is,
in a dc motor and converts the dc voltage to a variable frequency ac voltage.
however, relatively complex and is usually fabricated in a Darlington configu-
The inverter bridge is similar to the rectifier used in a dc motor drive except,
ration to reduce the demand on the base current driver. A small reverse voltage
because of the lagging power factor presented by the induction motor, the thy-
must typically be applied to the base of a transistor in the off state and at turn
ristors must be replaced with devices which are capable of being turned off as
off it is usually necessary to extract current from the base. IGBTs (insulated
well as on. The choice at present is:
gate bipolar transistors) and MOSFETs (metal oxide semiconductor field effect
Thyristors plus External Commutation Network transistors) are more recent additions to the transistor family and promise to
Bipolar Junction Transistors (BJTs) replace BJTs in lower power applications. They do not suffer from second
breakdown, u destructive loss mechanism in bipolar transistors that must be
MOS 1;icltl LXcct Transistors (MOSFETs) carefully controlled. Since both devices are turned off by field effect tech-
Insulated Gate Bipolar Transistors (IGBTs) niques, the current demand on the gate drive is minimal a@ they can even be
turned off with an integrated circuit chip. Overall losses, parts count, and
Gate Turn Off Thyristors (GTOs) driver cost are markedly reduced with these devices resulting in an increas-
MOS Controlled Thyristors (MCTs) ingly competitive product even though the devices remain more expensive
Since thyristors are not inherently capable of turn off, they require external than a BJT.
commutation circuitry not shown in Figure 1.3. The types of commutation cir- Gate turn off thyristors (GTOs) and MOS Controlled Thyristors (MCTs)
cuits are numerous but typically require one additional commutation thyristor
are relative newcomers to the scene and promise to have the benefits of both
and one capacitor (or pair qf capacitors) per phase. Since the capacitor needs to thyristors and transistors. GTOs are both turned on and turned off by applying
short gate current pulses. However, they require complex gate circuitry similar
have its polarity reversed in preparation for a commutation, a technique known
to BJTs. GTOs are presently available to 5000 volts and are becoming widely
as resonant reversal is used which requires at least one and often two resonant
12 I Introduction to AC Drives The Six Step Voltage Stiff Inverter (VSI) [2] 13

used in inverters operating at the utility medium voltage level (>I000V). Such
voltages are a necessary operating condition when the rating of the motor
exceeds a few hundred horsepower.
In addition to the turn off element denoted as an ideal switch in Figure 1.3,
each arm of the bridge normally contains an inverse parallel connected diode.
These diodes which are called the return current or feedback diodes, must be OIL , ,'a

provided to allow for an alternate path for the inductive motor current which
continues to flow when the main power device is turned off. When regenera-
tion occurs'the roles of main power device and diode reverse. The diodes now
return the regenerated power to the dc link while the thyristors carry the reac-
tive current. The return of power to the dc link will raise the link voltage above
its normal value and steps must be taken to absorb this regenerated power to
prevent a dangerous link voltage buildup. Typically a resistor is switched in
parallel with the dc link capacitor to absorb this energy or the input bridge is
made bidirectional by adding a second inverse parallel bridge in much the
same manner as Figure 1.1 (c).
The basic operation of the six step voltage inverter can be understood by
considering the inverter to effectively consist of the six ideal switches shown
in Figure 1.3: While it is possible to energize the motor by having only two
switches closed in sequence at one time, it is now accepted that it is preferable
to have three switches closed at any instant since this strategy produces a Figure 1.4 Illustrating the six connections of the induction motor when
higher output voltage under any operating condition [4]. This second pattern switched by a six step voltage stiff inverter
produces the voltage waveforms of Figure 1.4 at the terminals a, b and c
referred to the negative dc potential n. The numbers written on the level parts voltage drops are concerned, the machine may be represented by three equiva-
of the waveforms indicate which switches are closed. This sequence of switch- lent impedances as shown in Figure 1.4 for the six possible connections. Note
ing is in the order 561, 612, 123, 234, 345,456 and back to 561. The line to that a specific phase is alternately switched from positive pole to negative pole
line voltages VaC,Vb,.V,, and line to neutral voltages V,,,Vb,, V,, then have the and that it is alternately in series with the remaining two phases connected In
waveform shown in Figure 1.5. The line to line voltage contains an rms funda- parallel or it is in parallel with one of the two phases and in senes with the
mental component of other. Hence the voltage drop across the phase is always 113 or 213 of the dc
bus voltage with the polarity of the voltage drop across the phase being deter-
mined by whether it is connected to the positive or negative pole. A plot of one
of the three motor line to line voltages is also given in Figure 1.5. Note that
Thus a standard 460 V, 60 Hz induction motor would require 590 volts at the harmonics of order three and multiples of three are absent from both the line to
dc terminals of the motor to operate the motor at its rated voltage and speed. line and the line to neutral voltages (and consequently absent from the cur-
For this reason a 600 volt dc bus is quite standard in the US for inverter drives. rents). The presence of six steps in the line to neutral voltage waveform is the
Although the induction motor functions as an active rather than a passive reason for this type of inverter being called a sin step inverter. A Fourier analy-
load, the effective impedances of each phase remain balanced. That is, so far as sis of these waveforms indicates a square wave type of geometric progression
 I Introduction to AC Drives Six Step Inverter Speed and Slip Frequency Control Techniques 15
14

of the harmonics. That is, the line to line and line to neutral waveforms contain
115th of the fifth harmonic, 117th of the seventh harmonic, Ill lth of the elev-
enth harmonic and so forth.

1.7 Six Step Inverter Speed and Slip Frequency Control

Techniques
In general, the motor requires a nearly constant amplitude air gap flux for satis-
factory working of the motor. Since the air gap flux is the integral of the volt-
age impressed across the magnetizing inductance and, assuming that the air
gap voltage is sinusoidal,

Hence, the motor requires a constant ratio of volts to hertz (plus a small
additional boost to overcome stator resistance) and the dc voltage must be
adjusted in almost direct proportion to frequency. A variable direct voltage can
be readily obtained by using the controlled rectifier in Figure 1.3. In this case
the inverter continues to operate in the so called six step mode giving the same
motor voltage wave forms as shown in Figure 1.5 at all speeds. Ilowever, as
the frequency is reduced the amplitude of the waveforms can be reduced in
direct proportion (assuming ideal control) as shown in Figure 1.6. In general,
the large dc link capacitor prevents the amplitude of the ac voltage from being
increased as rapidly as the frequency, which is developed with practically no
delay by simply feeding triggering pulses to the inverter switches. Hence, it is
customary to slave the frequency to the dc link voltage as shown in Figure 1.7
to prevent the motor from ever receiving an inappropriate value of Volts/Hertz.
Of the two control 'handles', frequency control is by far the most critical as
small changes in frequency produce large changes of slip frequency and,
hence, large changes in current and torque. By slaving the frequency command
to the dc bus voltage, the rate of change of frequency is generally limited to a
value to which the motor can respond without drawing an excessive current or
without regenerating. However this only solves half the problem since the slip
frequency is also influenced by motor speed. A mechanical overload will
Figure I Line-teline and line-to-neutd voltages across the
induction motor phases when operating from a six step quickly increase the slip frequency and hence motor current. This overload
voltage inverter
16 I Introduction to AC Drives Six Step Inverter Speed and Slip Frequency Control Techniques 17

4-
3 Phase + Voltage
or AC/DC
I Phase Rectifier
AC ---- - Inverter

+I~.o
Limits , ;6
Voltage
J

Regulator

Integrator with
Adjustable Limit

Rate Adjust
dv/dt

Figure 1.7 Open loop speed control system with limits on acceleration,
frequency and current
Figure 1.6 Six step line-to-neutral inverter output voltage waveforms at The most common method for frequency control is to use a voltage con-
three different line frequencies trolled oscillator (VCO)to set the inverter frequency. This can be easily inter-
faced to the demanded speed signal via acceleration and decelerat~onlimits
must be detected quickly and the inverter frequency reduced accordingly so which prevent excessive slip as a result of a sudden change of speed demand.
that the slip frequency and current are kept low. An overload condition is most In cases where speed regulation is of concern, improved regulation can be
easily sensed by monitoring the dc link current. Similarly, an overhauling load obtained by increasing the frequency as a function of dc link current (i.e. load)
will reverse the motor torque and cause regeneration and a rise of dc link volt- thereby adjusting the slip frequency while keeping the rotor speed roughly
age if the regenerated power exceeds the losses in the inverter. If there is no constant. Speed accuracy approaching 1% can be readily achieved in this man-
provision to accept this regenerated power, the motor frequency must be ner. These controls are shown in Figure 1.7. Note that this type of motor drive
increased to prevent excessive regeneration. Thus, both link current and volt- is, in effect, a!n open loop speed controller since speed is not sensed explicitly.
age contribute essential sensed variables in any control scheme. Such drives are well suited to applications in which speed regulation require-
ments are modest (>I-2%). Since a costly tachometer and associated hardware
18 1 In~oductionto AC Drives The Pulse Width Modulated Inverter (PWM)[2] 19

is not required, this type of controller is in widespread use in industry as a gen-

'i
eral purpose drive.
If desired, slip frequency can be controlled directly if a digital speed trans- Stiff
Rectifier Inverter Induction
Motor
ducer is fitted to the motor shaft. The inverter frequency can then be generated AC ---
by the digital addition of the shaft speed pulses and slip frequency pulses
obtained from a VCO which determines slip frequency. This addition process
must be carried out at a high frequency so that each inverter period of 60
degrees is comprised of a large number of shaft pulses to which are added a
number of slip frequency pulses which is sufficiently large so that the inverter
periods are of reasonable uniform duration. If regeneration is required the
reversed sign of the speed error signal must cause the slip frequency VCO to
subtract shaft pulses from the shaft transducer on a pulse for pulse basis to
reduce the inverter frequency below synchronism and consequently produce
regeneration. A typical number of pulses corresponding to full slip is 20-30.
The additionlsubtraction process must correctly take into account coincident or
overlapping pulses. However, this is a relatively simple problem with modern
digital electronics. Figure 1.8 shows such an arrangement. Integrator with
Adjustable Limit
It should be noted that precision speed control can be obtained without the
need for a closed loop speed control if the induction motor is exchanged for a
synchronous reluctance or permanent magnet motor. Speed holding precision
is then totally dependent on the stability of the VCO. If the VCO is replaced by
Figure 1.8
P Rare Adjust

Closed loop speed control with digital addition of speed and

a crystal oscillator, divider and phase locked loop, then the precision can be as
slip frequency
good as that of the crystal. This technique has found extensive application in
the fiber spinning industry where a large number of motors must be speed-con- sinusoidally, to synthesize a more nearly sinusoidal motor current (called sinu-
trolled with great precision and in exact synchronism with each other. soidal pulse width modulation or sinusoidal PWM).

1 .a.1 Block Modulation

1.8 The Pulse Width Modulated Inverter (PWM) [2]
Block modulation is the simplest type of modulation and is closest to simple
The PWM inverter again utilizes the voltagk stiff dc link but combines both six step operation. Instead of varying the amplitude of the motor voltage wave-
voltage control and frequency control within the inverter itself. It operates form by variation of the dc link voltage, it is varied by switching one or two of
from a fixed voltage dc source, i.e. an uncontrolled diode rectifier or a battery the inverter thyristors at a fixed switching ratio and adjusting the notches of the
(or perhaps both). In this case the power switches in the inverter are switched resulting pulses to control the motor voltage to suit the speed. The number of
at a high frequency thus operating, in effect, as cho*rs. In general, modula- pulses is maintained constant over a predetermined speed range although the
tion techniques fall into two classes: those which operate at a fixed switching
ratio to the fundamental switching frequency (block or picket fence modula-
:, of pulses may change discretely at several prescribed speeds.
A'simple form of block modulation is shown in Figure 1.9 where the chop-
tion), and those in which the switching ratio is continuously changing, usually : <5,

ping is Limited to the middle 60 degrees of each device conduction period

resulting in minimum switching duty on the semiconductor switches. In spite
20 I Introduction to AC Drives The Pulse Width Modulated Inverter (PWM) 121 21

Figure 1.9 q p i c a l block modulation waveforms showing voltages

from line to negative dc bus and line to line voltage Vab
of the similarities between block modulation and the six step mode, the torque
pulsations at low speed are much less severe than for the six step inverter. The
harmonics typical of a six step inverter are present also with block modulation,
but there are also higher harmonics associated with the chopping frequency.
The motor losses and motor noise, however, are significant compared to more
Details of
Single
:
elegant modulation algorithms and this method is no longer widely used. Pulse

1.8.2 Sinusoidal PWM Modularion

The objective of sinusoidal PWM is to synthesize motor voltages to produce
currents as near to a sinusoid as economically possible. The lower order volt-
age harmonics can be greatly attenuated, leaving typically only 2 or 4 harmon- Figure 1.10 Illustration of sinusoidal pulse width modulation
ics of substantial amplitude close to the chopping or carrier frequency. The (PWM)technique
motor now tends to rotate much more smoothiy at low speed. Torque pulsa- sine waves are used with 120 degrees phase displacement, the switching pat-
tions are virtually eliminated and the extra motor losses caused by the inverter tern for the six inverter devices ensues.
are substantially reduced. To counterbalance these advantages, the inverter
- 1.8.3 Synchronous vs. Asynchronous Modulation
control is complex, the chopping frequency is high. typically 5W2500 Hz for
GTOs and up to 5,000 or more for IGBT transistors, and inverter losses are If the carrier and modulation are to be synchronized, there must be a fixed
higher than for the six step mode of operation. In one method for approximat- number of carrier cycles in each modulation cycle. If this number is chosen to
ing a sine wave a high frequency triangular wave is compared with a funda- give a 'good' sinusoidal current waveform at low frequency, say at 1 Hz, then
mental frequency sine wave as shown in Figure 1.10. When the low frequency the ratio of carrier frequency to fundamental frequency will be in the neighbor-
hood of 100-200. If the same ratio is then used at a motor frequency of 100 Hz,
22 1 Introduction to AC Drives The Pulse Width Modulated Inverter (PWM) [2] 23

the chopping frequency becomes 10-20 kHz. This switching frequency is far wise constant, i.e. sampled at the carrier frequency. An example of carrier and
too high for normal thyristor inverters and somewhat too high even for bipolar modulated waveforms for such a scheme is reproduced in Figure 1.12 for a
transistor inverters. Hence, a technique of 'gear changing' must be employed
which reduces the switching ratio in steps as the fundamental frequency
increases. As shown in Figure 1.11, a commercially available system employ-
ing a custom chip designed for such a purpose is now available 151. Hysteresis
must be included at each change in switching frequency ratio so that the sys-
tem does not cycle continuously between two differing switching ratios at cer-
tain speeds.

Chops
,
250
Asynchronous PWM

Per
Second 4 1000 --

fc/f = Chops Per Cycle

Line Frequencv,f ( H z )
Figure 1.1 1 Illustration of 'gear changing' in a pulse width modulated
inverter Figure 1.12 Switching pattern using regular sampling modulation
Whereas the carrier frequency must be a fixed ratio of fundamental fre- C

quency at moderate and high speed to prevent undesirable subharmonics from $witching ratio of 9 [ 6 ] .
p 6

appearing, asynchronous operation of the carrier frequency becomes accept- t'

V . 8 . 4 Pulse Dropping
able at low speeds where the effect of the differing number of carrier cycles per
modulation cycle is small. However, a change over to synchronous operation is Refemng to Figure 1.10 it can be noted that the center pulses of the PWM
required at some speed and some type of phase locking technique must be modulation are narrower than the pulses where the fundamental voltage passes
introduced to achieve the synchronization. through zero. These center pulses typically become progressively more and
While Figure 1.10 provides a good picture of the modulation process, once more narrow as the speed (and hence voltage) increases. When the conduction
the possibility of using a microprocessor is introduced, the 'analog' solution . pulses become very narrow there is insufficient time to reverse bias the off
obtained by the intersection of a triangle wave and a sine wave becomes only going power device in force comrnutated systems. Hence, if the voltage across
one of many possibilities. This so called natural modulation can be replaced to this device increases in a positive sense too soon, the device will again conduct
advantage by regular modulution in which the modulating waveform is piece and high losses will occur. Hence, if the modulation scheme calls for a pulse
24 I Introduction to AC Drives The Current Stiff Inverter (CSI) 25

width below this minimum time, then this pulse must be omitted altogether and the power switching device and losses in the commutation circuits themselves
a cancelling adjustment made to a nearby pulse to correct for the change in the (if any). The last three of these losses increase with inverter output frequency.
fundamental component. Once again, pulse dropping must be coupled with a An attempt to synthesize the l p t possible sine wave for the motor by selecting
small hysteresis to avoid cycling across the threshold condition. I a high carrier frequency may well create more losses in the inverter than it
saves in the motor. Overall efficiencies at full speed for an inverter fed ~nduc-
1.8.5 Changeover to Six Step Mode tion motor drive lie between 70% for the smaller industrial drives 10 over 90%
Without pulse dropping techniques it is apparent that sinusoidal modulation for large drives such as those used for traction.
prevents the motor terminal phase voltage from reaching its maximum value
(0.78 V,) since 'notches' of minlmum width imposed by the switching limita- 1.8.7 Torque Pulsations
tions of the solid state switching devices prevent full fundamental voltage from Torque pulsations, sometimes called cogging torques, are always present when
being reached. Operation at a reduced 'maximum' fundamental voltage is the motor current is non-sinusoidal, but for all but the lowest speeds the rotor
costly since the attendant motor must be designed for a lower voltage, imply- inertia effectively smooths them out to give a uniform speed. The six step volt-
ing in turn. higher currents for a given horsepower requirement. This increased age stiff inverter produces noticeable torque pulsations below about 3 Hz.
current duty then impacts on the choice of switching device used in the inverter However, with sinusoidal PWM these effects are virtually eliminated. In addi-
design and results in a more costly system. This disadvantage can be overcome tion to torque pulsations which are an inherent result of the harmonics in the
if the modulation process is allowed to 'saturate' at high speeds, so that the applied voltage waveform, a torque transient may occur whenever there is a
waveform takes on first several forms of block modulation and, finally, full six sudden change of motor voltage or frequency. The latter can be avoided elec-
step operation. During the intervening process when the transition is made tronically, but considerable care must be taken in the design of the modulation
from sinusoidal PWM through the various block modes to square wave opera- scheme, particularly when changes in pulse number occur through pulse drop-
tion, additional harmonics will exist which cause extra heating in the motor. ping, to ensure that the effective fundamental voltage before and after is unal-
However, since the motor is running near its rated speed, cooling is good and tered.
there is an overall benefit.
1.8.8 Induction Motor Speed Control Using a PWM Inverter
1.8.6 Losses Caused by Inverters [3]
.Figure 1.13 shows a speed controller based on use of a PWM inverter. Note
The non-sinusoidal output of the inverter creates extra losses in the motor. The that in this case the time delay introduced by the link capacitor can be avoided.
six step mode is worst with respect to voltage harmonic content with 20% of While the voltage and frequency can now be changed almost instantly, acceler-
the 5th harmonic. 14% of the 7th and so forth. Sinusoidal PWM reduces the ation and deceleration limits are again used to prevent changing the slip too
low order harmonics to very small proportions, but significant harmonics rapidly, which as noted previously, could result in an over current condition
remain at or near at the carrier frequency. Frequency dependent losses such as E

skin effect and iron losses often result in motor losses being greater under
1.9 The Current Stiff Inverter (CSI)
PWM operation than with a six step inverter. Different manufacturers give
varying degrees of derating, but typically the motor can give 80-90% of its Since induction motors have traditionally been designed to operate from a volt-
nameplate output at its rated speed without an increase from its rated operating age source, the voltage stiff inverter was developed and used first since it is, in
temperature. At lower speeds extra derating is required to counteract the poorer principle, an approximation of the waveform presented to the motor by the util-
cooling. The inverter also has internal losses caused by device conduction volt- ity. The current stiff inverter on the other hand is very different in concept.
age drop, losses in the auxiliary voltage supplies and gatehase drives, switch- Current fed inverter drives have been in use only for about twenty years. They
ing losses, losses in snubbers or other circuits to aid the switching capability of have, however, several properties which make them attractive as well as an
26 1 Introduction to AC Drives The Current Stiff Inverter (CSI) 27

the current is switched sequentially into one of the three motor phases by the
I.
top half of the inverter and returns from another of the phases to the dc link by
3 P/IUC the bottom half of the inverter. Since the current is constant, there will be zero
or
1 Phase voltage drop across the stator winding self-inductance and a constant voltage
AC
drop across the winding resistances. Hence, the motor terminal voltage is not
set by the inverter but by the motor. Since the motor is wound with sinusoi-
dally distributed windings the consequent voltages which appear on the termi-
nals of the motor are nearly sinusoidal. Ideally, the current waveform is an
exact replica of the voltage waveform of a six step voltage stiff inverter. In
practice, the motor currents cannot change instantaneously, and the transitions
in the current waveform have a finite slope. During these transitions the current
lntcgmor with transfers from one inverter thyristor to the next with the aid of one of six com-
Adjustable Lirnrr
mutating capacitors. A sketch of the motor voltage and currents is shown in
Figure 1.15.
Rate Adjust
A very attractive feature of a CSI is its ability to ride through a commuta-
tion failure and to return naturally to normal operation. This feature is in
Figure 1.13 Basic induction motor speed control block diagram using a marked contrast to most voltage fed inverters in which costly preventive mea-
PWM inverter sures are necessary. Another benefit of the CSI is its ability to regenerate into
inevitable number of undesirable effects. As the name implies the inverter the utility supply by simply reversing the polarity of the rectifier dc output
switches of a CSI are fed from a constant current source. While a true constant voltage. Again, this in contrast with the VSI where the current flow rather than
current source can never be a reality it is reasonably approximated by a con- dc voltage must be reversed, thereby requiring an additional inverse parallel
trolled rectifier or chopper with a current control loop as well as a large dc link connected six pulse bridge. This again is a desirable advantage over a VSI.
inductor to smooth the current. The circuit is shown in Figure 1.14. In this case Operation of the motor at negative slip automatically causes a reversal in the
P dc link voltage since the link current rather than voltage is the controlled vari-
able. Hence, power is automatically regenerated into the ac supply.
An important limitation in the application of a CSI drive is the fact that
open loop operation in the manner of a VSI is not possible. Figure 1.16 shows
the torquelspeed curve of a typical induction motor fed from both a voltage
source and a current source. A marked 'peaking' exists in the case of the cur-
3 Phase rent source inverter. Inspection of this torqudspeed curve suggests two possi-
A C Supply
ble operating points: one on the negatively sloped region which is usually
regarded as stable and one on the positively sloped region which occurs at a
speed below the point where the maximum torque is reached (breakdown
torque) and is generally unstable (depending, however, upon the load torque
vs. speed characteristics).
n Careful examination of the point on the stable side of the torque speed
Figure 1.14 Schematic of current stiff inverter (CSI) drive curve reveals that continuous operation is not feasible here because the work-
28 I Introduction to AC Drives The Current Stiff Inverter (CSI) 29

- Voltage Characteristics
4
No Saturation

ROTOR SPEED IN PER UNIT

Figure 1.16 Induction motor torque-speed curves operating from a
voltage source and a current source
current source) means that operation is impossible without some sort of feed-
back control which forces the machine to remain at this operating point. One
system is a motor voltage control loop which regulates the motor voltage by
controlling the input phase control rectifier (or dc link chopper). A representa-
tive block diagram of a system of this type is shown in Figure 1.17. Typically,
an internal current loop is used as shown with the voltage error serving as a ref-
erence signal for the current regulator. Some IR drop compensation is often
Figure 1.15 Illustration of the 6 connections of an induction motor when added and, since motor current and dc link current are proportional, this is eas-
switched by a six step current stiff inverter showing motor ily accomplished. Additional compensating circuits are usually employed to
line currents and one phase voltage improve system dynamics.
The size and cost of the ac comrnutating capacitors and the dc link inductor
ing flux in the machine at this point is high, resulting in saturated operation and
are the major disadvantage of this inverter. The capacitors are large because
excessive magnetizing current and iron losses. On the other hand at the other
they must absorb the total energy stored in the leakage field of the winding
point, the so called unstable point, the flux in the machine is near its rated value
when the current is commutated. In order to keep this energy interchange at a
and losses are not excessive. (Note that this point corresponds to the intersec-
minimum, motors of special design are advantageous. In particular, these
tion of the torquelspeed curve for rated voltage with the torquelspeed curve for
motors are designed to have as small a leakage reactance as possible. Unfortu-
rated current. At this point both voltage and current are thus at their rated val-
nately, this requirement is at odds with conventional motor design since small
ues.). Unfortunately, being on the unstable side of the torque speed curve (for a
leakage reactances imply an over designed motor from the point of view of
30 1 Introduction to AC Drives
High Performance Drives vs. General Purpose Drives 3I

Three tions. In this case, the three top switches are sequentially switched, one at a
PI~n.se
nc time, to feed the link current into the motor/capacitor network and one of the
Supply
bottom three switches returns the current to the link. The capacitors are now
selected to absorb only the harmonic currents and need not be sized to absorb
the total energy in the comrnutated phase during switching. The motor current
is much more sinusoidal than the square wave current impressed on the motor
in the conventional scheme. Pulse width modulation is again used to suppress
the low frequency 5th and 7th harmonic torque pulsations which are inherent
in six step operation. A major disadvantage of this scheme is the potential for
resonance between the capacitors and the motor inductances. Care must be
taken to avoid impressing current harmonics into the motor/capacitor network
which will excite one of the system resonant frequencies. This possibility can
Figure 1.17 Block diagram of CSI-induction motor drive with motor
voltage feedback be avoided by careful use of pulse width modulation. However, since the motor
parameters must be known to implement such an approach, the drive is pres-
continuous rating. Finally, the use of thyristors rather than transistors or turn
ently not popular for general purpose applications.
off thyristors in this circuit can be noted. While the CSI could potentially
employ switches with turn off capability, they would not enhance the perfor-
mance of the converter significantly since the rate of change of motor current 1.10 High Performance Drives vs. General Purpose Drives
and, thus, the voltage rise across the motor would remain fixed by the capaci-
I At present, the vast majority of ac drives manufactured in this country and
tors. abroad concern applications which do not require fast transient response.
'Turn off devices can be used in a current stiff link configuration if the Accordingly, the design of control circuits to service such ac drives are based
capacitors are freed from their cornmutating requirements and simply placed on the so called 'per phase' T equivalent circuit of the induction motor [7],[8].
across the terminals of the machine. Figure 1.18 shows a motor drive configu- Such an equivalent circuit, however, is derived assuming steady state condi-
tions. In the case of an induction machine, the resulting equivalent circuit
assumes that the output power can be represented by an equivalent rotor resis-
tor of (1-S)R2/S where S is the per unit slip. While this resistor is present in the
equivalent circuit during transients, it is a fictitious resistor which does not
truly represent the electrical behavior of the motor. Thus, the electrical time
constants as culculaled from this equivalent circuit are incorrect. Design ol'
speed controllers such as Figure 1.7 can still proceed since speed is set in open
loop fashion and the motor voltages and currents are not used except to per-
form a limiting function. Hence, the electrical time constants do not markedly
influence the control gains.
Figure 1.18 Current stiff inverter drive employing turn off devices When a closed loop speed control, such as Figure 1.18, was implemented
(GTO is shown as an illustration) on this basis, however, it became quite apparent that the motor time constants
,now began to influence the control gains [9]. When speed regulation require-
ration which is becoming increasingly popular in high horsepower applica- ments are not tight this control still yields adequate speed regulation of a few
32 1 Introduction to AC Drives References 33

per cent. However, continued increase in gain to improve speed regulation and from the general purpose drive controller. The essence of field oriented control
improve transient response leads to an instability and performance degradation is that both the flux producing component and torque producing component 01-
at low speed which can not be predicted from the per phase equivalent circuit the stator current are identified and then expressed as equivalent three phase
Clearly, additional factors are involved in control of an induction machine. current commands. Regulation is then used to ensure that the stator currents are
These questions were largely answered by K. Hasse in his classic Ph.D thesis faithful reproductions of the commanded values.
in 1969 which also pointed the way to improved controllers [lo]. In this work, While the principle of field oriented control is relatively straightforward, a
the essential need for the control of the spatial position of rotor flux and stator familiarity with motor behavior beyond that afforded by the simpler per phase
current was identified. It was emphasized that the electromagnetic torque is equivalent circuit is clearly necessary. In the next chapter we return to the
fundamentally the cross product of flux and current in much the same manner basics to derive a true transient model of the induction machine. In subsequcnt
as other ac and dc machines. This is in contrast to the conventional per phase chapters we use this model to develop a keen understanding of the principles
equivalent circuit approach which, in effect, views torque production as only involved in the design of a field oriented controllers for both induction and
frequency dependant and ignores spatial considerations. That is, torque pro- synchronous machines.
duction becomes equivalent to the power dissipated in a resistance inversely
proportional to per unit rotor slip [i.e. ( I ~ ) ~ R ~ / s ] . References
The requirement for accurate spatial positioning of the stator ampere turns N.D. Matheson, "Evaluation of Improved Induction Motor Control Schen~cs".Master of
with respect to the rotor flux has led to an emphasis on precise regulation of the Eng. Science Thesis, University of Sydney. Sydney Australia. 1988.
stator current based on information as to the spatial position of the rotor flux R.M. Davis. "Inverter-Fed Induction Machines". Proc. of DrivcslMotorslCo11trols'X2.29
density. This form of modulation, termed current regulated pulse width modu- June-l July 1982, Leeds U.K.. pp. 66-75.
lation or CRPWM will be discussed in Chapters 4 and 6. This type of approach D.W. Novotny. "A Comparative Study of Variable Frequency Drives for Energy Con-
servation Applications", University of Wisconsin Report #ECE-8 1 4 , 198 I .
to ac motor control is variously called jield oriented control or vector control
T.A. Lipo and F.G. Turnbull, "Analysis and Comparison of Two Types of Square Wave
depending upon whether the motor variables are viewed as distributed field
Inverter Drives". IEEE Trans. on Industry Application. Vol. IA-I I, No. 2, March/April
quantities (flux densities and MMFs) or circuit variables, (flux linkages and
1975, pp. 137-147.
currents). A simplified block diagram of a high performance drive is shown in
J.A. Houldsworth and W.B. Rosink, "Introduction to PWM Speed Control Systenl for 3-
Fig. 1.19. Note the presence of the inner current loop which is notably absent Phase AC Motors", Electroqics Components and Applications, vol. 2.. No. 2. February
1980.
S.R. Bowcs. "New Sinusoidal Pulsewidth-Modulatctl Invcrtcr". Proc. 114:. vol. 122. No-
vember 1975, pp. 1279-1285.
C.J. Amato, "Variable Speed with Controlled Slip Induction Motor", in Conf. Record of
IEEE Industrial Static Power Conversion Conf., Nov. 1-3, 1965, pp. 18 1-1 89.
J.T. Salihi. P.D. Agarwal and G.J. Spix. "Induction Motor Control Scheme for a Battery

racho &wgM-[4 Powered Electric Car (GM - Electrovair I)", in Conf. Rec. IEEE Industry and General
Applications Society Annual Meeting. October 1967, pp. 107-1 14.'
A.B. Plunkett, J.D. D'Atre and T.A. Lipo, "Synchronous Control of a Static AC Induc-
tion Motor Drive", IEEE Trans. on Industry Applications, Vol. IA-15, No. 4, JulyIAu-
gust 1979, pp. 430-437.
K. Hasse, "Zur Dynamik Drehzahlgeregelter Antriehe Mit Stromrichtergespeisten Asyn-
Figure 1.19 Simplified block diagram of high performance induction
motor drive incorporating field oriented control chron-Kurzschlublaufermaschinen". Ph.D. Dissertation, Tech. Hochschule Darmstadt.
July 17, 1969.
 2 d,q Modelling of Induction and
Synchronous Machines

2.1 Introduction
The induction motor is truly the workhorse of the industrial world. Until
recently, however, they were limited only to driving loads which required
nearly constant speed (or at best several discrete speeds) since the efficiency of
the motor suffered drastically as the speed was lowered below the rated speed.
With the advent of variable frequency power supplies this problem has been
largely overcome. However, costs associated with the inverter have thus far
prevented across the board replacement of dc motors by ac induction motor
drives. Analysis of induction machines is typically presented by means of the
so called per phase equivalent circuit. However, this model proves inadequate
when applied to dynamic conditions as encountered in a variable frequency
drive.
We will begin our study by first determining the coupled circuit equations
of the induction machine. While the result will be a formidable set of equations
we will learn that by using a complex vector representation these equations
reduce quite readily to a manageable form. The theory is then extended to the
synchronous machine including the effects of non-uniform air gap and non-
symmetric rotor windings typical of such a machine.

2.2 Winding Inductances

Let us begin by considering a simplified representation of the winding distribu-
tion of one phase of a three phase induction machine as shown in Figure 2.1.
For purposes of analysis we will assume that a uniform air gap induction
machine can be modelled initially as two concentric cylinders separated by a
thin air gap of constant radial length g. It will be assumed that the surface of
the stator is equipped with sinusoidally wound three phase windings. We will
neglect the effects of slotting so that the windings can be visualized as being of
negligible depth and existing in the air gap. Also, harmonic components of flux
36 -.
2 d.q Modelling of Induction and Synchronous Machines Windine Inductances 37

In the following discussion the permeability of the stator and rotor iron is
assumed to be infinite and saturation and iron losses, as well as end wind~ng
and slotting effects are neglected. For convenience, we will assume a two pole
construction. However, the theory can be readily adapted to multi-pole
machines at a later stage. A visual representation of the winding distribution
for a typical phase of either the stator or the rotor is shown in Figure 2.2. We

Winding Densiry 11 = ,sin 5 (TurndRudiun)

Figure 2.1 Placement of conductors of one stator or rotor phase around

the air gap of a two pole machine. The details of the slot
shape are not shown for clarity.

arising from the placement of the actual conductors in the discrete slots (har-
monic leakage flux) will be neglected.
In general, the rotor of the machine may be either a winding placed in slots
(wound rotor machine) or consist of a simple set of shorted bars placed in the
rotor slots (squirrel cage winding). In the case of the wound rotor machine the Figure 2.2 Idealized induction machine illustrating sinusoidal
distribution of one phase winding
rotor windings can be idealized in the same manner as the stator windings.
When the machine is equipped with a squirrel cage rotor the corresponding
idealizing assumptions are less evident. However, it is important to note that will designate the winding as phase 'a'. If the winding is sinusoidally distrib-
the flux density in the air gap, being spatial in nature, is sinusoidally distrib- uted, we can represent this fact by a winding density distribution of the form:
uted spatially even for arbitrary temporal (time dependant) values of stator 11 (5) = rl, sin5 (2.2-1)
winding current. Hence, rotor currents are always induced which are also sinu-
where we have located the reference point for the angle 6 at the magnetic axis
soidally distributed spatially even if the rotor consists of a shorted cage. As a
of the winding, i.e. the maximum value of the fundamental component of air
result, we can replace the actual winding distribution (actually an n phase con-
gap flux density produced by the winding. Since qm is not normally specified
centrated winding) with an equivalent three phase sinusoidalIy distributed
explicitly, it is first necessary to relate the peak value of this winding density
winding. For further discussion of the procedure for replacing the squirrel cage
by an equivalent sinusoidally distributed winding see Langsdorf [I].
 2 d,q Modelling of Induction and SynchronousMachines Winding Inductances 39
38

distribution q, to the number of turns per pole pair. If the number of tums is N,
then, clearly
lt

Iq,,
sintdt = N, (2.2-2)
0

which results in

The number of turns N, is called the effective number of turns and, in general,
is related to the actual total number of turns N, by
= kpkdk.sN~
(2.2-4)

where kp, kd, and k , are the winding pitch, distribution and skew factors respec-
tively.
Let us now locate a small portion of the winding defined by the differential
dN. If this differential portion of the winding is located at 6, we can write,

Figure 2.3 Differential number of winding turns of winding x

We evaluate first the effects of such a small incremental winding acting by

itself. For this purpose consider a line integral around the path 12341 shown in
Figure 2.3 By Ampere's Law we can write,
and

Since R must he continuous across the air-iron boundary and since pi >> pa.
We have dropped the subscr~pl' 12' for convenience. From symmetry it is clear
that the field intensity is uniform (i.e. constant) around the gap and, assuming
that positive polarity corresponds to the field intensity H directed from rotor to
Hence, H,,,,, can be neglected or, alternatively, + 'saturation factor' can be in- c
stator, it is negative during the interval n < < 2n. This solution is, in fact, the
field produced by what is termed a concentrated full pitch coil and corresponds
lroduced to later account for Hi,,. If the air gap is small and uniform, the field
to a what is essentially a square wave of flux density in the air gap.
intensity is constant over the paths 12 and 34, so that eqn (2.2-6) becomes
Since all field lines are perpendicular to the surface of interest the differen-
H I : ( ~ , + O R + H ~ ~ ( ~= ~d N - (O5 , R
)Ix (2.2-8) tial flux crossing the gap at any point along the air gap is
c
where the functional dependance of H 1 2 and H34 on 6, and are explicitly
shown. By symmetry HI2= H34. Thus, in general, for an arbitrary value of 5,
40 2 d,q Modell~ng-ol Induct~onand Synchronous Machmes
-- -
Winding lnduc~anccs
.- - -- -- .... ..- -- -.

Flux Packet &

4 Conductors linked by

which becomes, with the aid of eqn (2.2-5) and (2.2-9),

d @ = p o ~ x ~ x ( & ) s i n ~( docdd~l ) for 0< c i< x

(2.2-12)
Return Path of
Flux Packet d@
and is thc rlcgative of this vi~luefor clsewherc. Magnetic Axis of
Rcl'erring to Figure 2.4 i l is important to ohscrvc I h i ~ cach
t prrckct of flux d@
links a different number of turns. The number of turns linked by a differential
<
packet of tlux located at a position relative to 6 , is
So+{
N = I Nx
-i.~inudu (2.2-1 3)
5" - 6
where u is a dummy variable of integration. Equation (2.2-13) can be evalu- Figure 2.4 Showing winding portion linked by an arbitrary flux packet
ated as located at 6 + to
N = Nx ~ i n 5 ~ s i n r (2.2-14)
The total flux linkage is the sum of the linkages from all such differential
Assuming no fringing, the flux is evenly distributed throughout the length 1 of windings. The subscript on the variable 6, can now be dropped. The result is
the machine. The total number of flux linkages linked by all such packets of
flux d41is
1 x
dA, = k pN
_xrx's i n 2
kod(j d l l ~ sin
, jdc

Thus the magnetizing inductance associated with this winding is

where we have made use of the symbol ')c' to denote flux linkage (weber turns)
rather than flux I$ (webers). Note that the differential is still used since this In general, current in phase a also links the other windings located on either the
expression accounts only for the flux linkages produced by the differential cur- stator or the rotor. Let us now calculate the flux linking one of the other stator
rent sheet element located at to.The subscript 'xm' for the flux linkage variable windings whose magnetic axis is displaced counterclockwise from the first by
is used to remind us that this flux corresponds to the magnetizing component an angle a as shown in Figure 2.5. In this case the number of turns N linked by
rather than the total (i.e. self) component of flux linkages. the flux packet d@is given by a modification of eqn (2.2-14) which bccomcs
 2 d.q Modelling of Induction and Synchronous Machines System Equations in the Stationary a,b,c Reference Frame 43
42

Winding Density r(;$sin 5

W nding
From an analogous derivation it is not difficult to show that reciprocity is
maintained. That is

2.3 System Equations in the Stationary a& Reference Frame

The general result which we have obtained can now be used to calculate the
mutual inductance between any pair of stator or rotor windings. Let us now
proceed to calculate the state equations of our idealized three phase induction
machine. In particular, we will assume the winding configuration shown in
Figure 2.6. In this case the winding placement is only conceptually shown with
Figure 2.5 Sinusoidal current sheets for two windings displaced by an angle a the center line of the equivalent inductors directed along the magnetic axes of
the windings. Again an elementary two pole machine is considered. 'Balanced'
three phase windings are assumed for both the stator and the rotor. That is, all
three stator windings designated as the as, bs and cs windings are assumed to
Hence, the total number of flux linkages of winding 'y' linked by the flux have the same number of effective turns, N,, and the bs and cs windings are
c
packet dQ located at 5, + is symmetrically displaced from the as winding by 5120'. The subscript 's' is
n used to denote that these windings are stator or stationary windings. The rotor
N I
d h , = po-&-%rl ~ (6, - a)sincdc (2.2-21)
s i n @ ( l ~ sin I
windings are similarly arranged but have Nr turns. These windings are desig-
0 nated by ar; br; cr in which the second subscript reminds us that these three
windings are rotor or rotating windings.
N,N I The voltage equations describing the stator and rotor circuits can be written
= po- " x n r l ~ i n ~ o(E,,~ i-na)dS conveniently in matrix form as
2 h'
'abcs = 5 iabcs + phabcs (2.3-1)
The total f ux linkages of winding 'y' due to a current in winding 'x' is there-
fore,
N,N I where p represents the operator &dt and and kaks are 3x1 vectors
1,, = Jp,+nrlsint sin (5 - a ) d t vabcs, iabcs
R defined by
0

(2.2-23)
The mutual inductance between winding x and winding y is therefore
44 2 d,q Modelling of Induction and Synchronous Machines Determmat~on
of Induction Machme Induclances
- - .- - -- 45

where

br-axis

4----
'as

Note that as a result of reciprocity, the inductance matrix in the third flux link-
age equation, eqn (2.3-8). above is simply the transpose of the inductance
matrix in the second equation. eqn (2.3-7).

2.4 Determination of Induction Machine Inductances

While the number of inductances defined above is large, the task of solving for
Figure 2.6 Magnetic axes of a three phase induction machine
all of these inductances is straightforward. The self inductance of stator phase
as is obtained by simply setting a = 0, and by setting N, and Ny in eqn (2.2-24)
Similar definitions apply for the rotor variables vabcr, iabcrand i a b c r . to Ns or. alternatively, by setting N, = Ns in eqn (2.2-19). Whereby,
In general, coupling clearly exists between all of the stator and rotor
phases. The flux linkages are therefore related to the machine currents by the
following matrix equations. The subscript m is again used to denoted the fact that this inductance is a mag-
'abcs = 'abcs (s) + 'abcs ( r ) (2.3-4) netizing inductance. That is, it is associated with flux lines which cross the air
gap and link rotor as well as stator windings. In general, it is necessary to add a
'abcr = 'abcr(s) +%bcr(r) (2.3-5) relatively small, but important, leakage tern to eqn (2.4-1) to account for leak-
age flux. This term accounts for flux lines which do not cross the gap but
46 ....... -.
--.
2- -
d.q Modelling of Induction and Synchronous Machines
-. Determination of Induction.....
Machine Inductances 47

instead close with the stator slot itself (slot leakage), in the air gap (belt and Let us now turn our attention to the mutual coupling between the stator and
harmonic leakage) and at the ends of the machine (end winding leakage). the rotor windings. Referring to Figure 2.6 we can see that the rotor phase ar is
Hence, the total self inductance of phase as can be expressed. displaced from the stator phase as by the electrical angle 0, where 8, in this
'as = L ~ + s Lam
(2.4-2) case is a variable. Similarly, the rotor phases br and cr are displaced from bs
and cs respectively by 0, Hence, the corresponding mutual inductances can be
Since the windings of the bs and cs phases are identical to phase as, it is clear obtained by setting N, = Ns, N, = N , and a = 0, in eqn (2.2-24).
that the magnetizing inductances of these windings are the same as phase as so
that, also
Lbs = Lls Lbm +
(2.4-3)

Lcs = Lls + Lcm (2.4-4)

It is apparent that La,, Lbmand Lcmare equal making the self inductances also The angle between the as and br phases is Or+ 2 6 3 , so that
equal. It is therefore useful to define the stator magnetizing inductance
L"' = ,N
l Z ( ~I )
O S g 4
Finally, the stator phase as is displaced from the rotor cr phase by the angle 8,
so that - 2x13. Therefore,
'as = 'hs = C
'S = 1's + 'ms (2.4-6)

The mutual inductance between as and bs, bs and cs, and cs and us
are derived by simply setting a = 2x13 and N, = Ny = N, in eqn (2.2-24). The The above inductances can now be used to establish the flux linking the
result is stator phases due to currents in the rotor circuits. In matrix form,

or, in terms of eqn (2.4-5),

The flux linkages of phases as, bs, and cs resulting from currents flowing in the
stator windings can now be expressed in matrix form as
cos8, cos (0, + 2x/3) cos (Or - 2x/3)
coser cos (0,. + 2x/3)
cos or

(2.4-1 3)
The totai flux linking the stator windings is clearly the sum of the contribu-
11
iar
ibr

tions from the stator and the rotor circuits, eqns (2.4-9) and (2.4-13),
'abcs = 'abcs (s) + 'abcs (r) (2.4-14)
It is not difficult to continue the process to determine the rotor flux link-
ages. In terms of previously defined quantities, the flux linking the rotor cir-
'
cuits due to rotor currents is
48 2 d.a Modellinn of Induction and Svnchronous Machines Complex Vector Representation of Thrae Phase Variables 49

introduced simply as a different type of notation rather than a totally new trans-
formation of the basic phase variable machine equations.
For those readers with long experience using real variable transformations,
the additional abstraction associated with complex notation may seem unnec-
essary and perhaps counterproductive. However, in cases where the symmetry
of the d,q model is maintained, complex notation is ideally suited to exploit
this symmetry. It provides a compact notation, easy algebraic manipulation,
very simple graphical interpretations and often a very logical and direct devel-
opment of concepts which are difficult to develop using real variable analysis.
While il is certainly true that the use of complex variables can hc avoided
where LIr is the rotor leakage inductance. The flux linking the rotor windings thereby gaining the extra degree of generality of d,q theory, therc is, i l l thc casc
due to currents in the stator circuits is of induction machines, a great deal of manipulative and conceptual value i n
their use.
Consider, first of all, the equation describing the instantaneous position of
the stator air gap MMF. If phase as is sinusoidally distributed with a maximum
cos 0, cos (0, - 2n/3) cos (0, +
value located at 5 = a, then the MMF in the gap resulting from current flowing
~ 0 . ~ 0 cos
~ (0,- in phase as is
cos 0,

(2.4-16)
where p = 5 - a. Similarly, if currents flow in phases bs or cs which are spa-
Note that the matrix of eqn (2.4-16) is the transpose of eqn (2.4-13). tially displaced from phase as by 120 electrical degrees, then the respective air
The total flux linkages of the rotor windings are again the sum of the two
gap MMFs are:
components defined by eqns (2.4-15) and (2.4-16), that is,
' a b c r = 'abcr(r)
+ 'obcr(s)
(2.4-17) 5, = 52i bs
cos (p - 2x/3) (2.5-2)

2.5 Complex Vector Representation of Three Phase Variables

It is apparent that the extensive amount of coupling between the six circuits
makes the analysis of this machine a rather formidable task. However, we are The total air gap MMF resulting from all three currents flowing simultaneously
now in a position to determine if there is any simplification that can be is
expected in these coupled equations. The approach to simplification involves
the topic of axes transformations or reference frame theory and constitutes an
essential aspect of machine analysis. A very useful and compact form of the
machine equations for uniform air gap machines is obtained by using complex It is not difficult to show that if the currents, i,,, ibs, and ics form a balanced
variable notation. This concept was introduced by Lyon [2] and was initially sinusoidal three phase set of currents, then eqn (2.5-4) describes an MMF
treated as a formal transformation in the same sense as the d,q transformation wave having a constant amplitude and rotating at a constant angular velocity.
typically used for synchronous machine analysis. In this section the concept is It is useful to introduce the Euler relation wherein
 2 d,q Modellingof Induction and Synchronous Machines Complex Vector Represenlation of Three Phase Variables 51
--50--.-- ..--- --

eje = cos9+ jsine

Solving for cos 9 and sin 9, we have,

then
itabcs= 52 cia, + a 2 i b , + czi,,)
Equation (2.5-12) can now be written in the compact form

Using eqn (2.5-6), the MMF for three phase excitation becomes
Note that whereas the MMF Fabcs remains a real variable, it is represented
as the effects of two complex variables bbcs and jLbcs. Quantities defined in
the manner of eqn (2.5-1 3) are called complex space vectors. They should not
be confused with complex phasors which are normally used to represent sinu-
soidally alternating quantities in the steady state.
The inverse relations for finding the physical currents from the complex
vector is obtained by expressing eqn (2.5-13) in complex form

Collecting terms,

if the sum of the three stator currents equals zero (no zero sequence compo-
nent). Hence,
'a, = Re [iabcsl (2.5-17)
Similarly,
Defining

a -- ,j2n/3

and noting that

@2 = a-- I = ej4n/3 = e-j2n/3 = at
-
where t denotes the complex conjugate, eqn (2.5-9) becomes
So that.
Fa,,,
r
= - { [ i , , + a2ibS+ ni,] + [ i , + a i bs+ a2iCs]e-jp} ibs = Re
4
If we define Finally,
 2
52 2 d.a Modelline of Induction and Svnchronous Machines Complex Variable Model of Three Phase Induction Machine 53
- -
2

Note that an additional term must be added to these expressions if the three
phase currents contain a zero sequence component. The second component of stator flux linkages arises from the rotor current.
By repeated use of eqn (2.5-6), eqn (2.4-13) can be written as 1

2.6 Complex Variable Model of Three Phase Induction

Machine
Consider now the representation of the machine equations in complex variable
form. Transformation of the voltage equations is, indeed, almost trivial. If we
multiply the second row of the stator voltage matrix equation, eqn (2.3-1). by
(2.6-7)
'a' and the third row by a 2 , add the three rows and multiply the entire result by
2/3 we immediately obtain, Again, multiplying the second row of this matrix equation by g and the
- third row by a2 and adding the three rows results in
?abcs - 5 jabcs+ pLabcs (2.6-1)

where

and so forth for bbcs and L b c s . The complex vector expression for the voltage
equation of the rotor circuits is developed in a similar manner and written in However, it is readily shown that
complex form as 1+ ( J + g 2 = 0
-
?abcr - 7 abcr' phabcr (2.6-3) so that in terms of complex space vectors, eqn (2.4-13) can be written in the
The stator flux linkages were identified in Section 2.4 as having two com- equivalent compact form
ponents, $bcs(s) resulting from stator currents (eqn(2.4-9)) and L b C s ( , ) arising
from rotor currents (eqn (2.4-13)). Upon multiplying the second row of eqn
(2.4-9) by a and the third row by a2 and adding the three rows, the stator flux
where
linkages resulting from stator currents can be written as
2
&abcs ( r ) = 3 ('as ( r ) + g h b ~( r ) + a 2 h ~( r .) )~ (2.6-1 1)

and
iabcr= 52 (iar+ aibr + a2icr)
It is not difficult to demonstrate that
The total stator flux linkages are therefore

so that eqn (2.6-4) reduces to

54- 2 d.q Modelling o f Induction and Synchronous Machines Turns Ratio Transformation 55
. - - .-.-
- ---- -

In a like manner it can be easily shown that the rotor flux linkages can be
expressed in the form

Also, since the 312 term appears in all of the flux linkage expressionIS it is use-
2.7 lhrns Ratio Transformation ful to define the magnetizing inductance
It is rather apparent that the turns ratio NJN,causes these equations to be need-
lessly cumbersome. The appearance of the turns ratio is, of course, not unex-
pected since the induction motor is simply another type of coupled magnetic Combining the above results the equations for stator and rotor circuits become,
circuit and is related to the transformer. In fact we can view the induction finally
motor as equivalent to a transformer with a short circuited and rotating second-
ary. We can get rid of the explicit dependance of these expressions on the turns
ratio by referring the rotor circuits to the stator in much the same manner as
utilized for a transformer. Equations (2.63), (2.6-13) and (2.6-14) can be
manipulated to form
(2.7-1 1)
The rotor voltage yibcr is, of course, identically zero for the case of a squirrel
cage machine, but has been carried through this analysis for generality.
Employing the normal rules for differentiation, these two equations can be
written in alternative form as

We can now define the primed variables

where or = per = d- er
dt '

It is important to note the enormous simplification afforded by the complex

Space vector representation. It is a tedious but educational exercise to confirm
that eqns (2.7-12) and (2.7-13) generate the six scalar equations denoted by
eqns (2.3-1) and (2.3-2).
56 2 d,q Modelline, o f Induction and Svnchronous Machines Transformation to a Rotating Reference Frame - --.
57

2.8 Transformation to a Rotating Reference Frame qaxis

While the complex vector approach to writing the machine equations results in bs-axis

a compact form, the essential sinusoidal coupling between the stator and rotor
circuits with rotor position remains. This coupling can essentially be elirni-
nated, however, if the stator and rotor equations are referred to a common
frame of reference. This common frame of reference can be non-rotating in
which case it is associated with the stator and is called the stator or stationary
reference. Alternatively the d,q axes can be made to rotate with the same angu-
lar velocity as the rotor circuits, and is termed the rotor reference frame. On
some occasions it may even be useful to rotate these axes synchronously with
one of the complex vectors denoting stator or rotor voltage, current or even
flux. In this case the axes are said to be attached to the vector. In the most gen-
eral case the axes need not be specified resulting in an arbitrary or freely rotat-
ing reference frame [3].
Consider, for example, a rotating set of d,q axes as defined by Figure 2.7.
Figure 2.7 also shows axes of reference corresponding to the magnetic axes of
the three phase stator and three phase rotor circuits. Variables along the a,b and
c stator axes can be referred to the q- and d-axes by the expressions

cr-axis

Figure 2.7 Location of rotat'ing d,q axes relative to the magneLic axes of

f, = 2 [ f a , sine + jbssin(0 - F) + fcssin ( 0 + )]

the stator and rotor phases

where the symbol f is used to represent any of the three phase stator circuit
In this context the use of the symbol 0 in the subscript of the new variable is
variables such as voltage, current or flux linkage. It should be noted here that
used to designate a zero component normal to the d,q plane
the coefficient 2M in these two equations is somewhat arbitrary. The choice of
In practical applications such as those considered in this book, the machine
213 is usually selected so as to maintain the same length of the voltage and cur-
is connected either in delta or in wye without a neutral return path. Thus, the
rent vectors for sinusoidal steady state as we shall see in Chapter 4. Another
three stator currents sum to zero. As a result, the other significant three phase
choice is to use the coefficient m 3 in which case the power as calculated in
variables such as the phase voltages, the stator flux linkages and rotor currents
the d-q coordinate system is the same as in the a,b,c system. In this case the
also sum to zero. Hence, it is necessary to concern oneself only with projec-
transformation is said to be power invariant. Since there are three phases, then
tlons on the d- and q-axes. If eqn (2.8-2) is multiplied by j and the result sub-
in general, it is necessary to define a third new variable to obtain a unique
tracted from eqn (2.8-1) we have immediately, from Euler's Equation,
transformation. This third new variable is typically defined as the zero
sequence component,
58 2 d.a mod ell in^ o f Induction and Svnchronous Machines Transformation to a Rotating
-.-
Reference Frame 59

Thus, eqn (2.7-10) can also be written in the form

(2.84)
which can be written as,

where

This expression, combined with the zero sequence definition in eqn(2.8-3),

defines a general rotating axis complex vector transformation. When the zero Upon multiplying through by 213 and making use of eqns (2.8-5) and (2.8-6)
sequence quantity is zero, which is the usual case in machine analysis, the sin- we quickly arrive at
gle expression in eqn (2.8-5) completely describes the transformation. Note
that the rotation associated with the general d,q transformation is separately yyds = rsiqds+ (Lls+ Lrn)piqds+ L n l ~ i b d+r j m [ (Lls+ L,) i,,d,+ LmiidrI
described by the exponential function e-le. The g operator, in effect, handles
the transformation from three phase to two phase d,q variables. (2.8-10)
In a similar manner, rotor variables can also be transformed to the rotating In a similar manner it is not difficult to show that in a rotating d,q coordi-
axes by a complex exponential function. In this case, however, since the rotor nate system, the rotor circuit equations can be represented by
is itself already rotating, a relative rotational angle must be used rather than the
angle 0. It is common to choose the angle between the q-axis of the rotating d- y i d r = r: ibdr + (Lir + Lm + Lmpiqds
q frame and the a-phase magnetic axis of the rotor circuits. The equation of
+ j ( m - m r ) [ ( L ; r + L m ) i b d r + L m i q d s l(2.8-11)
transformation relating rotor variables t t ~equivalent variables rotating at an
arbitrary rotational speed deldt becomes As a final step, the zero sequence component must be determined. If the
three rows of eqn (2.3-1) are summed and the resulting equation simplified by
means of eqns (2.4-9) and (2.4-13), it is not difficult to show that the result is

vaJ. + vbJ + vCr = ( r s + 1)LlS)( i l l s + ibs + icJ

Multiplying this equation by 1 / 3 and making use of the definition of the zero
Equations (2.8-5) and (2.8-6) can now be used to transform the machine
sequence quantity, eqn (2.8-3). we have finally,
complex vector equations to a rotating reference frame. Multiplying eqn (2.7-
10) by e-je yields Vos = ( r.Y+ pLls) ins (2.8-1 2 )
En a similar manner it is easily shown that if a zero sequence current flows in
e - v ~ +~ ( L,,~ + L~ ~e-jOpiabcs
~ = ~rse-~eiObcs ) + ~ , e - j ~(ihbcr
p ejer) the rotor, the equation describing its behavior is given by
(2.8-7)
However, by the chain rule of differentiation.
It is important to note that while the form of these zero sequence equations 1s
:
$
correct, the parameten of the particular result we have obtained 15, in fact, due
60
-
-
2 d,q Modelling of Induction and Synchronous Machines Interpretation o f Complex Vectors 61

to our assumptions. In particular, the inductance of the zero sequence circuits is

not generally equal to the stator and rotor per phase leakage inductances. In
practicc, the stator and rotor leakage flux components consist of both compo-
nents which either cross or enter the air gap (zig zag and tooth top leakage flux)
and components which do not involve the air gap (slot, harmonic, skew and The equivalent circuit of an induction machine in a freely rotating refer-
end winding leakage flux). When the stator (or rotor) is excited with a zero ence frame is shown in Figure 2.8. Note that we have proceeded in a general
sequence current, their gap dependant flux components change (i.e. reduce) way so that voltages may also be applied to the rotor circuits. In such machines
drastically to continue to satisfy Gauss' Law while the non-air gap dependant the rotor is wound with coils in the same manner as the stator. Such machines
fluxes are hardly affected. Hence, the zero sequence inductance is typically less are normally termed wound rotor induction machines. If the machine has a
than the per phase leakage inductance being roughly 0.8 to 0.95 of this value. wound rotor with slip rings the rotor voltages arc typically dclincd by exrcrnal
Replacingp by dldr and collecting our results to this point, eqns (2.8-10) to circuit conditions. For example, consider the case of a three slip ring machine
(2.8-1 3) can be written in scalar form, for the stator: connected to external resistors R, vh, = - ( Ns/ N,) Rx iir , and K x v y r ' =
- (Ns/Nr) 2R, i i r with v;, = 0 . If the machine is a squirrel cage construc-
tion the rotor voltages vir ,v;, and v& are identically zero.
In the vast majority of cases the machine is connected in delta or wye such
that the neutral current does not flow. In this case the neutral axis voltages vos
and v;, remain identically zero. Hence, from eqns (2.8-19) and (2.8-25). the
currents iOSand ibr are identically zero and solution of these equations are
always trivial. The four remaining equations are often assembled in a single
matrix equation expressed as

For the rotor:

where L: = Lir + Lm and Ls = L,, + L,,,

2.9 Interpretation of Complex Vectors

Since the use of complex space vectors is not a familiar concept it is useful to
pause in this analysis to consider the complex vector concept more fully. From
Section 2.8 the equation relating stator a,b,c variables to stator d,q variables
can be expressed as
where 2 -.
fpds = T~ Je~fllS+~fbS+(12fC,J (2.9-1)
h i r = L;, iir + Lm ( ids + i;, )
62 2 d.q Modelling of Induction and Synchronous Machines Interpretation of Complex Vectors 63

When the d,q reference frame does not rotate we have 8 = 0. In this case the
reference frame is said to befired in the statoc Using a superscript s to denote
this fact, the equation relating a,b,c variables to d,q variables in a stationary
coordinate system is

fids = :
[ f a s +Qfbs + Q2fcsl (2.9-2)

Figure 2.9 illustrates the concept of a complex vector in terms of the stator cur-

d- axis equivalent circuit

2 d-axis / ios

q- axis equivalent ci muit

Figure 2.9 Transformation from three phase stator to complex
vector voltage

rents [i.e.f replace by i in eqn (2.9-2)]. In this graphical interpretation, the cur-
rent ia, becomes a vector in the direction of the as coil, sibs a vector in the
No Coupling
direction of the bs coil, etc. The projections of the complex vector i i d s on the
real and negative imaginary axes are the equivalent q and d axis two phase cur-
rents. For the case illustrated in the figure there is also a zero.sequence current
as indicated.
A pure rotational transformation, for example from stationary d,q axes to
0- axis equivalentci muit what is typically called the synchronously rotating axes, would be simply
expressed as
Figure 2.8 The d,q,O equivalent circuits of three phase induction
machine
where 8, = met when o, is a constant. The symbol s is used to denote a station-
ary set of axes while the subscript e denotes axes which are rotating electri-
cally with the applied voltage vector. Equation (2.9-3) has the graphical
64 2 d,q Modelling of Induction and Synchronous Machines Interpretation of Complex Vectors -~
65
.-

interpretation shown in Figure 2.10. At first sight the reverse rotation of the

Thus, the complex vector qds itself rotates ;it synchronous speed in the posi-
tive direction. When transformed to a synchronous reference frame using eqn
(2.9-3). the resulting complex vector ?:i is stationary because it rotates back-
ward at synchronous speed relative to iqds..
It is convenient to generalize the onginal result of eqn (2.9-1) to handle
either rotor or stator quantities by defining the general phase transformation

where x can be replaced with s or r to treat stator or rotor quantities. The gen-
eral rotation transformation is best stated as
d-axis f qdx = f ;dxeLJe (2.9-10)
Figure 2.10 Transformation from d,q variables in a stationary axes to
and interpreted as the transformation from a stator referred quantity to a freely
rotating d,q variables, (pure rotation)
rotating quantity. The advantage of this interpretation is that all angles are mea-
sured from the axis of phase a and hence the angle 8 is well defined in all
complex vector i& may seem incorrect since a synchronous frame is usually .. cases. For example, the transformation of any stator referred quantity to axes
visualized as rotating with the synchronously rotating machine quantities. rotating with the rotor is given by
However, the rotation in Figure 2.10 is correct since balanced, positive f idx = _/idx.-Ier (2.9-1 1 )
sequence current excitation results in
For a rotor quantity, this equation becomes
iaS =. IScosw,t (2.94)
fidr= ~ : ~ ~ e - j ~ ~ (2.9- 12)
i,, = ~,cos(o,r - 2j?) The initial transformation of a rotor quantity derived from eqn (2.9-9) to the

i, a Zj?)
= ~ ~ c o s (+ ~f

which, when expressed as a complex vector yields

66 -
2 d,q Modelling of Induction and Synchronous Machines Interprelation of Complex. Vectors 67
- - ..-..-

While the proper sign on the exponential term in eqn (2.9-13) can be obtained b-axis
by careful physical reasoning, the approach enforced by using eqn (2.9-1 1) as
the basic rotation transformation makes it automatic and error free.
The inverse of any rotation transformation is readily obtained by simply
transposing the exponential term. The inverse of the phase transformation is
more difficult. The inverse can be obtained by using the Q operator in rectangu-
lar form

. . a-axis

By taking the real part and using the definition of the zero sequence from eqn
(2.8-3)
.I' = ~w,l,;,,,l
11 1
(2.9- IS)

'I'lic. co~nyontlingresulrs for fh and f, are obtained i n u similar fashion

and are
b-axis
& = Re [a2f;dIl +for (2.9-1 6 ) negative
1.-axis
.fcx= Re [afGdx1 +fox
If the angle of the complex vector quantity is $
p;, = Iflei@
the inverse relations can be written as
= ISICOS@+~,,~
fCdX
far

Note: f is negative
CX

The diagram in Figure 2.11a illustrates the angles involved in these equa-
tions. This dingram also shows that these same angles can be defined by intro-
ducing a three phase reference system and measuring from -f:d! to these Figure 2.11 Graphical inverse of complex vector quantity
reference axes instead of rotating f- idx.
This leads to the graphical mterpreta-
68 2 d.a Modelline of Induction and Svnchronous Machines Power F I O ~in the d.0 Eauivalent Circuit 69

tion of the inverse illustrated in Figure 2.11b. Here the components of the
inverse are obtained by simply taking the projections o f f Gd!
on a three phase
reference system with phase a oriented along the reai axls. This graphical
inversion process is very useful in many applications. The zero sequence, if Note that we have immediately used the stator referred rotor variables in
any, must be added to the portion coming from f- &,*. eqn (2.10-3) since it is readily seen from Section 2.7 that the rotor voltagelcur-
rent product is independent of the turns ratio N,,jN, Transforming now to cl,q
variables in a rotating axis coordinate system wc have from eqn (2.8-3, for the
2.10 Power Flow in the d,q Equivalent Circuit .
stator variables
The powcr flow into the induction motor in terms of our newly defined d,q,O (2.10--4)
variables must now be examined. The power flowing into an n phase induction -f qds = e-J8fobcs

machine is clearly the instantaneous product of thc voltages across and cur-
rents through the n phases of the machine. In our case the machinc is assumed or, inversely,
to have three stator and three rotor phases. In order to express the power into f abcs = eJefqds
the machine we can begin by noting that if 't' denotes the complex conjugate,
then and, from eqn (2.8-6), for the rotor variables,
- e~(e-er)
f bbcr - f bdr
Thus in terms of d,q variables eqn (2.10-3) is written as

In scalar form

(2.10-1)
Note that the input powcr is independent ol' thc rcfcrence frame velocity o =
If the three currents sum to zero, this expression becomes, finally d9ldt as would be expected.
It is interesting to observe that the power into the d,q circuits is calculated
in a slightly different fashion to that in the actual circuits of the machine due to
the appearance of the 312 factor. This is a direct result of the 213 factor that we
Thus, assuming a three wire machine without a neutral return, the power into
selected as the proportionality constant between the n,b,c and ci,q variables.
the machine can be written as
eqn (2.9-1). It is sometimes stated that power is not conserved with this choice
of scale factor although it clearly is if we include the 312 factor in the i1.q form
70 2 d.q Modelling of Induction and Synchronous Machines Example - Stator a.b.c and Rotor d,q Model (Stator Referred) 71
~- ~ -.. ..-
pp

of the power equation. Other transformations are possible in which the power, which implies that
as calculated from the transformed variables, is the same as in the actual phase
vos - (rs + Ls P) ios = - Lm P ios
variable system. In this case the factor m3 is introduced rather than 2/3 . 1
We will have more to say about the significance of this factor in Section 2.12. Since ios = (ias + ibs + ics)
Since both types of systems are in widespread use the reader should be alerted
to this fact when reading the literature.

2.11 Example - Stator a,b,c and Rotor d,q Model (Stator

Referred)
In problems in which there are switching constraints imposed directly on the Phase b:
stator currents and voltages it is sometimes difficult to transform the con-
straints to d,q variables. For such cases it is often easier to carry out the analy-
sis directly in the three phase variables. However, to avoid the many
trigonometric coupling terms in the differential equations, it is possible to rep- vbs - vos = (rs + L,p) (ibs- ios) + Lmp
resent the rotor by a stator referred d,q model while leaving the stator variables
in their original form. Derive such an a,b,cld,q stator referred model (with the The same substitution for vos - (r, + LJJ )i, yields
qs-axis aligned with the us-axis) using the basic concepts presented in deriv-
ing the dq/d,q stator referred model. For generality, assume the zero sequence
stator quantities are not zero.
Solution - Starting with the d,q/d,q model. [eqn (2.8-26)], implicitly assuming and for c phase by symmetry
use of a turns ratio transformation (primed rotor variables) at this point,

For the rotor.

Since, i>ds= 52 (las

.
+ uibS+ a2ics) and i ids= iSqs - j i i s
In a stationary reference frame, o = 0 and assuming q and a axes are aligned,

vas = v 4s
S + vOS = (rS+ Lsp) i & + Lmpiir + v,,
But, i i s = ias - ios substituting,

Hence, v,, = ( r ,+ L p ) ias + L,,p iir+ vOs- (rs + L g ) ios

However, v,, = ( r , +(L, - L,,) p) io,
The real part is:
72
.--. -.- -. ...- -
2 d,q Modelling of Inductio~~
and Synchronous Machines The Electromagne~icTorque - - -. --
-.

2.12 The Electromagnetic Torque

v;, = ( r , + L r p ) i i r - wrLrildr + ~ ~ p (- j1ii b. si -~ ~
Up to this point we have treated the induction machine as simply an intcresting
The imaginary part corresponds to: type of coupled magnetic circuit and the electromechanical properties of this
device so far has been ignored. It is now time to consider where the energy
v.5,. = ( r , + L r p ) i i r + urLriSqr + Lmp-1 ( i o - ibs)
goes that passes into the machine via the stator and rotor terminals. Upon
J? inserting eqns (2.8-10) and (2.8-11) into the d,q expression for the input
Solution Summary power, eqn (2.10-8), we can write that
Stator a,b,c and Rotor d,q Model

- -
This expression in the right hand side can be arranged in the form

Power lost in Conductors Time Rate of Change of Stored Energy

Hence, the electrical power into the teminals of the machine can be segregated
into three terms. The first term clearly accounts for the power dissipated in the
stator and rotor resistances. The second term corresponds to the time rate of
change of the magnetic energy stored in the inductances of the machine. Since
Note that since Lo, = Ls - L, = L1,, and Lm = J L , , , ~
2 the remaining power must be going somewhere, it is natural to suspect that the
third term on the second and third lines account for energy conversion, that is
1 2
then Ls - - Lm = - Lm + Lls = Lms + L l S . the power being converted from electrical to mechanical form. Let us call this
3 3 term P,,, the electromechanical output power.
Examining the four terms of the energy conversion term itself, ~tis appar-
ent that the first and third quantities in this expression have only imaginary
74 2 d,q Modelling of Induction and Synchronous Machines The Electromagnetic Torque 75
-- -.-.- .- -

components. Hence, the energy conversion is accounted for by the second and
fourth term. We can write that,

It can be shown, however, that the expression for electromagnetic torque

has many equivalent forms. For example, the stator flux linkage is related to
the stator and rotor currents in complex form as
Sqds= (L,s + Lm 1iqds
+ Lm ibdr (2.1 2-9)
It can be easily shown that if &and h are complex numbers then
Solving for L, iidrand inserting in eqn (2.12-7) we find that
&t + gtb = 2 (arbr + aibi)
That is, it has no imaginary part. It therefore follows that the first term of eqn
(2.12-2) has no real part. The output electromechanical power reduces to the Since the first term has no imaginary part, this expression reduces to
last term in this expression,
3P
Te = 22'm CigdsBtqds 1 (2.12-11)
r e ,= 2 { jwrLm GASitid,1
or, in scalar form
or, alternatively,
3P
Te = 2 2 (?"dsiqs- Aqsids) (2.12-12)
P, 2 { o,~,,,
= 21, }
Alternatively, the rotor flux linkages are expressed as
which can be easily determined to be equivalent to hidr = (L;, + Lm)ibdr + LAiqds (2.12-13)
Upon solving for $,iqdsand inserting in eqn (2.12-7) results in
10 L [ iqsiir - id& )
Pe, =
2 3P
The electrical angular velocity of our equivalent two pole machine, w, is T, = 21m { qdri2ddr1 (2.12-14)
related to the actual mechanical speed by theI pole pairs. That is,
P (2.12-6) Hence, torque production in the machine can be visualized either as the rotor
("r = Tor" flux linkage vector interacting with the rotor current vector or, alternatively, the
stator flux linkage vector crossed with the stator current vector.
where P is the number of poles and om is the mechanical speed in radians per If we define the air gap or magnetizing flux linkages as
second. Since rotational mechanical power is defined as the product of speed
times torque, it is apparent that the electromagnetic torque impressed on the hqdm= Lm(iqds + q d r ) (2.12-15)
shaft of the machine can be expressed as ,solve for Lmiidr ,and substitute into cqn (2.12-7) we obtain

Another important expression for electromagnetic torque can be obtiuned if

solve eqn (2.12-1 3) for rotor current rather than stator current, whereupon
76 2 d,a Modelline of Induction and Svnchronous Machines The Electromagnetic Torque 77
.

ponent and therefore produce the same electromagnetic torqueh addition to

the equations describing the electrical behavior of the machines, additional
equations are, of course, necessary to describe its electromechanical behavior.
Substituting this result into eqn (2.12-7) yields In its simplest form, the electromechanical behavior is described by

where 4 = L;, + L, . Hence. torque production in an induction machine where O, is the mechanical angular velocity of the rotor in radls and TI is the
may also be viewed as the interaction of rotor flux and stator current. In future load torque which may itself be described by additional differential or alge-
chapters we will learn that this particular form of the torque equation is braic equations. Since the actual mechanical speed does not appear in the elec-
extremely useful for purposes of torque control. trical equations, eqn (2.8-26), it is common to write eqn (2.12-20) as
Still other expressions are possible. For example, the torque can be 2~~~
expressed as the cross product of stator and rotor flux linkage, i.e. T =+
-'T
Pdt
where w, = ~IB,.,,,/P.

where o = 1 - Li/L,L,. A table summarizing these relationships is given

below. Figure 2.12 shows the stator, mutual and rotor flux linkages as well as
the stator and rotor currents for a typical operating condition. We have dropped TABLE 2.1 Complex Vector Exprpssions for Electromagnetic Torque. $ denotes variable
used to calculate torque
the use of subscripts d,q since complex vectors on the d,q plane are already
implied. The d-axis is aligned along the rotor flux vector in order to iliustrate
those vector cross products involving hd, It is apparent from the construction Torque Expression
of Figure 2.12 that the flux linkage vectors produce the same d-axis flux com-

Figure 2.12 Illustrating torque production by vectors on the d,q plane

78 2 d,q Modelling of Induction and Synchronous Machines Analysis of Induction Motor Starting Performance Using d.q.0 Variables
-- -- 79

2.13 Analysis of Induction Motor Starting Performance Using

d,q,O Variables
As an example of the use of d,q,O variables we wiU first investigate the starting
behavior of a typical 5 horsepower (3.73 kW),3 phase, 3 wire, 230 V.,4 pole
squirrel cage induction machine with parameters:
r , = 0.53 1 R r2 = 0.408 R J = 0.1 kg-m2
L , = L2 = 2.52 rnH L, = 8.47 mH
The term three wire is generally used to denote that machine has no neutral
return path. We will assume that the machine is suddenly connected to a bal-
anced three phase 60 Hz, sinusoidal supply having rated line to line voltage of
230 V rms. The effect of source impedance in series with the supply is
neglected. The voltages across the phases are then defined as:

In the first example, Figure 2.13, the machine is simulated directly in terms
of phase variables. When the voltage is suddenly applied to the stator windings
the currcnt riscs rapidly to a Ii~rgcvalue, usuully termed the inrush current. The
elcctl.o~~~i~gnrtic torque develops u ul~idircctionulcomponent which ucls ta
i n a . the torqua ulso con~uinsa l u r p
~ I I Crotor of the ~ ~ ~ u d ~NOWCVO~,
I I C C ~ ~ C ~ 1112
pulsa~ingcomponent which decays as time progresses. When the speed
approaches rated speed (synchronous speed), the currents in both the stator and
rotor windings begin to decrease. Note that while the frequency of the stator
current is fixed, the frequency of the rotor currents change continuously as the ' Figure 2.13 Starting performance of 220 V, 5 hp squirrel cage induction
motor accelerates since the voltage induced in the rotor is always proportional motor with a balanced sinusoidal supply showing physical
to the relative angular velocity between the synchronously rotating stator variables. After reaching rated speed the motor is loaded to 0.83
MMF and that of the rotor circuits. times rated torque. Traces from top to bottom: v,,, phase a line
Since the motor is assumed as unloaded and second order effects such as to neutral voltage, ,i line current of stator phase as, ibr ,phase
windage and friction have been neglected, the simulated speed reaches its syn- current of rotor phase ar (referred to stator turns). T,, electro-
chronous speed. At this point the stator current decreases to the no-load value magnetic torque, TI,load torque, rotor speed in RPM; time axis
needed to excite the machine, i.e. magnetizing current. Since the rotor is rotat- 0.1 sldiv.
80 2 d,q Modelling of Induction and Synchronous Machines Analysis of Induction Motor Starting Performance Using d,q,0 Variables 81

ing synchronously with the stator MMF, the rotor current becomes zero. In A plot of the starting of the induction machine is repeated in Figure 2.14. In
practice, of course, the rotor would actually reach a speed slightly below syn- this case the motor is represented in stationary axis d,q,O variables (o= 0).
chronous speed defined by a no-load slip. For purposes of analysis the wind- Since the motor is c o ~ e c t e dwithout a neutral return, the neutral axis current is
age and friction can be considered as an external effect which can easily be zero and is not shown in Figure 2.14. The voltages applied to the equivalent
modeled if desired. Some time after achieving synchronous speed, the motor circuit, Figure 2.8. are from eqn (2.9-2)
model is loaded with a step torque equivalent to 0.83 times rated torque. The
motor responds to the increased torque by slowing down, whereupon the
torque increases to an equal (and oppositely directed) value to support the load
torque. The stator current increases in amplitude and changes phase with
respect to the stator voltage resulting in power passing from the electrical to
the mechanical terminals. The steady state currents for both the no-load and
the loaded condition can, of course, be correlated with results obtained by con-
ventional phasor analysis.
The pulsating torque observed upon switching on the stator voltages can be = - g 2 3 0 sin (3771)
explained by considering the transient currents which flow in any inductive cir-
cuit. In general, sudden switching of the voltage onto the terminals of the
machine results not only in ac current of the same frequency as the supply but
also in dc transient components which die out with roughly the rotor open cir-
cuit time constant. This stator current component results in a stationary (non-
rotating) component of MMF which is represented on the d,q plane as a sta-
tionary, non-rotating stator current vector. The flux created by this non-rotat-
ing MMF acts to induce currents in the rotor which rotate backward at rotor
speed (frequency) with respect to the rotor. These induced rotor currents pro-
duce a rotor MMF which becomes stationary with respect to the stator MMF Also. the zero seauence comDonent is
(represented as a non-rotating rotor current vector in the d,q plane) producing
a negative m q u e which acts to brake the machine. In addition, the rotor MMF
also interacts with the synchronously rotating stator MMF (represented as a
synchronously rotating stator current complex vector in the d,q plane). Since
torque is produced by the product of stator and rotor current, eqn (2.12-7). the Since the current i$ is obviously zero it is not plotted in Figure 2.14.
torque produced by the interaction of these two components oscillates essen- Observe that the behavior of the d,q stator currents is nearly identical to those
tially at line frequency (60 Hz). It should be noted that in practice this interpre- obtained in terms of phase variables. The rotor currents on the other hand
tation is very nearly but not exactly correct due to the fact that the rotor and appear quite different. Since both the stator and rotor circuits are represented
stator time constant are not completely independent. The rotor current vector on stationary d,q axes the rotor currents as well as the stator currents oscillate
(or rotor MMF), in fact. rotates very slowly with respect to the stator current at 60 Hz. Note that the torque and speed remain identical to Figure 2.13 since
vector (stator MMF) resulting in a 'beating' which can be observed superim- they must be invariant under the d,q,O transformation.
posed on the stator currents as well as in the electromagnetic torque.
~7 2 d.a Modellinn of Induction and Synchronous Machines Analysis of Induction Motor Stariing Performance Using d,q,O Variables 83-
- -

It is important to note the close similarity of the q-axis stator current $:i
with the current in phase as, that is ids. In particular, recall from eqn (2.9-2)
that

If in addition.

ioS + ibs+ ics = 0

then

This is to be expected since we can note from eqns (2.13-1) and (2.13-4) that
the voltage applied to the q axis circuit is also equal to the a s phase voltage. In
essence this feature is the reason why the factor "213" was chosen in the initial
definition of d,q variables, eqn (2.9-1). Although this introduces a 312 factor in
the torque and power equations, it has the advantage of making it easier to
interpret variables such as voltage and current. For consistency with the work
of previous researchers we will employ this so called scaled transformation
henceforth throughout this book.
In Figure 2.15 the same motor starting problem as studied in Figure 2.13
and Figure 2.14 has been solved in a rotating d,q,O coordinate system. In this
case the reference frame angular speed has been set equal to the rotor speed,
i.e. a rotor reference frame. Note that all of the d,q variables now assume a fre-
quency equal to the rotor slip frequency since the reference frame is now slip-
ping with respect to the synchronously rotating magnetic field. Since the torque
and speed are invariant under the d,q transformation they are again identical to
Figure 2.13.
In general, the angular velocity of the rotating axes can be freely chosen. In
the study of electrical machines, however, it is frequently useful to "attach" the
$ q axes to the vector describing the instantaneous spatial position of the motor
sl I terminal voltage, current or flux or even to some other vector variable. In most
Figure 2.14 Starting performance and Subsequent loading of 220 V, 5 hp ptactical cases, the voltage is considered as the independent vanable, (i.e. the
squirrel cage induction motor withsa balanced 60 Hz sinusoidal :input') so that the angular velocity of this vector is known in advance. It is
supply. Motor is modelled in stationary d,q axes. (Stationary ,.,en a simple matter to attach the reference frame axes to the voltage vector by
reference frame). Traces f r p top to bottom - & ,v& .iL,
isqs, simply setting equal to the angular velocity of this vector. When the phase
i'j, , :ir , rotor speed in RPM,T,.;time axis 0.1 sldiv.
84 2 d.q Modelling of Induction and Synchronous Machines Analysis of Induction Motor Starting Performance Using-~d,q.O Variables ..~... . 85

/
voltages are balanced and sinusoidal the angular velocity of vector is simply
-equal to 2n times the frequency of the sine wave. Hence, in ous casc, i f we take

0=377 radh (2.13-10)

and let

0 (0) = 0 (2.13-1 1)
then the q-axis is fixed to the voltage vector itself. It is not difficult to show
that the voltages to be applied to the equivalent circuit now become, from eqn
(2.9-3),

~$1 = 0 (2.13-12)

where the superscript sv refers to the 'stator voltage' reference fralnc.

The solution for the starting problem, this time investigated in a ~cference
B frame synchronized with the rotating stator voltage vector is given In Figure
x 2.16. Note that since the input is now 'dc', all variables take thc form of slowly
er- varying dc variables. In effect. $e dc offset and 60 Hz ac components observed
r in Figure 2.13 and Figure 2.14 have interchanged roles. That is, the dc current
offset due to sudden switching-on of the source voltages appear5 now as a
decaying 60 Hz ac transient while the ac inrush current obtained during start-
ing now becomes a slowly varying dc component. The electromagnetic torque
and speed are, of course, again invariant under this transformation. It is impor-
tant to note that while the d-axis voltage is zero, the corresponding current is
substantial since the d- and q-axes circuits are coupled via the dependant volt-
! age generators of Figure 2.8.
Figure 2.15 Starting performance and subsequent loading of 220 V, 5 hp
r In Figure 2.17 the starting behavior of the same motor is computed one
squirrel cage induction motor with a balanced sinusoidal final time. In this case the reference frame has been attached to the rotor flux
supply. Motor is modelled in d.q axes rotating at rotor speed. vector rather than the stator voltage vector. Since the rotor flux IS not cxpllcitly
(Rotor reference frame). Traces from top to bottom - ~ 2v&~, . known throughout the solution period, the angular velocity of the reference
frame which is needed to 'attach' the d,q axes to the flux vector must be solved
i'd,, i& , iZr , "i 9' ,rotor speed in RPM, torque T,.;time axis 0.1
implicitly. Since the angular velocity of the reference frame is known only
ddiv. implicitly, the stator voltages can now only be expressed in the form
 86 2 d,q Modelling of Induction and Synchronous Machines Analysis of Induction Motor Starting Performance Using d.q.0 Variables 87

Figure 2.17 Starting performance and subsequent loading of 220 V, 5 hp

Figure 2.16 Startlng performance and subsequent loading of 5 hp, 220 V squirrel squirrel cage induction motor with a balanced sinusoidal
cage Induction motor with a balanced sinusoidal supply. Motor is supply. Motor 1s modelled in d.q axes rotating synchronously
modelled in d,q axes synchronously rotating with the applied voltage with the rotor flux. (Rotor flux reference frame). Traces from
vector. (Synchronous reference frame). Traces from the top: vS&,
~ d, S , isvqS , i;;,
v ~ : I.sv i$ ,rotor speed in RPM,electromagnetic torque ,
the top: vvj! , v $ , ix , i$ ,::i . i"$ .
mtor speed in RPM.
electromagnetic torque T,;time axis 0.1 sldiv.
T,; tlme axis 0.1 sldiv.

.%
 88 2 d.a Modellinn of Induction and Svnchronous Machines -- Machines
Extension of d,q,O Theory to Analysis of Salient Pole Synchronous 89.
-.

Since the air gap of a salient pole synchronous machine varies along tlic
inner circumference of the stator, the machine is no longer symmetrical in the
sense of an induction machine. In particular, the stator self inductances repre-
sented by eqn (2.4-9) for the induction machine, now v q with rotor position.
Referring to Figure 2.18 it is clear that the self inductance of any winding must

v$ = 0 (2.13-17) bs-axis
where Od isthe instantaneous position of the rotor flux vector and where the
superscript rfhas been used to denote the fact that the d,q axes rotate with the
'rotor flux'. The necessary information to determine BJf can be obtained by
(Rotor y-axis)
means of feedback regulation [4]. In this case, the (/-axis is continuously
nlipnctl with ~llcrotor flux vector so h u t the y-axis ~'otorflux component is
always iden~icidlyzero. From eqn (2.12-18), the elecrromagnetic torque equa-
tion reduces to

- us-axis
+ "as -

Note that in this case the q-axis component of the stator current is always
at right angles to the rotor flux and therefore always corresponds to the torque
producing component of stator current. While the rotor flux in Figure 2.17 is
not constant, the similarity of the traces for the q-axis stator current and the fd-, kd-axes
electromagnetic torque is apparent. Since we can identify the orthogonal or (Rotor d-axis)
torque producing component of stator current in this reference frame, the deter-
mination of the instantaneous position of the rotor flux is essentially the key to
motor torque control. A discussion of the various means by which this is
Figure 2.18 Magnetic axes of a salient pole synchronous machine
accompIished together with their advantages and limitations is the subject of
much of the remainder of this book.

now pulsate once each time the rotor moves one pole pitch. Neglecting higher
2.14 Extension of d,q,O Theory to Analysis of Salient Pole order harmonics, the stator self inductance has a second harmonic component
Synchronous Machines in addition to the constant component represented by eqn (2.4-5). For exam-
The extensive family of synchronous type machines are another important type ple, the self inductance of phase as is ,)J ov&c,co
of machine for motor drive applications. For the remainder of the chapter we i
will examine d,q theory as related to a salient pole synchronous machine I
'as, as = 1's + Lo,- L2sc0s20r (2.14-1)
including the effects of non-uniform air gap and non-symmetric rotor wind- j where Lls again represents the leakage inductance of the phase. Using the same
ings typical of such a machine. We will then examine how these equations can approach as Section 2.2, modified for a machine with salient poles, it can be
be modified to mociel the behavior of other synchronous type machines. shown that
90 2 d,q Modelling of Induction and Synchronous Machines Extension of d,q,O Theory to Analysis of Salient Pole Synchronous Machines 91
- --

and

Inductances for the other phases are found in a similar manner. They are
Finally, the inductances corresponding to fluxes which link the q-axis damper
and the stator windings are

These inductances can be assembled in a matrix equation to represent all of the

The inductances corresponding to flux linking the field winding with the three fluxes linking the three stator windings. The voltages across the three stator
stator phases can be written as I phases of the synchronous machine can be written

n
where
'1

In like manner, the inductance corresponding to flux linking the d-axis

damper winding and each of the three stator windings is
92 2 d.q Modelling of Induction and Synchronous Machines Extension of d.q,O Theory to Analysls of Salient Pole Synchronous Machlnes 93
.- --

I
Lls + Los - LZScos20, -!L
2 0s
- L2s cos ( 2 0 r -- 2x/3) hkq = ( Llkq + Lm k q ) ikq- Lskq [iossin8, + ibssin (8, - 2n/3) (2.14-28)
8
L
!-
2 0s
- L2s cos (28, - 2W3) ~ ~ +~2x2.)
LIS + LOs - L 2 s (20, + i,, sin ( 9, + 2x/3)]
I I
--L
2 0s
- L 2 s C O ~ (20,+ 21t/3) 2 0s - L 2 s ~ ~ ~ 2 0 r
--L In these equations, it can be shown that [ S ]

+
I
L
sfd
L cose,
sfd LskdCOSOr -LskqsinOr

cos (9, + 2x13) Lskdcos (0, + 2x/3) -la,,.kqc~s

~ ~-~2n/3)
Lsfdcos (0 - 2x/3) Lskdcos (0,- 2x/3) - I , , y k q(0,
(0, + 2x/3)
1 [j
. ikd
'mkd = porlN:d(:)a
I
mrn

The quantities LYd, Llkd, Ll4 are the leakage inductances of the field, &axis
(2.14-22)
damper and q-axis damper windings respectively.
The voltages across the field, d-axis damper, and q-axis damper windings are
We can write the flux linkage eq"ation for the stator windings in the form

"kd = 'kd ikd + phkd (2.14-24)

"kq = 'kqikq+ phkq (2.14-25)

where,
h f d -- ( Ll f d + LmJd)iJd + Lfld ik;+LrJd [iUscos9,+ i h , r(9,-
~ ~2 ~d 3 )

+ i,,, cos(9, + 2 ~ / 3 ) ]

+ i,, cos (9, + 2d3)]

94 2 d,q Modelling of Induction and Synchronous Machines Extension of d.a.0 Theorv 10 Analvsis of Salient Pole Svnchronous Machines 95
--

This equation can be written in complex form as

3
- L skd ikd J
+2 2 skq ikq e

It is not difficult to show that the companion differential equation express-

ing the stator voltages is the same as for the induction motor. It can again be
written in complex space vector form as

The space vector concept has again permitted us to express the machine equa-
tions in a remarkably simple form. ,In the case of induction machine we were
able to transform the complex vector equations to a freely rotating axis. This
was accomplished by essentially multiplying the stator flux linkage equation
by e - ~ 9However,
. in this case the necessary symmetry does not exist. Simpli-
fication, however, is still possible if we let 8 = 8, That is, if we fix the refer-
ence frame to the rotor of the machine (rotor reference frame), then in this case
eqns (2.14-36) and (2.14-37) can be written,

Multiplying the second row of this equation by a and the third row by Q~ (3
babcse-JOr= L,, + ~ ~ ~ ) i ~ ~ ~ ~ e " ~ ~
and adding the result to the first row, one obtains after some simplification,
3 3 3 -15
-2 ~ ~ejOr~+ ?LSfdi
i fdb + i~~ s k~d i k-d~2 Lskqikqe '
(2.14-39)
Multiplying through by 213, using the chain rule of differcntiation and deti ning

+ -3 3
2L sfd i fd , j e , +2- ~skd i kd , J ' ~ - ?2 L skq i kq ej ( 0 , - 5) Y ids = V& - jv&
2 -ier
= j vobcse (2.14-40)

(2.14-35)
Using the basic definitions for the complex vector and its conjugate, eqns (2.5-
qds= hiS- =
2h
3 -abcs e-jOr (2.14-41)
13) and (2.5-14), this expression becomes finally.
96 2 d,q Modelling of Induction and Synchronous Machines

we arrive at 3Lskd (2.14-49)

kd = ( L I k d+ Lmkd)ikd + LFdi fd + 7[ i i+ (iids)'I
(2.14-50)

(2.14-44)
Note the use of the superscript r to designate the 'rotor' reference frame. (2.14-51)
One further simplification is obtained if we again refer the rotor circuits to
the stator by the turns ratio. If we define direct axis and quadrature axis induc- (2.14-52)
tance L,,,,J and Lm9 it can be shown that

when eqns (2.14-48) to (2.14-50) are referred to thc stator turns, they bccome,
1 .
yd = L k d i j d + L r n d { G d+ iid + -2[ l rqd s + (i;dJt1I (2.14-53)

1 (2.14-54)
Using the same manipulation as Section 2.7, eqn (2.14-44) becomes
h$ = L;, ild + Lmd iLd + ijd +5 [F+
(i&)+l 1

1 (2.14-55)
h~ = L;,, i b + L ~ , (i;, - j -2 [ i r d s -( i i d s ) tI~
9

In this case, however, it has become necessary to define

(2.14-56)

where the prime is used to signify the turns ratio transformation.

It is not difficult to show that the rotor flux linkage equations are also sim- (2.14-57)
plified by use of the complex space vector. However, the gain in this case is
only marginal because the complex vector cannot be utilized to represent the
rotor variables since the rotor is not 'symmetrical'. That is, the inductances (2.14-58)
and resistances of the d- and q-axis circuits are not the same

the absence of the 312 factor in eqns ( 2 . 1 4 4 8 )to (2.14-50). Also

98 2 d,a mod ell in^ of Induction and Svnchronous Machines Extension of d,q,O Theory ro Analysis of Salient Pole Synchronous Machines 99

hr9s = L,, ibs + Lmq(iis + i i q ) (2.14-71)

Since the necessary decoupling is only obtained in the rotor reference frame,
the use of the superscript for the stator d,q variables is unnecessary for syn-
The referred flux linkages continue to be defined as chronous machines and the superscript r is typically dropped. In addition, it is
important to mention that the damper cage is the equivalent of the squirrel cage
in an induction machme. Thus, the voltages vid and vid are identically zero.
We have again included these quantities as unknowns only for the purpose of
generality.
Note also that because of the use of the 213 term in the definition of the
primed rotor currents it has been necessary to define the stator referred rotor
2
resistances as rLd = -rkd ; 6
= ; and $ , = jr,,.
2 Finally, we have
been careful in this ciapter to use t e prime symbol to indicate when a rotor
winding has been referred to the stator by the turns ratio. Henceforth in this
book this turns ratio transformation will be tacitly assumed and we will also
The rotor voltage equations, eqn (2.14-10) to (2.14-25) can also be referred to
drop the use of the prime for this purpose.
the stator by the same turns ratio transformation.
Using a similar proccdure as for the induction machine it is not difficult to
When cqns (2.14-23) to (2.14-25), (2.14-43). (2.1444). and (2.1448) to
show that the power Input and torque output equations for the synchronous
(2.14-50) are written in scalar form they become the basis for Park's Equa-
machine are:
tiOtlS,

3
P, = 2- [vC;,iis + vQTSi&+ i>d ] (2.14-75)

The expression for power assumes that the 6 and q-axis damper c~rcuitvolt-
ages are zero.
When the torque equation is expanded by means of eqns (2.14-65) and
viq = r& i i q + p h i q (2.14-66), we can write
where 3P
T, = --
22
[ (L,, - Lqs) i i T i i 9+ Lmdi>d i i s + Lmdiidiis- L,,qi'kqiiT]
(2.14-77)
100 2 d.q Modelling of Induction and Synchronous Machines Extension of d.q.0 Theory to Analysis of Salient Pole Synchronous Machmes 101
- -

The first term in this expression is called the saliency torque which exists when
the machine has protruding rotor poles. The second term is called the excita-
tion torque or reaction torque which appears due to the excitation of the field
winding with dc current (which consequently 'reacts' with the stator current to
produce torque). The last two terms are recognized as the induction motor
torque which we previously obtained for the case of the induction motor. Since
this component serves to dampen oscillations in the power grid, this compo-
nent is also given the name damping torque.
Figure 2.19 shows a computer trace of the acceleration of a synchronous
machine when modelled in the rotor reference frame. The parameters of this
machine are as follows:

Pro,e, = 20 kVA Poles = 4 V = 230 V f = 60 Hz

rii = 0.1 R rId = 0.016 R rkd = 0.17 R rkq = 0.17 R
Lllld = 4.1 mH Lmq = 2.0 mH Lds = 4.89 mH Lq, = 2.79 mH
2
Lfd=4.48nrH Lkd=4.39mH Lkq=2.91nzH J=0.2N-m-s

Note that the machine takes considerably longer to accelerate than the
equivalent induction machine due to the relatively massive poles and weak
induction motor torque. While torque pulsations at line frequency are again
apparent, oscillations also appear at twice line frequency which is a result of
the saliency of the machine. Also note the slight dip in the average torque near
half speed. The effect, called the Gorges Effect after Hans Gorges, occurs
because of Ihc difference in the impedance of the d- and q-axis rotor circuits.
Because of this asymmetry a forward and backward wave is set up on the rotor
at slip frequency. While the forward wave rotates in synchronism with the sta-
tor MMF, the backward wave rotates backwards with respect to the rotor at slip
frequency or with respect to the stator at 1 - 2fSlip.Below half speed the wave
is rotating backward with respect to the stator windings. However, when the
rotor reaches half speed the wave begins to rotate forward with respect to the
stator. At the point just above half speed the backward wave exerts a strong
negative torque which could in severe case result in continuous operation near C-'-- _- _- _ J,
half speed, perhaps resulting in failure. Observe finally that when the machine Figure 2.19 Acceleration of d 20 kVA, 230 V salient pole synchronous
is excited as it reaches rated speed, the machine begins to oscillate strongly. machine from rest with a shorted field winding. Traces from
Such oscillations or 'swinging' are characteristic of machines operating from a the top: i,,, i;,, i L , ik ikd, qd, T,,, rotor speed in rpm;
fixed frequency supply. We will learn later in this book that such problems can time axis - 0.1 s/div.
102 2 d.q Modelling of Induction and Synchronous Machines References 103

be completely overcome when operating from a solid state converter by suit- Figure 2.20 shows a line start acceleration for permanent magnet machine
able control. An understanding of the source of such problems and the solu- having a starting cage. The parameters of this machine are as follows:
tions via control are the topic of the remainder of this book.
'rated= hp Vll(rms) = 230 V f = 60 h z Poles = 2
2.15 Extension of d,q,O Theory to Analysis of Permanent - 0.32 R
rs - rk,/ = 0.99 R rky = 2.0 R
Magnet Motors Lmd = 23 mH Lt,,q = 50 m H Lds = 26.2 m H Lqs = 53.2 m H
2
Permanent magnet machines form an increasingly important class of high per- Lkd = 29.4 mH L k q = 56.4 mH J = 0.01 N - m - s
formance ac motor drives. Fortunately, we need not repeat the analysis for this
machine since it is simply a special case of a salient pole synchronous
machine. In general, many types of permanent magnet machines exist which Note the marked increase in the fundamental frequency torque pulsation which
can be represented by different constraints on Park's Equations. The most gen- occurs due to the permanent magnet excitation. Since the permanent magnet
eral type is obtained by recognizing that permanent magnet excitation is sim- field cannot be 'turned off' the machine experiences a relatively strong braking
ply equivalent to a constant field current in Park's Equations. Defining the torque. In particular, both the excitation torque, saliency torque and the induc-
constant tion motor torque are separately plotted in Figure 2.20. Note that while the
induction motor torque is positive as expected, the saliency torque contributes
4Lf = L:nd iid (2.15-1) no net average torque while the excitation torque is actually negative. If the
Park's Equations for a permanent magnet machine equipped with a starting magnetic field of the magnet is sufficiently strong or, if the cage is not robust,
cage become the machine could actually 'hang up' at a subsynchronous speed. In any case,
the acceleration of the machine is prolonged which causes detrimental heating
of the rotor which is a potentially serious problem because of the temperature
sensitivity of the magnets.

References
A S . Langsdorf. "Theory of Alternating Current Machinery", McGraw-Hill Book Co.,
vLq = riq i i q + phiq New York. 1955.
W.V.Lyon, 'Transient Analysis of Alternating Current Machinery", John Wiley & Sons,
where
Inc., New York, 1954.
P.C. Krause andC.H. Thomas, "Simulation ofsymmetrical Induction Machinery", IEEE
Trans. on Power App:~t.arusand Systems, Vol. 84, Nov. 1965, pp. 1038-1053.
Y.K. He and T.A. Lipo. "Computer Simulation of an Induction Machine with Spatially
Dependent Saturation", IEEE Trans. on Power Apparatus and Systems. Vol. PAS- 103.
No. 4. April 1984. pp. 707-714.
T.A. Lipo. "Analysis 01' Synchronous Machines". Course Notes for ECE 5 1 I , University
of Wisconsin, 1994.
104
.-
2 d.q- Modclhng
- of I n--
d u c m ~dnd Synchronous Machmes

Problems
Problem 2-1 Reciprocity of Winding Inductances
Verify eqn 2.2-25 by direct application of the general expression for flux Imk-
age.
Problem 2-2 Complex Variable Coil Variable Equations
Verify eqn 2.6-14 by direct combination of the scalar equations describing the
rotor circuits.
Problem 2-3 Complex Variable 49 Equations
Derive the rotor circuit d,q equation given in eqn 2.8- 11.
Problem 2-4 Power in d,q Model
Derive the equation for the total three phase power in the stator windings ol' thc
uniform air gap machine model in terms of d,q variables including thc zcro
sequence quantities.
Problem 2-5 Source Voltages
Calculate the d,q voltages in stator, rotor and synchronous frames for a bal-
anced three phase, 460 volt, 60 hz system. Take the a-b line voltage as the ref-
erence quantity (i.e. zero phase cosine wave) and align the phase 'a' voltage
with the q-axis for the stator and rotor frames. Put all of the voltage in the q-
axis for the synchronous frame. Express the results in real variables and in
complex form.
Problem 2-6 Source Voltages-Unbalanced Supply
Repeat 2-5 for a 460 Volt, 60 Hz supply with vOb = vu, and vbc = 0 (put all
of the positive sequence in the q-axis for the synchronous frame).
Problem 2-7 Alternate Forms For Torque
Figure 2.20 Acceleration of 5 HP, 230 V. permanent magnet machine started Derive the torque equation in terms of the quantities listed below starting from
from rest. Traces from top to bottom: ,:i , q-axis stator current the basic form
(A.), iis d-axis stator current (A.), Trm,reluctance component
;,,

of torque (Nt-m), Tpm,PM excitation component of torque (Nt- a) airgap flux and rotor current,
m), Tim, induction component of torque (Nt-m), Tern =
Trm+Tpm+Tm, total electromagnetic torque (Nt-m), rotor speed b) airgap flux and stator current,
(RPM);time axis 0.1 sldiv. c) stator flux and rotor current,
lo6 -
- - .--
-
2 d.q Modelling of Induction and Synchronous Machines

d) stator flux and stator current, construct the

e) rotor llux and rotor current, d,q,Oequivalent circuit for an induction machine with iron loss.

f) rotor flux and stator current,

g) rotor Rux and stator flux.

Problem 2-8 Torque Expressed in Ph& Variables

Working backwards from eqn (2.12-12) derive an expression for the electro-
magnetic torque in terms stator 'phase variable' flux linkages and currents, i.e.
in terms of ins, ibs,i,, A,,,, AbS,and h,, . Can you provide a physical (or space
vector based) interpretation of your result? Note: Same circuit for phases b and c.

Problem 2-9 Balanced Capacitor Circuits in Expressed in a Rotating Ref-

erence Frame

Three balanced (equal) capacitors are wye connected across a three phase line.
Dercrrnine the equations and corresponding equivalent circuit expressing the
circuit behavior of these capacitors in a freely rotating d,q reference frame.

Problem 2-10 Incorporating Iron Loss Resistance Into the d,q,O equiva-
lent circuit.
Determine the d,q,Orotating reference frame equivalent circuit of the balanced
r-L circuit shown below. If rm and Lrn represent the iron loss resistance and
; 3 d,q Models for Solid State Power Converters
3.1 Introduction
For analysis of motor drives, it is sometimes convenient to characterize the
inverter in the same framework as used for the induction motor, that is, in
terms of a d,q,O components. As will be shown, the d,q orthogonal reference
system also yields converter models which are useful apart from their conve-
nience in combined machine-converter modeling. This is particularly evident
when a complex vector d,q representation is employed and the value of such
models becomes apparent in later divussion of. for example, current regulated
inverters.

3.2 d,q Model for Voltage Source Inverter

. The d,q model for a voltage source inverter (VSI)system is obtained by simply
applying the d,q transformation developed earlier in Chapter 2 to the inverter
constraint equations on a mode by d e basis. Recall that the d,q transforma-
tion from phase variables to d,q axes fixed in the stator is written in scalar form
as

where the symbol f represents voltage v, current i or flux linkage h.

To illustrate the development of the d,q equations which describe the six
k step voltage source inverter, consider first the connection 612 in Figure 1.4.
The voltages across the machine are then,
110 3 d.q Models for Solid State Power Converlers
d,q Model for Voltage Source Inverter 111
. -.
-. - -, . . .. . . ... --------. -

Vbc = 0 Applying the d,q transformation to the current constraints of eqns (3.2-6)
The current in phase as is

lus = li (3.2-6)
where ii is the instantaneous current supplied to the inverter from the dc link
and is defined in Figure 1.7.
To apply the d,q transformation, the voltage relations must be converted to
phase voltages which is readily accomplished by observing that the absence of i& = 0 (3.2-17)
a neutral connection forces the constraint that, Note that although zero current flows in the Ckomponent equivalent circuit,
(3.2-7) short circuit current will clearly flow in the d-axis circuit due to the speed volt-
ios + ibs + ics = 0 age which appear in the d,q,O equivalent circuit, see Figure 2.8. Equation (3.2-
For any balanced load (for example, a wye connected induction machine) the 16) cannot be reduced beyond its basic definition since the current which flows
line to neutral voltages are then constrained such that in this portion of the equivalent circuit is defined simply by induction machine
d-axis circuit conditions. The current i& clearly does not contribute to the
vos + vbs + vcs = 0 (3.2-8)
inverter current ii.
The phase voltages become, from eqns (3.2-4) and (3.2-5). If the same type of analysis is carried out for the remaining five modes
trated in Figure 1.4, the results summarized in Figure 3.1 are obtained for
the d,q variables. Since v& and i& are zero in all cases, they are not shown.
'1 The d,q relations for the VSI can be conveniently described by defining two
switching functions to express the constraint equations in Figure 3.1. The two
switching functions are illustrated in Figure 3.2 and permit writing the VSI d,q
constraint equations as
1
vcS = j ( v c u - I!bc ) =

Application of the d,q transformation then yields,

1
- -3
V. 1
(3.2-1 1)
V S
qs .
= 2- v , g S

2
1 qs
(3.2-1 8)

vis = ;
vi~is (3.2-1 9)

n.
- 1 , = i s gS
3 I qs qs + idssgjs (3.2-20)

V& = 0 (3.2-14) These expressions relate the instantaneous inverter input quantities vi and ii
the instantaneous d,q output quantities. The functions g and g& are inde-
Hence, modelling of the inverter during this time interval can be accomplished 9s
ndent time functions which express the inverter switching operations. Note
by applying 213 the inverter dc link voltage to the q-axis equivalent circuit of is related to the ideal phase voltage of the VSI and that is related
the induction machine while applying zero volts (i.e. shorting) the d- and 0- the deal line to line voltage. The choice of the quantities 7d3 and .&/6 as
axis circuits. e amplitude of g:s and g i s is arbitrary and have been chosen such that the
112 3 d,q Models for Solid State Power Converters

in the Fourier series for the two functions

Figure 3.2 The VSI switching functions g i s and gis

Figure 3.1 d,q equations for the six modes of a VSI The complex vector form of these d,q equations for the VSI take on very
simple forms and are very useful in visualizing and manipulating the equa-
114 3 d.q Models for Solid State Power Converters

tions. Combining eqns (3.2-18) and (3.2-19) yields the complex d,q equations
for the VSI voltages,

From the Fourier series for g& and g j s , the complex function g is given
qds
by B,qds = 5 (mode I )

t
- q-axis
&ds '~3 e ' ' ( m o d e 4 )

and is thus seen to represent a series of rotating complex vectors moving in

alternate directions at the speed a, and multiples of a,.The complex vector
form of the current constraint equation in eqn (3.2-20) is easily shown to be

An alternative and very useful form of the d,q complex vector voltage Figure 3.3 The six voltage vectors characterizing VSI operation
equation can be obtained by writing out the complex vector voltage expression
for each of the modes illustrated in Figure 3.1. Thus, in mode 1
d,q Model for PWM Operation
(mode 1) (3.2-26)
iques will alter the switch-
and in nlode 2 functions g i s and g& such that each type of PWM has its own unique
hing functions. Although it is a straight forward process to establish the
constraints and the associated switching functions, the results become
nvolved and are seldom actually required for analytical purposes. In most
cases the PWM inverter can be modelled in terms of its fundamental compo-
h harmonics treated as a separate issue. The fundamental component
The complex vector output ill bc presented in a later section.
vectors (60' apart), ere is one important aspect of PWM operation which distinguishes it
fference is the existence of
tional zero voltage states as illustrated in the switching mode diagram
sion for rlie complex vector voltage is re 3.4. In addition to the six possible voltage complex vectors associ-
d with the VSI (Figure 3.3), there are two zero voltage states associated with
1 three of the positive pole switches on or all three of the negative pole
onsidered as an eight state
This six vector concept is very useful in visualizing inverter operation and in
explaining various inverter properties.
 1 16 3 d,q Models for Solid State Power Converters
d,q Model for CSI System - 117

I
dq Equribns Complrx
Vector Current

The zero states are important in PWM operation and play significant roles, for
example, in the operation of current regulated PWM inverters.

Mode 7 Mode 8

Figure 3.4 Switch conditions corresponding to the zero voltage vector

state for a PWM inverter

3.4 d,q Model for CSI System

The duality of VSI and CSI systems implies that switching function models.
which are the direct duals of those of the VSI, must exist for the CSI. In gen-
eral, the exact dual of a VSI feeding a Y connected load is a CSI feeding a A
connected load. However, since it is more convenient to always consider Y
connected loads, the CSI model presented here is not the exact dual of the VSI.
There are, however, very close dual relationships between the analysis and the
results for the two types of systems.
The d,q equations for each mode are obtained by the same process as for
the VSI and are illustrated in Figure 3.5. The scalar equations can again be
written in terms of CSI inverter based switching functions h, such that

Figure 3.5 d,q equations for the six modes of CSI

where the CSI switching functions h4S, andhis are shown in Figure 3.6. The
Fourier series for the switching functions are,
1 I8 3 d,q Models for Solid State Power Converters
d,q Model for CSI System 119

where

hids = epll-
5
le-.i5@,1
7
+ 97@,f-
... (3.4-8)
>
Note the close relationship between the switching - functions for the VSI and the
CSI. In particular, note that in h4sdS only the signs of the harmonic terms are
altered. An alternate expression for the complex vector current can also be
I X
Y I I I YL
L Jc
. written based again on the fact that the complex vector current is constant in
each mode and shifts by 60" at each mode transition. The exuression is
w e ' -b

as can be easily verified by comparison with the last column in Figure 3.5. The
: six vectors corresponding to the switching of a CSI is shown in Figure 3.7.

I I

-
71: n -
3n
2
2n
2
we* +-
1-
Figure 3.6 The CSI switching functions his and h &

1 1
his = s i n m e t - - s i n 5 w e t + - s i n 7 ~ -
5 7
... (3.4-5)

In complex form these equations become

Figure 3.7 The six current vectors characterizing CSI operation
i,",, = 2.i,(h,",-
.b jh$) (3.4-6)
!' . It should be noted in passing that the idealized d,q model of a CSI as pre-
ented above is a less accurate representation (for inductive loads) than the cor-
v , = ~ R I [ If"
~ '] (34 7 ) esponding model for a VSI. The inherent commutation delay associated with
X . qds qds
120 3 d.q Models for Solid State Power Converters Inverter d.a Models in a Svnchronous Reference Frame 121

switching the current is much more significant than the commutation time
associated with switching the voltage in a VSI. In mosl systems the CSI output
currents will be decidedly trapezoidal instead of rectangular in waveform,
especially at frequencies at the high end of the inverter operating frequency
range.The errors associated with using the idealized CSI model are usually
quite minimal. Corrections for the commutation phase shift and for the change
in fundamental component can be incorporated whcre necessary. Note, also,
n?ii = Re [ i9ds
e eje. (g;ds) * ]
that the accuracy question would be reversed if h e load were capacitive;
another dual property of VSI and CSI systems.
= Re ~ i ; ~ ~ g ; ; ~ l

3.5 lnvcrter d,q Models in a Synchronous licference Frame where the new cornplcx switching function 8.qdr is defined ns
Although the stationary d,q reference frame is appropriate for many purposes,
a synchronously rotating reference frame offers some advantages. It is espe-
cially useful in simulation and in the development of small signal models and = s,', - j g i S
transfer functions. Written explicitly in terms of their components, eqns (3.5-5) and (3.5-6) are
The most direct development is to simply transform the stationary frame
therefore
models of the preceding sections to the synchronous frame using the basic
transformation equations from Chapter 2. Repeating, the pertinent transforma-
tion is

f & = f ; , sin 0 , + f jS COSO, (3.5-2) n- 1 .. = i e g

3 1 qs i s + i&G (3.5-10)
where 0 , is the electrical angle corresponding to the inverter operating fre-
quency we Using the expression for g 4S s given in eqn (3.2-24). the complex Fourier
series for ge
- gds is
r
ee = joedt + 0 , (3.5-3) e - ,i(a,l-eJ + ie-i(sael + 0.) - -1 ,j(7ae1 -et) - . . .
0 gqds - 5 7
The manipulations are more readily handled using the complex vector form of (3.5-1 1)
the transformation equations,
Expanding and separating into two components yields the scalar switching
functions
.e .s ,-ie<
lqds = lqds (3.5-4)
For the VSI, applying the transformation to eqns (3.2-23) and (3.2-25) respec-
tively,
122 3 d,q Models for Solid State Power Converters
Examples of Inverter-Induction Motor Models 123
-~

In the above two equations we have assumed that the rotating d,q axes have and the scalar switching functions are
been synchronized with the fundamental frequency of the inverter output volt-
age, i e . Oc = act. The time functions which generate these series are illustrated
in Figure 3.8.
12 24 .
his = --sin6~o~t- -s1nl2w~t+ ... (3.5-1 8)
35 143
The scalar equations corresponding to eqns (3.5-14) and (3.5-15) follow
directly. These functions are illustrated in Figure 3.9. Note that the functions

Figure 3.8 The synchronous frame VSI switching functions

The choice of phase reference in these equations (8& = 0) has placed all of
the effect of the fundamental frequency (dc in the synchronous frame) in the q-
axis. The harmonics are also shifted in frequency by a, such that only a sixth
harmonic ripple (plus multiples of the sixth) now exists in the inverter voltages
and currents. [
I wet -
Figure 3.9 The synchronous frame CSI switching functions
t
By the same method the synchronous frame CSI equations become I
k are the same as the switching functions for the VSI except for the fact that the
initial condition has been chosen differently for the CSI case.

3.6 Examples of Inverter-Induction Motor Models

To illustrate the application of the inverter d,q models of the preceding sec-
. the overall d,q model of a voltage stiff inverter and an induction motor is
tions,
.I(
presented here. All that is required is to combine the inverter models developed
above with the motor models previously developed. Figure 3.10 shows the sta-
tionary d,q model of a VSI driving an induction machine. The inverter current
where it is represented by two current sources dependent on the two motor d,q cur-
rents iSg, and ii . This circuit represents a complete dynamic model In which
the inverter voltage v, can be any independent time function and the Inverter
124 3 d.q Models for Solid State Power Converters Examples of Inverter-Induction Motor Models . - - .- .-.

q-axis

A q-axis
'i

(We-W,) A,,,.
iigds'ds iFg9s1 4,s
L,,, 'r

Figure 3.10 Stationary frame d,q model of a VSI driven induction machine I I

d-axis
switching is completely modelled including all harmonics. The model of the Figure 3.1 1 Synchronous frame d,q model of a VSI driven induction machine
machine and the switching functions can also be represented in the synchro-
nous frame. The equivalent circuit becomes that of Figure 3.11. Similar equiv-
alent circuits for the motor driven from a current stiff inverter supply are
shown in Figure 3.12 and Figure 3.13. The simplified equivalent circuit is illustrated in Figure 3.14. If we take note of
For many purposes, only the fundamental component of the Fourier series the fact that
representing the output waveforms of an inverter need be retained. In electric
machine analysis this is a common approximation since it is easily shown that
the harmonics result in very little average torque. Thus, in cases where only the
basic overall electromechanical response is of interest, the system can be mod- the need for the current source models to represent the inverter current disap-
elled with sufficient accuracy by simply neglecting the harmonics caused by pear entirely. In a similar maaqer the harmonics can be neglected when model-
inverter excitation. Examination of Figure 3.8 and eqns (3.5-12) and (3.5-13) ling a CSI driven machine. In this case, from eqns (3.5-17) and (3.5-1 8),
indicate that if we wish to neglect harmonics, the inverter can be modelled in ).
the synchronous reference frame by simply neglecting the effects of the 6th e ~1
h9s (3.6-4)
harmonic and multiples of the 6th.The switching functions become simply,

The corresponding simplified equivalent circuit is shown in Figure 3.15.

126 3 d,q Models for Solid State Power Converters Examples of Inverter-Induction Motor Models 127

Figure 3.12 Stationary frame d,q model of a CSI driven induction machine
gure 3.13 Synchronous frame d,q model of a CSI driven induction
The models represented by Figure 3.10 and Figure 3.11 can be used to machine
obtain the transient response of a VSI driven induction motor. Figure 3.16 and
Figure 3.17 show computer traces of the acceleration of a VSI driven induction cannot reverse. This effect is modelled with a simple diode. The equations of
machine in both the stationary and synchronous reference frames. Note that the the L C filter are
representations of v9,, i9, and iqr9arequite different in the two simulations,
whereas the remaining quantities are all identical. vr = vi + ( rlf + p Lf) ir i@ (3.M)
Figure 3.18 portrays behavior of the system when the effects of the inverter
vr = vi ir=O (3.6-7)
harmonics are neglected. It is apparent that these harmonics have little effect
on the basic electromechanical response of the system. It also should be noted 11
that these simulations assume that the dc link voltage is ideal, i.e. a constant. In vi = --(ir-ii) (3.68)
pCf
practice, this voltage can fluctuate substantially, producing, in reality, an oscil-
p again denotes the time derivative operator dldt and llp represents the
latory or even unstable operating condition [1].[2]. In such cases the dc link
components must be modelled in more detail. with respect to time.
Figure 3.19 shows the synchronous reference frame representation of the ' Observing the proportionality constants in eqns (3.5-8) to (3.5-lo), these

inverter and induction motor which also incorporates the effect of the dc link bations can be rewritten as
filter. For simplicity, the rectifier input, which may in fact be a single phase or
three phase rectifier, remains modelled as a voltage source in which the current 2 6
= - - ~ ~ + ~ ( )r (~~ +i ~p ) ~ i r > o (36-9)
ivr 7K f 3
128 3 d.q Models for Solid State Power Converters Examples of Inverter-Induction Motor Models 129
- ---

y-axis

d-axis

Figure 3.14 Simplified synchronous frame d,q model of VSI driven induction
machine Figure 3.15 Simplified synchronous frame d,q model of CSI driven
induction machine
the dc link capacitor is assumed to be completely charged before corimands
are given to the inverter to supply voltage to the motor. The results are nearly
the same as before indicating that the effects of the input supply impedance
have essentially been decoupled from the output by the action of the large filter
capacitor, a useful attribute from the point of view of ease of control.
These equatior; combine with eqns (3.5-8) to (3.5-10) and the induction It should be mentioned that the condition studied corresponds to a fixed
motor equations in the synchronous frame to form Figure 3.20. Note that now inverter frequency of 30 Hz. In prfictice, the motor is rarely, if ever, started
the filter model has been conveniently coupled directly to the q-axis machine under such conditions since the resulting inrush currents to the motor would
model and the inverter harmonics appear as a series voltage source equal to destroy the inverter switches. In mpst practical cases, open loop general pur-
(2/7t) v i (g: - 1) . When the effects of the inverter harmonics are neglected, pose drives such as the type modelled here are normally started w ~ t hthe
the two voltage dependant sources in Figure 3.20 now become short circuits inverter set at the permissible minimum frequency, usually only a few hertz.
and the equivalent circuit of Figure 3.20 results. In this case primed variables I
The frequency command is then ramped up to ensure that the motor continues
have been defined to represent the dc link variables which can now be consid- to operate with a slip frequency of only a few hertz as it accelerates up to the
ered as referred to the stator in much the same manner as the turns ratio trans- desired speed.
formation used to refer the rotor variables to the stator i n Section 2.7. The equivalent circuits for a CSI fed induction machme including the effect
To show the effect of the filter on the acceleration of an induction machine of the dc link inductor can be developed in much the same manner. Once again,
from rest, Figure 3.21 can be compared with Figure 3.16. In this computer run, scaled variables can be introduced on the dc side to correspond to the Inverter
130 3 d,q Models for Solid State Power Converters Examples of Inverter-Induction Motor Models 13 1
-- - -
.- .

Figure 3.16 Simulation of an induction machine accelerating from rest

using a VSI supply with the system represented in the
stationary reference frame. The inverter frequency is fixed at Figure 3.17 Simulation of an induction machine accelerating from rest
.s with a VSI supply with the system represented in the
30 Hz.VA = vas = v;, (V), IAS = ias = 1
9s
(A), IAR = synchronous reference frame. The inverter frequency is fixed
iClr = i Y r ' (A), TEM = Te (N-m), TLOAD = Load Torque
I
at 30 Hz.VQSE = v i S (V), IDSE = i6 (A), IQRE = i:r (A).
(N-m), SPEED = come,,, = 20/P (radls) TEM = T,, (N-m),TLOAD = TIoad (N-m), SPEED =
132 3 d.q Models for Sol~dState Power Conveners
--- -- -- -- Fundamental
- ..- - Component Approximation for Steady
-- Stale Operation
..-

switching relations, eqns (3.5-14) and (3.5-15). Figure 3.22 shows the result-
ing equivalent circuit. When the harmonics impressed on the motor are
neglected, eqns (3.64) and (3.6-5) apply and the equivalent circuit reduces to
that of Figure 3.23. Note that whereas the VSI is modelled by connecting the
scaled voltage dc bus to the q-axis and shorting the d-axis, Figure 3.20, the
CSI simplified circuit is obtained by connecting the scaled dc current bus to the
q-axis and open circuiting the d-axis.

Figure 3.19 Synchronous frame d,q model of VSI driven induction

machine including the effects of the dc link filter

Figure 3.18 Simulation of an induction machine accelerating from rest with

3.7 Fundamental Component Approximation for Steady State a VSI supply with the system represented in the synchronous
Operation reference frame neglecting the effect of the inverter harmonics.
It has already been noted that when the drive torque characteristics are the pri- The inverter frequency is fixed at 30 Hz. VDSE = v;,, (V),
mary consideration or when only the approximate overall characteristics are
VQSE = vi, (V), IDSE = i;, (A), IQSE = iiS (A), IQRE = i:,:
desired, neglecting all harmonics and representing only the fundamental com-
ponent behavior is quite acceptable. Incorporating the harmonics (or carrying (A), TEM = Te (N-m), SPEED = ( 2 a r ) /P radk
out time domain solutions) is justified only in cases where waveforms, torque
pulsations, efficiencies, or other internal details of operation are the desired The steady state output waveforms for six step VSI and CSI inverters are
results. The simplification resulting from neglecting harmonics is very signifi- repeated in Figure 3.24 with the fundamental component shown as shaded
cant and is the basis for the steady state inverter equivalent circuits presented curves. Although the phase shift of the waveform on a per phase basis is arbi-
in this section. trary, a typical phase and line voltage can be described by reference to Figure
134 3 d,q Models for Solid Stare Power Converters
-- - --
Fundamental Component Approximation for Steady State Operation 135

Figure 3.20 Simplified synchronous frame d,q model incorporating the

effects of the dc link but neglecting the effect of the inverter
harmonics
3.24. Considering phase voltage v, and line voltage -vb, as typical, the Fou-
rier series representing these waveforms for the VSI are:

where 0 = o,r + @ is determined by the inverter switching rate and the time zero
reference. The corresponding results for the CSI are obtained by replacing v
with V and i with I (to denote steady state). Thus, from Figure 1.15, using the Figure 3.21 Simulation of an induction machine accelerating from rest
instant of switching from interval 61 to 12 as the t = 0 reference, with an voltage source inverter supply. Simulation
i incorporates the effect of the dc link filter. The inverter
frequency is fixed at 30 Hz. VI = V i(V), VA = v,, = vS
9 s (V),

IA = i,, = i i , (A), IAR = :,i (A), TEM = Te (N-m),

TLOAD = Tload (N-m), SPEED = (2~0,)/ P (radls)
136 3 d,q Models for Solid State Power Converters Fundamental Component Approximation for Steady State Operation -- - 137

q-axis

d-axis

Figure 3.22 Synchronous frame d,q equivalent circuit of CSI fed induction Figure 3.23 Simplified synchronous frame d,q model of CSI fed induction
motor incorporating the effect of the dc link inductor. Link motor incorporating the effects of the dc link but neglecting the
variables referred to the induction motor stator effect of the inverter harmonics

?
? Whereas in transient analysis the peak values are of importance, it is con-
The output voltage waveforms for the PWM-VSI inverter described in
Section 1.7 are also shown in Figure 3.24. In general the Fourier series describ- ventional in steady state analysis to use rms quantities. Retaining only thc fun-
ing this type of waveform can be written in the form damental term yields the following relations between the rms line voltage Vi,,,,
and phase voltage Vphoseand the inverter dc voltage Vi in a VSI.

' l i n e , rms It (3.7-6)

where the PWM coefficients akand phase shifts Ok are dependent on the specific
type of PWM modulation employed. In addition to the odd harmonics shown in
eqn (3.7-5) some PWM algorithms may even contain non-triplen even har-
"phase, rms
"4
-- 7
monics (even harmonics not a multiple of three). For the simple case illustrated Similar relations for the CSI relating the rms line and phase currcnts Ilble and
in Figure 3.24, the values of akcan be readily written as a function of the length Iphase ( A connection) to the inverter dc current lican also be written
of the short circuit submode. Note that as defined, a , is always equal to or less
than one. 'line, rms
= @Ii
It
(a) (3.7-8)
138 3 d.q Models for Solid State Power Converters Fundamental Comwnent A~oroxirnationfor Steadv State Owration 139
- - -

= -.J2
Fundamental
Component 'phase. rms
Ii
Line Voltoge Vlim
For the PWM-VSI, the relations are similar to those for the VSI in eqn (3.7-6)
except that the PWM ratio a l is present as a multiplier as shown in eqn (3.7-10).
IN V i

i 2/3 V;

VSI Wuvejbrms
Phase Volrage Vphe
- J6.,Yi
'line, r m s - X

c
"phase, r m s - JZ.,V, (3.7-1 1)
The relations given in eqns (3.7-6) to (3.7-10) represent what can be
Line Current described as the basic or primary switching constraints of the three types of
inverters under consideration. They describe the dc to ac fundamental compo-
f nent transfer relation for the directly controlled quantity for each type of
$ inverter; voltage in the VSI, current in the CSI. There remains the questlon of
the transfer relation for the other (auxiliary) variable; current in the VSI and
voltage in the CSI. This auxiliary relation is not determined by the Inverter
. alone since it is clear that the switching constraints do not completely specify
CSI Waveforms the auxiliary variable. Because of the degree of freedom in the auxiliary vari-
able switching constraints, the nature of the load has an important impact on
the transfer relation. A detailed analysis of the auxiliary variable transfer rela-
tion is presented in 131. For our purposes it is much simpler to consider the
overall input-output balance and to determine the auxiliary variable transfer
mlation on this basis. The procedure is illustrated for the six step VSI; the other
'cases are handled in exactly the same way and only the results are presented.
1/3 V i For the VSI, the input-output voltage relation is known, eqn (3.7-6), but
I In! 2/3 V;
7-77
L
Phase Voltage Vph,
there remains the problem of determining the input-output current relation. If
we assume that the inverter behaves as an ideal dc to sine wave converter (i.e.,
assume the total dc input power is converted to fundamental frequency power
PWM-VSI Waveforms
in the output) we can write for a Y-connected load

r m s ~ ~ ~ $$ =. power factor angle (3.7-1 2)

V iIi = 3 Vphosr,rr,,,%llirle,
Figure 3.24 Output waveforms for basic inverters with constant dc
which when combined with eqn (3.7-6)(b), yields the following input-output
input
current relation
140 3 d,q Models for Solid State Power Converters
Fundamental Component Approximation for Steady State Operation ----- - 141
--

of the in-phase component of the load current. A simple equivalent circuit

which is described by eqns (3.7-16)'and (3.7-17) is shown in Figure 3.25.
The interpretation of this result is that only the in-phasc or power component of
the output current is reflected to the input of the inverter and hence, by implica-
tion, that the quadrature or reactive component circulates in the inverter in the
form of the short circuit current shown in the conduction mode diagram of Fig-
ure 1.4.
The corresponding result for the CSI, obtained with the same type of power
balance relation, is

Figure 3.25 Partial per phase fundamental component equivalent circuit

of three phase six step VSI
In this relation, and for the remainder of this discussion we assume Yxonnect-
ed loads on both the VSI and CSI. At the right hand end of the circuit are the inverter output terminals con-
For the PWM-VSI, the result corresponding to eqn (3.7-13) becomes nected to the load. The rms phasor voltage and current at the load are
Vpha,TeLOO = (&E) Vi and IlineL0. Connected in parallel is a variable
reactive element (a capacitor for an inductive load) which cancels the reactive
part of the load current. This element represents the effect of the inverter short
where the only change is the addition of the PWM ratio n l as a multiplying fac- circuit currents and is inherently continuously adjusted to be in parallel reso-
: nance with the load impedance in the steady state. To the left of this element,
tor in the equation. . - the current is the inverter input current referred to the ac side of the inverter
The input-output transfer relations given by eqns (3.74) and (3.7-13) for : which is given by the expression tlc/3&) Ii (eqn (3.7-17)). The input and
the VSI, eqns (3.7-8) and (3.7-14) for the CSI, and eqns (3.7-10) and (3.7-15) output voltage are the same and equal to the phase voltageVphose,r,ns given by
for the PWM-VSI can be conveniently expressed in equivalent circuit form if (&/n:) Vi, eqn (3.7-16). The inverter functions as a source of the reactive
all quantities of interest are referred to the ac side of each inverter. Again using power required by the load. continuously adjusting its supply of reactive power
the VSI to illustrate the method, observe that (adjusting the value of the shunt reactive element) to be exactly equal to the
1) the ac side equivalent of the inverter input voltage is given, on a per reactive demand of the load.
phase basis, by eqn (3.7-7) A more complete equivalent c4cuit including the resistance rl/ in the dc
link filter inductance supplying the inverter input terminals can be constructed
as shown in Figure 3.26. The ac* voltage drop is rlf Ii, but this must be
'phase, rms
-- 7
Jzvi referred to the ac side by multiply&g by the voltage transfer ratio &/n: to
2) the current constraint relation, eqn (3.7-1 3), can be rewritten as yield (&n) r g I i . Since the ac side inverter input current is (n/3&) Ii , the
referred value of rlf becomes

which expresses the ac side equivalent of the inverter input current Ii in terms
142 3 d.q Models for Solid State Power Converters Fundamental Component Approximation for Steady State Operation -- 143

The input voltage to the dc link V, can also be referred to the ac side by
multiplying by $2/n to complete the diagram. The power balance equation
for Figure 3.26

verifies that the circuit correctly represents the per phase power relations of the
inverter.
Figure 3.27 Per phase fundamental component equivalent circuit of
three phase six step CSI

Inductive
Load

lnduclive
Load

Figure 3.26 Per phase fundamental component equivalent circuit of three

phase six step VSI

Figure 3.27 presents the corresponding' equivalent circuit for the CSI. In I
Figure 3.28 Per phase fundamental component equivalent circuit of the
this case the reactive element is in series with the load current to represent the three phase PWM-VSI inverter
reactive voltage developed as a result of the open circuit inherent in the CSI
switching modes. As for the VSI, the reflected effect of dc link resistance is
included in the circuit for those cases where the source feeding the CSI is a 6 The variable ratio 1: a, transformer is a very compact and convenient means of
%
'!+
voltage source. If the input to the CSI can be represented as a current source, representing the effect of PWM-control.
* The circuits of Figure 3.26, Figure 3.27 and Figure 3.28 can be used to ana-
the resistance and equivalent ac side voltage source can be omitted and
replaced by the current source. lyze the steady state fundamental fnquency behavior of VSI and CSI inverters
For the PWM-VSI inverter, comparison of eqn (3.7-6)(b) with (3.7-10)(b)
and eqn (3.7-13) with (3.7-15) reveals that the influence of the PWM control
is to introduce the PWM ratio a1 in exactly the form of the turns ratio of a
1 feeding any type of balanced linear load. The variable reactive elements in the
circuits are the fundamental frequency representations of the inherent ability of
i- the inverters to supply the reactive power required by the load. Since the loads
transformer. This suggests the modification of the VSI equivalent circuit of I are assumed to be balance three phase loads, all that is really required is a
i
Figure 3.27 to produce the PWM VSI equivalent circuit shown in Figure 3.28. means of transferring the stored energy from phase to phase. For inductive
1 44 3 d,q Models for Solid State Power Converters References 145

loads this transfer mechanism is provided by the short circuit paths in the VSI Table 3.1 Duality of VSI and CSI Systems
and by the commutation of current from phase to phase in the CSI. For capaci-
tive loads the transfer occurs at commutation in the VSI and in the open cir- VSI I
I
CSI I
cuited phase in the CSI. Thus the accuracy of the representation depicted in the 1
7) dc bus current reverses in regen- 7) dc bus voltage reverses in
circuits is better for the natural cases of inductive loads on the VSI or capaci- eration regeneration
tive loads on the CSI. but can be applied to the other cases if care is used in
8) Immune to open circuits 8) Immune to short circuits
interpreting the results and the likely occurrence of commutation spikes associ-
ated with the reactive energy transfer is recognized.
References
3.8 Duality of VSI and CSI Systems [I ] T.A. LIFOand P.C. Krause, "Stability Analysis of a Recttfier-Inverter Induct~onMotor
Drive", lEEE Trans. on Power A p p ~ t u and
s System%Vol PAS-88. No I. Janun~y
The duality of VSI and CSI systems has been stressed in the preceding sec- 1969, pp. 55-66.
tions. In terms of the basic power circuit, the two systems are exact duals if we T. A. Lqo. P. C. Krause and H. E. Jordan. "Hmnontc Torque and Speed Pulsations In a
[2]
compare a VSI with a wye connected load with a CSI with a delta connected Rectifier-Inverter Induction Motor Diive", IEEE Tramon Power Apparatus and Sys-
load. The equivalent circuits of the preceding section do not exhibit exact dual- tems, Vol. PAS-88, No. 5, May 1969, pp. 579-587.
ity because a wye connected load is assumed in all cases. However, the duality [3] D.W. Novotny, "Switching Function Representation of Polyphase Invertors". 1975 In-
principle is still a valid and important concept. Table 3.1 summarizes the dual dustry Applications Society Annual Meeting, Conf. Record, Atlanta GA, Sept. 28-0ct.
properties of these two basic systems. 2, pp. 823-83 1.
[4] D.W. Novotny and T. King. "Equivalent Circuit Representationof Current Inverter Drw-
Table 3.1 Duality of VSI and CSI Systems
en Synchronous Machines," IEEE PAS,Vol. 100, No 6, June 1981. pp 2920-2926.

capacitor inductor
3) dc bus current proportional to 3) dc bus voltage proportional to
motor power and hence dependent motor power and hence dependent
( on motor power factor I on motor power factor I
4) Output contains voltage har- 4) Output contains
monics varying inversely as har- monics varying inversely as har-
monic order monic order
5) Prefers motors with larger leak- 5) Prefers motors with lower leak-
age reactance age reactance
6) Can handle motors smaller than 6) Can handle motors larger than
inverter rating inverter rating
 146 3 dq Models for Solid State Power Converters
Problems .. . . . 147

Problems Derive the Fourier series in eqns (3.5-12) and (3.5-13) from the complex
series in eqn (3.5-11).
Problem 3-1 Six Step Waveform (VSI)
For the six step waveform shown below (with each step of duration ld3 radi- Problem 3-5 VSI Waveforms
ans) For a VSI operating with a constant voltage input v, = 100 volts at a constant
fundamental frequency of 60 hz, sketch the steady state waveforms of i i d , i i ,
and i, for :
a) a balanced three phase, Y- connected resistive load of 10 a/$.
b) a balanced three phase, Y- connected inductive load of L=5 mW$.
c) a balanced three phase, Y- connected R-L load with r=10 R / $ and
L=SmN$.
d) a balanced three phase, Y- connected R-C load with -10 R / + and
C=50 pf /$ where 'I$' denotes 'per phase'.

Problem 3-6 CSI Waveforms

For a CSI operating with a constant current input ii = 10 amps at a constant
fundamental frequency of 60 hz,sketch the steady state waveforms of vi, ,viS
and v i for :
a) a balanced three phase, Y- connected resistive load of 10 R/$ .
b) a balanced three phase, Y- connected capacitive load of 50 pf /$
Find: c) a balanced three phase, Y- connected R-L load with R=10 R / + and
a) vq,S and v 2 as waveform sketches. L=5 mN$ .
b) expressions for v& and v& as ~Aurierseries. d) a balanced three phase, Y- connected R-C load with R=10 R/Q and
C=50 pf /Q where 'I$' denotes 'per phase'.
C)complex variable Fourier series for %dS.
d) complex variable Fourier series for %dS. Problem 3-7 VSI Harmonic Analysis

Problem 3-2 VSI Switching Functions For the conditions specified in problem 3-5, find the fundamental and 5th har-
monic components of i i T and s:i and the dc and 6th harmonic components of
Verify that the Fourier series in eqns (3.2-21) and( 3.2-22) are correct repre- 4.
sentations of the switching functions gi, and g:S shown in Figure 3.2.
Problem 3-8 CSI Harmonic Analysis
Problem 3-3 CSI Switching Functions
For the conditions specified in problem 3-6, find the fundamentaland 5th har-
Verify that the Fourier series in eqns (3.4-4) and (3.4-5) are correct representa- monic components of V$ and vjs and the d~ and 6th harmonic components of
tions of the switching functions hi, and hsd, shown in Figure 3.6. y.
Problem 3-4 Synchronous Frame Switching Functions
148 3 d,q Models for Solid State Power Converters Problems . ~ ~ 149

Problem 3-9 VSI Synchronous Frame Model tor with the VSI equivalent capacitor serving as the excitation source2 (reactive
The synchronous frame model of a VSI with harmonics neglected is given by power). Show that for no-load (Rdc + 0 0 ) , the relations (ueLm)= -rl r2
eqn (3.6-1) and (3.6-2). The zero reference for the synchronous frame for this must hold and hence the level of saturation (and output voltage) is determined
case is chosen such that the q-axis coincides with phase a at t=O. Derive the by the slip s and the machine resistances r , and r 2 .
general model corresponding to these equations where the q-axis is at an angle
Problem 3-13 CSI Constant y System
QOe from the a-axis. Sketch the general form of the equivalent circuit shown

in Figure 3.14. Derive the torque-speed characteristic for a system of the type in problem 3-
11 operated with constant y and constant $ . What type of dc machine has a
Problem 3-10 VSI Driven Synchronous Machine similar characteristic?
Derive a fundamental component equivalent circuit for a VSI driven non-
Problem 3-14 CSI Self-Excitation
salient pole synchronous machine. Use the circuit to derive the torque-speed
characteristic of this type of drive operated with a constant value of torque Find the relation for CSI self-excitation corresponding to the result given in
angle 6 (between V and E) and constant If .What type of dc machine has a problem 3-12 for a VSI (the dc terminals are short-circuited in the CSI case).
similar characteristic?

Problem 3-11 CSI Driven Synchronous Machine

Show that the fundamental component equivalent circuit for a salient pole
synchronous machine driven by a CSI is as shown in the figure below. The
angle y is the internal power factor angle (between E and I ) of the machine and
Req =(1/2) ( X q - Xd) sin2y

0' Ecos 6

Problem 3-12 VSI Self-Excitation

Consider a VSI driven induction machine in which the dc supply is replaced by
a load resistor. This system can be operated as a self excited induction genera-
cept in simplifying and manipulating the differential equations describing the
machine. The resulting model can be interpreted as a set of ordinary, real vari-

machine in specific situations. For each application a set of external constraint

relations describing the specific situation is developed and imposed on the gen-
to obtain the application specific model.
When the external constraints are unsymmetrical (i.e. single phase opera-
tion, unequal phase impedances) the analysis is usually best carried out in
terms of the real variable d,q differential equations. However, in those situa-
tions where the fundamental polyphase symmetry is retained, analysis can be
conveniently carried out in complex vector form. The result is that the four

complex variable model [1.2]. This reduction in order permits considerable

simplification in analysis. For example, in the case of constant speed operation

c equation [3,4]. Even in cases where a lack of symmetry requires that the

in terms of complex vectors.

ch correctly represent the solutions play a significant role in understanding

152 4 Complex Vector Analysis of Induction Machines Turns Ratio and Modified d.a Models 153

constant speed operation and these circuits are also readily derived using com- through the rotor equation (4.2-2) by the operator (p + j o ) 1 [p + j(w - a,)] to
plex vector modelling. obtain (with %dr = 0)

4.2 Complex Vector Equivalent Circuit

An equivalent circuit representing the arbitrary frame complex vector differen-
tial equations developed in Chapter 2 can be constructed if the equations are The equivalent circuit suggested by this result is shown in Figure 4.2. 11 has the
expressed as

Y9ds = ( r s + L , y ~ ) i q d sL+m ~ i q d r + j ~ h q d s (4.2-1)

j Or) hqdr
Vqdr= 4 p i 9 d s + ( r r + L r ~i)q d r +(a- (4.2-2)
where the flux linkages are given by

Figure 4.2 Complex vector operational equivalent circuit (zero external

hqdr = Lmiqds+ Lriqdr (4.24)
rotor voltage)
The complex vector equivalent circuit corresponding to these equations is
illustrated in Figure 4.1. Note that this circuit is of exactly the same form as the
merit of simplicity and clearly the ori8inal complex variable differential equa-
tions can be recovered by writing out the circuit equations. It is useful for cir-
cuit manipulations and as a simple representation of the dynamic model of a
w
1 V machine with zero applied rotor voltage (induction machine). In steady state
i qdr j(a-ar)? analysis it also can be used as a rapid means of obtaining the steady state
equivalent circuit for balanced operation and demonstrating several important
'qds 'qdr
,
' steady state results.

- - 4.3 Turns Ratio and Modified d,q Models

The selection of the 'turns ratio' between the stator and rotor circuits of the d-
Figure 4.1 Complex vector equivalent circuit
q model can be shown to be essentially completely arbitrary when the machine
is operated with zero applied rotor voltage. If we interpret the basic complex
real variable circuits. The speed voltages again &cur as dependent generators;
vector model as describing a machine in actual 'non-referred' parameters and
in this case both appear as counter emf voltages with thc sign reversal between
introduce an arbitrary constant b, the basic equations can be written in the form
the d and q circuits handled by the j operator. Clearly the two real variable cir-
cuits are easily obtained by inspection from the complex vector circuit.
An operational form of equivalent circuit can also be employed to repre-
sent the complex vector equations. This form is obtained by multiplying
154 4 Complex Vector Analysis of lnducrion Machines

r,,, = h ~( p,+ j ( o - o r ) )iqds+ b 2 [ r r + ~ ~+ j ((W-W,))

p I % (4.3-2)
These equations, with parameters,

and 'referred' variables,

Vqds' iqds,byqdr' iqd/b

wherc the 'referral ratio b is any constant we might want to choose (except b =
0 or w). Note that if we are willing to give up knowledge of the actual value
of the rotor current (and voltage), we do not need to know the value of b. Thus,
in induction machines where v+dr = 0,we can utilize this freedom to simplify
the model by letting b have a convenient value. For example:
I) Let L , = h2Lr or b = (LS/L,)''2. This selection yields equal self induc-
tances and results in a circuit model with equal leakage inductances.
N
-s
4) Physical Circuit b = Nr
2) Let bLm = b2L, or b = Lml L, This choice produces equal mutual and
rotor self inductances and results in a circuit model with all 'leakage'
located in the stator.
3) Let L, = bL, or h = L, l L,. S e l ~ t i n gequal mutual and stator self
inductance results in a model with all 'leakage' located in the rotor.
4) Let h = N, l N , Choosing b as the stator to rotor turns ratio results in
the "real" leakage inductance model based on the designer's computa-
tion o f leakage inductances. This is the most commonly used version of
the various choices.
The operational equivalent circuits resulting from these various selections
of h are illustrated in Figure 4.3. Circuit 4 is the most common selection and
the one which has been used in the basic model as given in Chapter 2. When
machine parameters are determined from laboratory tests, circuit 1 is often
used since there is no way to measure the actual turns ratio unless measure-
ments in the rotor circuit can be made. Circuit 2 is very convenient in systems
which control rotor flux, such as in field orientation and in switching transient
problems where a very widely used approximate method based on constant as obtained from various selections of
rotor Hux linkages is applicable. These situations will be examined in more
detail in later sections.
 Turns Ratio and Modified d,q Models 155

by,,, = bL,n(~+j(o-or))b+b2[rr+~,(p+j(o-or))l~
(4.3-2)
These equations, with parameters,

Vqds7 iqds, iqd/b

where the 'referral ratio b is any constant we might want to choose (except b =
0 or m). Note that ~fwe are willing to give up knowledge of the actual value
of the rotor current (and voltage), we do not need to know the value of b. Thus,
In ~nductionmachines where v+dr = 0,we can utilize this freedom to simplify
the model by letting b have a convenient value. For example:
I ) Let L,, = b2Lr or b = (LJLr)1/2.This selection yields equal self induc-
tances and results in a clrcuit model with equal leakage inductances.
L 4
2) Let bL, = b2Lr or b = L,,, 1 L, This choice produces equal mutual and 3) No Stator "Leakage"6 = A 4) Physical Circuii b = -
Lm Nr
rotor self inductances and results in a circuit model with all 'leakage'
located in the stator.
3) Let L , = bL, or b = L, l L,,,.S e l p i n g equal mutual and stator self
Inductance results in a model with all 'leakage' located in the rotor.
4) Let h = N, 1 N,. Choosing b as the stator to rotor tums ratio results in
the "real" leakage inductance model based on the designer's computa-
tion of leakage inductances. This is the most commonly used version of
rhc vanoils choices.
The operat~onalequivalent circuits resulting from these various selections
of h are illustrated in Figure 4.3. Circuit 4 is the most common selection and
the one which has been used in the basic model as given in Chapter 2. When
machine parameters are determined from laboratory tests, circuit 1 is often
used since there is no way to measure the actual tums ratio unless measure-
ments in the rotor circuit can be made. Circuit 2 is very convenient in systems
which control rotor flux, such as in field orientation and in switching transient
problems where a very widely used approximate method based on constant Figure 4.3 Modified d,q models as obtained from various selections of
rotor flux linkages is applicable. These situations will be examined in more the referral ratio b
detail in later sections. s;
E
156 4 Complex Vector Analysis of Induction Machines Steady State Equivalent Circuits 157

4.4 Steady State Equivalent Circuits Substituting these voltage expressions into the defining equation for v.;d,v .
For operation at constant rotor speed, the volt-ampere equations become linear eqn 2.9-2, yields
differential equations. In addition, the torque equation, which remains nonlin-
ear, does not have to be solved in combination with the volt-ampere equations
and can simply be evaluated as a last step after the currents are known.
For any periodic excitation voltage, an expansion in a Fourier series can be
carried out and an analysis on a tenn by term basis is valid because of the lin-
The bracketed term multiplying eJat' is the formal definition of. t11c posllivc
earity of the equations. The analysis for an arbitrary set 01' sinusoidal excitation
sequence voltage of steady state symmetrical component theory
voltages is therefore sufficient to handle any periodic input. The complex vec-
tor model will be employed in the analysis both as an illustration of the appli-
cation of this model and because the algebraic operations are easily carried out
in this form.
and if the second bracketed term is rewritten using thc fact that the con,juga~c
of a product is the product of the conjugates. i t .
4.4.1 Sinusoidal Excitation in G)n~plexVector I+wm
Consider a set of general three phase voltages of unequal amplitude and arbi- i t 1 -2t 2 = ( Z 1 Z 2 ) + (4.4-7)
trary phase (it is assumed that the zero sequence voltage is zero)
it becomes the conjugate of the negative sequence voltage of steady state sym-
metrical components

Vcs = Vcsc0s (met + 9,) The complex vector voltage can then I&
written as
To,facilitate evaluating the complex vector voltage associated with this
excitation, each of the voltages is written as the sum of two complex exponen-
tial~,for example vas can be written as If the excitation voltages form a balanced three phase set, v,, = 0 and the
vector voltage reduces to

ySqds =
-
V.spe
JV = v rn , j v (4.4- 10)
If the exponential term involving Vas is combined with the voltage ampli-
tude to form the conventional peak value phasor voltage where V, is the peak value of the input voltage and its phase angle (I, is arbi-
trarily taken as zero.
b- s -- vas ,j% The stator referred complex vector voltage can be transformed to an arbi-
trary rotating reference using
the equation for vas can be written as
yqds = y;ds e-j0

which yields
Similar expressions can be written for vbs and v,,.
158 4 Complex Vector Analysis of Induction Machines Steady State Equivalent Circuits 159

Teble 4.1 d,q voltages for sinusoidnl excitation

Table 4.2 presents the transformed voltage for the three most commonly
used reference systems. Note that in a stator reference frame both the positive
and negative sequence voltage are at the ftequency, we. The difference between
the two voltage sets is the phase sequence. In a rotor reference the positive
sequence voltage is at slip frequency, we - a, and the negative sequence at the
sum frequency, we + a, Again the phase sequence is reversed. In a synchro-
nous reference the positive sequence becomes dc and the negative sequence is
at twice frequency, 20,. Ref. Reference Frame Reference Frame Voltage Reference Frame Voltage
qpeSpeed & Angle Positive Sequence Negative Sequence
4.4.2 SfaforReferred Equivalent Circuits 0=ot+$ Peak = Vp Freq = 6.1, Peak = V,, Freq = a,
As a first method of developing the steady state equivalent circuits, the direct
approach of applying the excitation cons-ts of the previous section directly
to the complex vector equations will be employed. Using the basic stator Stator 0=0 - jo,r - + -jo,t

referred equations yields (with zero external rotor voltage)

4 =o yids = '5pe !ids = S
'n

vS45 = Vpc0s (met + $p) vSq5 = Vnc0s (wet + $,)

v ; ~ = -Vpsin (wet + $p) viS = Vnsin (wet + $,)

0 = L,,, ( p - imr) i i d s + [ r r + L~( P-jar)I 4;dr (4.4-14)

Rotor o = 0,
Since the equations are linear, the two components of the excitation can be
handled separately. For the positive sequence, the steady state solution will be $ 0=w,.r+$
of exponential form and will be symbolized by

iidSp= ?5pe'mc1, lqdrp - I rp ej%l (4.4-15) vrqJ = V p ~ ~ s [ ( ~ c - a , ) t + $ p - $ l vr45 = V , C O S [ ( ~ , + ~ ~ ) ~ + $ ~ + $ ]

Substituting these definitions and cancelling the common factor eJof' yields vis = -Vp sin [ (w,- o r ) t + gp - $1 viS = Vnsin [ ( a , + w,) t + gn + $1
2. Synch. w = we
fi.
0 = j (a, - o r ) LmiS,,+ [rr + j ( a e - wr) Lrl irp (4.4-17) e O=wet+$

which can be interpreted as a set of conventional phasor equations. Multiplying

the rotor equation by ja,/ j (we - o r ) results in
160 4 Complex Vector Analysis of Induction Machines Steady State Equivalent Circuits 161

which now has the same apparent frequency as the stator equation. These two
equations, (4.4-16) and (4.4-18), are the same as the equations of the conven-
tional equivalent circuit of the induction machine as illustrated in Figure 4.4
where the slip is defined as

+ d -I s p
iWls
-jw, I.,r

I ~ P
0 = -j (o,+ w,) L,Is,-t + [ r r - j ( o , + o r ) Lrl it,,,
Thc occurrence of the minus sign an each of ihc j q , tcrms is unusual in
(4.4-22)

terms of real variable machine theory. These minus signs occur becausc 01' thc
,-jo I
multiplier in the negative sequence portion of the complex vector volt-
age. To eliminate these minus signs, the entire set of equations can be replaced
-
by their conjugate set. Clearly there is no loss of information in this operation
and the negative sequence voltage V,, now occurs rather than its conjugate.
The result is

Figure 4.4 Positive sequence equivalent circuit

0 = j ( o ~ + o ~ ) ~ , ? , ~ + ~ r ~ + j ( ~ ~(4.4-24)
+ w ~ ) ~ ~ ~ 7 ~ ,
Multiplying the rotor equation by jye l j(o,+o,) yields the equations of the
conventional negative sequence circuit of Figure 4.5 where

Note that eqn (4.4-18) is not valid at we = 0 (dc excitation) since this would
mean we have multiplied by zero in obtaining the result. For dc excitation one
must return to eqn (4.4-17) and abandon the equivalent circuit model. The - The comment regarding zero frequency operation as for the positive sequence
rotor current is not zero for dc excitation as is easily seen from eqn (4.4-17). "
holds here also and again the interpretation for complex vector analysis d~ffers
The interpretation of eqns (4.4-16) and (4.4-17) as conventional phasor a from conventional phasor analysis.
equations is quite acceptable but is not the complex vector interpretation. In I
The steady state torque is obtained by direct substitution of the solutlon for
convptional phasor methods. to obtain the time function solution we multiply
the current expressed in complex vector form in the torque equatlon gwen In
by eJwc' and take the real part. In complex vector notation, both the real and
eqn (2.12-7).
imaginary parts have meaning; the real part being the q-axis solution and the
(negative of the) imaginary part the d-axis solution. One can, however, make
the association with ordinary phasor methods to have a familiar point of refer-
ence for the final result.
An exactly parallel development for the negative scqucnce can be carried
out. Defining the negative sequence complex vector components as
which yields, after expansion, two terms; the average torque
162 4 Complex Vector Analysis of Induction Machines Sleady Stale Equivalent Circuils 163

and a pulsation torque at 20),

These expressions can be converted to the conventional results in terms of

rotor currents only by using the relations between rotor and stator sequence
currents obtained from the equivalent circuits

Figure 4.5 Negative sequence equivalent circuit

Using these results yields

P = jwe (4.4-34)
In the operational circuit of Figure 4.2, the operatorp +jw of each inductor
therefore beconics
and
p + j w + jw,
:, and the operator of the rotor resistance becomes

which are the standard forms resulting from conventional symmetrical compo-
nent theory (except that rms values are normally used and one of the factors of
112 therefore does not occur in the leading factor). which immediately converts the general operational circuit to the positive
As an alternative approach to obtaining the equivalent circuits in Figure 4.4 sequence circuit of Figure 4.4.
and Figure 4.5, the operational circuit of Figure 4.2 can be employed. With By similar arguments, for the negative sequence
vSq,, known to be an exponential function of frequency we, the steady state
model is obtained by recognizing that the p-operator hecomes the complex
operator 3z jw, depending on the sign of the exponential frequency. Thus, for a
stator referred system, C and
f
p + jo + -jo,
w = o
and for the positive sequence
1 64 4 Complex Vector Analysis of Induction Machines Modified Equivalent Circuits 165

which (after conjugating) yields the negative sequence circuit of Figure 4.5. quency dependent inductors are always evaluated at o, and the resistance mul-
tiplier is always we/ (o, + a,) or a,/ (a, - or). The interpretation of the
4.5 Invariance of the Steady State Equivalent Circuit circuits as conventional phasor circuits is, however, not possible except in the
stator frame. For example, in a rotor frame, the excitation frequency is o, - or
The steady state equivalent circuits shown in Figure 4.4 and Figure 4.5 were
I (positive sequence) and yet the reactances in the circuit are evaluated at a fre-
derived in a stator reference system. They are, however, exactly the same in
; quency o,. The synchronous frame is especially interesting since here thc cxci-
any reference system. This is easily shown since the conlplex vector voltage in
tation frequency is zero or dc (positive sequencc) and yct we hnvc ;I co~~iplcx
en arbitrary rcfcrcnce is. eqn (4.4-12),
parameter equivalent circuit. The complex vector intcrprctation is straightliw-

'q/s
- p
- sp
ej(mel-W + v;,, e - i ( ~ t f + o )
(4.5-1) "j
ward. The real parts of the complex quantities arc thc cp-axis dc valocs and (hc
parts are the d-axis values. Physically, we can think of moving i n synchro-
Thus, for a frame rotating at a,the frequency of the positive sequence excita- i nism with the traveling waves in the machine so all quantities are stationary.
tion is o, - o and hence in the steady state Figure 4.6 illustrates a d,q axis diagram in the synchronous frame wherc the
!f time zero reference of the rotation transformation is sclected such that the air
P = j(o,-m) (4.5-2) i gap flux is in the d-axis. Note that while the diagram is labeled as a synchro-
The operators in the operational equivalent circuit become nous frame diagram, it is really indistinguishable from n stator frame diagram
where the complex vectors are interpreted as conventional phasors. Note also
p + j0+ j ( ~ , - 0 ) + j 0 = jo, : that if the time reference is chosen such that the positive sequence voltage has
and a zero phase angle

1
Gsp = crplejO (4.54)
then the reference for the rotating transformation would have to be
which are the same as previously obtained for a stator reference. Similarly for
thc ncgative sequence the frequency is -((I),, + 0 ) ) mtl
where cp is thc phase angle of V,, in Figure 4.6.

and 4.6 Modified Equivalent Circuits

The modified d-q models introduced in Section 4.3 can be used to obtain mod-
6 ified steady state equivalent circuits [S]. The simplest way to obtain these cir-
8
cuits is by converting the operational circuits shown in Figure 4.3 to steady
p+jO me
(4.5-7) "tate circuits by the method used in Section 4.4.2. With this approach, the mod-
p + j ( ~ - ~ , )we+q
ified circuits would be interpreted as resulting from turns ratio transformations
which again are the same as obtained for a stator reference. of the basic coupled circuit model.
Physically, the reason the circuits are all identical is, of course, that they all There is, however, another way to obtain these modified circuits which
represent the same machine. In terms of transformation theory, the change in offers a different and, in many ways,'a mdre useful interpretation of their ori-
the excitation frequency in different frames is exactly counterbalanced by the gin. This alternative approach is in terms of the spatial orientation of the rotat-
speed voltage arising from the motion of the reference frame so that the fre- ing frame d-q reference system and its relation to the flux waves in the actual
166 4 Complex Vector Analysis of Induction Machines Modified Equivalent Circuits 167

Rotor
Current

SmorJR Drop Stator Leakage Volroge

Stator Current

Figure 4.7 Complex vector diagram for positive sequence

synchronous reference frame - air gap flux reference
Figure 4.6 Complex vector diagram for positive sequence synchronous
reference frame - air gap reference
5, = (Ls -4,,)ls-iWrn
machine. The concept is most easily interpreted in a synchronous frame; the Similarly, the rotor flux vector is
resulting circuits are of course valid in any frame.
Consider the positive sequence synchronous frame complex vector dia- h, = ( Lr - Lrn)I , - jLrnIrn
gram shown in Figure 4.7 where the reference system is chosen to coincide The voltage equations in the steady state are
with the air gap flux. This diagram is simily to Figure 4.6 except that the rotor
and stator fluxes are also shown. From the figure it is clear that for this refer- II, = 5 !,+ joehS
ence system (the subscript p to indicate positive sequence is omitted)

I = -I (since hmq= 0) (4.6-1) S1r + j wr &r

0 = 1 (4.6-8)
SY rq
Using eqns (4.6-5) and (4.643, the voltage equations become
Isd = Im
+ Ird (4.6-2)
Thc components of the stator flux linkage vector are -V, = [ r , + ioe(L, - L,) 1 1, + ioeLrn(-jIrn) (4.6-9)

hsd = LsIsd + LrnIrd = (Ls,- Lm) + Lm (Isd + Ird) (4.w) which are clearly the equations of the positive sequence circuit of Figure 4.4.
I

from which the stator flux linkage vector can be written as

16R 4 Complex Vector Analysis of Induction Machines Modified Esuivalent Circuits 1 69

If we now consider the d,q axes reference shifted such that the d-axis coin- and the voltage equations become
cides with the rotor flux as illustrated in Figure 4.8, the new conditions for the
currents are 1 I L:
Yr = (rS+ jw& ) Is + j w - (-jl,,,) (4.6-1 7 )
Lr

= O
L~nlqs + ' r l q r (since hqr.= 0) (4.6-12)
The stator flux components become Multiplying the rotor voltage equation by L , 1 L , and using eqn (4.6-1 2) to
eliminate Iqr yields
!r

Equations (4.6-17) and (4.6-19) deskribe the equivalent circuit shown in Fig-
ure 4.9 which is the same as would be obtained from Figurc 4.3. (2) i~singa
'turns ratio' of L, 1Lr

d-axis
Figure4.8 Complex vector diagram for positive sequence
synchronous reference frame - rotor flux linkage
reference

- -

Figure 4.9 Complex vector equivalent circuit - rotor flux reference

This circuit, which places all of the 'leakage' on the stator side, is often
which results in the stator flux vector
called a rotor flux based circuit since it contains a magnetizing branch which
depicts the total magnetizing current producing the rotor flux. Although the cir-
cuit can be viewed as resulting from a particular choice of 'turns ratio', the der-
ivation from a rotor flux oriented, synchronous frame dc model is much more
The rotor flux vector is simply useful. This circuit has an important role in developing the concept of field ori-
ented control and will be employed in later chapters for this purpose.
170 4 Complex Vector Analysis of Induction Machines Electrical Transients at Constant Speed 171

If a stator flux oriented synchronous frame is employed then, as one might 4.7.1 Complex Vector Operational Impedances
expect, a circuit with all of the 'leakage' referred to the rotor is obtained. The The complex vector electrical equations in an arbitrary rotating reference are
circuit is the same as would be obtained from the operational circuit of Figure (eqns (4.3-1) and (4.3-2) 'with b = 1 )
4.3 (3) and is illustrated using d,q complex vector notation in Figure 4.10. The
derivation is left as an exercise for the reader.

0 = L , , , ( p + j ( o - ~ , ) ) i q d S +[ r r + L r ( p + j ( o - a r ) ) 1 i q d r (4.7-2)
To shorten the notation so we can focus on the basic nature of the system, the
complex operational impedances defined below are introduced:

Zss = rs + Ls ( P + j o ) Zsr = L,,, ( P + j a )

(4.7-3)
Z,, = L,,,(p+j(w-o,)) Zrr = r r + L r ( p + j ( a - a , ) )
The electrical equations can then be written as

Figure 4.10 Complex vector equivalent circuit - stator flux reference

Note that the complex operational impedances always contain the deriva-
tive operator p but that the frequencies present depend upon the selected refer-
ence system. Thus, in a stator reference ( o = 0) the impedances are
4.7 Electrical Transients at Constant Speed
There are many important practical situations where the purely electrical
behavior of a machine is of concern and over the time period of interest the
speed of the machine can be considered to be constant. This situation can arise in which the stator impedances are real and the only frequency present is the
because the electrical transients are simply very fast compared to the slow rotor speed. In a rotor reference ( a = or), the situation is reversed
mechanical transients which occur when the system mechanical inertia is large,
or it may be that we are only concerned.with very short time periods during
which the speed can be assumed constant. In any case, the constant speed con-
straint yields linear electrical differential equations with constant coefficients
for which general solutions are known. The electrical torque, although a non- but again only the rotor speed is present. Finally, in a synchronous reference
linear function of the electrical variables, can be evaluated after the solution for where o = oeall impedances are complex
the electrical quantities is completed. The following sections are restricted to
thc case o f uniform air gap machines where the external rotor voltage is zero
(cage i~~duction machines).
and the stator excitation frequency occurs in the stator impedances and the
rotor electrical frequency (slip frequency) occurs in the rotor impedances
172 4 Complex Vector Analys~sof Induction Machines Electrical Trans~entsat Constant Speed 173

4.7.2 Example - Sudden Application of Stator Current Source Excitation The eigenvalue is formally obtained from the characteristic equation
As an example of a constant speed switching transient, consider the sudden obtained by substituting ceb' into the differential equation with the source set
application of controlled positive sequence stator currents to an induction to zero (the reduced equation) which yields
machine turning at a speed y . The excitation will be taken as
rrCeh' + L, (5 + j ( a - or)) ~ e b '= 0 (4.7-15)
or

Since the only unknown current is iqdr only the rotor equation
Note that this forrnalrsrn is equivalent to setting p = in the tliffcl.cnlii~lcqua-
() = Zrsiqds+ Zrriqdr (4.7-9) tion with the source rcrnoved. Equation (4.7-16) yiclds the eigenvaluc :IS
must be solved. To illustrate the differences between solutions in the various
reference frames, the analysis will be camed out in a general frame moving at
a speed w In this general frame, the stator excitation becomes (all the follow-
ing analysis is for t 2 0 ) so that the complete solution can now be written as

.
lqds
-
-
isp e i ( o t - o ) f iqdr= Ce-f/Tre-i
-
( a- a,)1 j ,i
+
TI'
(Oc - O )1
(4.7-1 8)
and the rotor differential equation becomes To evaluate the constant C, the initial value of iqdr must be known. This
initial condition can be found from the knowledge that the rotor flux linkage
must be continuous and hence

The complete solution to this equation is

bdr= C& + i'P e ~ ( * t - * f (4.7-13)

Combining this result with eqn (4.7-18) evaluated at t = 0 yields
wherein C is a complex constant depending on initial conditions, & is the
complex eigenvalue of the (first order complex) system and jrp is the steady
state phasor solution for the rotor current. The steady state solution can be
obtained from the equivalent circuits of the previous section or directly from
eqn (4.7-1 1) by setting p =j(oe - a )

This expression can be simplified by observing that the final steady state value
of the rotor flux linkage is
174 4 Complex Vector Analysis of Induction Machines Electrical Transients at Constant Speed 175

equation given in eqn (4.7-16) yields the eigenfrequency as zero in a rotor ref-
erence as it should.
and defining Viewed in a stator frame (61= 0) the excitation frequency is a, and the
eigenfrequency is the rotor frequency a,.The stator voltage will contain a
voltage component at this frequency during the transient period. In a synchro-
allows writing the final solution as nous frame, the excitation frequency becomes dc and the eigenfrequency is slip
frequency, a, - or. In all reference frames the amplitlldes of the components
(coefficients of the complex exponential time functions) are the same. Graphi-
cally, the lengths of the rotating vectors are preserved, only the frequency or
speed of rotation changes as the reference frame is changed.
The torque can be evaluated from the basic torque equation The solution for the current is of interest also from the point of view that it
is possible to execute the switching operation such that the transient compo-
nent is zero. The necessary condition is that the initial rotor flux linkage be the
same as the final value (both magnitude and phase). This is, in fact, the scheme
and is easily shown to consist of the expected steady state component, see eqn employed in a field oriented controller as will be shown in later chapters. Note
(4.4-3 1) that with the transient component zero, the torque responds instantaneously to
the stator excitation current. This is the major goal of field oriented control - to
obtain a torque response which follows a torque command instantaneously.
This simple example actually illustrates a basic requirement for this type of
and a transient component given by response: that the rotor flux linkage must remain constant in amplitude and
phase.
As a final step in the solution, the d and q real variable expressions for the
rotor current will be obtained. From the general form given in eqn (4.7-24), the
which varies at slip frequency and decays with the rotor open circuit time con- real variable expressions are:
stant T,. Note that the reference frame speed does not occur in the torque
expression; the torque is not a transformed quantity and must clearly be the
same in any reference frame.
Returning to the expression for the rotor Current, it is clear that the frequen-
cies present in the current do depend on the reference frame. There are two fre-
quencies in eqn (4.7-24); one from the applied excitation, a, - CB,and one
from the eigenvalue, o - or. Both depend on the choice of reference frame.
The "natural" or real frame for the rotor current is clearly a rotor frame (i.e. 61
= a,).Viewed from the rotor, the excitation frequency is slip frequency, 61, -
w, and the elgenvalue frequency is zero. This implies that the transient compo-
nent of the rotor current is simply a decaying dc component which is 'trapped'
at the instant t = 0 when the excitation is applied. Note that the characteristic
176 4 Comolex Vector Analvsis o f Induction Machines The Constant Speed Eigenvalues 177

where Qro,QrJ and QrlL are the phase angles of the complex quantities plex conjugate pair produces the damped sine and cosine terms which occur in
hqdr( 0 ) ' hqdr(00) and IrPrespectively. the real variable results. A mathematical interpretation of this simplification is
In the special case where Sqdr( 0 ) is zero, hqdr( 0 0 ) can be expressed in that the complex vector model is only one half of the total model and the com-
terms of machine parameters by using eqn (4.7-23) and (4.7-14), and the com- plete complex vector model also involves the conjugate complex vector
plex current for this case can be written as (f + j f d ) . In this case there are four complex equations just as there are four
real equations. Because of the machine symmetry the two sets of two equations
are conjugates of each other and only one set need be retained; the other set can
be obtained by conjugation.

(4.7-30) 4.7.3 General Solution for Switching Transients

which involves only the excitation current, rotor speed and machine parame- ; In general, the solution for switching transients involves the simultaneous solu-
ters. If the rotor impedance is defined as tion of the stator and rotor differential equations. The example in the prcvious
section only required the rotor equation because the stator current was a known
zrejer = rr + j (ae- u r )L~ function. The basic approach to the solution is the same, however, and consists
this expression becomes of finding the steady state and transient parts of the solution and evaluating the
unknown coefficients from initial conditions. The steady state solution is again
available from the steady state equivalent circuits or by direct solution of the
appropriate steady state form of the machine equations.
The system characteristic equation is obtained from the machine equations
(4.7-32) with the source removed
from which the d and q components can be written as

I ~ S P I L ~rr -,/T,
i '7' = - 7q--[Zg cos ( ( 0 - u r )t + Q r + Q P ) (4.7-33)
. ,
again by substituting for iqdrand iqdran arbitrary complex exponential solu-
- ( 0 , - o r )sin ( ( 0 ,- 0 )t - @,. + Qp)
I tion. The result of this formalism is the characteristic equation

in which, as we have shown, the operatorp is replaced by the unknown eigen-

values symbolized by S . Once the eigenvalues are known, the solution pro-
ceeds as in the example in Section 4.7.2.

(4.7-34)
4.8 The Constant Speed Eigenvalues
where Q p is the angle of I,,, .
The real variable results illustrate that in real variable analysis the single The evaluation of the roots of eqn (4.7-37) can be carried out in a general ref-
complex eigenvalue which occurs in complex vector analysis is augmented by erence frame rotating at a speed 61 to obtain the general reference frame values
its conjugate value to yield a complex conjugate pair of eigenvalues. This com- of the system eigenvalues [3,4]. Carrying out the multiplication indicated in
 178 4 Complex Vector Analysis of Induction Machines The Constant Speed Eigenvalues 179

eqn (4.7-37), with p = h , the following quadratic equation with complex

coefficients is obtained

The other two eigenvalues of the complete real variable system are the conju-
gates of the two given by eqn (4.8-3) or (4.8-4).
As can be expected on the basis of transformation theory, the speed of the
(4.8-1) reference system enters as a simple linear shift in frequency. Thus, if the eigen-
where the couplingfactor a is values are known in a stationary reference, they can be transformed to any
other reference system by simply subtracting o from the frequency of the
eigenvalues (i.e., the same as for excitation frequency). A plot of the migration
of the eigenvalues in the stationary reference frame (w = 0) as a function of
Solving for the two complex roots of this quadratic yields, after some algebraic rotor speed is shown in Figure 4.11. The parameters of the machine used for
manipulation, the roots can be written in either of the following two equivalent the study are for a 5 horsepower four pole machine and are given as r, =
forms, 0.442, rr = 0.708 Q, L, = L, = 0.069 mH, and L,= 0.0668 mH.
It is interesting to observe that in a stationary reference the eigenvalues are
always complex except for zero speed where both ari'ieal. The induction
machine, therefore, always has complex roots (damped sinusoidal oscillations
in the transient response) except at zero speed. There is, in fact. one condition
for a machine with a = 1 for which the eigenvalues occur as a double complex
pair at the rotor speed which makes the square root terms in eqn (4.8-3) equal
to zero. In a reference system rotating at one half this speed the machine is also
characterized by real roots (four equal real roots in fact) but this is a single spe-
cial exception to the general rule of complex conjugate roots. Since all of the
motor parameters are involved in determination of both of the pairs of eigen-
values, it can be recalled that in the stationary reference frame, a speed voltage
proportional to rotor speed appears in the rotor circuit, see Figure 2.8 or Figure
4.1. Hence, one expects that the pole which migrates as a function of rotor
speed corresponds to the effects of the rotor circuit. Consequently, these poles
are termed the rotor poles. On the other hand, in the stationary frame, speed
T~ =
L
-r (rotor open circuit time constant) voltages do not appear in the stator circuit so that the poles remain near the real
'r
.axis and form what are termed the stator poles.
,, When the eigenvalues are calculated for the synchrpous reference frame
L
T =--S (stator open circuit time constant) the plot of Figure 4.12 results. In this case we have set = 377 since this cor-
r~
ponds to the usual sixty hertz operation. Note that since the reference frame
'ttr = b'rr (rotor short circuit time constant) appears only as a single linear term in eqn (4.8-3), the roots have simply
dergone a spectral shift. Upon analyzing our machine in the synchronous
= GTs (stator short circuit time constant) ame we would now expect to see damped sinusoidal oscillations of the cur-
180 4 Complex Vector Analysis of Induction Machines The Constant Speed Eigenvalues 18 1

Figure 4.11 Eigenvalue vs. speed loci in the stationary reference frame Figure4.12 Eigenvalue vs. speed loci in the synchronously rotating
reference frame

, quency of the rotor poles remain roughly constant. Examination of Figure 4.1 1
rents at sixty hertz when the rotor speed is zero. This tendency can be observed
and Figure 4.13 are both needed to reveal the physical nature of the stator and
in the transient solution of Figure 2.16. However, it is important to caution that
, rotor poles. In general, the effects of stator and rotor poles both show up in the
the eigenvalue approach assumes constant speed operation whereas Figure
stator current transients. However, a complete eigenvector analysis would
2.16 clearly pertains to variable speed. Nonetheless, these observations remain
1 reveal that the stator poles of Figure 4.11 show the predominant nature of the
roughly true when the acceleration of the rotor is not too rapid.
, physical stator currents while effect of the rotor poles on the stator current is
In Figure 4.13 the eigenvalues are plotted in the rotor reference frame. In
relatively weak. Figure 4.11 also shows the nature of the equivalent rotor cur-
this case the stator poles undergo a wide variation in frequency while the fre-
. rents which depend primarily on the rotor poles. However, it is important to
182 4 Complex Vector Analysis of Induction Machines The Constant Sped Eigenvalues 183

note that these rotor currents are not the physical currents which are, in fact, Two important properties of eqn (4.8-3) can be identified by examining its
rotating with the rotor. On the other hand, Figure 4.13 represents the pole posi- form [6]. Specifically if the two eigenvectors and are added, one obtains:
tions in the rotor reference frame and therefore shows the predominant nature
of the physical rotor currents which depend heavily on the location of the rotor
poles and weakly on the location of the equivalent stator poles. In each case the
pole dominating the physical circuit is located near the real axis and has only a Therefore the sum of the two eigenvalues add to a constant value which is
small frequency component which is fundamentally the result of the rotor rota- independent of rotor speed (or reference frame). Since the stator and rotor tran-
tion (or, equivalently stator rotation with nspect to the rotor). sient time constants can be readily determined from machine parameters, this
fact can be used either as a check on computations or as a means of calculating
the second eigenvalue, given the first. A sketch of the real parts of the eigenval-
ues as a fun
0

55 ,-
!
-3
I",

a
- -100
+
X
E
-
8
-
,%
X
z -200

-300
0 360 720 1090 1440 1800
Speed (rpm)
Figure 4.14 Real parts of &and and their sum vs. rotor speed
Figure 4.13 Eigenvalue vs. speed loci in the rotor reference frame
For the imaginary part one obtains:

Im [ h , + h2] = ur- 2 0
184 4 Complex Vector Analysis of Induction Machines The Constant Speed Eigenvalues 185

Given the imaginary part of one of the eigenvalues, the imaginary part of the
second can also be obtained. Note that when the reference frame speed is set to
zero or to rotor speed the result is +or and -or,respectively, indicating the
Since most conventional machjnes have a close to one and the inequality
symmetrical nature of the reference frame on the location of the stator and
is satisfied at all but very low s e d s , this is often a useful approximation
rotor poles. The result is essentially the same since the sum of the conjugates
for real machines. Note that a is a fundamental parameter in the sense
would simply reverse the signs of the answer. A plot of the imaginary parts of
that it is unaffected by the turns ratio transformations discussed in Sec-
the eigenvalues and their sum as speed changes is shown in Figure 4.15.
tion 4.3. An a close to one implies nearly equal rotor and stator I ~ R
losses.
2) rs = O

This limiting case implies no stator damping and as a result one of the
eigenvalues (i.e. the stator pole) has a zero real part. It can often be used
as an approximation for cases where only a short period of time is of
interest. It also has value as a limiting case in stability analysis.

3) r, = 0 , r, = 0

5, = -j0, 5, = j(0,-0) (4.8-14)

This is another limiting case where all damping is neglected. It is often

used as an approximation to the transient behavior for a very short period
0 360 720 1080 Id40 1800
following a disturbance. This approximation is directly related to the
constant rotor flux linkage transient model discussed in the next section.
Rotor Speed in rpm

Figure 4.15 Imaginary parts of bland & and their sum vs. rotor speed in
the rotor reference frame

There are several special or limiting cases for the eigenvalues which are
often useful as approximations in theoretical work or as illustrations of typical The limiting case of a = 0 imblies perfect rotor to stator coupling or
behavior. These include: LJr = L; and results in one pigenvalue at the origin. It is essentially
never a good approximation to real machine but does serve as a useful
theoretical result in stability analysis.
5) 0 < <1 1 +~BW,QZ,
186 4 Comdex Vector Analvsis of Induction Machines The Constant Speed Eigenvalues 187

the values given are only to indicate order of magnitude. The trends with size
are, however, well established and have a sound theoretical basis.
Table 4.4 Qpical Values of Machine Parameters and Eigenvalues

Eigenvalues
at O r 3 7 7
(4.8-17) (rad./sec.)
This is sometimes used as a low speed approximation for the eigenvalues
in real machines. Since o<< 1 is essentially always satisfied except in
very small machines, this approximation is nearly always applicable
below some particular speed.
Table 4.3 indicates typical ranges for the machine parameters involved in
determining the eigenvalues. The largest range is for the open circuit time con-
Table 4.3 Ranges of Machine Parameters fok Normal (NEMA -B)4-Pole Machines

I Parameter 1 Range I Comments I

0.2 - 0.04 Smaller in large or high
4.8.1 Example - Sudden Short Circuit of Unloaded Induction Muchine
0.5 - 2.0 Generally of the order of . As an illustration of a transient solution using the full set of eigenvalues, con-
1.5 with no significant size sider a sudden short circuit of an induction machine initially operating in the
influence

I OTr='=;
I 4 - 200 ms
1 Larger in large or high effi-
fiemy w h i n e s
$
9,
steady state at no load. Assuming the system inertia maintains the speed con-
stant, the transient stator current following the short circuit will be

I --
. r. r..
0.03 - 5.0 s Larger in large or high effi-
ciency machines where &, and h2 are the eigenvalues corresponding to the rotor speed. The
constants CI and C2 are found from the initial current and initial current
derivative
stant T, which varies over a 100:1 range from small to large machines. The
least variable parameter is the time constant ratio cr which perhaps varies over
a 3: 1 range with no relation to machine size. Table 4.4 also gives several sets of
typical values of machine parameters for different size machines. The corre-
sponding eigenvalues at a,= 377 radlsec ark also given in Table 4.4. Consider-
able variations from these typical values are possible for any specific machine;
188 4 Complex Vector Analysis of Induction Machines Thc Constant Speed Eigenvalues 189

Prior to the short circuit, the motor currents will be

iids
-- i sqds
'5
l q d r= 0
so the initial values are which completes the solution.
If we assume the speed is high enough so the approximation for the eigen-
values given in eqn (4.8-12) applies,

iidr(0) =0 (4.8-26)
To maintain constant flux linkage at t = 0 requires that these currents be
and the expressions for the constantslC, and C2 reduce to
continuous at the switching instant. The derivative of the stator current at t = 0,
pGds(0) can be found from the machine equations at t = 0 1
C =
-I
--1-0 s
a lqds9 c'2 = ; l i d s (4.8-35)

The short circuit current is then given by

which consists of a decaying dc component of magnitude

I-qds
dc component magnitude = -
"
a
I I (4.8-37)

and a rotor frequency component with a magnitude

ac component m a g n i h e = 9
1 ~,,:,c 1 (4.8-38)

Since a can be as small as 0.03 in a large machine, these short circuit cur-
rents can be as large as 30 times the no load steady state current. The no load
current is given approximately by
Substituting these results, eqns (4.8-25) and (4.8-3 I), into the expressions , I

;
for the constants CI and C2 yields

so the short circuit current components are approximately

190 4 Comolex Vector Analvsis of Induction Machines Transient Equivalent Circuils (Constant Rotor Flux Linkage) 191

Phidr = 0
from which the solution for the rotor flux is
each of which is also approximately equal to the steady state locked rotor cur-
rent. At the instant when the ac and dc components add in one of the machine krqdr = hrqdr ( 0 ) = constant = hidro (4.9-5)
phases, the peak current is roughly twice the locked rotor current. The actual Transforming to a general frame moving at speed o yields
peak value is
1' e-.i(o- a,)
&qdr = -qdro (4.9-6)
peak short circuit current = 2-01"
-x dsl
=
aeLs
which simply indicates that the trapped constant rotor flux generates sinusoi-
dally varying rotor flux linkages in any other frame of reference.
Both components of the short circuit current decay with the transient time con-
stant 09,. 4.9.2 Transient Equivalent Circuit
To utilize the constant rotor flux linkage expression in eqn (4.9-6) in the stator
4.9 Transient Equivalent Circuits (Constant Rotor Flux equation and obtain a circuit model, we employ the flux linkage equations
Linkage)
hqds = Lsiqds + Lrniqdr (4.9-7)
There are many instances in which the behavior of a machine for a very short
period following an electrical disturbance is of interest. Examples include eval-
hqdr = Lmiqds+ Lrfqdr (4.9-8)
uating worst case fault currents or peak transient torque and modelling the
sub-cycle behavior of PWM inverters or other fast switching controllers Eliminating the rotor current in these equations yields
applied to machines. In such cases it is possible to ignore the damping of the
transient solutions and greatly reduce the complexity of the analysis. This is
equivalent to neglecting the resistances of the machine and results in the sim-
ple eigenvalue expressions given in eqn (4.8-14). but

4.9.1 Constant Rotor Flux Linkage Model

A very useful conceptual model can be obtained for this situation by starting
from the flux linkage form of the machine equations where L: is the stator transient inductance which was originally introduced in
Section' 4.6 in connection with turns ratio considerations. Substituting these
.results in the stator equation, eqn (4.9-l), yields
O = r f q d r + [ ~ + j ( ~ - ~ ~ ) l h ~ ~(4.9-2)
~
With the rotor IR drop neglected, eqn (4.9-2) becomes uqds = 15 + L: ( P +ja) I b+( p + j a l ?hqdr (4.9-11)

[ P + j ( a - a r ) 1hqdr = 0 and finally using eqn (4.9-6)

or in a rotor reference
192 4 Complex Vcclor Analysis of Induction Machines Transient Equivalent Circuits (Constant Rotor Flux Linkage) 193

If we define the voltage behind transient reactance, Ebd to be

Figure 4.17 Stator referred equivalent circuit

Equation (4.9- 12) becomes

which describes the transient equivalent circuit shown in Figure 4.16. The
most commonly used circyit is the stator referred circuit shown in Figure 4.17,
obtained by setting KI = 0 in the general circuit. From this circuit it is clear that
in a stator reference, the voltage behind transient reactance Eid is a balanced
sinusoidal voltage oscillating at rotor frequency with respect to the stator d,q
frame. Figure 4.18 illustrates the evolution of the transient equivalent circuit
from the complex vector equivalent circuit of Figure 4.1.

Figure 4.18 Evolution of transient circuit from equivalent circuit. The

Figure 4.16 Transient equivalent circuit in reference frame rotating at symbol 'N' denotes the parallel combination of the two inductances.
speed w

Er = jWe(Lmf;ds+LrfSqdr) = j~,h;~~ (4.9-15)

4.9.3 Evaluation of Voltage Behind Transient Reactcmce
The flux linkage &;dr will become the trapped flux ktdroand hence
The voltage Eid is proportional to the rotor flux linkage which exists at the
instant the transient is initiated. Assuming the machine is in the steady state
prior to the initiation of the transient, the steady state rotor flux just prior to the
transient can be used to find E'9d In the steady state stator referred circuit of
'
Figure 4.19, the voltage across the rotor resistor is given by or in terms of the slip S
194 4 Complex Vector Analysis of Induction Machines Transient Equivalent Circuits (Constant Rotor Flux Linkage) 195

qd= Lpm- s ) E , (4.9-17) Er0 = = ~ d s

l ~ e ~ m cds4 =m
(4.9-1 8)

where Era is the value of Er at the switching instant. and from eqn (4.9-17). the voltage Eqd' is

Li
E;, =
S r
VSqds = ( 1- 0 ) !!ids (4.9- 19)

The transient equivalent circuit is as shown in Figure 4.21. Subsequent to

L:, /

Figure 4.19 Evaluation of Efqd from steady state circuit prior to

initiation of transient Figure 4.21 Transient equivalent circuit for example short circuit problem

4.9.4 Exctnzple - Sudden Short Circuit of Unloaded Induction Machine

the short circuit, the differential equation describing the short circuit current is
A similar example as was treated in Section 4.8.1 by eigenvalue methods will
now be examined using the transient equivalent circuit.
0 = L: pisqd, + ( 1 - 0 ) y i d s e (4.9-20)
The first step is to evaluate the transient voltage Ebd for the condition just
prior to the short circuit. Since the machine is operating at no load the slip is from which
zero and the steady state circuit takes the form shown in Figure 4.20The volt-
age Er is clearly (neglecting rs)

To evaluate the integration constant, the initial value of the stator current

Figure 4.20 Steady state equivalent circuit at S = 0 for example problem is used with the result
196 4 Complex Vector Analysis of Induction Machines References I97 -.
-- . ---

121 K.P. Kovi~cs,"Synmctrical ComponcnQi in Polyphase Mi~chines".(in G C ~ I I I MI.)I.IIc!~

I).
von Rosch and Co.. Bern. 1962.
[3] D.W. Novorny, and J. H. Wouterse, "Induction Machine Transfer Functions and Dy-
namic Response by Means of Complex Time Variables," IEEE PAS, Vol. 95. No. 4
JulyIAug. 1976, pp. 1 125-1333.
[4] R. Stern and D.W. Novotny, "A Simplified Approach to Induction Machine Dynamic
Response." lEEE PAS, Vol. 97, No. 4, JulyIAug. 1978, pp. 1430-1439.
[S] R. DeDoncker and D.W. Novotny, "The Universal Field Orienred Conrrollcr." I H X
and the total solution is IAS Trans.. Vo1. 30. No. I . JanuarylFebruary 1994, pp. 92-100.
161 l3.T. Ooi and T.H. B;irton. "The ~ o d aApproach
l to Intluclion Motor Dyniimic\. I'XI 1
nntl 11" Papers C72-045-8, C 72 046-6, IEEE Winter Mccliny. New York. Jim {O I:cII.
4. 1972.

which is the same as eqn (4.8-36) except for the exponential decay term.

4.9.5 Comments on the Transient Equivalent Circuit

Although in the development of the transient equivalent circuit model the sta-
tor resistance can be retained as was done in Section 4.9.2, the time constant
associated with this resistance is not a correct description of the time behavior
of the system. Since the rotor resistance was neglected in the development of
the circuit, the time dependence of the rotor flux is neglected. As has been
shown in the section on constant speed eigenvalues, there is a strong interac-
tion between the rotor and stator damping and to ignore one and retain the
other is not generally valid. If one is very long compared to the other (i.e. if a
is very large or very small) there is some validity in neglecting the smaller time
constant. Since, in general, the two time constants T, and 7, are comparable,
there is no valid reason to retain only one or the other. Thus, the transient
equivalent circuit should be viewed as a means of evaluating the undamped
behavior of the machine and the stator resistance should be neglected. It is left
in the model as developed in Section 4.9.3 only as an indication that at the ini-
tiation of a transient, the difference between the terminal voltage and the tran-
sient voltage does appear across the stator resistance and the transient
inductance.

References
[I] W.V. Lyon. "Transient Analysis of Alternating Current Machinery", John Wiley and
'
Sons. Inc.. Ncw York. 1954.
198 4 Complex Vector Analysis of Induction Machines Problems 199

Problems
Problem 4-1 DC Excitation of Induction Machlne
Consider a three phase induction motor in which stator phases b and c are con-
nected to a dc current source of amplitude I d c . Phase a is left disconnected.
Use a stationary reference system with $ = 0 and find:
a) the stator d and q currents;
b) the stator d and q steady state voltages if the rotor speed is constant at
wro(radsec);
c) the rotor currents for the condition of part (b);
d) the torque as a function of the dc current, the rotor speed and the
machine parameters.

Problem 4-2 DC Excitation of Induction Machine

Repeat problem 4-1 using a rotor reference.

Problem 4-3 AC Current Source Excitation of Induction Machine Problem 4-5 T o q u e Pulsation Caused by Unbalanced Excitation
Repeat problem 4-1 if the current source is an ac supply of amplitude I , and For a 100 hp, 460 volt, 60 Hz, 4 pole, three phase induction machine with
frequency w e . Discuss the advantage (or disadvantage) of a synchronous refer- equivalent circuit parameters (in per unit):
ence for this situation.

Problem 4-4 Single Phase Excitation of Induction Machine

For a three phase induction machine operated with phase a open and a voltage
vbc suppled to terminals b-c: operated at rated voltage and frequency and a slip of S = 0.0175 under the
single phase condition of problem 4-4, find (assume the inertia is large enough
a) derive a d, q, 0 stator referred differential equation model which
to maintain constant rotor speed):
explicitly includes vbc and ib variables. Give both the voltage-current
equations and the torque equation; ! a) the phasor current ib (take cbc as the reference quantity, i.e. zero
phase angle);
b) for an input voltage vbc = V C O S Oand ~ ~ a constant rotor speed ore,
derive the steady state stator referred phasor equations and the torque b) the positive and negative sequence average torque and the total aver-
equation; age torque;
C)show that these equations can be represented by the (symmetrical c) the peak value of the second harmonic pulsating torque.
component) equivalent circuit shown below and express and& in
terms of the phasors representing vbc and ib ;
200 4 Complex Vector Analyv\ o f Induct~onMachines

Problem 4-6 Induction Machine Constant Speed Switching Transient the slip equal to rated slip (SR = 0.0247). Locate all of the fundanientiil volt-
age in the q-axis.
The 100 hp, 460 volt, 60 Hz, 4 pole, three phase induction machine of problem 2
6
4-5 is operatmg at full load (S = 0.0 175) at rated V and f . A Y -connected set d
of 10.0 pu resistors is in parallel with the stator. At t = 0 the source supplying Q
4!
the motor and resistors is disconnected.
a) Find the values of rS and rr (in ohms) as well as L , L,, andL, (in :
henries) for this machine;
b) Find the constant speed eigenvalues which characterize the switching
transient I'ollowing removal of the source;
c) Find the initial values of the complex vectors representing the rotor
and stator fluxes in a stator reference;
d) Find the time domain expression for thc stat01 voltage v;',, (stator
reference) following the switching operation assumlng the speed remains
constant.
Approximetcly how long will it take for this voltage to drop to 10% of rated
voltage?

Problem 4-7 Transient Current and Torque by 'Plug' Reversal

For the 100 hp, 460 volt, 60 Hz, 4 pole, three phase induction machine of prob-
lems 4-5 and 4-6 operated at rated voltage and frequency and no load (assume
the inertia is large enough to maintain constant rotor speed):
a) use the constant rotor flux linkage transient model to calculate the
transient current immediately following the instantaneous reversal of
two line wires (plugging) if the switching occurs at a voltage maximum;
b) find the maximum instantaneous transient torque resulting from the
switching operation.

Problem 4-8 VSI-Induction Machine Harmonic Analysis

For a VSI driving a 100 hp induction machine with pu parameters
rg = 0.015, xIS = 0.10, x, = 2.0, rr = 0.020, xlr = 0.10 with a constant
input voltage V, such that fundamental output voltage is 1.0 pu, find the funda-
mental and 5-th harmonic components of iqs and ids and the dc and 6th har-
monic components of i i . Take the fundamental frequency equal to 1.0 pu and
 5 Principles of Vector Control and Field
Orientation

5.1 Introduction
In this chapter the basic concepts of torque control and field orientation are
introduced based on steady state considerations for both synchronous and
induction machines. A basic set of three requirements which enable direct con-
trol of machine torque production is developed and used to illustrate the type
of controller needed in both types of ac machines. These controllers are gener-
ally referred to as vector controllers because they control both the amplitude
and phase of the ac excitation. The vector control of currents and voltages
results in control of the spatial orientation of the electromagnetic fields in the
machine and has led to the t e m j e l d orientation [1,2,3]. Usually, this term is
reserved for controllers which maintain a 90' spatial orientation between criti-
cal field components and hence we will adopt the termfield angle control or
simply angle control for systems which depart from the 90" orientation. The
dynamic performance of the systems outlined in this chapter is considered in
later chapters.

5.2 DC Machine Torque Control

Before proceeding with the development of the principles of vector control and
field orientation, a brief review of torque control in dc machines will be pre-
sented. The close parallel between the physical torque control principles in a dc
machine and in ac vector controllers will be emphasized in later sections.
A dc machine consists of a stationary field structure utilizing a stationary
dc excited winding or permanent magnets and a rotating armature winding
supplied through a commutator and brushes. This basic structure is schemati-
cally illustrated in Figure 5.1 along with the resulting orientation of the arma-
ture MMF and the field flux. The action of the commutator is to reverse the
direction of the armature winding currents as the coils pass the brush position
such that the armature current distribution is fixed in space no matter what
204 5 Principles of Vector Control and Field Orientation DC Machine Torque Control 205
-

rotor speed exists. As shown in the figure, the field flux and armature MMF are
maintained in a mutually perpendicular orientation independent of rotor speed.
The result of this orthogonality is that the field flux is unaffected by the arma-
ture current except for second order. nonlinear effects.
The electromagnetic interaction between the field flux and the armature
MMF results in two basic outputs: an induced voltage proportional to rotor
speed,
Torque

and an electromagnetic torque proportional to the armature current

Orientation of Armature MMF and Field Flux where omis the rotor speed in mechanical radians per second, P is the num-
ber of poles and Aaf is the flux produced by the field current which links the
armature winding. (The constants in the two equations are unit dependent but
are identically equal to PI2 in SI units). The flux linking the armature is related
to the total field flux linkages hf by the expression

Lap LF and Lf are the mutual inductance between field and armature windings,
the field leakage inductance and the fieid self inductance respectively. The
torque can then be written in the alternative form

Torque = Constant x Field Flux x Armature Cur-ent

Inclusion of the armature resistance (and inductance for transient currents)
completes the model as illustrated in Figure 5.1. It is important to note that the
simplicity of the model is strongly dependent on thc mutually pcrpcndicular
orientation of the Hux and MMF. If this orthogonal~lywere disturbctl (by shift-
ing the brushes for example) two major complications occur:
Figure 5.1 DC machinc model 1) the field flux is no longer independent of the armaturc currcnt since
there will be an MMF component in the f~eldaxis, and
2) the voltage and torque relations in eqns (5.2-1) and (5.2-2) will be
206 5 Principles of Vector Control and Field Orienvation Requirements for Toque Control 207

modified by addition of an angle dependent function (sine of the an-

gle between axes to a first approximation). Flux Command
Adjustable speed operation is normally attained by operating with a fixed
1;
field flux and a varying the armature voltage. The no load speed is then deter-
mined as the speed where the induced voltage is essentially equal to the
applied voltage. Torque production requires armature current and results in an
armature IR drop. This IR drop must come at the expense of a smaller induced
voltage which necessitates a small speed reduction. Thus, every armature volt-
age has an associated torque-speed curve and the actual speed of operation is ac or dc
Power Voltage
determined by the armature voltage and the torque required to supply the load.
Input
The situation is analogous to variable frequency, constant voltshertz operation
of an induction machine. In both cases the torque, current and speed are deter-
mined by the load with the torque-speed characteristic set by the excitation.
Adjustable torque operation is readily attainable in the dc machine by the
simple action of controlling the armature current instead of the voltage. As
expressed in eqn (5.2-4), with a constant value of field flux, the torque is
directly proportional to armature current. Thus, the torque can be adjusted as
accurately and as rapidly as the armature current can be adjusted and con-
trolled. In practice this is accomplished by using a feedback current regulator
Figure 5.2 Elements of a torque controlled separately excited dc motor drive
and a power electronic supply. If both positive and negative values of torque
(and speed) are desired, the supply must be capable of bidirectional current
(and voltage). The armature voltage required to supply the controlled current is 2) an independently controlled or constant value of the field flux;
always the sum of the speed dependent induced voltage and the IR and L(dI/dt)
3) an independently controlled orthogonal spatial angle between the flux
voltages of the armature circuit. Very rapid current changes thus require large
axis and the MMF axis to avoid interaction of the MMF and the flux.
transient armature voltages to overcome the inductive effects and operation at
high speed requires a large average voltage to supply the speed dependent If all three of these requirements are met at every instant of time, the torque
induced voltage. However, the torque will exactly follow the current and to the will instantaneously follow the current and instantaneous torque control will
extent that the current regulator follows its reference input, the torque will be result. If, as can occur in certain systems, these requirements are only met for
proportional to the current reference to the regulator. Figure 5.2 illustrates the steady state conditions, only steady state torque control will result. During the
elements of a typical torque controlled dc motor drive. transient period, the torque will not follow the current exactly and this may or
may not be satisfactory for the specific application.
In the dc machine, requirements 2 and 3 are assured by the commutator and
5.3 Requirements for Torque Control
the separate field excitation system (dc winding or permanent magnet). In ac
Following the concepts outlined for the dc machine, the requirements for machines,
, . these two requirements must be achieved by external controls and
torque control are: thus the situation is more complex and somewhat more difficult to understand.
1) an independently controlled armature current to overcome the effects gor this reason, the development which follows first examines torque control in
of armature winding resistance, leakage inductance and induced voltage; synchronous motors where the separate field winding and known position of
 208 5 Principles of Vector Control and Field Orientation Svnchronous Machine Vector Control
--

the flux axis on the rotor are easily visualized as similar to the dc machine situ-
ation.
DC Link CSI Synchronous Machine

5.4 Synchronous Machine Vector Control

The CSI driven synchronous machine is a natural starting point since it com-

E
bines a number of features which suggest vector control and angle control.
These include: Field
1) the CSI is a current supply which can be controlled in both amplitude
and phase,
2) the field winding of the machine is physically available and can be

+
controlled as in the dc machine, and L I

3) the space position of the rotor dc field is clearly located in space by

the position of the rotor.
4
Voltaae
4'
SCR Gate Field
These electrical and physical features suggest the possibility of controlling ~onr&l Signals Control
Signals Signals
the CSI - synchronous machine combination so as to emulate a dc machine.
The resulting brushless dc machine or cornmutatorless dc machine has been
utilized for many years, especially in Japan [4,5]. Figure 5.3 CSI-synchronous machine system

5.4.1 CSI - Synchronous Machine (Commutatorless DC Machine)

The basic open loop CSI - synchronous machine system is illustrated in Figure I1Stator Field , ,
5.3. The CSI can be force commutated or in many high power applications is
load commutated by operating the synchronous machine with leading current
(overexcited). In either case, the concept of sensing the rotor position to locate
the field winding axis and using this rotor position information to control the
firing of the SCRs in the inverter and hence control the field angle can be
employed. The scheme is illustrated in Figure 5.4 and the analogy to a dc
I Rotor Position
9 hcder

motor is apparent in the sense of producing a fixed space angle between the (Phase of Srator Current)
field winding and the stator winding MMF. Clearly the system is self-synchro-
Figure 5.4 Commutatorless dc motor utilizing direct feedback of rotor
nous since the frequency of the inverter will always be the shaft frequency.
position to control phase of stator current
Steady state performance of the machine for the fundamental component
can be treated by the standard equivalent circuit approach as illustrated in Fig-
ure 5.5. For simplicity, a non-salient pole machine is considered since saliency In the circuit of Figure 5.5 the voltage E, is the internal voltage produced
. does not affect the basic system behavior and only serves to obscure the under- by the dc field current and a, is the speed of the rotor in equivalent electrical
lying concepts to be considered here. Saliency effects are considered later radians per second and, under steady ,state conditions, is equal to the electrical
when dynamic behavior is presented using d,q modeling. angular frequency a,. The amplitude of E, is proportional to the rotor speed
and the flux produced by the field current
 210 5 Principles of Vector Control and Field Orientation Synchronous Machine Vector Control 21 1

needed to create a leading power factor at the machine terminals as shown in

Figure 5.6.

Figure 5.5 Phasor diagram and torque equation for cornrnutatorless dc

motor with positive y
Figure 5.6 Phasor diagram for larger value of y showing current leading
voltage

just a s liw thc counter EMF in a dc machine, and the time phase of E, is Thc analogy of the system to a dc machine can be extended to include the
directly related to the rotor position. Thus, the rotor position feedback loop can armature counter EMF if we consider the dc side variables of the inverter.
be viewed as directly controlling the angle iy between the phasors of Ea and la i Equating the dc power input to the power associated with the fundamental
as shown in Figure 5.5. In this figure, yis @own as a small positive angle with + component on the ac sidc yiclds
I , leading E,. Note that for this condition, idding in the IJs and IaRs voltages
i V,II = ( V,-I,Rdc)Il = 3V,INcose (5.4-4)
yields a terminal voltage which results in a:lagging current at the machine ter-
minals. + As shown in Chapter 3, the Inverter current I, is directly related to the
'
The machine torque can be evaluated from the power input to the voltage motor current (fundamental component) by
E, divided by mechanical speed

and from the phasor diagrams in Figure 5.5 or Figure 5.6, the voltage compo-
Substituting for E, from eqn (5.4-1) yields nent Va cos 8 can be expressed as

V,cos 0 = E, cos y + I, Rs (5.4-6)

Combining these three equations, the following relation between the recti-
which is identical in form to the corresponding equation for a dc machine if y is fier output voltage VR, the armature current I, and the back EMF E,cos y can
zero. Note also that the torque per amp&. can be maximized by choosing
y = 0". However, if load commutation is desired, a relatively large value of y is 2 be obtained
i;
212 5 Principles of Vector Control and Field Orientation , Synchronous Machine Steady State d.q Model 213

This result is illustrated in the form of an equivalent circuit in Figure 5.7.

Current Lagging Voltage Current Leading Voltage

Marimurn Torque/Arnpere Self Cornmutorion Possible
Optimal Decoupling Armature and Field Interactive

Figure 5.8 Influence of internal angle

Figure 5.7 Equivalent circuit showing analogy to dc motor if y = constant

5.5 Synchronous Machine Steady

, State d,q Model
When y is constant, the circuit is identical to the armature circuit of a dc
To facilitate the transition to the dyn&c analysis to be presented rn Chapter 6,
machine. Note that the system of Figure 5.4 can be operated in an exactly anal-
it is convenient to introduce the steady state d,q model of a xynchronous
ogous manner as a dc shunt motor; the speed of-operation will be essentially
machine. The notation employed in the dynamic analysis of Chapter 6 w ~ l l
set by the rectifier voltage VR with the system in steady state equilibrium when
the counter voltage E, cosy plus the IR drops is equal to (x/3&)vR. The
$1 then correspond to the steady state notation employed in this section.

speed-torque curves will be straight lines just as for the dc machine. !I 5.5.1 Steady State Conditions in d,q Variables
5
5.4.2 Torque Control and Choice of y To relate the steady state concepts of Section 5.4 to the d , q model, ~t is only
necessary to specialize the d,q equations to the steady state. The construnts for
For direct torque control, eqn (5.4-3) indicates that just as in the d.c. machine
steady state are:
the requirement is to be able to control the stator current I,. As noted previ-
ously, the choice of y = 0" is attractive in terms of maximizing the torque per
: 1) Constant amplitude and phase qf stator current - mplies iq, = I,, and
ids= Ids are constants,
ampere. As will be shown subsequently, this choice is also advantageous in
terms of decoupling the transient response of the system. One drawback to 2) Constant rotor flux 11nkages- implies having zero currents in lhc dampel
choosing y = 0" is the inescapable fact that the terminal power factor will be windings, iqr = idr = 0,
lagging. This is unacceptable in large hp drives where load commutation is 3 ) Constant field current - implies if, = I f .
necessary for other reasons and hence a relatively large y (40"-60°) is often i
used ir?these cases. Figure 5.8 summarizes these properties of the angle y. i' The basic torque equation from Chapter 2 is
214 5 Principles of Vector Control and Field Orientation mchronous Machine Steady State d.q Model

r e =23-P2- ( d~
h i9s - h4s i d s ] (5.5-1)

or, by substituting for the Hux linkages in terms of currents using the flux link-
age expressions in Chapter 2,

where we now consider, for generality, a salient pole machine.

Since the damper currents are zero, the steadv state toraue heromm

Salient Pole Machine

which consists of the torque produced by the field winding

3pXmd
3% Iqs
Reaction Torque = -- -I I
22 wb J s s
and the reluctance torque

Reluctunce Torque = - -
~ P X ,- x*
22 0, 'Ids Ips

The steady state stator flux linkages are

hqs = L'7s I qs (5.5-6) ;

' I
'ds= Lds'ds %d If
+
(5.5-7) Non-Salient Pole
d-axis Machine
and the steady state stator voltage components are @ = 0 in the steady state)
Note: All Reactances Evaluated at o = w e

Figure 5.9 Steady state vector diagrams showing internal voltage E,,
terminal voltage qdsand the internal angle y

5.5.2 d,q Variable Vector Diagrams

I?
The voltage equations are illustrated in Figure 5.9 in the d,q axes. The voltage 3
component
216 5 Principles of Vector Control and Field Orientation Torque Control Implementations- Synchronous Machines 217

is the induced voltage E, of steady state theory and the diagram in Figure 5.9
for the non-salient pole machine (Xdu= Xqs, Xmd = Xmq) is clearly the same as
the phasor diagrams presented in Section 5.4$.
The diagrams in Figure 5.9 are d,q diagrams representing constant d,q
quantities in the two axes, i.e. the complex vector interpretation of the scalar d-
q axis quantities. They are, however, identical to the (time) phasor diagrams of
conventional steady state theory. Note that the angle y which appeared as a
phase angle in conventional steady state theory is now interpreted as a space
angle between the q-axis (where E, is located) and the vector position of the
current hds The two interpretations are a result of the shift in reference sys-
tem; a stationary reference for conventional steady state theory where all elec-
trical quantities vary at stator frequency and a reference system rotating with I

the rotor where all quantities are dc. I

The torque can be rewritten in terms of the angle y i :d-axis

3P 1 Figure 5.10 Steady state vector diagram for field orientation (y = 0, Id, = 0)
Reaction Torque =-- -EaIqdsCOSY
22w,

3P 1 5.6 Torque Control Implementations - Synchronous Machines

Reluctance Torque = ( X d s- Xqs) l i d scosysiny (5.5-12)
2 2 0,
The implementation of the angle control and field orientation concepts devel-
These torque components are the same as found from steady state theory. The , oped in the preceding sections calls for control of the magnitude and phase of
amplitude of the vector current 4 d S is clearly the same as the amplitude of the the stator current with respect to the location of the field winding axls. In gen-
stator (peak) current as can be seen by comparing the phasor diagrams of Sec- eral, this vector control of the stator current must be maintained for both steady
tion 5.3 with the vector diagrams of Figure 5.9. state and transient conditions. Several possible implementation schemes are
An important special case is illustrated in Figure 5.10 where the stator cur- now presented.
rent is all in the q-axis (y= 0). This is the situation usually referred to asjeld-
orientation. Note that for this situation the field current in the d-axis and the 5.6.1 Torque Control Using Field Orientation with a CSI
stator current in the q-axis are 90" apart as is the case in the dc machine. Note Figure 5.11 suggests a direct implementation of field orientation (y = 0) using
also, that because of the absence of d-axis stator current there is no reluctance absolute rotor position sensing and il CSI. With y = 0, the stator current is
torque and only the q-axis reactance is involved in finding the terminal volt- entirely q-axis current and is equivalent to a torque command as illustrated.
age, i.e. there is no direct magnetization or demagnetization of the d-axis, only
the field winding acts to produce flux in this direction.
i The rotor position information is directly utilized to set y = 0 by controlling the
inverter firing times. Because of the commutation delay inherent in the CSI,
some form of compensation is necessary to maintain y = 0 for different levels
of current and different operating freqpencies. This compensation 1s suggested
- -
in the figure by the commutation delay compensation block and by the inclu-
$. E, is usually expressed as an rms quantity in steady slate analysis but represents the peak
value when using (1.q variables. sion of a phase regulator block. In principle. a y* command could be entcred in
218 5 Principles of Vector Control and Field Orientation Torque Control Implementations - Synchronous Machines 219

*t* Encoder
Torque Command I l l
r* Rotor to Stator Rotor
Position

Commutation
tos 0,
Delay
Compensation Figure 5.12 Torque control using a CPRWM. Field orientation requires
i,=O, ( y = 0 )

Figure 5.1 1 Torque control via field orientation using a current regulated CSI convert the dc signals representing the torque command ,:i and the field com-
ponent command ia, into ac signals which become the current commands for
the phase regulator in order to operate at other than y = 0. In this case the input the CRPWM. In complex vector form the rotation transformation is simply
current would no longer represent a torque command since it would contain
both a q-axis and a d-axis component. An jmplementation for such a scheme = ,jer ir*
qds qds (5.6-1)
will be presented after introducing the conFpt in connection with the current 1, since the commands are in a rotor reference and must be converted to a stator
regulated based implementation of angle cofitrol in the next section. frame. The two phase stator current commands i$ and iss are then converted
to three phase current commands iS,*, , ig andic and supplied to the
5.6.2 Torque Control Using a CRPWM ,
CRPWM. The actual equations implemented in the rotor to stator transfonna-
The regulation of the stator current by means of a fast switching power con- tion block in Figure 5.12 are then the equivalent of the combined rotating to
verter providcs a conceptually simple means for implementing torque control stationary transfonnation and the two phase to three phase transfonnation
with independent q-axis and d-axis current inputs [6]. Figure 5.12 illustrates resulting in the expressions
the basic system typically called a currentregulated pulse width modulated I

(CRPWM) inverter. In essence, all that is required is to use absolute rotor posi- ! i 6 = i$ ~ 0 . +~i$ 0 siner
~ (5.6-2)
tion information to convert the i;, and i& commands in the rotor reference
frame to a stator reference. The stator referred currents, at stator frequency,
become the current commands for the CRPWM as shown in the figure. Normal
field orientation is obtained by simply setting i;%, = 0 . Other choices for i*d,
allow for controlling the motor power factor or other performance features.
The rotor to stator transformation indicated in block form in Figure 5.12 is
the same basic transformation discussed in Chapter 2. The purpose here is to
220 5 Principles of Vector Control and Field Orientation Toraue Control molem men tat ions - Svnchronous Machines 22 1

For field orientation (y = 0) the flux command current i'd', would be zero. In 5.6.4 Torque Control Requirements - Synchronous Machine
later sections dealing with induction machines, the angle 8, will be chosen as The vector controlled synchronous machine systems of Figure 5.11 and Figure
the position angle of a particular flux vector in the machine but the same equa- 5.12 can be interpreted in terms of the basic torque control requirements out-
tions are applicable with the new choice of angle. lined in Section 5.2.
For the field oriented case of y = 0,examination of Fig. 5.1 1 reveals that:
5.6.3 Magnitude -Angle Resolver and Use in CSI Torque Control
1) the current amplitude regulator provides the controlled current of
The CSI torque control shown in Figure 5.11 can be generalized to the case of
requirement 1;
independent y- and d- axis input currents by introducing the concept of a mag-
nitude-anglc (or vector) resolver. The concept is illustrated in Figure 5.13 and 2) thc field winding is the exact counterpart of the clc m i ~ c l \ i ~licld
~ c wintl-
ing; I

3) the rotor position feedback loop controlling the phase of the stator cur-
rent provides the field orientation of requirement 3.
This system is readily seen to be a close relative of the dc machine. It is, in
fact, a machine with an 'electronic commutator' provided by the rotor position
controlled inverter switching times. h e physical commutator of a dc machine
provides exactly the same function df rotor position controlled current switch-
ing. In actuality all three torque control requirements are provided in very
much the same way as in the dc machine.
Figure 5.13 d-q to magnitudeangle resolver The only conceptual difference Gtween the system of Figure 5.12 and the
CSI system of Figure 5.1 1 is that theseparate amplitude and phase controls of
is easily understood in the context of thinking of the d.q set as a vector. From the CSI are combined into a single instantaneous current controller and both
this perspective, the resolver simply expresses two ways to describe a vector; functions are handled simultaneously Thus the CRPWM handles both the
orthogonal components or polar form. independent current excitation (requirement 1) and the field orientation
Using a resolver as an input element for the CSI system of Figure 5.11 per- (requirement 3) necessary for torque control in a single operation. In this sys-
mits treating the input quantities as i$ and iss as in Figure 5.12. The magni- tem it is not possible to isolate the electronic commutator and the current con-
tude output of the resolver would become the current command of the CSI trol function from each other in the same physical way as in the system of
I
(which would no longer be a torque command). The angle output would Figure 5.1 1.
become a y command and would be fed to the phase regulator to be combined The current reference expessiods in eqns (5.6-2) to (5.W). however,
with the rotor position information. The combined 8, signal and the signal + offer a new viewpoint which provides further insight into the torquc control
from the resolver would set the phase of the current to give the proper position concept. Consider that for any fixed speed a, the rotor angle can be expressed
to the current vector & in the rotor reference frame. With this approach the as
CSI system can be made fully equivalent to the system of Figure 5.12. Again
compensation for the commutation delay would be required to obtain equiva-
lent performance from the two systems. In both cases, setting iZs = 0 results where p is simply a position reference (position at t = 0). Using this result, the
in field orientation (y = 0) which would be the preferred option for torque con- expressions in eqns (5.6-2) to ( 5 . 6 4 ) clearly indicak the frequency and phase
trol and would be the obvious selection unless there is a need to regulate the of the current references and demonstiate that the stator currents are always at
power factor or some other terminal variable. synchronous frequency and at a fixed phase with respect to the field axis. Note
222 5 Principles of Vector Control and Field Orientation Brushless DC Machines 223
- -

that at zero speed the currents are dc currents and no special problem is
3 Phase
encountered. Note also, that when the speed is varying, the frequency (and 60 Hz
hence phase) of the currents is also varying so as to maintain the required fixed
space angle between the MMF and the field axis.

5.7 Electrical Measurement of the Rotor Field Angle 8, Current

Regulator
There are many situations in which the use of an encoder or resolver on the
shaft of the machine is undesirable for cost or reliability reasons. It is possible

1+*
to avoid the need for direct shaft position information by making electrical
measurements from which the rotor position (more specifically, rotor flux posi-
Toraue 'ndr
tion) information can be calculated. Such measurements generally require sig-
nal processing to obtain the desired information and the computations require

A
Resolver
knowledge of the stator parameters of the machine. The presentation of actual
measurement schemes and the associated computational requirements are
L
Field
-
Componenl
deferred to the next chapter on dynamic analysis.
Figure 5.14 illustrates the nature of an angle control or field orientation -- Electrical
Motor
( ij:, = 0 ) system based on electrical determination of rotor field angle.
Although the illustration shows a CSI based system, the same techniques are
I
CFO
V
7.. Stator

applicable using a CRPWM. Many high power, load comrnutated inverter

drives (LCIs) utilize a measurement and computational scheme like that illus-
trated in Figure 5.14 and do not employ an encoder or resolver.

5.8 Brushless DC Machines

1 Figure 5.14 Field orientation (i& = 0) or angle control in synchronous
The principles developed in the preceding sections have been widely used to
machine using electrical estimation of rotor flux angle
create machines which possess properties similar to dc machines but do not
employ ;I commutator and brushes. These machines are referred to collectively
as brushless dc machines but there is, in fact, a great deal of variation in the 1 Sine Wave Machines wii!hCurrent Contro
details of how they are constructed and controlled 17.81. While they all share Permanent magnet machines with distributed windings resulting in sinusoidai
the basic concept of field orientation and most utilize a rotating permanent :back EMF voltages are used to produce high performance brushless dc
magnet field and a stationary armature winding, there are major differences in r machines for demanding servo applications [6,8]. These drives typically
flux distribution, position sensing and current control which result in a variety
i employ a high resolution encoder or resolver and a CRPWM power supply
of performance characteristics. very much as illustrated in Figure 5.12 and are true field oriented drives.
: Because of the PM excitation a field component of current typically has little
" effect and is seldom used except to produce a very moderate increase in speed
educed torque loads.
224 5 Principles of Vector Control and Field Orientation Brushless DC Machines 225

With a zero value of the field component of current, the armature MMF and
field flux are orthogonal and the torque is directly proportional to current. With
a PM field, the cross magnetization caused by this armature current has very
little effect on the total flux and hence the linearity of the torque vs. current Phuse A Currm

characteristic is excellent for a wide range of current. Since machines with 210'

nearly sine wave back EMF are readily built and with modern CRPWM sup- W 150'

plies easily capable of excellent sinusoidal current waveforms, the torque rip-
ple in this class of drives is generally very low. They therefore offer very i
nearly ideal dc machine chalac~eristicswith bcttcr torquelineriia and torque
ripple properties and wider spccd ranges than brush type dc machines.

5.8.2 Trapezoidal Wave Machines with Current Control W 3311

Phus 8
In a surface magnet PM machine with concentrated coil windings the wave- UO' 270'
form of the back EMF is typically quasi-square wave, or more accurately, trap-
ezoidal [8,9]. This results from the nearly constant magnet MMF and the
resulting nearly constant flux density along the rotor surface. An essentially
constant torque can be obtained by controlling the current to be distributed to
the individual phases so as to coincide with the back EMF as shown in Figure
5.15. Note that it is only necessary to keep the 120' current pulse centered on
the back EMF wave. Since the 60" intervals of zero current correspond to the
regions where the EMF is changing polarity, it is only necessary to have the
constant voltage region be approximately 120" in width. The rise time of the
current and any asymmetries in the structure create torque ripple.
Control of the machine is much simpler than for sine wave machines since
only an indication of the center of the back EMF wave is needed. The most Figure 5.15 Back EMF and current in trapezoidal flux brushless dc
common method is the use of three Hall sensors to provide timing pulses for i machine
the three phi~ses.In many machines of this type threc liming pulses are used to
provide gating information for iln inverter which uses 120" switch operation as
in a CSI as illustrated in Figure 5.16. A dc link current regulator completes the 5.8.3 Other Brushless Machines
control. Z There are many other variants of the fundamental concept of field orientation
Machines of this type are field oriented in the sense that the field flux and which are included under the term brushless dc machine. To illustrate, consider
armature MMF are held to a fixed relative spatial position. However, lirnita- the machine in Figure 5.17 where the rotor position information is obtained by
$
tions of the control (current rise time, etc.) can cause variations in this spatial sensing the EMF zeros. The inverter in this case is operated with 120" gating so
relationships and create torque pulsations and reduced average torque. f there is always one phase off and the voltage of that phase is the back EMF. In
ka' this drive no current loop is used (except for current limit) and the speed is set
: by regulating the voltage applied to the motor. Drives of this type are generally
limited to small sizes.
 226 5 Princi~les
of Vector Control and Field Orientation Induction Machine Vector Control -Steady State 227

I I I
lnvener
Con1rol

Figure 5.16 Trapezoidal EMF motor used as brushless dc machine

Figure 5.18 Conventional induction motor equivalent circuit showing

I I rotor induced voltage E,

Speed t Mi,
Inverter EMF
current I, is shown in a direction opposing the stator current; this is opposite to
the usual steady state convention.
Control Zeroes
The torque is represented in the equivalent circuit as being proportional to
the air gap power which is the power in the resistor r,/S
Figure 5.17 Brushless dc machine using voltage sensing to control spatial angle
(no current loop)

5.9 Induction Machine Vector Control -Steady State where w, is the stator electrical frequency. This equation can be rewritten in
With synchronous machine vector control as background, the question of terms of the voltage magnitude E,. across the resistor rdS to obtain
inductton machine torque control can be approached from the perspective of
asking how vector control of induction machine stator current can be employed
to dircctly control torque [I]. The following development utilizes conventional
steady state induction machine theory but attempts to parallel the development which is a similar form to the torque expression in a field oriented synchronous
used in synchronous machines. Transient analysis using d,q modeling is treated machine (y = 0).In the synchronous machine, the induced voltage E, was con-
in the next chapter. trolled directly by the field current; hence to parallel the synchronous machine,
some means for independently controlling Er in the induction machine is
5.9.1 Co~tventionalEquivalent Circuit Cofisiderations needed. If this can be accomplished, eqn (5.9-2) implies that torque control
Figure 5.18 is the conventional induction motor equivalent circuit containing can be achieved in a similar manner as the synchronous machine. Note that the
two series reactances normally referred to as the stator and rotor leakage reac- phase angle between E, and 7, is automatically zero, corresponding to the
tances. To parallel the synchronous machine model and d,q theory, the rotor field oriented synchronous machine (y = 0), so that this aspect of torque control
228 5 Principles of Vector Control and Field Orientation Induction Machine Vector Control - Steady State 229

is automatically achieved in the induction machine (for steady state condi-

tions).

5.9.2 Mod@ed Equivalent Circuit

In order to illustrate h e similarity in torque production between the dc and
induction machines, the conventional equivalent circuit is not well suited to
our purposes. It has been shown in Chapter 4 that there are many possible
cquivitlent circuits for steady state operation. The very general model with an
arhitriuy rcl'crr;~lratio a is shown in Figure 5.19. It can hc rccalled that the

L~
Ls - = L :=
~ stator tnmsient inductance
Lr
.-
Figure 5.20 Induction motor equivalent circuit without rotor Ic;~k;~gc
referral ratio u = LJL,
rent in the conventional equivalent circuit and the new rotor voltage is L,,IL,
times the usual rotor voltage. Note hso, that the new magnetizing reactance
has this same voltage across its terminals and can therefore be directly associ-
Figure 5.19 General equivalent circuit showing arbitrary value of referral r ated with the flux producing the voltake E, In effect, the new circuit represents
ratio a ( a = NJNr yields the conventional circuit of Figure 5.18) f behavior in terms of rotor flux wherek the conventional circuit emphasizes air
gap flux. This is important for torquecontrol because it places in evidence the
magnetizing component of current responsible for the rotor flux and E,. rather
;
choice of a is completely free (except for a = 0) and hence an infinite number than that creating the air gap flux and Em as in the conventional circuit. This
of circuits can be obtained from Figure 5.19 by choosing different values of a. same circuit was derived in Chapter 4 by starting from the synchronous fmme
The conventional circuit is obtained by choosing a to be the stator to rotor model with the d-axis fixed to the rotor flux.
effective turns ratio. The new circuit is redrawn in Figure 5.21 with new labels on the circuit
An especially useful form of the circuit for torque control analysis is elements and current components. The stator side reactance is identified as the
obtained by choosing the referral ratio such that the series reactance in the stator short circuit transient reactance
rotor branch is zero. Setting this element equal to zero in the general circuit of
Figure 5.19 yields the vdue of the required referral ratio as

a = -="I (5.9-3) which is a well known induction machine transient parameter. More important,
r the stator current is shown divided into two components; one through the new
With this choice for a the general circuit reduces to the circuit shown in Figure magnetizing branch called I,$ and one through the new rotor resistance called
5.20. Note that the new current in the rotor branch is LJL, times the rotor cur-
230 5 Principles of Vector Control and Field Orientation Induction Machine Vector Control - Steady State 23 1

tTT.
These are the two components of stator current which respectively control The torque component of stator current is immediately identified as
the rotor flux and torque as is demonstrated by the following analysis.

5.9.S Torqrte Control in Terms of Is# and IsT

The voltage E, has been identified as the voltage drop across r j S . It is there- and the torque developed can be expressed using eqns (5.9-2). (5.9-6) and
fore equal to the time rate of change of the rotor flux

Er = juehr (5.9-5)
From the circuit of Figure 5.21, the cun'ent Is$ is given by

which demonstrates the desired torque control properties in terms of the cur-
rent components I,+ and IsTNote the similarity to the field oriented synchro-
nous machine (and the dc machine) with I,$ playing the role of field current
and ISTthe stator (or armature) current. A phasor diagram illustrating the cur-
rent components is given in Figure 5.22.

Figure 5.2 1 Equivalent circuit showing torque component (IsT) and rotor flux
component (Isq) of stator current

1.
Figure 5.22 Phasor diagram showing torque and flux components of stator
current

It is important to note that the circuit of Figure 5.21 contains one more rela-
Combining eqn (5.9-5) and (5.9-6) yields rr -
tion involving the torque component of current and E, In particular
- -
hr = LmI,$ (5.9-7)
which clei~rlydemonstrates that the rotor flux is controlled by Is@.
232 5 Principles of Vector Con~roland Field Orientation Induction Machine Vector Control - Steady State 233

using rotor flux; in particular, the two current components are not orthogonal
components and the decoupling for transient behavior to be demonsfra~cdin
the next chapter is not complete. In effect, what happens is that a torque
increase in this system requires a build up of the rotor leakage flux whereas the
rotor flux orthogonal component system already has the proper flux level in the
Combining eqns (5.9-6) and (5.9-10) yields a relation between IsT and Is@ rotor circuits. The controlled air gap flux system can be considered as some-
what like a synchronous machine angle control scheme in which y # 0; there
will be a torque transient in response to a torque command instead of the
instantaneous torque response associated with systems having y = 0. It is, how-
ever, possible to design a compensated field oriented controller based on air
which is a consequence of the fact that the voltage across the magnetizing reac-
gap flux as will be shown in Chapter 6.
tance and the voltage across the equivalent rotor resistance are equal. This is an
extremely important result since it implies that for torque control in terms of 5.9.4 Terminal Behavior in Terms of lsgandlsT
,
Is$ and IsT as expressed in eqn (5.9-9), there is also a unique value of slip fre-
quency associated with each set of values of the current components. Rewrit- While torque control focuses on the relation between the current components
ing eqn (5.9-1 1) as a slip expression in terms of current magnitudes and the flux and torque, the terminal behavior of the machine also must be
evaluated. The required terminal voltage and stator power factor are important
quantities with respect to machine and converter ratings. Figure 5.23 is a pha-

emphasizes this relationship and the slip variation which must accompany
torque control via the current Is+ and IsT.
This relation is simply an expression of the fact that specifying stator cur-
rent and slip frequency in an induction machine completely determines the
torque. The special feature of the current components Is+ and IsT is that they
specify both rotor flux and torque (and they are orthogonal). Note that once Is+
and IsT are chosen, eqn (5.9-12) determines the one and only slip frequency
value which will yield the proper torque and flux. In fact, this represents a
means of steady state control; choose Is@ and IsT and compute Swe from (5.9-
12) to attain the proper operating point. This same concept will be shown to
hold for transient conditions and is the basis of indirectjeld orientation [I].
It is also interesting to note that steady state torque control can be achieved
in terms of air gap flux and the normal magnetizing current. Conceptually, this
involves the conventional equivalent circuit of Figure 5.18 and the resolution
of the stator current into components representing I,, and I, in Figure 5.18. Figure 5.23 Phasor diagram showing rotor induced voltage and terminal
There is an expression like eqn (5.9-12) for the slip associated with these two voltage
current components to produce a torque expression in terms of I,,, and I ,
Although this approach is often used it has several disadvantages compared to sor diagram based on the equivalent circuit of Figure 5.18. Thc ncgativc of the
234 5 Principles of Vector Control and Field Orientation Induction Machine Steady State d.q Model 235

rotor induced voltage E, multiplied by LJL, is shown 90" ahead of the rotor The most appropriate d,q model uses a synchronously rotating frame with
flux vector. To obtain the terminal voltage, it is only necessary to add the stator the d-axis fixed to the rotor flux. This model was developed in Chapter 4 and is
7, rs and i , ~voltages
, exactly as is done in a synchronous machine. However, repeated in Figure 5.24 using slightly different notation to correspond more
there is an important difference; in the induction motor the reactance is Xi
whereas in the synchronous machine the reactance is X,. The transient reac-
tance XI is approximately equal to the total leakage reactance

and is usually of the order of about 0.2 per unit (pu). The synchronous reac-
tance of a synchronous machine is a self-reactance and can be as large as 1-1.5
put. Thus, a typical induction machine will have an I,X: voltage of relatively
small size even at base speed and torque while a synchronous machine can
have a much larger reactance voltage. As can be seen from Figure 5.23, a larger
reactance voltage leads to a larger terminal voltage and poorer power factor. Figure 5.24 Equivalent circuit using complex steady state d-q currents -
synchronously rotating reference centered on rotor flux vector
Thus the inherently low value of Xi is an advantage for the induction
machine. In synchronous machines, a low value of X4,is desirable for the same
reason; better power factor and lower stator voltage. closely with the dynamic d,q analysis presented in later chapters. While this
Note that although the torque and flux relations involving Is$ and Is=are equivalent circuit has the same form as the stator referred circuit of Figure
independent of rotor speed and frequency (a requirement of torque control), the 5.21, the interpretation is quite different. In Figure 5.21, the circuit represents
terminal voltage and power factor relations are obviously speed dependent. At the relationships between conventional phasors; the phasors represent the mag-
low speed the 7, rJ voltage becomes dominant and the reactance voltage i , ~ : nitude and phase of stator frequency sinusoidally varying time hmctions. Fig-
and induced voltage E, become very small. Torque control via Is$ and IsTis ure 5.24 is a complex vector circuit diagram representing the,nl&tionsbetween
independent of these variations so long as the cdrrent controller can continue to dc quant~tiesin the d- and q- axes. This is, usually, a much.more useful con-
supply the commanded currents at the required terminal voltage and power cept since the variables in this circuit can be directly associated with the dc
factor. control signals for Iqs and Id, which are involved in the implementation of vec-
tor control. The vector diagram in Figure 5.25 shows these dc quantities and,
5.10 Induction Machine Steady State d,q Model using the obvious similarity between the stator referred and synchronous frame
circuit models, illustrates the relationships between the d,q quantities Iq, and
The circuits and phasor diagrams of Section 5.9 have thus far been interpreted Id, and the phasor magnitudes IsTand I,$*.
as conventional (stator referred) phasor models of the machine. It is much
more common to use d,q modeling in treating vector control and it is therefore 5.10.1 Example - Induction Muchine Field Orientation
useful to inrmlucc the steady state d,q model of the induction machine at this
A 100 hp, 460 volt, induction machine is operated from an ideal controlled cur-
point.
rent electronic converter using field oriented control. The pu parameters of the
machine are, in cor~ventionalper-phase symbols: (at rated frequency)
t . In a salient pole synchronous machine the appropriate reactance is Xq, which is smaller
than Xd,< (typical range 0.2 < Xy,i0.8). The per unit system of units is d~scussedin Appendix
A. *. Phasors are usually expressed as rms quantities whereas d.9 variables usually peak val-
C
ues. Thus. the magnitudes ~ f l , ~ ~ r ntypically
d 3.
differ from I,,! and Id,
236 5 Principles of Vector Control and Field Orientalion Induction Machine Steady State d,q Model 237

a) Find the required values of iq8 and id, and the slip relation relating So,,
iqs and hdrto produce operation at rated torque and rated speed if the terminal
voltage and frequency are held at their rated values.
The rotor flux linkage is: xdr= L,id, + Lm ids (id,. = 0 in steady state).
Hence, Ad, = Lm ids. Therefore,

Rearranging.

I
I

: d-axis Also, by definition,

Figure 5.25 Steady state d,q currents - synchronously rotating reference

centered on the rotor flux vector

l'ysI2 = 1'sI2 - /Ids(' '

Combining expressions,

The rated slip frequency is 0.0248. Express all answers in real units.
a) Find the required values of iqs and id, and the slip relation relating So,, To find I~,,use the steady state equivalent circuit
iqs and Ad, to produce operation at rated torque and rated speed if the terminal
voltage and frequency are held at their rated values.
b) Find the final steady state torque and slip frequency if ids is reduced to
one half its value in part a while the stator current amplitude is held constant at
the value in part a. Find the terminal voltage and stator frequency if the new
speed is twice that in part (a).
Solution - The solution assumes the following:
1) The machine has 4 poles
2) The per unit base is 100 hp, 460V
The solution to part a also uses the standard steady state induction motor
model (two leakages). It is definitely acceptable (and in many ways, easier) to
use the motor model that separates torque and flux components. The most com-
mon error in so doing is failure to scale the components (r, and Lm) appropri-
ately.
 238 5 Principles of Vector Control and Field Orientation Induction Machine Steady State d,q Model 239

where '11' denotes the parallel combination of the two reactances on either side
rr I'qsl
of the double vertical line SO, = - -
L r (Ids(

wbase= 377 radls

b) Find the final steady state torque and slip frequency if id, is reduced to
one half its value in part (a) while the stator current amplitude is held constant
at the value in part (a). Find the terminal voltage and stator frequency if the
new speed is twice that in part (a).
6.78)Ids12= I .61 - lIds12
1
lIdsl = (0.456) = 0.228 pu (30.2 A peak)
ll(,sl= 0 . 4 5 6 ~ ~
Base Current: \Iqs/ = = = 1.25 pu (165 A peak
Find zb,,,
'base = ( loo hp) (746 W1hp) = 93.6 A rms (132 A peak)
A (460 V)
\Ids( = (0.456 pu) (93.6 A) = 42.7 A rms

llllBl= 42.72/2 = 60.3 A peak

Find Torque

Then,

T , = 214.8 N-m
Find slip frequency
11,1 = l l l f i = 157 Apeak

Swe = (0.0248) (377radls) = 9.35

Checking,
240 5 Principles of Vector Control and Field Orientation Implementation of Field Orientation in Induction Machines 24 1

( S o , ) = 19.6 rads implementation is illustrated in Figure 5.26. 'fhc only esscnrial differcnce

hlduction
I Motor

Find terminal voltage and stator frequency when the speed is doubled.
i,
1 .
, *
--c
* Torque Command

CommMd
Syt~chrotloirs
To Sfator
Tran.$or~uo~ion ws
J
or .

Figure 5.26 Basic induction motor field orientation system using a

tirid / o r
Mivhnt~icul
Si,~yrals

or = 2 ( o p - S o , ) = 2 (377 - 9.35) = 735 rads CPRWM

The frequency is
between the two types of machines is that the field winding (rotor flux) posi-
a, = o r + (So,) = 735 + 19.0 = 754 radls tion angle in a synchronous machine is directly available as a measurable
mechanical quantity while the position angle of the rotor flux in an induction
machine is not mechanically measurable. Just as in the synchronous machine,
two basic approaches to determining the rotor flux position angle have
Let lqds= \Iqs\ - jlldsl and Ids = -jlldsl. Remember the frequency is evolved; direct schemes which electrically determine the angle from flux mea-
754 rads (= 2.0 pu). This affects the reactance value proportionately. Substitut- surements and indirect schemes which measure the rotor position and utilize
ing the slip relation to compute the angle of the rotor flux relative to the rotor. The
indirect method. which is analogous to rotor position measurement in a syn-
chronous machine, is treated first.
vqds = (1.25 - jO.228)
5.11.1 Indirect Field Orientation
vqds = 1.12 - j0.0256 pu
Indirect field orientation is based on the slip relation, eqn (5.9-13). repeated
vqds = 11.12 - jO.02561 (460) = 514 volts 1-1 rms here for convenience in d,q notation

5.11 Implementation of Field Orientation in Induction

Machines
This method makes use of the fact that satisfying the slip relation is a necessary
It should be apparent that the implementation of vector control an and sufficient condition to produce field orientation, i.e. if the slip relation is
.tation in induction machines can be carried out in the same manner as for syn- satisfied, Id, will be aligned with the rotor flux. At present this has only been
chronous machines if the position angle of the rotor flux is known. The basic
shown for steady state operation but the dynamic analysis presented in the next
242 5 Principles of Vector Control and Field Orientation 243
- - Implementation of Field Orientation in Induction Machines

chapter confirms that it is also true (in a slightly altered form involving the
rotor flux linkage hdr instead of the current I*) under transient conditions.
Figure 5.27 illustrates the basic structure of an indirect field orientation

---I
-w- . 1 ,
Induction
Motor

4
sin 0
t
cos 0
t t
sin0 c o s 0
Overall Cwrdina te Tramformation (Vector)Rotation P h w Conversion

47
. Torque Command
.e* T-~
'ds Flux C'ornmand
I -

Figure 5.27 Indirect field orientation using a CRPWM

sysrcrn u s i ~ ~a gcurrent regulated PWM inverter. The commanded currents I&

and I$y are converted to stator referred reference currents by a synchronous to
stator transformation just as in a synchronous machine. The field angle 8,. is,
i i = c o s esine ~ I [ . ]fj
- s i ncos0 k ] = [ ~-sine ~ i ~ fe ] [ " l
~ ~ ecos0
howcvcr, generated by summing a rotor position signal and a slip position sig-
nal. Rotor position information is obtain+ directly from a shaft (incremental)
encoder and the slip position is derived from a slip calculator implementing
eqn (5.1 1 -I). The conversion of the slip signal to an angle describing the posi-
tion of the rotor field relative to the rotor (8: ) and the summation of the two
position angles 8: and 8, is normally implemented digitally to avoid drift and Stationary to Rotating Rotating to Stationary
provide the needed accuracy. The transformation T I which appears in Figure
5.26 is delined in Figure 5.28 and this .notation will be used through the
remainder of the text. Figure 5.28 Summary of transformations used in field orientation
244 5 Principles of Vector Control and Field Orientation Implementation of Field Orientation in Induction Machines 245

The necessity to use angles in the summation in Figure 5.27 does not stem
Induction
from the steady state theory presented in this chapter. In fact, the system will Motor
function to provide field orientation if the rotor position encoder is replaced by
a tachometer and the calculated slip frequency is simply added to the tachome-
60 Hz
ter signal to obtain a command for the stator frequency a,. The conversion to
rotor flux angle would then occur just prior to the transformation to stator vari-
ables. This conclusion is also true in the transient state as will be demonstrated
later. Such a system could be implemented using an analog tachometer and an
analog slip calculating circuit and, except for offsets and drift in the analog cir-
cuits, would function as a field oriented system. The exception in the previous
sentence is, however, an extremely important one. For example, an offset error
in the tachometer (or in the analog circuitry) will immediately create an error
in the resulting slip frequency in the motor. Recognizing the inherently small
slip frequency associated with high efficiency induction motors, it is easy to
see that the normal levels of offset and drift which occur in analog circuits
(caused by temperature changes, for example) would create intolerable errors
in the actual motor slip frequency. For these reasons, the addition of rotor
speed and slip frequency is essentially always done with digital circuitry. This
requires a digital signal describing the rotor speed. However, since it is ulti-
mately an angle which is required in the transformation to stator variables,
most systems use an encoder, convert the slip frequency signal to digital form
and perform the addition of angles rather than of speeds.
Assuming the implementation is correctly (exactly) done, the system of
Figure 5.27 is still subject to error because the rotor time constant T, directly
affects the calculated slip. In the figure, the parameters Lr and rr are entered
Figure 5.29 Indirect field orienwon
i using a current regulated CSI
with carets to indicate they are estimated values as opposed to actual values. To
the extent that & /t differs from I!+ /$, the computed slip will be in error and
correct field orientation will not be achieved. This can be a serious problem are not contained in the slip frequency command signal coming from the slip
and will be examined in detail in Chapter 8. calculator.
The basic structure of an indirect field orientation system using a CSI is Figure 5.30 illustrates the need fot the separate addition of phase changes.
illustrated in Figure 5.29. As in Figure 5.27, the torque and flux command cur- For a change in the torque command Prom I&, ~OI;,~ (with a fixed I:, ) the
rents are used to generate a slip frequency command using a slip calculator. In current vector should instantaneously change from I,*,, to I&2. The sllp fre-
addition, the current commands are converted to amplitude and phase com- quency should also change instazltaneously. The resolver does give the correct
mands using a resolver. The current amplitude command is directly employed amplitude Izd, and the new slip frequency will be provided by the slip calcula-
as the reference for the CSI current regulator as shown in the figure. The phase tor. However, unless the phase change (frequency) d@ldtis added as shown in
command is passed through a lead circuit such that phase changes are added Figure 5.29, the stator current command will correspond to the vector IT in
into the inverter frequency channel since these instantaneous phase changes Figure 5.30 and there will be a phase error in the field oriented system. This
246 5 Principles of Vector Control and Field Orientation Implementation of Field Orienvation in Induction Machines 247

form of adjustment of the control to account for motor parameter changes asso-
ciated with flux level variations or temperature variations.

5.11.2 Direct Field Orientation

Direct field orientation in induction machines [3] is the exact counterpart of
synchronous machine field orientation based on electrical measurement of the
rotor field angle. As for synchronous machines, there are a variety of methods
which can be employed to obtain the required rotor fluangle. Figure 5.31
illustrates the basic concept using a CRPWM. As in the case of the synchro-
nous machine, if the measurement system also provides the rotor flux magni-
tude, additional (optional) loops can be added to regulate the torque and flux as
shown in the figure. The consideration of the details of the measurement sys-
tem is deferred to the next chapter on dynamics of induction machine field ori-
Figure 5.30 Phasor diagram showing required phase correction for system entation.
of Figure 5.29
Direct orientation can also be accomplished using a CSI. The basic system
configuration is shown in Figure 5.32. Note that the phase angle of the stator
would result in an instantaneous loss of field orientation which produces a very
current must be added to the rotor flux angle to properly orient the current (to
sluggish response (in both flux and torque). Such a system will ultimately
avoid the problem illustrated in Fig. 5.27).
attain the correct steady state torque and flux. The transient period is very long
however, and is quite unacceptable for high performance applications. An anal-
5.11.3 Torque Control Requirements
ysis of the effects of incorrect orientation (detuning) is given in Chapter 7 and
the instantaneous phase error described above can be considered as a form of If we examine the induction machine field orientation systems of Figure 5.27
detuned operation. and Figure 5.29 or Figure 5.31 and Figure 5.32 to determine how the torque
It is important to note that the necessity for separate consideration of phase control requirements outlined in Section 5.3 are being met, the following
changes does not exist in the PWM system of Figure 5.27. The synchronous to observations can be made:
stator transformation used to generate the three phase current commands cor- 1) Independent current control of the stator current is employed to over-
rcctly includes the instantaneous phase changes associated with changes in come the stator resistance, stator leakage inductance and induced
lorque cornmand. voltage in the same manner as the dc machine.
The indirect field orientation systems depicted in Figure 5.27 and Figure 2) Independent control of the flux is obtained by controlling Id, = &Is9
5.29 arc clearly very closely related to the slip command systems which have
independently of Iqs = &IsT. The stator cumnt has two indepen-
been employed for many years (especially in CSI drives). They differ primarily
in that most other slip command systems are based on maintaining constant air dently controlled components.
gap flux while field orientation maintains constant rotor flux. They share with 3) The spatial orientation of the rotor flux with respect to the current
all slip command systems the basic problems of parameter sensitivity and the I qs = &lsT is maintained by using the rotor field angle Bg which
need for high accuracy in the tachometer loop. Digital techniques are adequate
is either obtained from flux measurement or computed from rotor po-
for the accuracy problem; the parameter sensitivity problem requires some
sition and slip frequency.
248 5 Principles of Vector Control and Field Orientation : A Basic Difference Between Synchronous and Induction Machine Field Orientalion
. .. --249

3 Phase
3 Phase 60 Hz
60 Hz

I
Basic Field I
Orientation I
System Inside Dashed
Lines I

Torque

I l
I
Flux
I
Regulator Motor
I
CFO Electrical
Sigtrals
f I
Torque
+ - I
Calculator Motor Current Feedback
i
9s cs

Figure 5.32 Direct implementatiori of induction machine field orientation

using a CSI (torque and flux regulators optional)

Figure5.31 Direct implementation of induction machine field

5.12 A Basic Difference Between Synchronous and Induction
orientation using a CRPWM (torque and flux regulators
optional) Machine Field Orientation
Thus the electronic commutator concept introduced in connection with the While the similarity between synchronous and induction machine field orienta-
synchronous machine drive carries over to the induction machine. The only tion has been emphasized in this chapter, there is one important difference
major difference is that while the rotor flux position is directly measured in a which should be noted. In a synchr&ous machine, the stator current creates a
synchronous machine, it is calculated in the induction machine drives of Figure rotor magnetic field which is orthogonal to that produced by the field current
5.27 and Figure 5.29. Comparing these figures with Figure 5.11 and Figure (hir in d,q notation). This effect is analogous to the cross magnetizing arma-
5.12 for the synchronous machine reveals the similarity. ture reaction in a dc machine and can cause a reduction in field flux as a result
250 5 Principles of Vector Control and Field Orientation 25 1

of saturation (demagnetizing effect of armature reaction as in a dc machine)

[lo]. hy, "Fast Response Reversible Brushless DC Drive with Regenerative Brak-
In contrast, the induction machine under field oriented control always ing". IEEE-IAS Annual Meeting Conference Record. 1984, pp. 445450.
maintains zero flux in the axis orthogonal to the rotor flux (this is, in fact, the [lo] D.W. Novotny, and P.L. Jansen. "Field Oriented Control of Synchronous Machines."
basic concept of induction machine field orientation). The field oriented induc- Journal of the Institution of Electronics and TelecommunicationEngineers, Vol. 37, No.
tion machine is thus analogous to a dc motor with a compensating (pole face) 1, 1991. pp. 46-56.
winding. There is, therefore, no demagnetization caused by the torque compo-
nent of stator current (qs component in d,q notation). Physically, the induced
rotor current (I, in Figure 5.23) functions as the compensating winding of a dc
machine, preventing any net cross magnetization of the field (rotor) flux.
Thus, while the field oriented synchronous machine may exhibit a second
order reduction of field flux at high torque, this effect is absent in field oriented
induction machines. In surface magnet PM synchronous machines, the demag-
netizing effect is likely to be small because of the long air gap associated with
the magnets. The cross magnetizing ampere turns can, however, cause magnet
demagnetization at high currents and must be carefully considered in motor
design.

References
K. Hassc, "Zur Dynamik Drehzahlgeregelter Antriebe Mit Stromrichtergespeisten
Asynchron-Kuzschlublaufennaschinen" (On the Dynamics of Speed Control of a Static
AC Drive with a Squirrel Cage Induction Machine), Ph.D. Dissertation, Techn. Hochs-
chute Darmstadt, 1969.
F. Bl;~schke,"The Principle of Field Orientation - the Basis for the Transvector Control
ol'Thrcc-Phase Machines", Siemens Zeitschrift, Vol. 45, No. 10, pp. 757-760, 1971.
F. Bleschke, "Das Verfahren der Feldorientierung zur Regelung der Drehfeldmaschine"
(Tlic Mcthod of Field Oricnti~tionfor Control of Three Phase Machines), Ph.D Disserta-
tion, University of Braunschweig, 1973.
S. Miyairi and Y. Tsunehiro, "The Characteristics of Commutatorless Motor with SCR
Inverter", Journal of IEE-Proceedings of Intermag Conference, April 1963.
H. Sasajima, T. Tsuchiya and M. Naito, "Shunt Commutatorless Motor". Elecuical En-
gineering in Japan, Vol. 88, No. 11, Nov. 1968, pp. 28-35.
G. Kaufrnan. L. Garces and G. Gallagher, "High-Performance Servo Drives for Machine
Tool Applications using AC Motors", I E W A S Annual Meeting Conference Record,
10x2, 1111. 469475.
I<. S . Colby, "Classification of Inverter Driven Permanent Magnet Synchronous Mo-
kws", IIIEE-IAS Annual Meeting Conferenct Record, 1988, pp. 1 4 .
I). C'. I I ; I I I ~ ~ I I"Drushlcss
I I ~ I I , Permanent Magnet Motor Design". (Book). McGraw-Hill,
252 5 Principles of Vector Control and Field Orientation Problems 253

Problems C) repeat "a" at we = 0.0.

Problem 5-1 Constant hp Operation Problem 5-4 Induction Machine Field Orientation
For a 60 Hz induction motor with per unit parameters A 100 hp machine with the parameters below is operated from a controlled
current electronic converter using field oriented control.
r,%= 0.02 x , ~= 0.08 x,, = 2.2 rS = 0.015 pu x , , = 0.10 pu x,, = 2.0 pu

r,. = 0.020 pu x l r = 0.10 pu r,,, = 50 pu at rated frequency

a) find the input current at base frequency, voltage and torque;
b) find the torque at o, = 1.5 and me = 2.0 if V is held at 1.0 pu and I is
held at the value found in (a); a) Find the values of IST and I and the slip relation relating So,,Is+
$4'
and IST to produce operation at rated torque and rated speed if the termi-
c) find the approximate upper frequency limit of the constant power region
nal voltage and frequency are at rated values;
of this machine.
For each case below, find the current components, the slip frequency, the stator
Problem 5-2 Torque-Slip Frequency Characteristics
frequency and the teninal voltage:
For the induction machine of problem 5-1:
b) 0.5 rated torque, rated speed, rated rotor flux (rated speed=O.9752 pu);
a) find and plot the torque-slip frequency curve for operation at rated
C)0.5 rated torque, rated speed, 0.5 rated rotor flux;
current for
d) zero torque, rated speed, rated rotor flux;
O<Sme< Ion;
e) rated torque, 0.5 rated speed,'rated rotor flux;
b) find and plot, on the same sheet, the torque-slip frequency curve for
f) rated torque, zero speed, rated rotor flux;
rated voltage, rated frequency operation;
c) repeat (b) at 5 Hz for rated voltshz. g) zero torque, zero speed, rated rotor flux.

Problem 5-3 Synchronous Machine Field Orientation Problem 5-5 Field Oriented Operation of a PM Synchronous Motor with
Voltage and Current Limits
A small permanet magnet synchronous machine has parameters,
For the PM synchronous motor of Problem 5-3 (parameters listed for conve-
nience)

The machine is operated from a controlled current electronic converter using

rotor position feedback such that the internal power factor angle y is always
operated above base speed subject to voltage and current limits of
held at zero. Find:
a) the stator currents and terminal voltages to give output torques of 1.O, V = 1 . 0 ~ ~I = 1 . 2 5 ~ ~
0.5 and 0.0 when the stator frequency is 1.0;
find:
b) repeat (a) at o, = 0.5 ;
a) the maximum speed attainable if field orientation is maintained;
254 5 Princi~lesof Vector Control and Field Orientation Problems 255

b) the current and torque at we = 1 . 2 0 if~ field

~ orientation is main- The diagrams should include hqdr,Iqdr,hqds,!qds and Vqds (where lqdris
tained; shown as an equivalent field current for the PM machine).
C)the maximum speed attainable and the corresponding value of y if Use an induction machine with per unit parameters
phase advance of the current is utilized and y is chosen to maximize the r , = 0.03 x , = 0.10 x, = 2.0
no load speed;
r2 = 0.03 x2 = 0.10
d) the maximum torque attainable add the corresponding value of y if
phase advance of the current is utilized to maximize the torque attain- and let the PM machine have xs = 0.3, r , = 0.03, x , = 0.10, x2 = 0.10 and
able. an internal voltage such that it has the same power factor as the induction
machine for rated conditions (assume the field has the same number of turns as
Problem 5-6 Field Oriented Operation of a n Induction Motor
the stator when you show Iqdr).
Tests on a field oriented induction machine yield the following data:
Discuss your results concerning the similarity (and differences) of the
speed frequency stator current stator voltage
steady state operating points.
radls Hz amperes volts
1) at rated flux-zero torque 188.5 60 21 200
2) at rated flux-rated torque 188.5 62.5 40 218

a) Use this data to find the rotor time constant, IsTand I,+ for rated
torque and rated flux;
b) Find the stator current and frequency if the machine is adjusted to run
at a speed of 377 rls at 50 % of rated torque and 50 % of rated flux;
C)Estimate the stator voltage for p& b. Explain your estimate.

Problem 5-7 Comparison of Field Orientation Drive Characteristics

Comparc the steady state characteristics of field oriented control as applied to
an induction machine and to a PM (permanent magnet) synchronous machine
in which y = 0 . Sketch vector diagrams in the synchronous frame compar-
ing the two systems assuming each system is adjusted to operated at rated con-
ditions at or = urBand include diagrams at:

a ) o r = urB b)o, = 0 c ) o r = 2wr8

T, = TR T, = TR T, = 0.5TR (half flux)
and and and
Te = 0.0 T, = 0.0 Te = 0.0 (half flux)
 6 Dynamics of Vator Control and Field
Orientation

6.1 Introduction
The steady state analysis in Chapter 5 has been used to demonstratc thc basic
concepts of vector control and field orientation in both synchronous and induc-
tion machines. The underlying motivation in Chapter 5 was to obtain torque
(and flux) control independent of motor speed. Although the steady state argu-
ments of Chapter 5 cannot provide information concerning the dynamic (tran-
sient) response, it was suggested that the same basic concepts as developed
from steady state theory also produce dynamic decoupling and torque control
in the dynamic state [1,2,3]. For high performance systems the dynamic
response characteristics are as important as the steady state performance and
this chapter is devoted to extending the concepts introduced in Chapter 5 to the
. dynamic state.
While it was convenient to treat the synchronous machine first in steady
state analysis, the opposite is true for the general dynamic case. The symmetry
of the induction machine results in a simpler dynamic model and will be
treated first. It is also true that virtually all induction machine vector controllers
employ a 90" spatial angle between the critical flux and MMF and are thus
field oriented systems by our definition. There are, however, important practi-
cal applications of synchronous machines where other than a 90" spatla1 angle
is employed in the vector controlled system. Such systems are not field ori-
ented by our definition and do not result in transient state torque control. The
analysis of such systems is more involved and is deferred until the end of the
chapter.

6.2 Dynamics of Induction Machine Field Orientation

I
The vector control concepts leading to torque control and field orientation in
induction machines as developed from steady state concepts in Chaptcr 5 will
now be examined using a full d,q variable, transient state model. 1t will be
258 6 Dynamics of Vector Control and Field Orientation

shown that the basic concepts carry over to the transient state and that the only we to be the instantaneous speed of hd,
and locking the phase of the reference
essential difference between steady state and transient behavior is the existence : system such that the rotor flux is entirely in the d-axis, resulting in
of a significant lag in the response of the flux to a flux command.
hi, = 0 (6.2-10)
6.2.1 Induction Machine d q Model with Axis Rotation at an Angular
This expresses the licld orientation concepi in d,q variables. With this
Velocity we
choice of reference, the d,q currents and rotor flux appear as shown in Figure
The d,q axis model of an induction machine with the reference axes rotating at 6.1. Note the obvious similarity of this d,q axis diagram to the phasor diagram
synchronous speed we is

where d-axis
Figure 6.1 d,q currents with reference axes oriented to the rotor flux

showing the phasors I,$ and IsT of steady state theory in Figure 5.22. Clearly
the orientation of the d,q axes to the rotor flux produces a one to one corre-
spondence between I ; , and is^ and between i i and I,(+. Assuming the
machine is supplied from a current regulated source so the stator equations can
e e be omitted, the d,q equations in a rotor flux oriented (field oriented) frame
he4' = L,'iZr + L, ( i q s + iqr) (6.2-9) f become:
The torque equation listed is one of many possible forms and has been selected
for the specific purpose of this section. 0 = rri;, + ( we - w,) h i , (6.2-1 I )
i,
6.2.2 Rotor F11uReferred IGpations
The torque and flux control concepts developed in the previous chapter imply
--
that the currents supplied to the machine should be oriented in phase and in 7
L? t. Again ~t 1s common to u w rms values for sleady state phd50rS and peak values for d,q quan-
quadrature to the rotor flux vector kdrThis can be accomplished by choosing
tities.
260 6 Dynamics of Vector Control and Field Orientation Dynamics of Induction Machine F~eldOrientat~on 26 1

Conibi~iirlg(6.2- 18) and (6.2-12) to climinatc I;,, yields lllc cquallon

relating i;, and hzr (lYqand h,)
6.2.3 Dynamic Response of Field Oriented Induction Machines
( r r+ L,P) A;, = r r ~ , , i ; s (6.3- 19)
Equations (6.2-11) to (6.2-14) describe the dynamic response of a field ori-
ented induction machine. The torque equation, eqn (6.2-14), clearly shows the or h;, can be eliminated to find the equation for iir
desired torque control property of providing a torque proportional to the torque
command current ie4 5 ' Note that when expressed in the form of eqn (6.2-14)
(in terms of and i;,) the torque equation is the same for transient and This last relation clearly shows that izr exists only when i& is changing
steady state conditions.
(changes in flux) and that in the steady state
A direct (ampere-turn) equilibrium relation between the torque command
current ieq s and the rotor current i'& follows immediately from eqn (6.2-13)

which demonstrates the correspondence between i;, and IS9.

The close parallel to the steady state concepts of the preceding chapter is
This is the same as the relation between IsT and the steady state rotor cur- clear. Figure 6.2 emphasizes this correspondence in terms of torque produc-
rent found in Chapter 5 . The slip relation follows immediately from eqn (6.2- tion. In this figure the relation between the flux command current iis and the
rotor flux A:, is shown as a first order transfer function with a time constant T,
11)
where,

or by using eqn (6.2-15), In the steady state the relation betweeri izs and h;, reduces to eqn (6.2-2 I ) and
the torque diagram of Figure 6.2 reduces to the steady state case of Chapter 5.
rr An alternative model for torque production is illustrated in Figure 6.3. Here
SW, = -- the rotor flux is suppressed and the torque is represented entirely in terms of
Lr A:~
motor currents. Following transient changes in the flux command current i;,,
which is the same as the steady state slip relation if the steady state result is an induced d-axis rotor current is produced and has a transient influence on the
expressed in terms of the flux Asr. The major departure from the steady state torque. During steady state operation the d-axis rotor current is zero and thc
results of Chapter 5 occurs in the behavior of the rotor flux h&. Equation (6.2- torque is proportional to the product of izs and i;, .
12) shows that in the steady state when phi, is zero, the rotor current compo- The nature of the dynamic response for changes in torque command is
nent i$r is zero. However, during flux changes, iZr is not zero but is given by illustrated at the left in Figure 6.4. with the flux command held constant, a
eqn (6.2-8) change in i;, is followed instantaneously by a corresponding change in iGr.
The result is an instantaneous change in the torque. The situation is more com-
 262
..---.----- h Dynamics of Vector Control and Field Orientation
-
Dynamics of Induction Machine Field Orientation 263

Fig~ire6.2 Torque production for field orientation in terms of rotor flux

and y-axis stator current Change in
k:ally

[
Change in
va1.e
zero
grows tofinal
I
Torque Command F l u Command
Figure 6.4 Illustration of response to step changes in torque
command and flux command
In summary, the only difference between the steady state and transient
behavior of a field oriented induction machine is the existence of a transient
rotor d-axis current whenever flux changes are made. The flux accordingly fol-
lows an exponential change with time constant L, / r, as the transient d-axis
current decays. Compensation for this effect can be included in the flux regula-
tor but will require large pulses of d-axis stator current to attain adequate com-
pensation. In effect, such compensation must overcome the tendency for the
rotor flux to remain constant and this is a fundamental transient property of a
closed circuit of low resistance.

6.2.4 Block Diagram of Field Oriented Induction Machine

Figure 6.3 Torque production for field orientation in terms of currents Figure 6.5 shows a complete block diagram for a current fed induction
plex for a change in flux command. In this case, a change in iZS induces a tran- machine in the rotor flux oriented reference frame defined by eqn (6.2-10). At
sient rotor currcnt isr ( 1 ) which subsequently, decays with the rotor open the left side the input currents in a stator reference are indicated. A transforma-
circuit timc constant L , / r , The induced current prevents any sudden change in tion to rotor flux oriented synchronous frame variables using the rotor flux
tlux and the tlux builds up to the new value exponentially with a time constant angle 8@is shown. The rotor flux angle is obtained by adding the rotor speed
of L, / r, This response is identical to that following a change in ?in a dc orand the slip speed So, and integrating the result. The main portion of the
machine; the induced eddy currents in the field poles being the equivalent of diagram is based on eqns (6.2-15), (6.2-17). (6.2-19) and the torque equation,
the induced rotor current i;,. eqn (6.2-14). The difference between the electrical torque T, and the load
torque TL is then shown as acting on the system inertia to produce the rotor
264 6 Dynamics of Vector Control and Field Orientation Indirect Controllers for Induction Machine Field Orientation 265

6.3.1 Indirect Controller with $1 and i9se * as Inputs

The simplest modification of the controller of Chapter 5 retains the currents
izz and i$ as inputs and only modifies the slip calculator to maintain orienta-
tion during flux changes. It employs the relation between the rotor flux h2r and
the d-axis current given in eqn (6.2-19). In operational form, this relation is

Using this result in the slip relation of eqn (6.2-17) yiclds

which shows that to maintain field orientation the lag in the flux response must
be incorporated in a nonlinear slip calculator based entirely on current com-
mands. The hat or caret in eqn (6.3-2) reminds us that the value used in the
Figure 6.5 Complete block diagram of current fed induction machine in computation is not the actual value but only our best estimate of thc actual
rotor flux oriented refercnce frame (field orientation) rotor time constant. Figure 6.6 illustrates a controller based on this result.
Note that this controller is only a slightly modified form of the controller
s p e d or This signal is then used to complete the diagram. With the perspec- introduced in Chapter 5; the modification being the lag element in rhc slip cal-
tive used in constructing this machine model, the machine is viewed as inter- culator branch involving the flux command current is:. For a fixed value of
nally consisting of a nonlinear slip frequency loop and a non linear torque loop. isf (constant flux operation), this lag has no influence and the controller per-
An indirect field oriented controller must perform the inverse calculations of forms exactly as the simpler controller of Chapter 5. During the transient
the slip loop in order to obtain the field angle Bg from which the appropriate period following a change in iz:, the lag element delays the influence of the
current commands for the machine (in a stator reference) can be obtained. change in iz: to match the actual change in flux in the machine. The actual flux
response follows eqn (6.2-19) and for this reason this controller is sometimes
referred to as an uncompensated$ux response controller.
6.3 Indirect Controllers for Induction Machine Field
Orientation 6.3.2 I n d i r a Controller with k;:and ie* as Inputs
9s
The indirect controller described in Chapter 5 is based on steady state consid- A controller having A.5: and i$ inputs has a fundamen~aldifference from
erations and will not properly maintain field orientation during transients the preceding controller in that the eontroller must calculate the correct value
which involve changes in the flux level. There are several versions of indirect of i$ as well as the slip. This is again accomplished using eqn (6.2-19); in this
controllers which correctly handle flux variations. These systems are described case the relation is used in the opposite sense as in eqn (6.3-1) by solving for
in this section [4]. i-j: in terms of A.2:
266 6 Dvnamics of Vector Control and Field Orientation Indirect Controllers for Induction Machine Field Orientation 267

Figure 6.6 Indirect field orientation controller using input current Figure 6.7 Indirect field orientation controller using flux and torque current
commands (uncompensated flux response) commands (compensated flux response)
*
ure 6.5. With this interpretation the goal is to feedforward the proper Brf to
control hfl,and the torque (via is: ) in the presence of a system disturbanck, in
this case the rotor speed a, (or position 8 , ). Since in this case the disturbance
The slip relation of eqn (6.2-17) is used directly and the resulting controller is
is measurable (via an encoder), feedforward theory would employ the inverse
illustrated in Figure 6.7. of the system transfet function and the measured disturbance to calculate the
This controller is substantially different from the controller of Figure 6.6 in correct 8;. Thus, the inverse of the lag between i:s and A;, is employed
that changes in flux command immediately alter the slip frequency calculation along with the slip frequency calculation to incorporate the rotor position (dis-
and also give rise to a compensation component of i;; . If the required i$ is
turbance) in the feedforward controller. The result is the controller of Figure
within the capability of the .stator voltage supply (and assuming ideal current 6.7. Note that with this interpretation the rotor position signal is not part of a
control) the flux exactly follows the flux command. Clearly, the rate of change feedback controller but is instead a feedforward of information about a system
of the flux command must be limited in any practical system. Note that for a
disturbance.
constant flux command this system is again identical to the simple system of
Chapter 5. 6.3.3 Other Indirect Controllers
The system of Figure 6.7 can be interpreted as an application of feedfor-
There are other forms which can be employed to implement indirect control.
ward control theory to the block diagram of the induction motor given in Fig-
Most of these are variations of the systems in Figure 6.6 or Figure 6.7 altered
268 6 Dynamics of Vector Control and r I:? f ~rientation Indirect Controllers for Induct~onMachine held Ur~entat~on LOY

to provide simpler implementations or modified to compensate for practical

problems associated with real controllers. For example, some controllers pat- e*
z
terned after the system of Figure 6.6 employ the measured currents in the slip ds c I f>J to current
regulator
calculator rather than the command currents. This allows extending the field
oriented mode somewhat into the region where the current regulator is reach- phase lag
ing its practical limits (either amplitude or frequency). A second example is the compensation
insertion of a limiter after the lead compensator for i$ in Figure 6.7 and shift-
to CSI gating
ing the input to the slip calculator to the output of this limiter. A lag element in circuits
the flux branch of the slip calculator is then required to maintain correct field 4 +

orientation. This variation is shown in Figure 6.8.

Figure 6.9 CSI implementation of field oriented control (replaces T-I

block of CRPWM systems)
input labelled phase lag compensation in the figure is included to emphasize
that the commutation delay inherent in CSI operation must generally be com-
pensated to achieve acceptable field oriented control. Note that the implemen-
tation shown in Figure 6.9 is equivalent to the scheme shown in Figure 5.26
where the addition of the angles is carried out in terms of frequencies. The
mangement shown in Figure 6.9 is well suited to digital implementation and
provides more precise control of the angle of the stator current vector relative
to the flux.

6.3.5 Indirect Field Orientation Start Up Transient

The issue of how the rotor flux builds up and becomes correctly oriented with
the indirect field orientation controller from a start with no initial rotor flux and
with only a relative sense of position is an interesting and informative subject
for study [5]. A convenient means for carrying out the analysis is illustrated in
Figure 6.8 Indirect field orientation controller using flux and torque current Figure 6.10 in which the machine is modelled in the actual machine rotor flux
commands with flux command current limiter frame and the controller is modelled in the controller's estimated rotor flux
frame. This figure is essentially a composite constructed from the field oriented
machine diagram of Figure 6.5 and the field oriented controller diagram of Fig-
6.3.4 Indirect Control Using a CSI ure 6.6. The error angle, Ae$ is the angle between the machine and controller
Although the indirect controllers shown in Figure 6.6 to 6.8 are illustrated as axes and will become zero when the machine reaches the correct field oriented
developing command signals for a CRPWM inverter, they can also be state. It is clear from Figure 6.10, and from the development in previous sec-
employed using CSI systems. In this case the transformation block T -1 is tions, that the transient flux build up process is independent of rotor speed and
replaced with a resolver and an angle summation as shown in Figure 6.9. The thus the analysis can be carried out at zero speed for simplicity.
 270 6 Dynamics of Vector Control and Field Orientation Indirect Controllers for Induction Machine Field Orientation 27 1

axis and results in a nearly equal and opposite induced rotor current. With zero
slip frequency the applied dc current remains on the d-axis and the rotor flux
builds up according to the rotor open circuit time constant and aligned with the
flux command current. The build up of the flux is associated with the induced
rotor current which continues to oppose the stator dc current and dies out as the
flux grows. Ideally all of this occurs with no transient in torque, which says a
free rotor would not have moved during the entire process.

Case 2: Sudden application offux and torque command P # 0

Next consider the sudden application of stator current containing both
torque and flux components. While this case is similar to that of only a flux
command, there are distinct differences the most important of which is the
existence of a finite commanded slip frequency. As in Case 1, at the first instant
the imposed stator current excites a nearly equal and opposite rotor current
vector as illustrated in Figure 6.1 l(a). The difference between the two collinear
current vectors creates a small flux in the initial direction of the stator current
vector which, therefore, becomes the initial motor d-axis. The error angle,
Figure 6.10 Induction machine indirect field orientation system for current at the first instant is the angle between the current vector, which is also the
source excitation showing error angle ABg machine d-axis, and the d-axis of the controller. This angle is illustrated in
Figure 6.1 1(a).
In analyzing this transient process the ratio of the applied torque and flux As soon as this initial flux is established it interacts with the commanded
command current components is an important parameter. It is convenient to dynamic slip frequency to create a rotor voltage and an induced current 90"
define the commanded ratio of iq& as P, where away in space from this initial d-axis as a result of the cross coupling between
axes in an induction machine, see eqn (6.2-11). As shown in Figure 6.11(b),
this current interacts with the applied stator current and its induced, collinear
rotor current and tends to rotate the resultant rotor flux and hence the machine
Case 1: Sudden application offlux command only, P = 0 reference frame in the direction to more nearly align it with the controller ref-
erence frame. At the end of the transient period the machine reference frame is
Consider the start up condition where only a flux component (iqse* = 0) of aligned with the controller reference frame as shown in Figure 6.11(c) and only
commanded current is suddenly applied to the machine. Initially both esti- the q-axis component of rotor current remains. The actual transition can be
mated and actual rotor flux angles are taken as zero and since iq/* = 0 and overdamped or underdamped depending on the value of P and the transition
speed = 0, dc is applied to the machine. The rotor flux is also initially zero and time is governed by the rotor open circuit time constant as will be shown later.
cannot change instantly. Mathematically at time zero there is a divide by zero Figure 6.12 and Figure 6.13 illustrate the results of simulations of the tran-
and this results in an undefined '010' slip frequency. However, with the first sient response following the sudden simultaneous application of q- and d-axis
increment of time a finite amount of rotor flux is built up and the slip frequency currents with varying ratios of iq/ids(P=O, 112, 1, 1.5 and 2) and with a constant
becomes zero, 'Olfinite = 0'. Physically, the rotor is at rest and the applied exci- 1.0 pu flux command. The rotor speed was held at zero throughout the simula-
tation is dc, therefore, the first application of current defines the machine d- tions. Observe, in the expanded views of Figure 6.12, that from t = 0 to 20 ps
 272 6 Dynamics of Vector Control and Field Orientation Indirect Controllers for Induction Machine Field Orientation 273

Induced
Rntnr
Current
Induced Induced
Induced Rotor Rotor
Rotor Current Current
Current (decaying) (cornrant)

Rotor f l

(growing)

C o m d
vector Vector
(constant) (comtalu) -
Time Microsecond5

Dashed Arrows Solid Arrows

- -- - - b ____)
Controller Reference Machine Reference

Figure 6.11 Motion of rotor flux and current vectors upon sudden
application of flux and torque commands with the machine
initially at rest
the rotor flux angle quickly builds up as a result of a very high dynamic slip
frequency and then levels off with low slip frequency thereby first producing
and then maintaining predominantly d-axis current. These curves (Figure 6.12) -
Time Micmsecond5
are the simulation equivalent of an impulse and a step function respectively
and are a result of the specific initial conditions chosen for the simulation. Figure 6.12 "Switching transient" associated with invalid initial conditions
The mechanism creating these switching transients is as follows. Since in simulation of field orientation start up transient
both q- and d- components are commanded at the start, the commanded slip
exceeds the commanded rotor flux angle. The difference angle then becomes
frequency is a non-zero value resulting -in a ramp in commanded rotor flux
negative and shifts the commanded stator current vector towards the machine's
angle. The initial conditions in the motor are all taken as zero and hence the
rotor flux d-axis. In doing so, the machine q-axis current decreases and the
rotor flux and rotor flux angle are initially zero. The actual slip frequency in the
result is a reduced machine slip frequency. This shifting of the machine rotor
machine is initially an undefined '0/0'. This can be seen from the machine slip
flux angle is an inherent effect that causes the applied current to shift into the
frequency calculation in the block diagram of Figure 6.10. Assuming that the
machine d-axis to create rotor flux.
controller initially has a zero commanded rotor flux angle, the applied current
As previously noted, this result occurs as a result of the choice of zero ini-
at the first instant is located exactly as commanded and the difference angle is
tial conditions for both the commanded and actual rotor flux angles. The
initially zero. However the slip frequency in the machine immediately
dynamics shown in Figure 6.12 are therefore the dynamics of the simulation
becomes very large (approximating an impulse) as finite q-axis current is
routine and not the true dynamics of the machine and controller. Note that with
applied with initially zero rotor flux. Given this large slip frequency, the actual
the more physical point of view of the previous paragraph, the initial position
rotor flux angle quickly integrates up (approximating a step increase) and
 275
274 6 Dynamics of Vector Control and Field Orientation Direct Controllers for Induction Machine Field Orientation

of the machine d-axis is immediately established as the direction of the initial

current vector; no impulse of slip frequency or step in rotor flux angle is
required. The difference in the two interpretations occurs as a result of the
choice of the initial value of machine rotor flux angle; equal to zero in Figure
6.12 and equal to 80 in Figure 6.1l(c).
Following this initial switching transient the actual dynamics of the initia-
tion of field orientation occur. As illustrated in Figure 6.13, during this period
the q-axis currents are held low, the rotor fluxes build up, and the commanded
rotor flux angle catches up to the machine rotor flux angle after 1 to 2 seconds, I I I I
all according to the rotor open circuit time constant. As the commanded rotor
flux angle catches up, the applied currents gradually shift towards their com- I I I I
I I
-1.2 I I I
manded values, the commanded and the actual rotor flux angles equalize and 0 1 2 3 4
-
Time Seconds Time -(b)
Seconds
steady state field orientation is ultimately achieved. It is important to note that (4
although the commanded slip frequency (which is the actual applied frequency 5.0L
- -+ - -I- - -1- - -4
1 1 1 1
at zero speed) is constant, the actual dynamic slip frequency varies with time
(Figure 6.13(d)). The difference between the two is a result of the transient
swinging of the machine rotor flux axis. The correct definition of slip fre-
quency is the difference between rotor flux speed and rotor speed, not between
steady state synchronous speed and rotor speed. The distinction is crucial dur-
ing transients while the two are identical in the steady state. -
P
Note that the higher ratios of exhibit some overshoot corresponding to $ o , 2 ~ z1 ~ -1i - - -1 ~ - 1- ~ - - ~
the applied vector swinging past the commanded position and then ringing
back in. The commanded slip frequency varies directly with the ratio 0 and it is -1.0 - - - I - - -1, I - J-I - -I I
L I
thus reasonable to expect that the startup transient behavior would also depend 0 1 2 3 4
strongly on this ratio. As shown previously, the case of = 0 results in zero Time - Seconds
. -.- -Command (a
error and the flux builds up exactly where commanded. As the ratio is
increased, the increased commanded slip frequency causes the rotor flux error 100 ,yp, 460volt, 60 Hz, 4 Pole Induction Motor
angle to respond more quickly during the transient period. As a result the tran- Per Unit Parameters
sient tends to become more oscillatory as the ratio P is increased, eventually Stator Resistance = O.lOpu Statorhokage Reacmnce = 0.10
Rotor Resistance = 0.15 pu Rotor Leakage Reactance = 0.10
showing substantial overshoot in flux and torque as illustrated in Figure 6.13. Ma~netizin~ Reactance = 3.0 lnenia Constant (H)= 0.5 Second
Except for this overshoot and ringing at high values of P the plots in Figure
6.13 are in basic agreement with the physical interpretation given previously Figure 6.13 Indirect field orientation start up transient - 100 hp machine
and illustrated in Figure 6.11.

6.4 Direct Controllers for Induction Machine Field

Orientation
As mentioned in Chapter 5, measurement (or calculation) of the rotor flux
 277
276 6 Dynamics of Vector Control and Field Orientation Direct Controllers for Induction Machine Field Orientation

angle directly from machine electrical variables is also possible [2,3].In these
direct controllers, the measurement of rotor position is eliminated by making
other measurements directly on machine electrical quantities. The slip relation
is no longer directly employed.
The advantage of such systems is, of course, the elimination of the rotor
position encoder. Unfortunately, this eliminates the direct knowledge of a sig- Rotor F l u Computer Field Orienter
nificant disturbance to the system, i.e. the motor speed and rotor position.
While there are schemes which can successfully overcome this loss of infor-
mation, they themselves require new sensors which are in many ways less
desirable than the encoder (Hall elements in the air gap, for example). While
schemes which only require voltage and current measurements are also feasi- Motor
ble, they are severely limited at low speed and introduce new and very trouble- Signal
some parameter dependencies. Inputs
The following sections describe the basic principles of direct controllers
and describe several important and useful variations of direct control. Flux Computer & Field Orienter

6.4.1 Direct Determination of Rotor Flux Angle Figure 6.14 Rotor flux computer and field orientation
Conceptually, the measurement of the flux angle requires two steps as illus-
trated in Figure 6.14; computation of the rotor flux components from measured
quantities (rotor flux computer) and computation of the magnitude and angle of
the rotor flux vector (field orienter). The field orienter is simply a resolver,
exactly the same as the resolver introduced in Section 5.5 (Figure 5.13). The
rotor flux computer takes on various forms depending on the motor quantities
chosen as the inputs.
I
J
k r = GdrI
6.4.2 Measurement of Air Gap Flux
Figure 6.15 illustrates direct determination of A, and O6 by sensing air gap flux 0 = A n g ~ ~ Ar AirGap 1
rf qdr
with flux sensing coils or Hall elements. The rotor flux equation in a stator ref- Flux
0
A9 m
erence is d

ydr= 4 C d S+ ~ ~ i ; ~ ~ (6.4-1) Figure 6.15 Field angle determination using flux sensors
The measured air gap flux is given by

hqdm-
- LmqdS
+ i;dr) (6.4-2)
Eliminating the rotor current between these two equations yields
278 6 Dynamics of Vector Control and Field Orientation Direct Controllers for Induction Machine Field Orientation 279

The only computation required is a correction for rotor leakage flux as indi-
cated by the equation given in the figure (plus the resolution to magnitude and
phase). This is the most direct approach to direct field orientation and has the
considerable advantage of requiring only two motor parameters, the rotor leak-
age inductance Ll, which, (except for closed slot rotors), is substantially a con-
stant value independent of temperature or flux level and L,/ L, which is only
moderately affected by saturation of the main flux paths in the machine. The
disadvantage is, of course, the need for special sensing elements in the motor.
If flux sensing coils are used there are also problems with integrating signals at
low frequencies (near zero speed). A special problem occurs in machines with Figure 6.16 Field angle determination from terminal voltage and current
closed rotor slots (a typical construction in cage rotor machines) since the rotor
leakage inductance then is strongly dependent on rotor current, especially at
low values of rotor current. The resulting error in flux angle can be quite large
if some form of adaptation is not employed. This parameter dependency is dis-
cussed in a later chapter.
The rotor flux linkage can then be written as
6.4.3 Voltage and Current Sensing
Special sensors can be avoided by the scheme shown in Figure 6.16 [5]. Since
the measurement now involves the stator voltage, the stator equation must be
employed to solve for the stator flux linkage. In a stator reference, this equation
is

.S
?ids = rslqds + phidS
from which, in operational form
The major difficulty in this case is the need for three motor parameters, r,,
1 s
hids = p (yqds - rsi;ds) L,' and L, / L,,,. The stator resistance is a significant problem because of tem-
perature dependence (and skin effect in large machines), the two inductance
The stator flux linkage expression is parameters are only moderately affected by saturation. There are also problems
with integrating low frequency signals and with the fact that the stator 1R drop
g& = g d S+ L,<,, becomes dominant at low speed. These limitations preclude use of this scheme
which when solved for the rotor current yields at low speed. It is, however, practical over a reasonable speed range and is used
in many implementations. Again a special problem occurs in machines with
closed slot rotors where Ls' becomes strongly dependent on rotor current
because of the influence of saturation on Ll, as noted in the previous section.
 280 6 Dynamics of Vector Control and Field Orientation Direct Controllers for I n d u ~ ~ m
hlachine
i Field Orientation 28 1

A scheme for eliminating the dependence on the stator resistance has been
proposed and implemented [6]. The method involves a winding tap at the mid-
point of the phase belt as shown in Figure 6.17 [7]. Since the two winding sec-

\n3 coil
a-axis

Inputs

Figure 6.18 Direct field orientation scheme

b-axis c-axis
angle adder of Figure 6.9 would replace the transformation block in Figure
6.18.
Figure 6.17 Field angle determination using tapped windings
If the CFO employed provides rotor flux amplitude as well as the angle, the
flux amplitude signal can be used as feedback for a conventional flux regulator.
tions are -resistively identical, taking the difference eliminates the stator resis- In this case the flux compensator lead element can be omitted as was illustrated
tive drop. However, because of the spatial separation between the two winding in the direct field orientation block diagram in Chapter 5 (Figures 5.31 and
sections, there is an induced voltage which can be related to a flux nearly equal 5.32). The flux amplitude and angle can also be combined with current feed-
to the stator flux. A relation similar to eqn (6.4-9) is then employed to evaluate back to provide a torque signal and a closed loop torque regulator can be
the rotor flux linkage. implemented as shown in the Chapter 5 diagrams. Figure 6.19 illustrates the
torque computation in block diagram form.
6.4.4 Implementation of Direct Field Orientation The direct field orientation controllers illustrated in Figure 6.18, 5.31 and
The direct computation of the rotor flux angle as outlined in the previous sec- 5.32 can utilize any of a wide variety of computational schemes in the CFO
tions eliminates the need for employing the slip relation and a measured rotor used to obtain the rotor flux angle. The three examples presented in this section
position signal. The complete controller thus only requires the flux compensa- are intended only as illustrations; new schemes will be developed as other sen-
tion portion of the indirect controller plus the CFO to obtain the rotor flux sors become available. Unfortunately, at present, only Hall sensors provide
angle as illustrated in Figure 6.18. For a CSI based system, the resolver and useful signals near zero speed; the others require integration and are unwork-
 282 6 Dvnamics of Vector Control and Field Orientation Induction Machine Field Orientation Using Air Gap Flux 283

ler is also useful in illustrating the existence of theoretical (and practical) limits
to the torque capability and transient response of field oriented induction
machines.
Motor
The following sectiohs present the analysis and development of an air gap
Signal flux based field oriented controller. Stator flux based field orientation is treated
Inputs in Section 6.8.

6.5.1 Air Gap Flux Referred d,q Equations

The basic synchronous frame d,q equations given in eqns (6.2-1) to (6.2-9)
can be modified to incorporate the air gap flux by using the defining equations
for the air gap flux

Figure 6.19 Torque computation using CFO output and current feedback .e e (6.5-2)
hirn = Lm( lds + id,)

able at low speed. Since Hall sensors in the air gap are considered to be unreli- The stator and rotor flux linkages are then expressed as (in complex vector
able, direct field orientation has not been widely applied for servo drives where form)
zero speed operation is necessary. However, where continuous operation at or
near zero speed is not required, direct field orientation using voltage and cur-
rent sensing is very attractive and will certainly see increasing application.
Some of the problems associated with parameter variations when using this b;,, = &id,+ &iidr (6.54)
method are considered in Chapter 8. where LI, and LI, are the stator and rotor leakage inductances

L,, = Ls- L,, (6.5-5)

6.5 Induction Machine Field Orientation Using Air Gap Flux
The field orientation controllers presented in Sections 6.3 and 6.4 obtain L,, = L, - L, (6.5-6)
instantaneous torque control by controlling the spatial angle between the rotor and the torque can be rewritten in terms of the air gap flux
flux and the torque command current. In the process, the amplitude of the rotor
flux is also a controlled quantity. While this is the usual (and original) choice 3P .e (6.5-7)
for field orientation, it is also possible to create systems which work with a flux Te = 53 ( ~ ; m ~ -9 shimi;,)
other than the rotor flux [4]. If we now align the d,q axes such that the air gap flux is entirely in the d-
Such systems are more complicated than the rotor flux based controller but axis, we can write
have the advantage of working with flux quantities which are directly measur-
able, i.e. air gap flux or stator flux. It is also normal to relate saturation levels in
the machine to air gap flux and hence a controller based on air gap flux is better
suited to treating saturation effects. The study of anon rotor flux based control-
 284 6 Dynamics of Vector Control and Field Orientation Induction Machine Field Orientation Using Air Gap Flux 285

which is the counterpart of eqn (6.2-10) for a rotor flux oriented system. To
distinguish the resulting equations from those of Section 6.2.2, we use a super- (rr+ Llrp) i;s
0,- Or = SO, = f-m
script m to indicate the reference is now the main or airgap flux instead of the
rotor flux. Again assuming the machine is supplied from a current regulated
source, the d,q equations in an air gap flux oriented frame are:

The flux relation is obtained from eqn (6.5-lo), (6.5-14) and (6.5-15)

or collecting terms and solving for pXym

6.5.2 Dynamic Response of Air Gap Flux Controlled Induction Machine

Equations (6.5-9) to (6.5-13) describe the dynamic response of a machine Note that unlike the rotor flux oriented system the slip relation (eqn (6.5-17))
field oriented to the air gap flux (compare with eqns (6.2-1 1) to (6.2-15) for and the flux relation (eqn (6.5-19)) are coupled equations. Note also that these
rotor flux orientation). The torque equation clearly shows that instantaneous equations reduce to the rotor flux oriented equations if the leakage inductance
torque response is possible if the air gap flux is held constant. L1, is taken as zero (but L,,, and L, must still be considered as unequal)!
A development very similar to that of Section 6.2.3 can be carried out to Because of the coupling between the slip relation and the flux relation, a
obtain the field orientation form of these equations with iYs and iX
considered torque production block diagram analogous to Figure 6.2 must include both the
as the input (controlled) quantities. From (6.5-ll), we obtain slip and flux diagrams. Rewriting eqn (6.5-19) in the form

.m
im = -1
'9 9s
and from (6.5-12)

in combination with eqn (6.5-17) leads to the block diagram in Figure 6.20.
m .m
ldr= -- 'ds (6.5-1 5) Clearly the coupling leads to a much more complicated conceptual picture of
Lm torque production based on air gap flux in comparison with the simple, decou-
Using these results, eqn (6.5-9) can be solved for O, - or(the slip relation) pled conceptual diagrams associated with rotor flux control. However, instan-
taneous torque control can still be realized with a controller which maintains
the air gap flux idmconstant as is easily seen from the diagram of Figure 6.20.
 286 6 Dynamics of Vector Control and Field Orientation Induction Machine Field Orientation Using Air Gap Flux 287

Flux Relation Thus, even in the steady state, the coupling persists. The most significant fact is
that the d-axis current Idm, does not independently control the flux as was the
case for rotor flux control. This is easily seen to be the result of the fact that, in
the steady state equivalent circuit of Figure 6.21, the rotor current has a d-axis
component because of the leakage reactance of the rotor. The situation is illus-

Figure 6.21 Steady state complex vector equivalent circuit for air gap flux
control
Slip Relation
trated in the vector diagram of Figure 6.22 where the magnetizing current Gs
Figure 6.20 Torque production in air gap flux controlled induction machine clearly has two components, Idm, and jldm,. Thus the stator d-axis current must
supply the magnetizing current plus a component to cancel the rotor d-axis
component. This d-axis rotor current increases as the slip frequency increases
and hence the stator d-axis current must also increase if the magnetizing cur-
6.5.3 Steady State Interpretation of Air Gap Flux Control
rent (and air gap flux) is to remain constant.
Before outlining a controller for air gap flux control, it is useful to examine the The steady state relations of eqns (6.5-21) and (6.5-22) can be easily
steady state behavior of the system of Figure 6.20. The steady state forms of derived from the steady state equivalent circuit of Figure 6.21. The exercise is
the slip and flux relations are left to the reader.

6.5.4 Indirect Field Orientation Controllerfor Air Gap Flux

An indirect (feedforward) controller for air gap flux control must clearly
implement the slip relation of eqn (6.5-17). The second requirement is to
m*
implement a means of evaluating the d-axis control current ids necessy to
control the flux hym.This can be accomplished by considering the flux hym as
an input and solving the flux relation, eqn (6.5-20), for the required &*.
From eqn (6.5-20), the result is
 ~ 8 8 6 Dynamics of Vector Control and Field Orientation 289
Induction Machine Field Orientation Using Air Gap Flux

Im
- qdr

L zm
h - qdr

Figure 6.22 Complex vector diagram for air gap flux control Figure 6.23 Indirect field orientation controller for air gap flux control

needed for the computation of the command value of d-axis current. The CFO
computation, however, no longer requires the rotor leakage inductance; in the
case of Hall sensors in the air gap, no calculation is required at all: For stator
flux sensing, the air gap flux is obtained simply as

The resulting indirect controller is shown in Figure 6.23.

The air gap flux controller is clearly much more complicated than the rotor
Since the rotor leakage inductance is a special problem in machines with
flux controller of Figure 6.7 and requires two additional parameters; the leak-
closed rotor slots, the air gap flux based controller has an advantage with
age time constant 21, and the ratio of the rotor leakage inductance to the rotor
regard to them*CFO computation. The rotor leakage inductance does enter into
self inductance. Note that if these two parameters are set equal to zero, the air
the slip and ids computation, but these computations are less important at light
gap flux controller of Figure 6.23 reduces to the simpler rotor flux controller of
load where the rotor leakage inductance tends to be highly variable. A closed
Figure 6.7.
loop flux regulator utilizing the flux magnitude from the CFO as the feedback
6.5.5 Direct Field Orientation Controllerfor Air Gap F l u signal can also be employed to help reduce the sensitivity to changes in rotor
leakage inductance.
If direct control is implemented using air gap flux, the summation of slip angle
0, with measured rotor angle Or is replaced by the output of the CFO produc-
ing the rotor angle Orf . In this case the slip relation is still required since it is
 290 6 Dynamics of Vector Control and Field Orientation Dynamics of Synchronous Machine Vector Control and Field Orientation 29 1

6.6 Dynamics of Synchronous Machine Vector Control and

Field Orientation
With the background of the previous sections and the basic vector control con-
cepts presented in Chapter 5, the dynamic response of vector controlled syn-
chronous machines will now be examined. Because there are practical systems
which do not employ the 90" spatial relationship of flux and MMF used in field
orientation, the influence of departures from true torque control will also be
considered.
The dynamic model employed is the d,q machine model in a rotor refer-
ence frame (Park's equations). A full salient pole machine with dampers is used
to allow the most general treatment.

6.6.1 d,q Model of SynchronousMachine

The d,q equivalent circuit model of a salient pole synchronous machine is illus-
trated in Figure 6.24. From Chapter 2 the circuit is defined by (the superscript r
to denote the rotor reference frame and the prime to signify the turns ratio
referral are omitted for simplicity)

~ i g u r 6.24
e Equivalent circuits of three phase salient pole synchronous
machine

which places in evidence the three torque components

hfr = Lfr ifr + Lmdids+ Lmdidr 1) Lnul(if,+ id,.) iqs - field plus d-axis damper winding torque
(6.6-5)
2) Lmqiqr ids - q-axis damper winding torque
Note that each of the flux linkage equations as well as the voltage loop differ-
ential equations can be written by inspection of Figure 6.24. 3) (Lds - L ~ idr~ iqs) - reluctance torque (saliency torque)
The torque equation is
6.6.2 Vector Control and Angle Control
The concept of rotor position feedback and vector control of the machine stator
current to maintain the space angle between the field winding and stator MMF
or by substituting for the flux linkages in terms of currents
results in stator currents which translate to controlled values of iqs and id, in
the rotor reference frame. This is a result of the instantaneous control of the
phase of the stator current to always maintain the same orientation of the stator
MMF vector relative to the field winding in the d-axis of the d,q model. The
 292 6 Dynamics of Vector Control and Field Orientation Dynamics of Synchronous Machine Vector Control and Field Orientation 293

resulting axis currents are illustrated in Figure 6.25. The concept of angle con- 6.6.3 Dynamics of Synchronous Machine Field Orientation
The dynamics of the field oriented case (y = 0) are extremely simple if we
q-axis assume:
1) the stator current is the independently controlled input variable, and
2) the orientation of the stator d,q currents is maintained for all speeds
including transient changes.
With p O , there is only a q-axis stator current component, i.e. ids = 0 for all
transient conditions. Under these conditions the rotor voltage equations and
flux linkage relations reduce to:
q-axis damper circuit
d-axis

Figure 6.25 Synchronous machine currents in d,q axes d-axis damper circuit

= (6.6-9)
rdridr+ phdr = 0 'dr Lmdifr + L d d r

trol (and field orientation) is that the controlled current supply to the machine field circuit
maintains this condition for transient changes in machine speed as well as
r frifr +phfr = O If. = Lmdidr+ fri fr
(6.6-10)
under steady state conditions.
Compared to induction machine vector control, the currents iqs and id, in and the torque expression reduces to
the synchronous machine are analogous to iis and iis in the induction
machine. The rotor flux axis in a synchronous machine is the field winding axis
which is also the d-axis of the rotor referred Park equations. The current id, is,
therefore, the stator current component in the rotor flux axis. The current iqs is
the torque component of stator current just as in the induction machine. 6.6.3.1 Constant Field Current Operation (Constant Torque Region)
As shown in Chapter 5, under steady state conditions the currents iqs and With a fixed dc voltage applied to the field winding, the solution of Eq's. (6.6-
ids are constant quantities and the damper currents iqr' and idr are zero. Dur- 9) and (6.6-10) is simply
ing transients the currents may vary with time and, in general, the damper cur-
rents are non-zero. The field current can also vary with time.
In the remaining subsections of this section it is assumed in all cases that
and
the stator current to the machine is directly controlled such that iqs and idsare
independent variables. The first case treated is field orientation; subsequent
sections deal with the situation existing when the 90" spatial orientation
between the d-axis and the stator MMF is purposely altered.
This is the same as for steady state conditions and is a direct result of the
absence of any d-axis stator current. The d-damper circuit and field circuit are
 294 Dynamics of Synchronous Machine Vector Control and Field Orientation 295
6 Dynamics of Vector Control and Field Orientation

simply a pair of coupled coils in the d-axis without any interaction with the
stator as a result of the "field orientation". There is complete "decoupling" of
the d-axis from the stator windings.
The q-axis equations can be combined and written in a form to emphasize
iqs as an input quantity to yield

Figure 6.26 Torque production - field orientation (y = 0) with constant

field excitation

Thus, a change in iqs will induce a transient q-axis damper current which
6.6.3.2 Variable Field Excitation (Field Weakening Region)
will have an initial value of
If the field current is changed, there will be an induced d-axis damper current
which can be evaluated by solving eqn (6.610) for id, with if, treated as an
input. The result is
where Aiqs is the change in iqs. This transient q-axis damper current will decay
with the self-time constant (open circuit time constant) of the damper,
L 9 r / $ r . There will, however, be no torque produced by this transient q-axis
damper current because there is no d-axis stator current. The torque for the This d-axis transient damper current will affect the torque since it will react
case of constant field excitation is simply with the stator q-axis current. A block diagram representation is given in Fig-
ure 6.27 along with a sketch of the response to a step increase in if,Once again

a result identical to steady state (since Ld I f is equal to E 1 or). Thus, the

torque response for field orientation is instantaneous and follows the com-
manded value of iqs exactly. This result is represented in block diagram form in
Figure 6.26 [6]. Note that this is the same as for a dc machine - in fact it is a
result of exactly the same physical phenomenon, the reaction between two
fixed current distributions. In the dc machine both are stationary. In the field
oriented synchronous machine both rotate in fixed relation to one another at
rotor speed.
Figure 6.27 Torque production for a change in field current in a field
oriented synchronous machine (H,ids=O)

this is identical to the situation in a dc machine since a step increase in dc

 296 6 Dynamics of Vector Control and Field Orientation ~ ~ n a m iof
c sSynchronous Machine Vector Control and Field Orientation 297

machine field current will induce currents in the field pole iron which must die
away before the field flux can rise to its new value. This is precisely what hap-
pens in the field oriented synchronous machine except the induced current is
treated as being in the damper circuit. Note that the lag in the torque response
is in addition to any lag in buildup of the field current resulting from an imper-
fect field current regulator.

6.6.4 Dynamic Response With ')stO (Angle Control)

To illustrate the more complex situations that result if the 90" orientation of
stator MMF and field MMF is not maintained, consider the case where y is
controlled to an angle other than zero [8]. In this case ids will not be zero. To
keep things relatively simple, consider the case where id, is held at a constant
value

ids = Ids (6.6-19)

-
and the torque is still to be controlled by controlling iqs. Under these condi-
tions, the expressions for iqr and id, in eqns (6.6-15) and (6.6-18) are still
valid. The torque expression, however, becomes

Figure 6.28 Torque production with y #0 for the case of constant lh

The q-axis transient current iqr(t) now does produce torque and the q-axis
The resulting block diagram is shown in Figure 6.29. Note that this dia-
stator current produces reluctance torque ((Lds - Lq,) iqs Ids). Figure 6.28 illus-
gram can be considered as the general torque production model for a current
trates the torque production in block diagram form.
fed synchronous machine; all of the other cases can be obtained by specializing
It is interesting to note that for the practical case of a negative Id, (associ-
this general model for the particular case of interest.
ated with operation at leading stator power factor for load commutation or field
It is important to note that operation with ids + 0 results in a strongly non-
weakening for example) both the saliency torque and the q-axis damper torque
instantaneous torque response only for changes in ids (or if,)When both ids
would be negative causing a lag and a reduction in the torque response result-
and i$. are held constant, the instantaneous torque becomes
ing from an increase in iqs.
If the d-axis stator current is also considered as a variable, the torque pro-
duction block diagram is even more interactive than shown in Figure 6.28. The
additional term represents the transient d-axis damper current resulting from
which is linear, except for the relatively small term involving iq,, and yields
changes in id,. From eqn (6.6-9) with the d-axis stator current added in to the
flux linkage expression for 3Ld, (see eqn (6.64))
nearly instantaneous response for changes in iqr Thus angle control (w)
retains this aspect of field orientation when the d-axis currents are held con-
stant, even for a salient pole machine.
298 6 Dynamics of Vector Control and Field Orientation Dynamics of Synchronous Machine Vector Control and Field Orientation 299

c) Same as (b), but yo = -20".

d) Find the torque Te(t) following a step input of iq: (rated value) for cases (a),
(b), and (c). Express time in seconds.
Solution:
a) At y = 0, ids = 0
the steady state terminal relation is
E + jXdIds + jXqsIqs + RsIqs+ RsIds = V
and the phasor diagram is

Figure 6.29 Torque production with y # 0 for the general case

From the phasor diagram

6.6.5 Example - Synchronous Machine Field Orientation
A 100 hp, 460 volt, 3 phase, 4 pole synchronous motor with the following per lV12 = (XqslIqsp2+ (IEl +rspqsl)2
unit parameters The general torque relation is, in per unit, Te = (Ads iqs - hqs ids)
XlS= 0.1 XV=0.2 rs = 0.04 but, since Ad, = LIdridr,+Lmd(ids + idr + if)
X d 1.1 xldr= 0.1 rdr= 0.04 and id, = 0 in steady state 4 Ads = Ld $(recall Id,= 0)
X, = 0.3 Xlqr=0.15 r, 0.08
rk = 0.01 ATe = LmdlfIqs

is operated as a field orientation controlled machine. For this example assume But IEl=XdIfsincem= l.Opu, IEl=mLmdIf 4 H=Lmd$
an ideal current regulator. T , = IEI lIqsJ
.a.

a) For field orientation ( y = O), find the rated current (i:, = I), rated internal Substituting,
voltage E, and the terminal power factor for operation with rated terminal volt-
age, rated current, and rated frequency. IVl2 = ~ x q s l I q sIqsl
l~~+[~+rslIqsl~2
b) Assume the encoder is incorrectly aligned such that yo = 20". Find and plot Given: V= 1.0, Te = 1.0, Xq = m(Lmq+ Lls) =1 (0.3 + 0.1) = 0.4
T, vs. I& from zero to rated value, also find the terminal voltage at rated
I = I;s.
 300 6 Dynamics of Vector Control and Field Orientation Dynamics of Synchmnouo Machine Vector Control and Field Orientation 301

To find the torque expression,

Te = LmJf 1, + (Lds - Lqs) Ids',

0 = 0. 1616)1qs14+ (-0.92) 1IqS1+ 1
Solving,
Te = LmdIfl;, cosy,+ (L,-Lqs) (I& )2sinyocosyo

Lmd = 1.1 If = 0.751 Lds = 1.2 = Lmd+ Lids

L, = 0.4 = Lmq+ Llqs

(
Since, E = 1- 1.0= 0 . 8 2 0 p u
~,)~E=m This can be plotted for 0 <
To find the voltage,
1iS< 1.22 pu as required.
The power factor is
- = E + jXdsIds + jXqsIqs+ rSbs+ rsIqs
V
Substitute

V = 0.82 + j ( 1.2) (- jl& sin yo) + j (0.4) ( I*,,cosy,)

+ (0.04) (-jl& sinyo) + (0.04) (I& cosy,)
[Note: Neglecting resistance will lead to a >lo% error in E and in II,I in this
Y = 0.82 + (1.2) (P, siny,) + (0.04) (1; cosy,)
example]
+ j [0.41$ cosy, - 0.041:, siny,]
b) Assume the encoder is incorrectly aligned such that yo = 20". Find and plot T
vs. I*,, from zero to rated value, also find the terminal voltage at rated For yo = 20°, s:I = 1.22, V= 1.44 L 17.9'
I = I.,, .
I f is the same (given) 4 If = o'820
= 0.751 pu

Ids= Is
: sinyo and Iqs= I& COSY,

V = 0.82 + 1.21*9, sinyo + 0.041& cosy,

+ j [0.41*, cosy, - 0.04qs siny,]
4
V = 0.599 L 52.5"
-
d) Find and sketch T,(t) following a step input of iis (rated value) for cases (a),
@), and (c). Express time in seconds.
Length is 1
;
The relevant equations are

vqS = rsiqs + pLqs orhds Lqs = L,iq, + Lm, (iqs + i,,)

302 6 Dynamics of Vector Control and Field Orientation Dvnamics of Svnchronous Machine Vector Control and Field Orientation 303

Since

-Lmd~ids + -Lmd~if
ldr =
'dr + L d r ~ 'dr + Ldrp
In general, in pet. unit,

Te = ' d s iqs - ' 4 s ids

Te = [ LlSids+ Lmd( i d s + idr+ i f1I iqs- [ Llsiqs+ Lmq( iqs+ iqr) I ids
I Vector Diagram for yo = 20"
Te = Lmd ( i f + id') iqs- Lmq (iqrids) + ( L m d - L m g ) idsiqs
Let iqs = i& COSY,U ( t ) and ids = izs sinyou ( t ) and note that p i f = 0 .
In Laplace Transform notation:

i:s cosy, i& sin yo

I q r ( ~ )= s Id'(s] =
Solve for I&), Idr(s)

-Lm9s iq*s cosy, -Lmds ig*, sinyo

'4' ('1 = r
9'
+L~,S S Idr(s) = rdr + Ldrs S

Vector Diagram for yo = -20"

-Lm& COSY, -LmdizS sinyo
Iqr ( s ) = rqr+ L4' s IdrW =
'dr + Ldrs
Inverting these Laplace Transforms

R
L .-Art --
Rd',
Lmd
i9' ( t ) = - Q i * C O S Y , ~ idr( t ) = --i* 9s sinyoe Ldr
L4' 9s Ldr
This result fits the torque expression. Substituting,
 304 6 Dynamics of Vector Control and Field Orientation 305
Field Orientation Using Voltage as the Controlled Variable

D-Damper torque
where the stator transient inductance

sinyocosyo e Ldr Q-Damper Torque

-"'
+ (Lmd- Lmq)i$ sin yo cosyo Reluctance Torque has been introduced to simplify the notation. Using these expressions in the
The time constants are: stator voltage equations results in the desired voltage equations in terms of
rotor flux linkages.
Ldr (p") 'base/@base - Ldr (PU) -- (1.1 +0.1) = 79.6 ms
rdr (PU) 'base @baserdr (pU) (377 S ~ C - ' ) (0.04~)

6.7.2 Decoupling Equationsfor Field Orientation

6.7 Field Orientation Using Voltage as the Controlled Variable
For a field oriented system these equations can be simplified since the field ori-
Although the stator currents are the variables which must be controlled to
entation forces
implement field orientation, it is possible to employ the stator voltage equa-
tions and indirectly control the currents by controlling motor terminal voltage.
To accomplish this task it is necessary to 'decouple' the voltage equations so as
to be able to independently control the two components of stator current. With this constraint the stator voltage equations become

6.7.1 Stator Voltage Equations in Terms of Rotor Flux - Induction Machine

The first step in developing the voltage decoupling equation is to rewrite the
stator equations given in eqn (6.2-1) and eqn (6.2-2) in terns of rotor flux.
This is accomplished by solving the rotor flux linkage equations (eqns (6.2-8)
and (6.2-9)) for the rotor currents and substituting the results into the stator
flux linkage equations (eqns (6.2-6) and (6.2-7)) to yield Since the currents izs, iis and the flux IZr are command inputs or calculated
from command inputs, these two equations can be used to calculate the
required voltages. The computational scheme is illustrated in Figure 6.30.
 307
306 6 Dynamics of Vector Control and Field Orientation Field Orientation Using Voltage as the Controlled Variable

can be used in eqn (6.7-7) to yield

(6.7-10)

= ( rs + ~ ; p s:i) +o ~ L ~ I ~ ~
and eqn (6.7-8) reduces to

The corresponding computational block diagram is shown in Figure 6.3 1. Note

'ds

Figure 6.30 Computation of voltage commands for voltage controlled field

orientation in induction machines (voltage decoupler)

The computation to decouple the voltages as shown in this figure is appli-

cable to both direct and indirect field orientation. The only difference is in the
method by which the various command quantities are calculated. In indirect Figure 6.31 Computation of voltage commands for voltage controlled
control with e*
and iqs as input commands, 6.): and €I*are obtained from field orientation in induction machines with constant flux
:f (constant flux voltage decoupler)
the slip relation whereas in direct control they are obta~ned from the CFO.
Thus, although hardware implementation will offer different practical prob-
the close similarity of this model to a dc machine operated with constant field
lems, the general concept is independent of the method employed in the basic
current; the only difference is the cross coupling term o,~;i;, in the d-axis
field orientation scheme.
equation.
For systems in which the flux is constant (no field weakening option), the
decoupling equations are greatly simplified. With a constant flux, the steady
state relation
 308 6 Dynamics of Vector Control and field Orientation 309
stator Flux Based Field Orientation
/

6.7.3 Examples of Field Orientation Using Voltage Controlled Inverters need for knowledge of the stator resistancelhis system could again be imple-
To illustrate the overall system configuration of field orientation controllers mented using a six step VSI inverter rather than a PWM inverter.
using voltage controlled inverters, two examples are presented. Figure 6.32 is a
block diagram of an induction machlne indirect controller combining the basic
indirect field orientation controller of Figure 6.7 and the voltage decoupler of Voltage

Figure 6.30. The flux compensation block and the slip calculator are retained to
Decoupler

Fig. 630
Regularor

b
Decoupler Vbs -
4
as
from
of 'b
e* *
vCS
- T
.e
j current
i Fig. 6.31
9s i sensor
I- S

*
' e" 4 A*
0
from

P rf CFO

- G
[
1 7 7 , *
Slip
Calculator
see - . 0; Figure 6.33 Induction machine direct orientation controller using flux
sensors and a voltage controlled inverter
P

Figure 6.32 Indirect field orientation controller using voltage

controlled inverter
6.8 Stator Flux Based Field Orientation
provide the required input quantities for the voltage decoupler. If only constant Rotor flux field oriented induction machine drive systems have been shown to
flux operation is desired, the flux compensation block can be omitted and the offer high performance as well as independent control on torque and flux.
simpler voltage decoupler of Figure 6.31 can be used. The system of Figure Efforts have been made to eliminate flux sensors in a direct field oriented
6.32 could be used with a six step VSI inverter controlled to provide funds- (DFO) system or position sensors in an indirect field oriented (FO)system
mental frequency output voltages corresponding to the voltage commands v,, , without sacrificing performance. A commonly used method is to estimate rotor
'bs and 'cs . flux horn terminal quantities. For example, the rotor flux can be estimated by
The second example, shown in Figure 6.33, is an induction machine direct the following equations:
controller with external flux and torque regulators. Because of these external
loops. the decoupler can be simplified by omitting the stator resistance-tran-
sient time constant blocks in the current feedforward paths thus eliminating the
310 6 Dvnamics of Vector Control and Field Orientation Stator Flux Based Field Orientation 31 1

stator flux. Figure 6.34 illustrates the spatial relation between the two reference
frames.
where Rs is the estimated stator flux vector, a, is the estimated rotor flux vec- To obtain the machine equations in the stator flux reference frame, one can
mathematically eliminate the rotor flux as a variable from eqns (6.2-1) through
tor and
(6.2-9) and then force the q- component of the stator flux to be zero similar to
the procedure used for air gap flux in Section 6.5. Assuming again that the
machine is effectively supplied by a current source, the resultant equations can
be written as follows:
is the leakage factor.
Equation (6.8-1) shows that the accuracy of the estimated stator flux
depends on the accuracy of the estimated stator resistance. Usually, the stator
resistance can be measured with reasonable accuracy and it is relatively easy to
adapt to its slow variation with temperature. Therefore, it is reasonable to
assume that the stator flux can be estimated accurately at least at relatively high
speeds where the stator IR drop is relatively small compared to the amplitude
of the applied voltage. Here, the superscript es is used to designate that the quantity is expressed in a
On the other hand, eqn (6.8-2) shows that estimation of rotor flux requires synchronous reference frame fixed to the stator flux vector. The quantity o is
knowledge of the inductances of the machine, especially the leakage induc- the couplingfactor as defined by eqn (6.8-3).
tance. It has been reported that for a standard induction machine with closed
rotor slots, the conventional no load and locked rotor tests do not give the cor- 6.8.2 Design of a Decoupler for Direct Stator Flux Oriented System
rect values for machine inductances. It is also observed that due to the closed Comparing eqn (6.8-5) to (6.2-12) indicates that there now exists a coupling
rotor slots, the inductances vary with operating conditions [9],[10]. Hence, it is between the torque producing component of the statorescurrent i:: and the sta-
difficult to tune the leakage parameters in the rotor flux estimator and adapt for
their variations. Some error in the estimated rotor flux can always be expected.
tor flux linkage. As a consequence, any change in iqS without changing
accordingly will cause a transient in the stator flux. In order to overcome this
iz
Structurally, the flux estimator is in the feedback path of the closed loop sys- disadvantage, a decoupler can be designed to compensate the coupling effect.
tem. From the view point of control theory, the performance of a feedback con- This approach is depicted in Figure 6.35.
trol system relies on the accuracy of the feedback signal. To obtain accurate The action of the decoupler can be expressed as follows,
and robust control characteristics, the most accurate signal should be chosen as
the feedback signal. Hence, it is natural to consider using the stator flux as the
feedback signal, which leads to the implementation of a stator flux oriented
system. es
where idq is the output of the decoupler and it is assumed that a proportional
plus integral (P-I) regulator is used to control the stator flux linkage. Substitut-
6.8.1 Mathematical Model of a Stator Flux Oriented Induction Machine ing eqn (6.8-7) into (6.8-5) yields:
In the mathematical model of a rotor flux oriented system (eqns (6.2-11)
through (6.2-14)), all variables are expressed in a reference frame oriented to
the rotor flux. With the stator flux chosen as the measured quantity, it is more
convenient to express the machine model in a reference frame oriented to the
 312 6 Dynamics of Vector Control and Field Orientation
- References 313

In order to decouple the d-axis stator flux from the q-axis stator current, it
qs-axis is required that

ds-axis

where SO, is determined from eqn (6.84),

dr-axis

Figure 6.34 Rotor and stator flux linkage reference frames

Equations (6.8-10) and (6.8-1 1) describe a decoupler which, when prop-

. es
- erly tuned, generates the correct value of i z so that for a specified value of
T;-f?Fl+
A D
I '9s

Synchronous
ii:, the flux linkage A: = As is not altered when torque changes are
demanded by the torque command Te* . The concept is similar to the decou-
to pling relation in eqn (6.5-23) and Figure 6.23 for air gap flux field orientation.
Stator Frame
es Decoupler ~ ~ ~ ~ Note that f since "p''~occurs in ~the numerator
~ of eqni (6.8-ll), ~the decoupler
~
and
* CRPWM requires the use of a differentiator which introduces noise into the current regu-
hds lator. However, since most controllers are today implemented in digital form,
bS d modem numerical algorithms can be used to moderate this effect.
4

Stator Flux
References
Estimator
111 K. Hasse, "Zur Dynamik Drehzahlgeregelter Antriebe Mit Stromrichter gespeisten
Asynchron-Kuzschlublaufermaschinen" (On the Dynamics of Speed Control of a Static
Figure 6.35 Block diagram of stator flux based field orientation system AC Drive with a Squirrel Cage Induction Machine), Ph.D. Dissertation, Techn. Hochs-
chule D m s t a d t , 1969.
[2] F. Blaschke, 'The Principle of Field Orientation -the Basis for the Transvector Control
of Three-Phase Machines", Siemens Zeitschrift, Vol. 45, No. 10, pp. 757-760, 1971.
131 F. Blaschke, "Das Verfahren der Feldorientierung zur Regelung der Drehfeldm-
aschine"288288 (The Method of Field Orientation for Control of Three Phase Ma-
chines), Ph.D Dissertation, University of Braunschweig, 1973.
[4] R. DeDoncker and D.W. Novotny, 'The Universal Field Oriented Controller," IEEE-
IAS Trans., Vol. 30, No. 1, JanuarylFebruary 1994, pp. 92-100.
[51 N.R. Garrigan and D.W. Novotny, "Start-up and Sensor-Parameter Error Transients in
314 6 Dynamics of Vector Control and Field Orientation problems 315

Field Oriented Induction Machines," Proc. IEEE-IAS Annual Meeting, October 1995.
P.L. Jansen and R.D. Lorenz, "A Physically Insightful Approach to the Design and Ac-
Problems
curacy Assessment of Flux Observers for Field Oriented Induction Machine Drives",
Problem 6-1 Induction Machine Field Orientation
Proc. IEEE-IAS Annual Meeting, October 1992, pp. 570-577.
D.Zinger, F. Profumo, T.A. Lipo and D.W. Novotny, "A Direct Field Oriented Control- A 100 hp, 460 volt, induction machine is operated from an ideal controlled cur-
ler for Induction Motor Drives Using Tapped Stator Windings," IEEE Trans. on Power rent electronic converter using field oriented control as illustrated in Fig. 6.6.
Electronics, Vol. 5, No. 4, October 1990, pp. 446-453. The pu parameters of the machine are:
R. Lorenz and D.W.Novotny, "A Control Systems Perspective of Field Oriented Control xls = 0 . 1 0 ~ ~ x,, = 2.0pu (at rated freq.)
r, = 0.015pu
for AC Servo Drives", Proceedings of Control Expo 88, June 1988, pp 18-1 to 18-12.
T. Irisa, S. Takaka, R. Ueda and T.Sonoda, "On Reliability of Induction Machine for
High Performance Based on Parameter Characteristics", IEEE-IAS Annual Meeting r,. = 0.020 pu xlr = 0. lOpu rm = 50 pu (at rated freq.)
Conf. Record, 1983, pp. 547-554. The rated slip frequency is 0.0248. Express all answers in real units.
R. Ueda, T. Sonoda, K. Fujitani, Y.Yoshida and T. Irisa, "Investigation of Induction Mo-
tor Characteristics by Means of Vector Control", IEEE-IAS Annual Meeting, Conf. a) Find the required values of i,, and ids and the slip relation relating
Record, 1985, pp. 578-585. So,, i and hdrto produce operation at rated torque and rated speed if
QS.
the termnal voltage and frequency are held at their rated values.
b) Find the final steady state torque and slip frequency if ids is reduced
to one half its value in part a) while iq, is held constant. Find the terminal
.
voltage and stator frequency if the new speed is twice that in part a).
C)Find the transient response of the torque for part b) assuming the cur-
rent regulator can change ids instantaneously.

Problem 6-2 Synchronous Machine Field Orientation

A 200 hp, 460 volt, 3 phase, 4 pole, 60 Hz synchronous motor with the follow-
ing per unit parameters

X,, = 0.1 Xu = 0.2 r, = 0.03

is operated from a current regulated variable frequency supply. For this prob-
lem assume an ideal current regulator.
a) For field orientation 0 = (y) with rated terminal voltage V=1.0 pu,
.S*
current = i qs = 1.0 pu, and rated frequency f = 1.0 pu, find the
torque, internal voltage E, and the terminal power factor.
316
6 Dynamics of Vector Control and Field Orientation
-
b) For unity power factor operation at 60 Hz, V=l.O pu, with the same
,
torque as in part a), find the required E, I, is* , r.s*ds and y.
c) Find and sketch T (t) following a step input of is:$ (step from zero to
that value which yields rated torque) for parts a) and c). Give the torque 7 Current Regulation in Power Converters
in pu but express time in real units. Assume is*s remains constant in
both cases.
7.1 Introduction
The power converter in a high performance ac drive essentially functions as a
power amplifier, reproducing the low power level control signals generated in
the field orientation controller at power levels appropriate for the driven
machine. Typically the input power to the converter is three phase, fixed fre-
quency, fixed voltage power from the utility power grid. The output must be
variable frequency, variable voltage power with the frequency and voltage
range appropriate to the intended operating range of the drive.
At present, except for very low speed drives where line commutated cyclo-
converters are employed, the power converters used in high performance
drives are dc link inverters. Developmental work on direct converters using
forced commutation and bilateral switches as well as resonant link converters
utilizing a high frequency single phase link rather than the usual dc link are
underway. However, application to commercially available drives is several
years away. The material in this chapter will be primarily concerned with dc
link inverters with only passing mention of other types of power converter
arrangements.
Although field orientation controllers which provide voltage commands are
feasible (see Section 6.8), most systems provide current commands and thus
require power converters which function as controllable current sources. Such
systems contain inner current loops very much like the current loop in a high
performance dc drive. The primary difference is that the current control
required in an ac drive must provide both amplitude and phase control. This is,
partly, the origin of the term vector control as applied to high performance ac
drives.

7.2 Current Regulated Inverters

As noted in the introduction, high performance drives utilize control strategies
which develop command signals for the ac machine currents. The basic reason
for the selection of current as the controlled variable is the same as for the dc
 318 7 Current Regulation in Power Converters Current Regulated Inverters 319

machine; the stator dynamics (effects of resistance, inductance and induced

emf) are eliminated. Thus, to the extent that the current regulator functions as
an ideal current supply, the order of the system under control is reduced and the
complexity of the controller is significantly simplified.
Current regulators for ac drives are more complex than for dc drives
because an ac current regulator must control both the amplitude and phase of 1 CSI
I ib

the stator current. In addition, the steady state currents are ac currents not dc
currents so the straight forward application of conventional proportional-inte-
gral control (PI control) as applied in dc drives cannot be expected to provide
performance comparable to that of a dc drive current regulator. As in the dc
drive, the ac drive current regulator forms the inner loop of the overall control-
ler. As such, it must have the widest bandwidth in the system and should have
zero or nearly zero steady state error. Achieving these goals in ac current regu-
lators has proven to be a challenging task and it is only recently that satisfac-
tory, fundamentally sound techniques have been developed.
Both CSI and PWM inverters can be operated in controlled current modes;
the VSI is not readily applicable and would offer no basic advantage over a
PWM inverter. The CSI is a 'natural' current supply and can readily be adapted
to controlled current operation. The PWM inverter requires more complexity
in the current regulator but offers potentially much higher bandwidth and elim- Figure 7.1 CSI inverter with dc link current regulator as a three phase
ination of current harmonics as compared to the CSI. Since the concepts regulated source
involved are much simpler in the CSI it is treated first with the current regu-
lated PWM inverter treated subsequently as a solution to some of the inherent treated in many references and will not be considered in detail here. While a
problems associated with the CSI. diode input rectifier and dc chopper could be used in place of the phase con-
trolled rectifier, this would eliminate the inherent regeneration capability of the
7.2.1 The Current Regulated CSI Inverter system. Although the chopper could potentially provide increased bandwidth,
The fundamental implementation of current regulation in a CSI is very simple other inherent problems of the CSI would not be overcome and the loss of
in concept requiring only the addition of a dc link current regulator as illus- regeneration capability is a serious drawback. For these reasons, diode recti-
trated in Figure 7.1. The command input I: is the reference input correspond- fier-chopper systems do not appear to be an important option.
ing to the amplitude of the stator current. The phase of the output current is While, in principle, phase angle control via thyristor firing time offers
determined by the firing times of the CSI thyristors and thus the phase current instantaneous phase control, there is a significant limitation imposed by the
command, ~ j ,: is shown as a direct input to the CSI inverter block. commutation delay inherent in the CSI. Unfortunately, this delay is both cur-
Implementation of the dc link current regulator is essentially the same as rent amplitude and frequency dependent. Thus, implementation of compensa-
for a dc drive current regulator since the current being regulated is a dc current tion for this delay is a significant problem. The principle is illustrated in Figure
and regulation is accomplished via the firing angle of the input phase con- 7.2 where compensation inputs for both amplitude and frequency are shown.
trolled rectifier. Ordinary PI-control can be applied since the regulated quan- Although reasonable compensation can be achieved, the non-linearity inherent
tity is dc in the steady state. Regulator performance and design is adequately in the commutation delay makes this task a difficult one.
 320 7 Current Regulation in Power Converters Current Regulated Inverters 32 1

CSI systems with six-phase (30" displacement) machines can be employed to

produce 30" steps in the mrnf wave.
Because of the commutation phase delay and the rectangular waveforms,
the controlled current CSI has serious limitations as a high performance cur-
3 Phase rent supply. However, in high power applications where response time is less
60 HZ ic significant, it provides the possibility of achieving reasonable current control
-
while retaining the practical advantages of CSI operation. This is especially
significant in synchronous motor drives of high horsepower where load com-
mutation is employed and the controlled current CSI is widely applied for such
Current
Amplitude
I li drives. Very high horsepower induction motor drives are also available utiliz-
ing the current controlled CSI.
Command 1 lS*I
Current
Regulator 7.2.2 The Current Regulated PWM Inverter
A current controlled PWM inverter operated with a switching frequency in the,
kHz range can function as a regulated current supply with the potential for
good dynamic response and low harmonic content. It is the most commonly

Y Commutation Delay
Compensation
t used power converter in the low to medium power range of high pe~formance
drives. Unlike the CSI, the PWM is a natural voltage source and conversion to
current source operation requires closed loop control with feedback directly
from the controlled ac currents. The feedback current sensors must therefore
Figure 7.2 Current regulated CSI inverter with commutation delay have a wide bandwidth, from the lowest fundamental frequency to be con-
compensation
trolled (usually dc) to somewhat above the PWM carrier frequency. Figure 7.3
The CSI system of Figure 7.2 also has the problem of producing significant illustrates the basic system. Usually only two current sensors are employed
harmonics in the output waveform since the CSI is basically a b s t e p system. since in the absence of a neutral connection the three currents must add to zero.
In effect, the essentially rectangular wave output current results in stepwise 60" It is, in fact, advantageous to use only two current sensors since this avoids
phase advances in the stator mmf of the connected machine. While this may error signals containing a zero sequence component resulting from sensor
not have serious effects at higher frequencies, it will generally result in large errors. The nature of the controlled output current is illustrated in Figure 7.4.
torque pulsations and associated rotor speed oscillations or cogging at low fre- As shown, the output consists of a reproduction of the reference current with
quencies. Although PWM techniques can be employed to reduce this cogging high frequency PWM ripple superimposed. This system has several fundarnen-
behavior at low speeds to acceptable levels, the fundamental problem of the tal differences from the CSI system including:
60" mmf steps and the associated torque pulsations remains at higher speeds. 1) The requirement of an actual time domain (waveform) reference instead
This is a result of the inherent CSI commutation delay and hence necessarily of the amplitude and phase references required in the CSI,
low PWM carrier frequency which can be employed. Above perhaps 10-20
2) The possibility of reducing harmonic content to arbitrarily small values
Hz, the system must necessarily return to normal bstep CSI operation with as the PWM frequency increases,
essentially rectangular (or trapezoidal) current waveforms. Twelve pulse, dual
3) The need for a current regulator operating with ac signals as opposed to
the dc regulator of the CSI,
322 7 Current Regulation in Power Converters Hysteresis Regulators 323

Phase
EMF (
(volts) f
3 phase
60 Hz 100.0 20.0
I I
, Current Reference
I

66.7

33.3

0.0

-33.3

-66.7

Figure 7.3 PWM system with current regulation to produce a -100

controlled three phase current source 0.0 0.71 1.43 2.14 2.86 3.57 4.29 5.0
-2
Time in Sec. *10
4) The need for wide bandwidth current sensors as described earlier.
While the sensor requirement and the development of good regulators has Figure 7.4 Simulated waveforms of a PWM inverter with current
required considerable developmental effort, the excellent response and low regulation using three independent hysteresis controllers
harmonic content of the PWM system make it the best currently available reg-
ulated current supply. A number of different types of regulators have been pro-
trated in Figure 7.5. The controller simply applies the current error signal to a
posed and use; and will be discussed separately. Basically the regulator must hysteresis element, the output of which supplies the logic signal to gate the
function to convert the time domain current error signals to firing signals for positive or negative inverter switching element. A lockout circuit is normally
the PWM inverter. The dual function of error processing and subsequent con- incorporated to allow for inverter switch recovery time and thus avoid short
version to gate drive signals is required in all types of regulators.
circuits across the dc link.
The subject of current regulators has been the subject of intense activity. In
While this system is very simple and provides good current amplitude con-
general, current controllers can be classified into three groups: hysteresis regu- trol, hysteresis regulation has the major disadvantage.of producing a highly
lation, ramp comparison and predictive controllers. variable PWM switching rate. Low frequencies appear in the spectrum regard-
less of the switching frequency. Unfortunately, the variation of the switching
7.3 Hysteresis Regulators rate is also opposite to the needs for good current control with the highest
The earliest and simplest type of PWM current regulator to implement is the switching rates associated with the lowest reference frequencies. As will be
hysteresis regulator [1],[2]. The basic concept of this type of controller is illus- discussed subsequently, the hysteresis controller also has the somewhat unex-
324 7 Current Regulation in Power Converters
- ~ ~ ~ t e r eRegulators
/
sis 325

Irn

a) Current Vectors in the Complex Plane.

Figure 7.5 Hysteresis current controller for one phase

pected property of limiting the current error to twice the hysteresis band rather
than to the band itself. Nonetheless, this controller is in active use in many
applications.
The behavior of the hysteresis controller can be explained in terms of a
complex plane switching diagram 121. Figure 7.6(a) shows the reference cur-
rent vector if the actual current vector is and the current error vector Ai in
-S
the complex plane along with the a, b and c axes of a three phase reference sys-
tem. The line current errors Ai,,, Aibs and Ai,, are the projections of Ai on
these three axes. The hysteresis controller will switch the a-phase i n v e z leg b) Switching Lines for Phase a.
when Ai,, exceeds the hysteresis band as represented in Figure 7.6(b) by the
two switching lines drawn perpendicular to the a-axis. The switching lines are
located a distance h, equal to the hysteresis band, from the tip of the current
reference vector. Similar switching lines can be drawn for phases b and c; the
resulting complete switching diagram is shown in Figure 7.6(c). The entire dia-
gram moves with the current reference vector with its center remaining fixed at
the tip of the vector.
The typical or expected behavior of the controller is to confine operation to
the interior, hexagonal region of the switching diagram. Thus, whenever the
current error touches one of the switching lines, that inverter leg is switched
driving the current error in that leg in the opposite direction. Note, however,
that the current error can be carried to one of the switching lines by motion of c ) Complete Switching Diagram.
the current vector is ,or by motion of the current reference vector $ on the
Figure 7.6 Hysteresis controller switching diagrms - three
switching diagram. Thus, the situation depicted in Figure 7.7 can occur, where independent controllers
326 7 Current Regulation in Power Converters Hysteresis Regulators 327

Figure 7.7 Hysteresis controller current trajectory resulting in error Figure 7.8 Hysteresis controller trajectory illustrating possible
equal to twice the hysteresis band high frequency limit cycles
the (-A) switching line is encountered with the inverter in the state (a+, 6-,c-). counter emf are assumed to be small. The current vector hits the +c switching
The system then can 'coast' out to the tip of the switching diagram before a line causing inverter leg c to switch and produce the new voltage vector ya
new switching of the inverter occurs. This demonstrates that an error of 2h, (a+, b-, c+). Next the current vector will hit the -a switching line producing
twice the expected value equal to the hysteresis band, can occur on an almost the voltage vector v5 (a-, 6-, c+). Continuing this reasoning, the six possible
random basis. non zero voltage vectors are applied repeatedly and a high switching frequency
The switching diagram can also be used to demonstrate the existence of results if the inductance is low and the hysteresis band small. Note that the
high frequency limit cycle oscillations when the load counter emf is low (low magnitude of the current error vector is not reduced to zero during the limit
frequency operation). In general, the vector voltage equation relating the cycle. The dashed line in Figure 7.8 represents a possible limit cycle when
inverter voltage ys, the load current is and the load counter emf es (voltage there is a non-zero counter emf.
behind transient inductance of the induction machine) is The limit cycle will be occasionally intempted by the intermittent occur-
rence of a zero voltage vector. A zero voltage vector will occur when one of the
ys = rsis+ L: pis + es (7.3-1) switching lines in the sequence is skipped as a result of the load counter emf or
where r, and L: are the load resistance and transient inductance per phase. If resistance, or the movement of the switching diagram caused by changes in the
es is small and the resistance is neglected, this equation becomes reference current. The application of a zero voltage vector will greatly reduce
the inverter switching frequency when the counter emf is low since the veloc-
vs = L: pi, (7.3-2) ity along a trajectory is proportional to the total applied voltage v - e (see eqn
(7.3-1)).
which indicates that the change of the current vector is in the direction of the
Figure 7.9 and Figure 7.10 illustrate the results of a simulation of a hyster-
voltage vector.
esis controller which shows the behavior described above. Figure 7.9 is the
Figure 7.8 shows a current trajectory, indicated by the solid line, which can
current error signal for an unloaded induction motor operating at 0.5 pu speed.
occur and represents a high frequency limit cycle. The initial voltage vector y ,
The occasional occurrence of a current error in excess of the 1.5 ampere hys-
(a+, b-, c-) (mode 612 in Figure IS), forces the tip of the current vector to
teresis band is clearly shown; several of these are very close to twice the band.
travel in the same direction as the voltage vector since the resistance and
The controlled current in Figure 7.10, which is for a stalled motor, clearly
 328 329
7 Current Regulation in Power Converters Hysteresis Regulators

Current
Error

-4.5
0.0 0.71 1.43 2.14 2.86 3.57 4.29 5.0
TIME (Secs. ) * 10-* 0.0 0.83 1.67 2.50 3.33 4.17 5.0
Time in Sec. *I0-2
Figure 7.9 Simulation results for hysteresis controller driving
a typical 10 HP induction motor at 30 HZ. and no Figure 7.10 Simulation results for hysteresis controller driving a
load showing random large current errors typical 10 hp induction motor at zero speed showing
sporadic
current high frequency limit cycles in the motor phase
shows the sporadic limit cycles which can exist for low frequency operation.
Note the nearly one half cycle of quite low switching frequency which also
exists; the limit cycles are essentially random occurrences. distance = Ai, = 6 h
The oscillation frequency within the limit cycles shown in Figure 7.10 can The inverter switching frequency is then
be estimated by dividing the velocity of the current trajectory by the distance
travelled in one complete period of the limit cycle. For zero counter emf, the velocity Vdc
velocity in amperes per second is approximately fs=-=- distance 9hL:

i s 2"dC
For the system illustrated in Figure 7.10,the inverter bus voltage was 300
velocity = d- E --
volts, the machine transient inductance was 3.36 mh and the hysteresis band
dt 3L;
was set at 1.5 amperes yielding a switching frequency of
and the distance (in amperes) travelled in a complete limit cycle is approxi-
mately
330 7 Current Regulation in Power Converters Ramp-Comparison Controllers 33 1

which was found to be very close to the value obtained in the simulation. between phases is eliminated by making the load appear to have a neutral con-
Clearly the highest switching frequency occurs during limit cycles with essen- nection back to a mid point in the dc link capacitor. This trick is accomplished
tially zero counter emf and the inverter must be designed to operate success- by calculating the zero sequence component of the stator voltage and then sim-
fully under this condition. Since high switching frequencies are not needed ulating the behavior of the zero sequence circuit. From Chapter 2 the zero
when the counter emf and hence reference frequency are low, this is a serious sequence voltage is given by
limitation of the simple hysteresis controller. In particular, the difficulty of
working with variable switching rates with respect to the filtering of measured
signals for feedback purposes argues against such controllers and has been
addressed by several authors [3]-[5]. The zero sequence circuit equation is
In Ref. [4] switchings still occur when the error vector crosses the bound-
ary of the hexagon. However, the voltage vector that is selected differs from
what a conventional hysteresis controller would select. When a segment of the
hexagon is crossed the voltage vector perpendicular to that segment and point- where Lo, represents the zero sequence inductance and is ideally equal to, but
ing toward the origin is selected. Such a scheme would produce high frequency in actuality somewhat smaller than the per phase stator leakage inductance LI,.
limit cycles of the type discussed earlier under all conditions if some means Solution of this equation enables the calculation of the value of the zero
were not used to introduced the zero voltage state occasionally. In Ref. 3, the sequence current that would flow if a neutral were connected. This computed
zero voltage vector is selected when the magnitude of Ai is less than a certain current is added to the current feedback signals as shown in Figure 7.11, so that
fixed amount and when the value of the back emf indicates that the Ai vector the individual hysteresis controllers appear to control phase currents that
will be driven in a favorable direction. The back emf iS calculated u>Ag mea- would flow if there were a neutral, thereby decoupling the three current con-
surements of rotor position and frequency. As this paper concerned a synchro- trollers.
nous motor drive, frequency and rotor speed are essentially the same so that
this problem is less difficult than for the induction motor. The use of the zero 7.4 RampComparison Controllers
- vector is outside the hexagonal region for a
voltage vector is disabled if the Ais
long period of time to allow faster transient response. At present, most systems employ other forms of current regulators in which the
In Ref:[4] switchings are again initiated when the error vector crosses the switching frequency is either nearly constant or at least known and under good
boundary of the hexagon. In this case, however, the voltage that is selected control. Constant switching frequency can be most easily obtained by a so
when a segment is crossed is modified such that the association between hexa- called ramp comparison controller. The ramp comparison controller is a direct
gon segments and voltage vectors is decided based upon the instantaneous carry over from dc machine current regulators employing the four transistor bi
direction of the back emf. For any one value of back emf only three of seven directional chopper or H bridge. Figure 7.12 illustrates the basic concept which
voltage vectors are used. These are the three vectors which are closest to the is essentially to employ a synchronous sine triangle PWM with the current
back emf vector, and automatically includes the zero voltage vector. Only a error providing the sinusoidal input. Although not shown in Figure 7.12, it is
simple estimate of the emf angle and not amplitude is required. An estimator common in dc systems to employ a PI controller ahead of the triangle compar-
for this quantity is described in the paper. The controller incorporates regula- ison to enable high dc gain and a controlled roll off of the high frequency
tion of the average switching frequency by a feedback controller that adjusts response.
the hysteresis band size. The basic operation of the ramp comparison controller is very simple; the
Reference [ 5 ] uses a fundamentally different approach to solving the prob- processed current error is compared to a triangular waveform and if the error
lems associated with the hysteresis controller. In this case the interaction amplifier output is greater (less) than the triangle waveform the inverter leg is
 332 7 Current Regulation in Power Converters
- cornparis par is on Controllers 333

Switching below the inverter dc bus voltage) [6].If the switching frequency is high com-
Synchronization pared to the frequency content of the input modulating signal (error amplifier
output voltage), the proportional relationship between the input and the funda-
Filter mental component of the output voltage across one phase of the motor can be
expressed by

= KAvin
*
'as
where K A = (2/n;) (Vi/V,) , 4 is the inverter dc input voltage and VAis the
peak value of the triangle signal. The input voltage vin is, effectively, the volt-
age at the input of the summer leading to the hysteresis block. The ratio of the
input voltage to the peak of the triangle voltage is typically termed the modula-
tion index mi. Hence, in terms of the modulation index eqn (7.4-1) can be writ-
ten,

When the instantaneous amplitude of the input voltage exceeds the triangle
Figure 7.1 1 Modified hysteresis current controller of Malesani and wave, intersections of the input voltage and triangle wave are eliminated and
Tenti [5] consequently PWM pulses are dropped. In this case it can be shown that, the
output voltage can be expressed by the describing function [6]

* (7.4-3)
A sketch of the variation of the fundamental component of the output voltage
as a function of mi is given in Figure 7.13. Note that the amplitude of the input
voltage vin must exceed a factor of 5 times the amplitude of the triangle wave
'as for full output voltage to be reached, i.e. square wave voltage operation. In Fig-
Figure 7.12 One phase of basic ramp comparison controller ure 7.14 the transfer function between the input voltage and the fundamental
component of the inverter output phase voltage is expressed as a nonlinear
gain. Note that the gain falls off rapidly as pulses are eliminated from,the
switched to the positive (negative) polarity. The inverter switches at the fre- PWM pattern indicating a loss of control of the stator current as the PWM
quency of the triangle wave and produces an output voltage which is propor- inverter approaches square wave operation.
tional to the error amplifier output signal (assuming the output voltage is well
334 7 Current Regulation in Power Converters -Stationary Frame Regulators 335

inverter leg if the time rate of change of the input voltage exceeds that of the
ramp. It is usually omitted when the input voltage rate is controlled.
In the typical dc current controller, a single ramp comparison controller is
employed with the output used to control the four transistors of the H bridge in
a complimentary fashion, i.e. either a+, b- (positive output) or a-, b+ (negative
output). The current error signal, as noted previously, is usually conditioned
using a simple PI controller and very satisfactory static and dynamic perfor-
mance Can be attained. An attempt to extend this basic concept to three phase
current controllers by using three ramp comparison controllers, one for each
phase, immediately poses a problem since the three PI controllers result in an
0 1O
. 2.0 3.0 4.0 5.0
attempt to regulate three independent states when only two exist (because the
three phase currents must add to zero). This immediate problem can be over-
Modulation Index - m come by using only two PI controllers and slaving the third phase, an approach
which is used in some cases. Alternatively, a zero sequence current can be syn-
Figure 7.13 Fundamental component of phase voltage versus
thesized in much the same manner as Figure 7.11 and fed back to the three reg-
modulation index mi
ulators to decouple them. An intellectually pleasing as well as very effective
approach is to view the current control problem in d,q coordinates, which
immediately indicates the necessity of using only two regulators and algebra-
i
v ically establishing the three phase inverter gating signals.
~ V A

Gain 7.5 Stationary Frame Regulators

Figure 7.15 illustrates the overall system diagram of a three phase ramp com-
parison controller based on stationary frame d,q variables. The regulator illus-
trated employs simple PI controllers and is a commonly used system as a result
of dc machine experience. It is easily shown that this simple system has inher-
ent regulation problems compared to its dc machine counterpart by considering
1.0 2.0 3.0 4.0 5.0 the steady state response. In the dc machine case, the steady state response is
Modulation Index - m i characterized by zero current error because of the integration in the forward
path. However, for the ac controller, the steady state condition requires sinuso-
Figure 7.14 Fundamental component gain versus modulation index mi idal output at the reference frequency and clearly the forward path integration
in the PI controller does not produce zero current error because of the fre-
Although eqns (7.4-1) to (7.4-3) are derived for sinusoidal inputs, it is quency dependent PI controllers.
common to apply them to arbitrary variations of vi, provided the ratio of The steady state error for the system of Figure 7.15 can be evaluated using
switching frequency to input frequency is sufficiently high. The hysteresis ele- the normal frequency domain steady state model derived in Chapter 4. From
ment shown in Figure 7.12 is included to prevent multiple switching of the Figure 7.15, the frequency domain relation between the d,q axis output voltage
and the current error can be written as
 336 7 Current Regulation in Power Convertem Stationary Frame Regulators
- 33 1

where

Again, L: is the stator transient inductance. Substituting these results in the

stator equation (eqn (7.5-2)) yields

I u I 'ds

Figure 7.15 System diagram of stationary d,q frame ramp comparison Eliminating the ir term in eqn (7.5-3) by use of eqn (7.5-5) results in. for
current regulator using PI controllers the rotor voltage equation,

(ii*- i", KAGc( j o , ) = rS, (7.5-1)

where, for generality, & Cio,) is the frequency domain transfer function of the In the stationary reference frame eqn (7.5-8) together with eqn (7.5-9) can
current error amplifier in Figure 7.12 rather than the PI regulator in Figure 7.15 be written as
and the complex quantities &* , i,iS, and v; are the steady state frequency
domain values of the currents and voltages.
For steady state operation, the normal equivalent circuit of the induction
machine (or synchronous machine) can be used to relate L'S, and r,S . It was
shown in Chapter 4 that in the general rotating reference frame the equations
for the stator and rotor circuits of a three phase, three wire induction machine
can be written, where

r
'r = 5
0 = rrir+ [ p + j ( o - o r ) ]hr
Upon eliminating the rotor flux linkage from these two equations yields.
where the flux linkages are
finally

hr = ~,i~+~,.r~ (7.5-5) where,

Eliminating the rotor current in these equations yields
 338 7 Current Regulation in Power Converters Stationary Frame Regulators 339

In eqn (7.5-13), the operator p should now be interpreted as the equivalent of this control problem quickly becomes very difficult since Zi,is essentially a
the Laplace operator s. The use of another variable for the Laplace operator is linearly increasing function of o, (for constant So,) and hence approximately
clearly necessary to avoid confusion with the per unit slip S. In this context the a pure lead network.
operator p is frequently called the Heaviside operator after a famous British An alternative representation of the machine in t e r n of the transient
engineer who exploited many of the benefits of operator notation long before model (voltage behind transient reactance) can also be employed to illustrate
Laplace supplied the mathematical formalism. the problem. This useful conceptual model can be obtained for this situation by
The block diagram representing these equations is shown in Figure 7.16, again starting from the complex space vector form of the stator equations, in
the stationary reference frame, that is
S
ys 1 s
is
zi,(j a , , j S W , )

t- In the frequency domain we can write this equation as

y: = [rs + jo,L: ] t", +e", (7.5-16)

This approach, which is also valid for the sub cycle behavior associated with
the carrier frequency response, leads to the model shown in Figure 7.17. Here

Figure 7.16 Steady state frequency domain block diagram for the current
regulator of Figure 7.15 using the induction motor equivalent
circuit model

from which the transfer function from current reference to output current can
be written as Note: Zt (iw, ) = rs +j m ,L' ,

Figure 7.17 Steady state frequency domain block diagram for the current
regulator of Figure 7.15 using the 'transient' model of the
machine
Note that the circuit shown in Figure 7.16 represents the steady state version of
the dynamic results expressed in eqn (7.5-14).
the machine impedance is the stator transient impedance and the counter emf,
From this expression it is clear that the system of Figure 7.15 can result in
%GoejSoe), is the voltage induced in the stator by trapped rotor flux linkage.
significant phase and magnitude errors if the frequency curve of G, Go,) does
In Figure 7.17, the counter emf is shown as a steady state complex quantity
not maintain sufficiently high gain at higher frequencies. It is also clear that
representing the steady state voltage behind the equivalent transient impedance
this control problem quickly becomes very difficult since Zin is essentially a
z,.
 340 7 Current Regulation in Power Converters Synchronous Frame Regulators 34 1

With the viewpoint of Figure 7.17, the system more closely resembles the ware for implementation because of the requirement to transform the measured
dc machine case in which the machine counter emf enters as a disturbance to currents to a synchronous frame and subsequently to transform the error ampli-
the controller. This suggests the possibility of employing feedforward compen- fier outputs back to a stationary frame to drive the ramp comparison controller.
sation for the counter emf as is sometimes done in dc drives. In the dc case, the These transformations require explicit knowledge of the frequency 61,.
counter emf is calculated from knowledge of the field flux and rotor speed and To examine the differences between this synchronous frame regulator and
added as a correction at the output of the error amplifier. Similar approaches the stationary frame regulator of Figure 7.15, it is convenient to transform the
are possible in ac drives; the details are left until the overall system has been stationary frame regulator to the synchronous frame so a direct comparison can
described since implementation of the feedforward compensation is applica- be made. The transformation is illustrated in Figure 7.19 which shows both the
tion specific. original regulator and its representation in the synchronous frame. The trans-
formation is readily carried out using vector notation. From Figure 7.19(a), the
7.6 Synchronous Frape Regulators equations of the stationary regulator in the stationary frame are
The inherent problems of regulating ac signals in'the stationary frame were
first recognized by Schauder and Caddy [7]. In their early work they showed
that the stationary frame ramp comparison regulator is only one of an infinite (7.6-2)
px", K (is,* - iS, )
collection of such regulators. In effect, regulators like that of Figure 7.15 but
implemented in other reference frames exhibit quite different characteristics. Using the transformation equation from stationary to synchronous reference
That this is so is quite clear if one considers that the frequency of the current is frame variables, (eqn 3.5-42), the stationary variables are replaced by
different in different reference frames and hence the regulator performance (if
it is frequency dependent) will also be different.
From this perspective, a synchronous frame seems especially appropriate to yield the synchronous frame variables. The result is
since the steady state currents are dc currents and a simple PI controller will
result in zero steady state error. A system diagram for a synchronous d,q frame
regulator is shown in Figure 7.18. It is more complex and requires more hard-

Rewriting this last equation as

suggests the synchronous frame equivalent of the stationary frame regulator as

shown in Figure 7.19(b).
The most obvious aspect of the synchronous equivalent of the stationary
frame regulator is the existence of frequency dependent cross coupling
between the d- and q-axes. If the regulator is constructed in the synchronous
Figure 7.18 System diagram of synchronous frame ramp comparison
frame as shown in Figure 7.18, no such cross coupling exists. There are three
current regulator using PI controllers
important effects of this cross coupling.
1) For steady state operation the cross coupling produces a response in the
342 7 Current Regulation in Power Converters Synchronous Frame Regulators 343

3) There is the possibility of strong resonance phenomena since the cross

coupling terms themselves form a local oscillator at the frequency a,.
The first two of these effects are the synchronous frame equivalents of the
steady state error problems described in the previous section. In effect, they are
previously anticipated results viewed from a different frame of reference. They
are completely absent in the synchronous frame regulator of Figure 7.18 since
no regulator cross coupling exists in the synchronous regulator.
The third effect is quite different in that it is not at all apparent from the sta-
tionary frame model. The existence of the possibility of local oscillation can be
demonstrated by writing the equations around the cross coupling loop. These
equations are, from Figure 7.19(b)
a ) Stationary Frame

from which, by eliminating xeh, there results

p2x:; = oexeq s
The solution to this linear oscillator equation is

and
.'e U x$s = A sin (oet+cp) (7.6-1 1)
'ds
b) Synchronous Frame If these synchronous frame sinusoidal oscillations are transformed back to the
stationary frame, the result is
Figure 7.19 Models of the stationary frame PI controller current
regulator in a) the stationary frame and b) the synchronous
frame
which corresponds to a dc signal in the actual stationary regulator. Since a dc
cross channel which increases in importance as the frequency increases,
error will be regulated to zero in the actual stationary regulator, the oscillation
i.e., the q-axis current will not be zero even when the q-axis command
does not exist in the steady state. However, during transients such oscillations
is zero if there is a non zero command in the d-axis channel.
can be initiated and they will die out according to the damping properties of the
2) For steady state operation there is an error in the direct channel which
regulator. One would therefore expect the possibility of small signal reso-
increases with frequency since the cross coupling feeds back into the
nances at low frequencies for the stationary frame regulator.
original channel as well, i.e., the q-axis current will not follow the q-
axis command.
 344 345
-
7 Current Regulation in Power Converters Feedforward compensation

All of these effects are absent in the synchronous frame regulator since no
cross coupling exists. The performance of the synchronous frame regulator is
therefore greatly superior to the stationary frame regulator in all aspects. It is,
however, as noted previously, considerably more hardware intensive, requiring
two transformations between stationary and synchronous variables.
A much simpler hardware implementation has been suggested [6],[8] in
which the synchronous regulator of Figure 7.18 is itself transformed to the sta-
tionary frame. This is, in effect, the opposite of the operation carried out to
derive the circuit of Figure 7.19(b). It is, however, used to develop a simpler
implementation of the synchronous frame regulator rather than as a means of
analysis. Since the development is very much the same as before, only the
result is given; the stationary frame version of the equations of the synchro-
nous regulator in Figure 7.18 are

Figure7.20 The stationary frame equivalent of a PI controller

synchronous frame regulator
The corresponding stationary frame equivalent of the synchronous regula-
tor is shown in Figure 7.20. Once again, cross coupling terms are introduced. Thus far, all of the regulators that have been discussed fall into the category of
However, in this case the cross coupling is very useful and significant since the series compensation. That is, whether implemented in the stationary or syn-
local oscillator formed in the cross coupling provides the required ac signal to chronous frames, both hysteresis and ramp comparison methods rely on a com-
allow the current error to go to zero in the steady state. The performance of the pensator placed between the error of the regulated quantity (motor current) and
regulator is, of course, identical to the synchronous frame regulator of Figure the output of the amplifier representing the PWM inverter. These regulators
7.18. Note, however, that only two multipliers are required for implementation can be viewed generically as shown in the block diagram of Figure 7.2 1. Note
beyond that needed for the ordinary stationary regulator. Reference [8] pro- that, neglecting the practical realities of scaling, the PWM inverter can be con-
vides an in depth study of the synchronous regulator and detailed comparison sidered as an ideal unity gain amplifier. This assumption, of course, becomes
with the stationary regulator. invalid as pulses are dropped when the inverter ceiling voltage is approached
and the inverter changes to six step square wave operation. Note again, that for
purposes of analysis, the induction motor can be viewed as a counter emf in
7.7 Feedforward Compensation series with the stator resistance and stator transient time constant. The feed-
From the perspective of control'theory, regulators can be grouped into three back block giving e, (p) in terms of is ( p ) follows directly from eqns (7.5-
categories; 10) and (7.5-11).
Series compensation, Other types of current regulation using feedback or feedforward compensa-
tion, either acting either alone or in conjunction with a series compensation
Feedback compensation,
method have been proposed in the literature. However, it is not necessary to
Feedforward compensation. concern ourselves with feedback compensation (which is not frequently used)
 346 7 Current Regulation in Power Converters Augmented Feedforward Compensation 347

age from the stator current. This portion of the problem can be eliminated by
calculating the rotor flux linkage from some other variable. For example, the
time rate of change of air gap flux linkage can be measured from flux coils
placed around the slots of the stator. When this quantity is integrated and pro-
cessed by the complex vector equation,

the rotor flux linkage can be obtained. Since this equation is algebraic, it is
Figure 7.21 Block diagram of series compensated current regulated
much simpler than the caIculation shown in Figure 7.22.
induction motor

since a nonunity feedback compensator H @ ) can be replaced by a series com- 7.8 Augmented Feedforward Compensation
pensator K(p) = H(p) and a feedforward compensator l / H @ ) .
The method of feedforward compensation is a type of open loop control since
When all of the parameters of the machine are accurately known, it is in
the quantity to be regulated is not measured or fed back to the input. While
principle possible to regulate the stator currents without any feedback of the
such an approach cannot, in itself, qualify as a high performance controller, an
motor current by using a feedforward compensator that is the exact inverse of
estimate of the current can be used to advantage in conjunction with one of the
the motor transfer function. Such a scheme is shown in Figure 7.22. Unfortu-
many series compensators as shown in Figure 7.23. In this case, the feedfor-

Figure 7.22 Block diagram of induction motor current controller with

ideal feedforward compensation. The carot symbol Figure 7.23 Block diagram of current regulator using series compensation
denotes an estimate of the quantity augmented by feedforward compensation. The symbol
denotes an estimate of the quantity
nately, regulation of the currents will typically be unacceptably poor due to
ward compensator is used to provide only a rough estimate of the voltage com-
uncertainties in the motor parameter values. In Figure 7.22 the upper block fol-
mand signal. The remainder of the required voltage is then contributed by the
lowing the current command synthesizes the back emf while the lower block
series compensator. Since the major portion of the voltage command is pro-
accounts for the stator voltage drop. Note that the major difficulty with this
vided by the feedforward compensator, the series compensator is not called
scheme is the computation involved in the block to compute the rotor flux link-
upon to slew as rapidly, thereby permitting a reduced gain in the series com-
 348 7 Current Regulation in Power Conveners
- predictive Control 349

pensator. This reduced gain allows the circuit to become less sensitive to ran-
7.10 Predictive Control
dom errors and injected noise, thereby increasing the reliability, as well as the
accuracy, of the overall control system. In practical applications, feedforward Recently, use of powerful digital signal processors (DSPs) have become cost
of only the counter emf is sufficient and the feedforward of the stator transient with more conventional analog control circuits. While DSPs
impedance drop is frequently omitted [9]. Alternatively, since the stator resis- remain too slow to replace analog type hysteresis or ramp comparison methods
tance is small and not reliably known due to heating, the feedforward of the for most applications at the present time, it will be only a matter of time before
transient impedance drop can be replaced by only an estimate of the drop con- the full power of these computing chips can be utilized. In general, use of
tributed by the transient inductance, i.e. by replacing (rS+pL:)Cs by DSPs permits the on line computation of the condition of the circuit and
pL: itin the feedforward path [lo]. enables the selection of more optimal strategies [ll]. These methods are typi-
cally based on the complex space vector approach to the analysis of rotating
machines presented in Chapter 2. If the supply voltages were sinusoidal, this
7.9 Augmented Feedforward Compensation with Decoupling vector would rotate with constant amplitude and fixed angular velocity. How-
As has probably been surmised by the reader, it is possible to combine the ben- ever, because of the discrete switching nature of the inverter, the vector can
efits of the cross coupled regulator of Section 7.6 wifh the augmented feedfor- attain only six spatial positions plus the two zero vector positions (Figs. 3.3
ward approach of Section 7.8. The combined controllqr can be represented as and 3.4). Switching of the inverter results in discrete jumps of this voltage
shown in Figure 7.24. In this case the use of complex vectors in the block dia- space vector rather than uniform rotation.
If the stator resistance is neglected, an approximation for the current trajec-
?(P)
.'1 A S . . S I tory is

If the switching interval T is sufficiently small, eqn (7.10-1) becomes

s *
Consider now the case where the desired value of i s (T) is known at the
Figure 7.24 Block diagram of current regulator using cross coupled series beginning of the switching interval. Since it is desired to force the current error
compensation, augmented by feedforward compensation. The to zero over the interval [O,T], then clearly
symbol '"' denotes an estimate of the quantity
is(T) = :Si (T)
gram is retained so that the proportional plus integral block of the conventional
series compensator, K[1+1/(~p)], must be replaced with a frequency shifted so that, the required voltage over the interval is
.
compensator K 11 + 1/ (T( p - jo,)) ] In effect, this series portion of the
compensator represents the current regulator of Figure 7.20. It should be
apparent that augmented feedforward can also be added to the current regulator
in the stationary frame, Figure 7.18. where it is assumed that iss* (T) is known at r = 0. The counter emf pSs can be
calculated by a feedforward compensator in the same manner as in Figure 7.22.
 350 7 Current Regulation in Power Converters References 351

Although eqn (7.10-3) defines the approximate vector value of inverter 7.11 Conclusion
voltage needed to drive the current error to zero, the inverter can output only
Current regulation of inverters is clearly an interesting and important area of
six discrete nonzero and two zero voltage vectors. Fortunately, minimum devi-
research in motor drives. This chapter has attempted to present only the high-
ation of the current error is, in general, accomplished only if voltage states
lights of a very dynamic and rapidly evolving technology. In addition to the
adjacent to the instantaneous position of the reference vector vs*s (0) are used
concepts we have discussed, such advanced technologies such as expert sys-
to synthesize the required voltage vector y: (0). Assume for example that the
tems and neural networks are beginning to be explored. We can look forward
position of the command $,*(0) is instantaneously located between vectors
,
y and y2 as shown in Figure 7.25. Since all of the terms on the right hand side
to many interesting new approaches to this problem in the years ahead.
References
D.M. Brod and D.W. Novotny, "Current Control of VSI-PWM Inverters", IEEE Trans.
Ind. Appl., Vol. IA-21, no. 4, pp. 562-570, May June 1985.
D. M. Brod, "Current Control of VSI-PWM Inverters". M.S.E.E. Thesis, University of
Wisconsin, 1984.
G. Pfaff, A. Weschta and A. Wick. "Design and Experimental Results of a Brushless AC
Servo-drive," IEEE Trans. Ind. Appl., Vol. IA-20, no. 4, pp. 814-821, JulyIAugust
1984.
A. Nabae, S. Ogasawara and H. Akagi, "A novel Control Scheme for Current-controlled
PWM Inverters," IEEE Trans. Ind. Appl. Vol IA-2, no. 4, pp. 697-701, JulyIAugust
1986.
Figure 7.25 Space vector synthesis of a voltage gs . L. Malesani and P. Tenti, "A Novel Hysteresis Control Method for Current-controlled
VSI PWM Inverters with Constant Modulation Frequency," IEEE Trans. Ind. Appl.,
of eqn (7.10-3) are constant it follows that over a switching interval T,
Vol. IA-26, no. 1, pp. 88-92, Jan./Feb. 1990.
T. Rowan, "Analysis of Naturally Sampled Current Regulated Pulse-width Modulated
T T T,+T2 T
Inverters", Ph.D. Thesis, University of Wisconsin, 1985.
jy:* (0) d t = j q d t + y2dt + y7dt (7.104)
C.D. Schauder and R. Caddy, "Current Control of Voltage-source Inverters for Fast
o o T~ TI 7-2
+
Fourquadrant Drive Performance," IEEE Trans. Ind. Appl., Vol. IA-18, no. 2, pp 163-
where y , and y2 are complex constants corresponding to the two adjacent 171, MarcMApril 1982.
switching states and y7 equals zero. Hence, T. Rowan and R. Kerkman. "A New Synchronous Current Regulator and an Analysis of
Current-regulated PWM Inverters," KEE Trans. Ind. Appl., Vol. IA-22, no. 4, pp. 678-
690, JulyIAugust 1986.
H. Nagase, Y. Matsusa, K. Ohnishi, H. Ninomiya and T. Koike, "High-performance In-
duction Motor Drive System Using a PWM Inverter," IEEE Trans. Ind. Appl., Vol IA-
where y7 represents one of the two zero voltage switching states. The voltage
20, no. 6, pp. 1482-1489, Nov.lDec. 1984.
necessary to drive the current to zero has thus been defined in terms of the
R.D. Lorenz and D.B. Lawson, "Performance of Feedforward Current Regulators for
inverter switching states over three subintervals. In effect, the two adjacent
Field-oriented Induction MachineControllers," IEEETrans. Ind. Appl., Vol. IA-23, no.
vectors are time weighted to produce the desired output voltage. In practice, 4, pp. 597-602, JulyIAug. 1987.
the zero switching state selected is the one involving changing the state of the J. Holtz and S. Stadtfeld, "A Predictive Controller for the Stator Current Vector of AC
least number of semiconductor devices in order to minimize losses. A vector Machines Fed from a Switched Voltage Source", Int. Power Electronics Conf. -Tokyo,
diagram representation of eqn (7.10-5) is shown in Figure 7.25. March 27-31.1983, pp. 1665-1675.
352 7 Current Regulation in Power Converten

Problems
Problem 7-1 Stator Frame Ramp Comparison Current Controller
Consider a stator frame ramp comparison current controller as shown in
Fig.7.15 used to supply a 100 hp machine with parameters 8 Parameter Sensitivity and Saturation Effects
r, = 0.015, x,, = 0.10, xm = 2.0, r, = 0.02, x,, = 0.10, all in pu. in Indirect Field Orientation
Assume the machine is rated 460 volts (line to line) and that K A = 10.
a) Select the PI controller parameters K and 2 such that the PI controller zero 8.1 Introduction
coincides with the motor stator transient time constant L',/r, and that the
In the indirect field oriented systems of Figures 6.6 or 6.7, the slip relation
steady state phase error of the current controller is 3 degrees at a stator fre-
given in eqn 6.2-17 is employed to obtain the correct subdivision of the stator
quency of 10 hz.and rated slip frequency.
current into the torque and flux components i$ and i$ . If the rotor open cir-
b) For the PI controller parameters of part a, calculate the steady state ampli- cuit time constant L i r , is not correctly known, or if it changes because of
tude and phase error of the current controller for frequencies of 0, 10, 20,40 motor heating or other variations, this subdivision will not be correctly
and 60 hz for rated slip frequency and for no load. attained. The result will be detuning of the controller and a loss of correct field
orientation [1],[2].
The overall effect of this detuning is to incorrectly calculate the rotor field
angle. In general, this causes the actual stator current components to be incor-
rect with the result that:
(1) the flux level is not properly maintained,
(2) the resulting steady state torque is not the commanded value, and
(3) the torque response is not instantaneous.
These variations are explored in the following sections.

8.2 Phasor Diagram for Detuned Operation

The steady state effect of detuning is easily visualized by writing the rotor loop
equation in the conventional equivalent circuit

Solving this equation for 7, yields

354 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation Phasor Diagram for Detuned Operation 355

where os= So, and L,,, + LLr= L,. Figure 8.1 [3] illustrates the situation exist-
I, = ?,,

Figure 8.2 Indirect field orientation when calculated slip frequency is

Figure 8.1 Indirect field orientation with correct value of slip frequency too small showing overexcitation and loss of orientation

8.2.1 Field Orientation Torque Characteristics

ing when the slip calculator has the correct value of er/pr. The specified val-
ues of isT and isqlocate is which ib turn locates the phasor -jws*L,, 5,. If Determination of the field orientation torque characteristic for detuned opera-
the slip is correct, the phase angle of rotor self impedance, cp, = tan-los L$rr tion is most conveniently made by returning to the conventional equivalent cir-
will be exactly the value necessary to put 7, = I in phase opposition to isT cuit of Figure 5.18. Since field orientation is a controlled current mode of
qr operation, the important torque characteristic of the machine is the torque vs.
as shown in Figure 8.1 [3]. The rotor flux, which is orthogonal to the rotor cur-
. The figure is also labeled
rent in the steady state, is therefore aligned with is,+, slip frequency curve for a constant value of the stator current. From Figure
with the corresponding d,q axis currents (the h
difference in scale between 5.18, the torque can be expressed as [4]
conventional phasors and conventional d,q axis currents is ignored) and the
vector sum Lm Gds + Lr Iq r is shown summing to hqdr = hdr . PC rr
T , = 3---
2 2 So,
The situation existing when the slip is too small (ir/ir? Lr/r,) is illus-
trated in Figure 8.2. In this case, the phasor -jos*LmIs (or the vector Using the rotor open circuit time constant
-jos*Lm Iqds) has the same position as in Figure 8.1 but is shorter. The angle
cp, is, however, too small and the rotor current does not lie on the negative q-
axis in opposition to Ig, (or isT).As a result the rotor flux hqdr is not entirely
in the d-axis but is shifted by an angle 6 as shown in Figure 8.2. Note that the and the current division equation relating I, to Is the torque can be written as
magnitude of the mtor flux is also incorrect; in the example of Figure 8.2, hqdr
is considerably larger than the correct value corresponding to the correct slip m
Figure 8.1. Clearly the error in flux angle and amplitude will also cause an
error in torque, i.e., the torque will not correspond properly to the torque com-
mand Iqs(or isT).
 356 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation phasor Diagram for Detuned Operation 357

Figure 8.3 illustrates this characteristic curve. It is important to note that

this constant stator current torque curve:
substituting the magnitude relation in the torque equation yields
1) is dependent only on slip frequency, it is independent of the actual
motor speed,
2) has a peak value proportional to 1; at the slip frequency where
Soezr = 1 ,
as a general torque expression describing field oriented operation. If the slip
3) will exhibit strong saturation effects at low slip frequency where the relation is used in eqn (8.2-8), the result is
flux becomes large as the constant stator current is forced into the mag-
netizing branch.

which demonstrates the correct linear relation between torque and torque com-
mand. This situation is illustrated in Figure 8.4 where the slip relation is used
to draw jeld orientation lines which determine the field orientation operating
points on the various constant stator current torque characteristics. Each spe-
cific choice of Isq determines the slope of a unique field orientation line and
also determines the unique value of IsT which corresponds to each constant sta-
tor current torque characteristic where it intersects the field orientation line
based on the magnitude relation. Two such lines are shown in the figure along
with the appropriate scaling for IsT on the torque curves.
If a normalized torque scale is used, the entire collection of field orientation
lines can be reduced to a single line as illustrated in Figure 8.5. The normaliza-
tion uses the coefficientof Is+ISq in the first form of eqn (8.2-9) as the base for
the torque. In this case changing Isq does not alter the field orientation line but,
instead, changes the value of the torque base, To, and hence the values of
torque on the torque scale. The various constant stator current torque curves
are now described by specific values of the ratio Thus, a change in Isq
changes the torque scale and the numerical value of the ratio Is+ISq for each
Figure 8.3 Torque vs. slip frequency characteristic for constant stator current
torque curve, but there is only one field orientation line. It should be noted that
these types of normalization are useful in the linear case (saturation neglected)
Field oriented operation may be examined by employing the steady state but cannot be used when saturation is included.
field orientation constraints which are the slip relation Note that with the point of view of Figure 8.5, field orientation may be con-
sidered as a means of linearizing the highly nonlinear constant stator current,
torque slip frequency curve. The cancellation of the 1 + ( S o , 2,) term in eqn
(8.2-8) by the 1 + ( I s T / l s , $ ) term illustrates the means by which this linear-
and the magnitude relation ization is achieved.
358 8 Parameter Sensitivitv and Saturation Effects in Indirect Field Orientation Phasor Diagram for Detuned Operation 359

Field Orientation Line

Line

Figure 8.4 Field orientation lines superimposed on constant stator

current torque characteristics

If the controller is detuned by having an incorrect value of ir in the slip

Figure 8.5 Field orientation line superimposed on constant stator current
calculator, the field orientation lines on Figure 8.5 become curved lines. This normalized torque characteristics
can readily be observed by introducing an error into eqn (8.2-6) by rewriting it
as

where T,. is the correct time constant and ir is the one used in the slip gain cal-
where a is the error in slip gain and is equal to culation. Using (8.2-10) in (8.2-8) yields
 360 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation phasor Diagram for Detuned Operation 361

Each of the curves is also labeled with the value of relative stator loss which is
simply the normalized value of the stator current squared given by 1 + P*.
Figure 8.6 clearly shows that the ratio of IsTto Isq (P, the normalized
torque command) is the most significant parameter in determining the amount
of impact a specific amount of detuning will have on torque production. Gener-
which when plotted as a function of ZsT/Zs,+, is clearly not a straight line ally speaking, the smaller the value of P, the less effect detuning tends to have,
unless a = 1. The nonlinearity is a result of the nonlinear relation between although for errors in which the slip is too small there are some anomalies near
torque and slip frequency. The only condition which produces linearity is the the peaks of the curves. These curves can be viewed as representing a single
original slip relation of (8.2-6) or (8.2-10) with a = 1. machine at various levels of torque command or as a means of comparing dif-
Since curved field orientation lines are both undesirable and inconvenient ferent machines at, say, rated torque command. The following types of infor-
for analysis, it is useful to rearrange the method of viewing the results when the mation can be obtained for any specific value of Is.I/ISq:
system is detuned. Equation (8.2-12) can be rewritten in a more compact form
1) torque and slip frequency for correct field orientation,
by defining
2) relative stator PR loss,

T = -312 P
O
L2
LL =
2 2 s , + , L,
Field oriented torque at 5'.,1 3) torque and slip frequency for detuned operation for gain error ratios
form zero to three,
Is,
4) required change in IsTto compensate for a specified gain error.

8.2.2 Examples
To illustrate the application of the curves in Figure 8.6, consider the following
to yield examples.
1) A large machine with = 4.0 at rated torque. With correctly tuned
field orientation, the normalized torque is 4.0, the slip frequency is 4 . 0 1 and
~~
the relative stator loss is 17 (times the stator PR loss at zero torque). All of
these values are read on the a = 1 line using the p = Isp'Iscp= 4.0 curve
If TJT, is plotted as a function of a with P as a parameter, we have the very
With a slip gain error of 2.0, the corresponding torque is 2.1 (53% of the
convenient result that the vertical line at a = 1 is the field orientation line and
commanded value), the slip frequency is 812, and the relative stator loss is
all other vertical lines are detuned field orientation lines for various specific
unchanged. These values are read at the a = 2 line using the P = ISfiIScp = 4.0
slip gain error multipliers a. Figure 8.6 [4] illustrates these results for slip gain
curve.
errors between 0 and 3. The curves on this figure, which have the appearance
To bring the torque back up to 4.0 would require increasing the torque
of motor torque vs. slip frequency curves, are plots of eqn (8.2-15) for specific
command p. Since no larger curve is available, eqn (8.2-15) can be used with
values of p = ZsT I/s., The peaks of these curves occur at a = 1/P as can be
T, = 4.0, a = 2
seen in the figure. Note that along the vertical line a = 1, the normalized torque
is exactly equal to p = ZsT /Is,+, as it should be for properly tuned field orien-
tation. The variation of torque for any degree of detuning can be found by
looking along the vertical line for the appropriate value of slip gain error a.
from which P must be increased to nearly 8.0 or about double the original
Value. Note that this would increase the relative stator loss to 65.
362 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation Phasor Diagram for Detuned Operation 363

/ Field Orientation Line increases the torque whenever operation takes place in the vicinity of the peaks
of the constant p curves.
It is easy to show from the curves of Figure 8.6 that large values of P = I,fl
Isq (which are typical of larger, more efficient machines at rated torque) lead
to:
(1) field oriented operation at slip frequencies larger than the slip frequency
at peak torque,
(2) large torque loss (or gain) for detuned operation,
(3) large changes in stator PR loss for changes in IsT.
Clearly, operation at lower values of I S f i q results in less sensitivity to
detuning, less change in stator PR loss as IsT changes and tends to increase the
torquelampere available from the machine for field oriented operation. This
suggests that there are advantages to operating at higher values of Isv (higher
flux) but this conclusion is application dependent as will be demonstrated.

8.2.3 Field Orientation F l u . Characteristics

As illustrated in Figure 8.2, detuning also results in a change in the flux ampli-
tude. This change is, in fact, a primary cause of the torque changes described in
NormalizedSlip Gah a = the previous section. An expression for the rotor flux amplitude can again be
IST/~S@ readily derived from the equivalent circuit of Figure 5.18 following the same
Figure 8.6 Field orientation torque characteristics (saturation neglected) pattern as used to derive the torque relation of eqn (8.2-8). The result is

2) For comparison, consider a smaller machine having a larger magnetizing

current and,hence a value of Is,+,/IsT = 1.0 at rated torque. With correct tuning
the normalized torque is 1.0, the slip frequency is 1 . 0 1and
~ ~ the relative stator
loss is 2.0. Once again, with proper field orientation (Sonr = I s T / I s q ) the undesired
With a slip error of 2.0, the corresponding normalized torque is 0.8 or 80% variation with slip frequency cancels and the flux is constant at the value L,
of the commanded value, the slip frequency is 212, and the relative loss is Is? Figure 8.7 illustrates the flux variation caused by detuning using the same
unchanged. slip frequency scaling as employed in Figure 8.6.
To bring the torque back up to 1.O, the value P only needs to be increased to As can be seen from this figure, detuning such that the slip frequency is
about 1.5 with the relative loss increased to 3.25. smaller than the correct field oriented value causes the flux to become larger
3) To illustrate the complex nature of detuning, observe that for a slip error than the correct value. This is a result of forcing more of the stator current to
of 0.5 and a small P = IsflISq = 1.0, the torque drops from 1.0 to 0.8 as a result flow in the magnetizing branch as the effective rotor impedance increases due
of detuning. However, for the same detuning with P = IsflISq =2.0, the torque to the smaller slip. The curves are monotonic since an increase in slip results in
changes from 2.0 to 2.5 as a result of detuning. Thus, for small torque com- more rotor current and less magnetizing current and flux. Note that, as for
mands the detuning reduces torque but for large torque commands detuning
 364 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation 365
phasor Diagram for Detuned Operation

In addition to limiting the flux change, saturation also changes the value of
L, and hence of the rotor time constant 2, These changes in L, and 2, have
important effects on the torque vs. slip frequency characteristics; reducing the
peak torque and increasing the slip frequency at which the peak occurs. On the
generalized curves of Figure 8.6, this results in lower peak torques for higher
values of IsflISq and moves the peaks closer to the field orientation line at these
higher values. The actual torque vs. slip frequency curves in Figure 8.6 would
now only apply to a specific choice of Is,+,;a different set exists for every other
value. In the saturated case it is therefore better to return to a characterization
like that of Figure 8.4 and to use actual torque (not normalized to Is,+,).For any
specific motor, one such set of curves exists representing the unique saturation
characteristic of the machine.
Figure 8.9 represents the torque curves including saturation for 7.5 hp
machine of Figure 8.8. Four field orientation lines are indicated on the figure
representing four different choices of Isv The field orientation lines are still
very nearly straight lines even with saturation since field orientation produces
constant rotor flux operation. Each line, however, has a different value of 7,
because of the change in Lm resulting from saturation. Note that, overall, the
influence of saturation is to make the effect of detuning less pronounced; both
Normdized Slip Gain a = the flux variation and torque variation are less severe. Note that saturation also
I ~ 11s
T 9 tends to move the peak torque points more in line with a high flux field orienta-
tion line. This suggests it is possible to operate close to maximum torque per
Figure 8.7 Field orientation flux characteristics (saturation neglected)
ampere, by operating at higher than normal flux, for nearly all values of torque
command. In low speed applications (servo drives) this is generally an advan-
torque, larger values of IsflIs,+,
cause greater flux changes, i.e., greater sensitiv- tageous operating mode.
ity to detuning.
8.2.5 Power Loss Considerations
8.2.4 Influence of Saturation
The changes in slip frequency and flux level associated with detuned operation
The flux curves of Figure 8.7 show very large increases in flux for detuning to
will cause changes in the PR losses and core losses in the machine. In general,
smaller than correct slip frequency values. Clearly these large flux changes will
detuning to lower than correct slip frequency will reduce the PR loss and
not occur in a real machine because the magnetic core will saturate and limit increase the core loss. The opposite effects occur if the slip frequency is too
the flux change. Figure 8.8 illustrates the nature of the flux variation including large. The quantitative changes depend upon the motor parameters and operat-
saturation effects. The figure presents a set of calculated flux vs. slip frequency ing speed and can be quite different for different motors or different speeds of
curves for a 7.5 hp, 230 V NEMA-A induction machine. The dashed curve for
operation.
100%current indicates the calculated curve with a constant value of Xm equal At low speed (servo applications), the core losses are relatively low and the
to the value at rated flux (rated slip equal to 0.022). The very pronounced influ-
major concern is the winding PR loss. The rotor PR loss clearly increases as
ence of saturation is clearly evident [5],[6].
the slip frequency increases; hence operation at greater than the correct field
366 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation Phasor Diagram for Detuned Operation 367

I
1
&, = Constant
'+1 ,rn% cwmt

SIip Frequency - pu
Figure 8.8 Flux vs. slip frequency for 7.5 hp induction machine -
saturation included
oriented slip produces high rotor loss. The variation of rotor loss is shown in
Figure 8.lOusing the same normalization as in the general torque and flux char-
acteristic curves of Figure 8.6 and Figure 8.7. Although the curves indicate rel-
atively minor increases in rotor I2R loss as the system is detuned to higher slip
frequencies, the actual situation is considerably worse as a result of the torque
Slip Frequency - pu
loss for high slip frequency detuning. In most cases a low speed servo applica- Figure 8.9 Torque vs. slip frequency for 7.5 hp induction machine -
tion must be examined at a specific torque output. Detuning to high slip fre- saturation included
 368 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation 369
phasor Diagram for Detuned Operation

quency will cause a torquelampere loss and hence will require an increase in which create high values of flux are to the left on these curves). The percent
stator current. Thus, both stator and rotor losses will be increased and motor change in the motor input power is labeled along the axis. Since the curves are
overheating is a serious problem. The converse is true for detuning to lower for rated torque and rated speed, the output power is constant and the changes
slip frequency until the detuning is sufficiently large to cause operation past the in power input represent changes in motor losses. To the left of the tuned point
peak of the torque-slip frequency curve. Past the peak, the stator flR losses the flux is high and the core losses become large. To the right, the flux is low
increase causing very large increases in total losses. and the PR losses become large. Note that in both cases there is a maximum
Field Orientation Line efficiency point at a somewhat larger slip than the rated slip point where the
~valueof~/I., !J I 1 Value of IsT/Ism systems of Figure 8.1 1 have been tuned to operate with field orientation.
The curves in Figure 8.11 illustrate several trends which are true in general
for detuned operation. These include:
1) low slip detuning causes high core loss and low PR loss,
2) high slip detuning causes low core loss and high PR loss,
3) tuning to rated flux will not, in general, produce minimum losses since
most machines have their maximum efficiency below full rated slip, and
4) small, low efficiency machines (high rlx ratios) are much less sensi-
tive to detuning than large, high efficiency machines (low rlx ratios).

8.2.6 Example - Detuned Induction Machine Field Orientation

For the 100 hp, 460 volt, induction machine in the example in Chapter 6 with
pu parameters:
rs = 0.015 pu xis = 0.10 pu xm = 2.0pu
at rated frequency
r,. = 0.020 pu xlr = 0.10 pu rm = 50 pu
Figure 8.10 Field orientation rotor 12R loss characteristics The rated slip frequency is 0.0248. Express all answers in real units. Neglect
r, except when making efficiency calculations.
For operation at substantial motor speed the core losses become important Assume the torques calculated in parts a and b of the example in Chapter 6
and the situation is more complex. In general, detuning to high slip frequency are the values required to supply the loads. Find the required value of iqs to
results in high P-R loss but low core loss and the converse is true for low slip yield this torque if the slip calculator time constant Gr is equal to half of the
frequencies. Figure 8.11 131 illustrates two typical cases for operation with a correct value of 2, in parts a and b. In each case calculate the copper and core
fixed flux command cuirent and rated output torque. Since these curves include losses and the efficiency when detuned and when correctly tuned.
saturation effects, the rotor time constant is different for different amounts of
Solution -The solution assumes the following:
detuning. The horizontal scale is, therefore, labelled in per unit values of the
ratio of the rotor resistance used in the slip calculator ir to the actual rotor 1. The machine has 4 poles
resistance r, The curves show the variation of the rotor flux and the required 2. Core losses are calculated by using the Vm(air gap voltage) found from a
torque command current as the system is detuned (low values of slip frequency model that ignores core losses, CL = V; / R, .
370 8 Parameter Sensitivitv and Saturation Effects in Indirect Field Orientation Phasor Diagram for Detuned Operation 37 1

3. Efficiency is dominated by core losses and 1 2 losses.

~
~ F i e l d ~ O r i e n t a t i oLine
n
Part a:
I
I I
I I
First compute values for the tuned case:
1.5 ---- I I
1 - - - - - - - I- --- From the Chapter 6 example, is = 1.O53 - jO.7 105 = 1.27 angle -34" pu
I I Torque Component of
I
I
/&Stator Current From current division in the rotor loop
I

0.5
I
I
- - - - JI - - - - - - -II- - - . /%
Rotor Flux ir = -is[?
- + jx,
] = (1 -27 angle -34)
j2.0
0.020 + j2.1
0.0248
Change in Motor
I -I, = 1.129 angle -193.0"
-6% 5% 16%20% Power Input
Copper losses are
0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Controller /Actual Rotor Resistance r; /rr I ~ =R 1,2 r,+ 1,2 r,
~ l+ (1.129)~(0.020)
= (1. 2 7 ) 2 ( ~ .5)
a ) 1 hp Machine
R 0.0497 pu
I ~=
/Field Orientation Line Or, on a 100 hp base,

R (0.0497)(100 hp)(746 Whp) = 3.71 kW

I ~=
Core losses:
Torque Component of In per unit, core losses are assumed to be ~Zlr,, where Vmis the air gap volt-
Stator Current
I
age.
a

Find V,,,:

- I
- - - - JI - - - - b
L
I
I
t
- - - - - - 0.5
120%10%5% -6% 5% 10% 20%
I
> v
oChange in Motor
Power Input
-
Vm = 1.0 angle 0" - (1.27 angle - 34")(0.015+ jO.lO)

?, = 0.918 angle -5.9"

from which the core losses are,
Controller /Actual Rotor Resistance r,*/r,.

b),100 hp Machine
Figure 8.11 Field orientation rotor I 2R loss characteristics
372 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation 373
-phasor Diagram for Detuned Operation
(The error caused by inserting rm while keeping the same Vm is negligible. It
affects the third significant figure).
or on a 100HP base,
Power output:

CL = (0.0169) (100 hp) (746 E) = 1.26 kW I (1 - rr - ( 1.19) ( 1 - 0.0260) (0.020)

hp Pout =
S 0.0260
If losses are assumed to be core loss and I ~ loss
R only,
Pout = 1.061 p u = 79.1 kW
Pout -- 74.6 kW
'=Pout+12R+CL 74.6kW+3.71kW+1.26kW Efficiency:

Pout

- The tuned case for part (b) is similarly computed. Recall that
' - 79.lkW
= pout+ I2R + CL 79.1kw + 3.92kW + 1.34kW
Is = lqds= ( 1.25 - j0.228) pu. With the change iq frequency, the equivalent 17 = 0.938 = 93.8 % (The same as part A)
circuit becomes:

Detuned Operation: Part a

Torque is unchanged; speed is unchanged. The slip calculator value of Z, is
112 of the actual value. Therefore,

Since,

Te -
- - 1+P2 where p = 1st
- = iqs
-
To 1 + (43)' Ist Ids

Ir = -Is
j4.00 and To = --
= -1.1 9 angle 0.040' pu
j4.00 + j0.20 + -
0.020
0.0260 we need to find P to give the same value of T,/To as in the tuned case.
T, 1+p2
-
0.020 + jo. 1755) = 0.946 angle 14.6' PU
377
In the tuned case, a = 1 and - =
To l + ( l ) P 2
(1)( P ) = P

I'R losses: l:rs = 1; rr = (1.27)' (0.025)~+ (1.19)'(0.020) = 0.0525 pu = Therefore,

3.92 kW
Core losses:
374 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation phasor Diagram for Detuned Operation 375

For the detuned case, we keep torque the same,

Find new 0

+ 4 p 2 ) = 2 p + 2p3
2.60 ( 1
Therefore, 2p3 + ( - 1 0 . 4 ~+~ 2p
) + (-2.60) = 0
Solving

p = 5.053,0.147 fj0.502
Since p must be real, p = 5.053
Then, since P = = fpi, iqs = P ids with ids unchanged
1st Ids
iqs = (5.053) (60.3)= 305 A peak

Since

then
-
V m = 0.549 L 20.1"
I ~ losses:
R

= 0.181 pu = 13.5 k W
or= Same = 377-9.35 = 367 radk
Core losses:
we= 36.3 + 367.6 = 403.9 radts
The slip is,

Efficiency:

Pout -
- 7.46 kW
= 0.842
The equivalent circuit is:
' = p o u t + 1 2 ~ + C L 7.46kW+1.35kW+0.45kW
 376
-
8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation Transient Response
/
377

Part b
The torque and speed are unchanged;
a = 2 (Same as detuned Part a)
1% losses:
-
0 tuned = 165 = 5.46 = -
30.2
T
1; rS+ Z$ rr = (2.48) (0.015) + (2.36) (0.020)
To
Therefore, 5.46 = 1 + p 2
(Nv 1 1 2 losses
~ = 0.204 pu = 15.2 kW
1+ (ap)2
Solve for p: P = 10.83 Core loss
Since id, is constant = 30.2 A peak

Then, Sw, = -
2p = (
2, 0.279sec
= 77.6 rad/s Efficiency

- 79.lkW
' Pout
= pout + 1 2 +~CL
-
79.1 kW + 15.2 k W +0.797 kW
Therefore, a, = 735 + 77.6 = 8 13.6 rads

8.3 Transient Response

In addition to the steady state errors described above, the loss of field orienta-
tion caused by detuning will result in a modification of the transient response.
In particular, the instantaneous torque response to changes in torque command
will no longer hold [1],[3],[7].

8.3.1 Torque Response with Slip Gain Error

TO evaluate this situation, the transient equations describing the torque
response for a change in torque command will be derived. The analysis
assumes:
(1) the only parameter error is Er/pr # L,/rr

Let is = $ - jIds = 327 - j3O.2 = 3281-5.3' A peak = 2.48 L- 5.3' pu (2) the flux command ied:, = 1% is constant

and the following error variables are defined:

A =A -1 = error in q-axis rotor flux linkage

Ah;, = ,:A -h z = error in d-axis rotor flux linkage

378 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation
-Transient Response 379

as the final error equations for the error fluxes. The torque error is given by
Er L r
ATr = Tr - Tr = -- = error in rotor time constant
'r 'r
o,*= command slip frequency
The two equations for the flux errors are a pair of linear first order differen-
Eliminating the rotor currents from the rotor voltage equations yields tial equations which exhibit oscillatory response. The eigenvalues which char-
acterize the response are easily shown to be given by

and the natural frequency and damping ratio are therefore

Rewriting these equations, in a reference frame oriented to h$. , in terms

of the error variables results in (recognizing that ideally A$.= 0 )
1
(8.3-12)
<=
This last expression shows that the response becomes progressively more
poorly damped as the torque command (and o :) increase. Note that the input
to the error equations is the i$ term in eqn (8.3-7)

The slip relation and the steady state flux equation provide the following
control equations.

and that with AT, = 0 or i$ = 0 there is no error response. For any specific
torque command the flux errors are proportional to AT, although the presence
of o: in the equations also influences the error amplitudes. Note also that the
settling time of the torque response is proportional to T, only; it is independent
Led*, = L,I: (8.3-6)
of the error and the slip frequenq w:
.
Using these two control equations to eliminate Led*, from eqns (8.3-3) and A small signal transfer function relating the output torque to the torque
(8.34) yields command current can be derived using standard methods. The result is

where
-=
Ai& 2 2 Lr
a,s2+a,s+(ag+a4)
s2+a5s+a6 I (8.3-13)
380 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation Transient Response 381

and the quantities with zero as the last subscript represent the operating point at
which the linearization is carried out.
The poles of this transfer function are those given in eqn (8.3-10). The
zeros are also complex and exactly cancel the poles when the system is cor-
rectly tuned (2, = zr ) yielding instantaneous torque response. Detuning
causes the zeros to move away from the poles and results in non-instantaneous
response. Figure 8.12 illustrates this pole zero pattern for the same two
machines as in Figure 8.11. Note that the pole zero separation increases as the
torque output increases and that the larger machine (100 hp) has greater separa-
tion than the small machine (1 hp). Figure 8.13 gives several torque step
responses for these same machines to illustrate the nature of the small signal
transient caused by detuning. '

Although detuning eliminates the desirable instantaneous torque response

of a field oriented system, the presence of the outer control loop (speed or posi-
tion control) will often compensate for the less responsive detuned behavior.
Unless the detuning becomes very severe, the transient effects of detuning are
less serious than the steady state power loss increases described in the preced-
ing sections.
Figure 8.12 Pole - zero plot of the torque vs. torque command transfer
function with a slip gain error of 0.50 a) for a 1 hp machine b)
for a 100 hp machine
I

382 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation Transient Response 383

8.3.2 Block Diagram Representation of Slip Gain Error

An alternative formulation of the slip gain error problem in terms of a block
diagram offers a different, more control system oriented means of visualizing
the system behavior [8].
A block diagram with i;, and ied, as inputs and Legr and Lzr as outputs
can be constructed based on eqns (8.3-1) and (8.3-2). The result is shown in
Figure 8.14 where the slip frequency a, is taken as a constant value about

I I I I 1
0.20 0.40 0.60 0.80 1.00
Time - (s)

Figure 8.14 Block diagram representation of rotor voltage equations

which the system is linearized. As indicated in the figure, if the slip relation is
utilized in an indirect field oriented controller, the net input into the q-axis
rotor flux summing point is zero if the rotor time constant 2, and the slip gain
calculation time constant ir are equal. In this correctly tuned case the outer
loop containing a, is compensated and is effectively opened. The block dia-
gram then reduces to the upper input path fed by i z and the first order loop at
the right side with the output A:,, as is expected based on earlier results in
Chapter 6. A

If, however, there is detuning in T , , the cancellation described above will

Time- (s) not take place and the outer loop will not be rendered inoperative. With i z , ir

Figure 8.13 Torque step response for open loop operation of field oriented control
a) with a I hp machine b) with a 100 hp machine
 384 Transient Response 385
8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation

and h s all constant, the slip frequency a*, will also be constant. The charac-
, teristic equation associated withthe outer loop in Figure 8.14 is

or solving for the roots

T , I I I
I
I I I
+ 30%
I
Torque

1
JI::-30% Torque
I
I
I

which agrees with the earlier result in eqn (8.3-10).

8.3.3 Simulation of Transient Response with Slip Gain Error

The transient response of a field oriented induction machine with an incor-
rectly tuned slip calculator can be conveniently simulated using the concept
outlined in Figure 6.10. In this figure the machine is modelled in the actual Time - (Seconds)
rotor flux frame of the machine and the controller is modeled in the controller's
estimated rotor flux frame. The error angle, Age is the angle between the
machine and controller axes and is zero when the machine is in a correct field
oriented state. It is clear from Figure 8.10, and from the development in previ- 1.4$ I '
! Start Up Transient + 30% Torque
I
I - 30% Torque
I
I
ous sections, that all transient or steady state behavior is independent of rotor
speed and thus simulations can be carried out at zero speed for simplicity.
Figures 8.15 and 8. I6 illustrate the behavior of a machine for four different
values of the slip gain (time constant ratio) a = z r / G r , including the ratio 1.0
which represents correct tuning [9]. The machine is the same 100 hp, 460 volt
machine used for the simulations in Figure 6.13 in Section 6.3.5. The simula-
tions were made for a commanded torque current to flux current ratio P of 2.0
with a 1.0 pu flux command.
The torque step responses shown in Figure 8.15 are in agreement with the
corresponding curves for the 100 hp machine in Figure 8.13. Note that higher
A
- Time - (Seconds)
values of the slip gain a = z r / z r result in a more oscillatory response as was
shown in the linearized analysis of Section 8.3.1. The flux responses in Figure
8.15 also show this behavior. The start up transients shown in the figure illus- 100hp, 460 volt, 60Hz,4 Pole Induction Motor
trate that the initial start up is also affected by the slip gain in the same way, Per Unit Parameters
with larger values yielding less well damped responses. Note also that unlike Stator Resistance = 0.010pu Stator Leakage.Reactance= 0.10pu
Rotor Resistance = 0.015 u Rotor Leakage Reactance = 0.10 pu
the step response behavior where field orientation yields a unique, transient Magnetizing Reactance = 80 pu Inertia Constant (H) = 0.5 Second
free, response, there is nothing unique about field orientation during the start Figure 8.15 Influence of detuning on torque and flux - 100 hp machine
up transient. The field oriented case ( a = 2 , / 2 , = 1 .O) is simply one mem-
386 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation Transient Response 387

ber of a family of start up transients which fits into a continuum of responses

with no special characteristic associated with field orientation. For the particu-
lar machine of Figure 8.15 the start up transient is underdamped for the choice
of applied torque and flux command current components (P = iqse*l idse* = 2.0)
used in the simulation. For smaller values of p, the start up transient is better
damped and becomes non-oscillatory when = 0.5 as can be seen in Figure
6.13.
Figure 8.16 illustrates the slip frequency and error angle responses corre-
I I I sponding to the torque and flux responses of Figure 8.15. The slip frequency
I I I
characteristics in Figure 8.16 clearly show the linearly proportional effect of
I I I I I I I I
the slip gain a = T,/T, on the steady state slip frequency, which is, at least in
part, the origin of the tern s l i p gain. High values of the slip gain a clearly yield
I I I I I large values of overshoot in the slip frequency during the start up transient and
I I I I I I I I I I
0 I I I I I I I I I J this high gain can be viewed as the cause of the poor damping associated with
0 1 2 3 4 5 6 7 8 9 1 0 the start up transient.
Time - (Seconds)
The error angle characteristics shown in Figure 8.16 illustrate the influence
I ' I I I I I I I I I of slip gain errors on the alignment of the machine and controller reference
,,I A istart Llp Transient
I
I + 30% Torque I
I
I
- 3b% Torque I axes. Again the oscillatory nature of the response increases with increasing val-
1 a= 2.0 1 I I I I I I I
ues of the slip gain. Note that while there are zero steady state errors between
the estimated and actual slip frequencies (as is forced by the controller), there
are finite steady state errors in the rotor flux angles. This implies that the actual
motor currents differ from their commanded values in such a way that the
actual slip frequency, calculated with the actual value of T, agrees with theA

commanded slip frequency calculated with the estimated value of 7,. This
requires that the actual ratio of iq//ihe differ from the commanded value of
iqse*/idse* by exactly the value of the slip gain error

1
0 1
I
I
2
I
I
3
I
I
4
I
1
5
I
I
6
I
I
7
I
I
8
I
I
9
I
I
l
I

d
Time - (Seconds)
This result allows a solution for the steady state error angle in terms of the two
100 hp, 460 volt, 60 Hz, 4 Pole Induction Motor
parameters a and p
Per Unit Parameters
Stator Resistance = 0.010 pu Stator Leakage Reactance = 0.10 pu
Rotor Resistance = 0.015 u Rotor Leakage Reactance = 0.10 pu
Magnetizing Reactance = r.0 pu Inertia Constant (H) = 0.5 Second
Figure 8.16 Influence of detuning on slip frequency and error angle - 100 from which the steady state value of the actual motor currents can be found us-
hp machine
388 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation Optimal Selection of Flux Level 389

ing 8.4.1 Maximum Torque Per Stator Ampere

Combining eqns (8.2-9), (8.2-6) and the flux equation hdr = Lmids allows
rewriting the torque of a field orientation controlled motor as

8.4 Optimal Selection of Flux Level

The preceding sections clearly show that the selection of flux level has impor-
T,= [ ---
; ; L y S ] SO,= [;"I SO,

tant impacts on a variety of operating characteristics from efficiency to the Thus, for constant flux operation, a field oriented machine produces a torque
effects of detuning. Optimal operation involves the proper selection of the flux which is linearly related to the slip frequency. The field orientation lines
level to satisfy the particular optimal requirements of specific applications.The described by Eq. (8.4-1) can be superimposed on the motor torque-slip fre-
influence of magnetic saturation in this selection of flux level is very signifi- quency characteristic to relate the field orientation and the motor characteristic
cant and must be incorporated in the selection process to obtain valid and use- as was done in Figure 8.4 and Figure 8.9. One such line can be drawn for each
ful results [ 5 ] . value of flux selected in the field orientation controller (via flux command).
While individual application requirements can create a large number of Figure 8.17 [5] illustrates this concept for three different values of flux.
specialized optimization criteria, the following four optimal modes are of gen-
eral interest and will be considered in some detail.
Peak Torque per Stator Ampere - In servo applications, when continu-
ous operation at low speeds where core loss is not significant is required, peak
torque per stator ampere operation is important. This allows obtaining mini-
mum response time for the particular peak current available from the motor
power supply.
Peak Drive Efficiency -Where long periods of continuous, relatively high
speed operation are required, drive efficiency may become an important con-
sideration.
Reduced Sensitivity to Controller Detuning - Temperature changes
cause the motor resistance to vary and this can result in detuning of the slip cal-
culator. Flux changes cause L, and L, to change which also affects the slip cal-
culation. In some cases, reduced sensitivity to this detuning is an important
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
issue.
Slip Frequency ( p.u.)
Optimal Utilization of Converter - The power converter supplying the
motor will always be limited in voltage and current capacity. Proper coordina- Figure 8.17 Field orientation lines of constant rotor flux plotted on
constant stator current torque curves
tion of the motor and converter to meet the overall requirements of the drive is
always a significant consideration. Each of these criteria will now be exam- Selection of the flux level for maximum torque per stator ampere is readily
ined. visualized from Figure 8.17. Maximum torque per stator ampere corresponds
to the peak of the motor characteristic and hence the field orientation line
should be selected to intersect the peak of the curve corresponding to the
selected current level where peak torque is desired. Note that, because of satu-
390 8 Parameter Sensitivitv and Saturation Effects in Indirect Field Orientatinn Optimal Selection of Flux Level 391

ration and the effect it has on the location of the peak of the motor characteris-
...............................................................
tic, there is one field orientation line (and hence flux level) which is nearly
optimal for all motor currents. This is in sharp contrast to the ideal, non-satu-
rating motor where the peak torque per stator ampere always occurs at the
i -- optimil i
..............................................................
.- :Conventional I
same slip frequency. Saturation, therefore, tends to make one single flux level
the optimal choice for all levels of stator current and in this sense is a 'good' .........
effect. It also, of course, has the usual bad effect of reducing the peak torque
for larger values of current. ..............................................
8.4.2 Maximum Drive ESJiciency
The efficiency of an induction motor is a relatively complex function of speed,
torque, and flux. Optimal efficiency operation requires the proper balance
between the motor I ~ Rlosses and the core losses. This balance can be con-
trolled by selecting the flux level in relation to the torque and speed of the
machine. Thus, for any specific torque and speed, there is a specific flux level Speed - (pu)
which minimizes the total losses and maximizes the efficiency. Unfortunately, Figure 8.18 Efficiency comparison for conventional and optimal constant
this flux level is different for each combination of torque and speed. power operation -rated power, 5 hp high efficiency motor
In general the optimal flux level varies directly with torque and inversely
with speed. At low torque the flux should be set low to minimize the core and
magnetizing losses (and also the motor acoustic noise). At higher torque levels
the flux must be increased to avoid large 12R losses resulting from low torque
per ampere. The flux level must also be reduced at high speed to avoid exces-
sive core loss.
An illustration of the improvement in efficiency which can be attained by
optimal flux selection is shown in Figure 8.18 [lo]. This figure shows the effi-
ciency of a high efficiency 5 hp machine operated above base speed with both
conventional constant voltage control (flux varying as 1lfl and with the optimal
choice of flux. In both cases the power output was held constant at rated value
(torque varying as llspeed). Note that a significant efficiency improvement is
possible, especially at high speed, and that there is a 'best optimal' speed for
rated power at about 1.6 pu for this machine. Figure 8.19 [lo] shows the
inverter voltage and kVA required to achieve the optimal curve in Figure 8.18. 0.5 1.O 1.5 2.0
The results in Figure 8.18 and Figure 8.19 are typical of high efficiency
Speed - (pu)
machines. In general, the reduction of flux level with speed to attain optimal Figure 8.19 Voltage and kVA for optimal efficiency at rated load - 5 hp
efficiency is less rapid than the llfvariation normally suggested for constant hp high efficiency motor
operation. This is especially true for high efficiency machines. In fact, these
machines typically have better efficiency at higher than rated flux for all
 392 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation Optimal Selection of Flux Level 393

speeds. In contrast, standard design motors are often better at less than rated torque load), the system will always seek the maximum efficiency flux level.
flux over a significant range of speeds. The existence of a 'best optimal' oper- The flux level changes are made slowly relative to the response time of the
ating point is also of some interest since, given the choice, this point corre- drive except when a large speed error is detected. In the latter case, the flux is
sponds to the highest attainable efficiency for that particular output power. This increased to a level which produces hgh torque per stator ampere to assist in
speed always is well above rated speed for rated power output. responding to the speed error.
Unfortunately, there are no general analytical results which provide infor-
mation on the best flux level for specific torque and speed conditions. The core 8.4.3 Reduced Sensitivity to Controller Detuning
and copper losses depend strongly on the saturation characteristic of the The results presented in Figure 8.6 and Figure 8.7 clearly show that a small
machine and hence the problem is strongly nonlinear. Calculations based on value of the ratio Is*$ results in a low sensitivity to detuning. While this
measured saturation and core loss are quite feasible, but direct measurement of observation is generally useful in understanding how sensitive a specific
motor performance is probably the best guide to proper selection of flux in machine is to detuning, the curves in Figure 8.6 and Figure 8.7 can also be
each individual case. used to help select a machine which exhibits less detuning sensitivity over a
A closed loop controller which automatically assures maximum efficiency load range or how to select the flux level Is$ or slip gain setting to reduce the
operation has been demonstrated. Figure 8.20 is a block diagram of this system detuning sensitivity. For example, with a machine having an IsTII,$ ratio of 1.0
used in a speed regulated drive [l 11. a detuned slip gain setting with a = 1.3 on the cold motor will result in virtu-
ally no torque gain or loss as the motor heats and thus virtually no detuning
Motor
Current effect. Similar gain settings can be found for machines with other values of the
Rectljier Regul ' ratio
I T Inverter
PW The single most significant factor which increases detuning sensitivity is to

E-7Ilrt;tFP
r-lf Eficiency Field
increase IsflIs4. Thus, for example, detuning effects are much more significant
in the field weakening region where Is$ is much smaller and hence Is7/1,$ is
much larger. Any changes which reduce IST/1,$ will be helpful in reducing
detuning effects.
Oriented Saturation of the air gap flux also helps to desensitize a field oriented con-
pin Controlle
Controlle troller to detuning. Generally, the lower slip gains which are needed to achieve
-u
I high flux produce a reduced sensitivity to parameter errors. This is because

':*zp
Signal
Overri
de*
although errors in the rotor time constant still cause incorrect positioning of the
Speed stator current vector, the amplitude of the rotor flux does not change apprecia-
bly. Thus, the output torque deviations are limited to the effects of vector mis-
positioning. Figure 8.21 [6] shows the effects on torque linearity for both the
saturating machine and a constant parameter, non-saturating machine for a 3: 1
Figure 8.20 Block diagram of optimal efficiency controller detuning range.
While the 3: 1 range of detuning is excessive, it demonstrates the beneficial
The system employs Hall effect power sensors to measure the drive input effects of saturation (low slip gains). At very low flux levels (high slip gains),
power. A microprocessor based controller makes small step changes in flux the differences between the saturating machine and the constant parameter
command and seeks the flux level which results in minimum power input. machine are seen to be negligible.
Since the power output is held constant by the speed regulator (for a fixed
394 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation Optimal Selection of Flux Level 395

Initial

Linear
FIW;/J Final
Flux

0.5 1.O 1.5

Torque Command - f pu)
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
Figure 8.21 Detuning sensitivity characteristica showing that saturation model Slip Frequency ( p. u.)
demonstrates less sensitivity to detuning than a constant parameter Figure 8.22 Operating point selection for an amplifier current limit which
model, a = normalized gain, see eqn (8.2-1 1) is lower than the machine's thermal limit

8.4.4 Optimal Utilization of the Converter

8.4.4.2 Converter Current and Voltage Limits
In general, flux selection may be determined by machine limits (primarily ther-
mal) or by converter limits (primarily current and voltage). If the thermal limits For high speed applications, the back emf voltage requirements of the machine
of the machine dominate, then the operating flux may be chosen to optimize may also limit the operating flux selection. Figure 8.23 [5] shows the change in
torque per ampere as presented in Section 8.4.1 above. However, the converter maximum torque per ampere operating point as required by both a converter
ratings for current and/or voltage are often below those required to operate at current and a voltage limit.
the machine limited values. Thus, it is appropriate to consider the selection of To evaluate this limit the stator voltage equations for field oriented induc-
flux under converter limited conditions. tion machines must be invoked. In scalar, steady state, synchronous frame
form they are:
8.4.4.1 Converter Current Limits
Figure 8.22 shows the change in the maximum torque per ampere operating
point corresponding to a converter current limit which is less than that the
machine's thermal limit. v;, = rs I;, - w e ~ t s l i s (8.4-3)
It is apparent that saturation reduces the effective change in torquelslip Given a maximum limit on the stator voltage and current amplitudes, and a
operating point as compared to the non-saturating machine model. Saturation value of slip frequency Sw, as defined by eqn (8.2-6) it is possible to compute
thus makes the system less sensitive to this stator current limit. the magnetizing current and rotor flux iteratively. As a reasonable approxima-
796 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation References 397

Initial
References
F l u/ L.J. Garces, "Parameter Adaptation for the Speed-ControlledStatic AC Drive with a
Linear Squirrel Cage Induction Motor", IEEE-IAS Transactions, Vol. IA-16, No. 2, March/
Machine
Model Final April 1980, pp. 173-178.
M. Koyuma et al, "Effects of Parameter Changes on Coordinate Control System of In-
duction Motor". Conference Record of the International Conference on Power Electron-
ics, March, 1983, pp. 684-695.
K.B. Nordin. D.W. Novotny and D.S. Zinger, "The Influence of Motor Parameter Devi-
ations in Feedforward Field Orientation Drive Systems," IEEE-IAS Transactions, Vol.
IA-21, No. 4, July/Aug 1985, pp. 1009-1015.
J. Loehrke, R.D. Lorenz and D.W. Novotny, "Torque Characteristics of Field Oriented
Induction Machines," Conference Record, First Annual Conference on Applied Motion
Control, Minneapolis, MN,June 1985.
R.D. Lorenz and D.W. Novotny, "Optimal Utilization of Induction Machines in Field
Oriented Drives," Journal of Electrical and Electronics Engineering, Australia, Vol. 10,
No. 2, June 1990, pp. 95-100.
Slip Frequency ( p.u.) F.M.H.Khater, R.D. Lorenz, D.W. Novotny and K. Tang, "Selection of Flux Level in
Field Oriented Induction Machine Controllers with Consideration of Magnetic Satura-
Figure 8.23 Operating point selection for an amplifier current limit which is tion Effects," IEEE-IAS Transactions, Vol. IA-23, No. 2, MarcMApril 1987, pp. 276-
lower that the machine thermal limit, and an amplifier voltage 282.
limit which is matched to the back emf voltage K.H. Bayer and F. Blaschke, "Stability Problems with the Control of Induction Motors
Using the Method of Field Orientation", 2nd WAC Symposium Conference Record,
tion, the magnetizing and maximum torque producing currents may be esti-
1977, pp. 483-492.
mated respectively by the following equations:
R. D. Lorenz and D.W. Novotny, "A Control System Perspective of Field Oriented Con-
trol for AC Servo Drives," Proceedings of Control Expo 88, pp. (XVllI-I)-(XVlJI-12).
June 1988.
N.R. Garrigan and D.W. Novotny, "Start-up and Sensor-Parameter Error Transients in
Field Oriented Induction Machines," Proc. IEEE-IAS Annual Meeting, October 1995.
F.M.H. Khater and D.W. Novomy, "Efficiency Optimization for Constant Horsepower
Operation of Induction Machines," Proceedings, International Conference on Evolution
where: and Modem Aspects of Induction Machines, July 1986, Torino, Italy, pp. 9-16.
V- = maximum amplifier voltage D.S. Kirschen. D.W. Novotny and T.A. Lipo, "Optimal Efficiency Control of an Induc-
tion Motor Drive," IEEE Trans. on Energy Conversion, Vol. EC-2, No. 1, March 1987,
I, = maximum arnplifier current pp. 70-76.
or= rotor electrical speed
oe= or+Soe
This optimization may be viewed as using flux to impedance match the
machine to the source (converter).
 Problems 399
798 8 Parameter Sensitivity and Saturation Effects in Indirect Field Orientation

a) Find IST and Is# for operation with rated terminal voltage VR = 1.0 pu,
Problems rated current IR = 1.17 pu, rated frequency fR = 1.0 pu, and rated torque TR =
1.02 pu.
Problem 8-1 Detuned Field Orientation Operation in the Steady State
For an induction machine with per unit parameters rs = 0.03, XI, = 0.10, x,,, b) Find the terminal voltage Vs required to maintain the same Is,, Is4 and
= 2.0, r, = 0.03, XI, = 0.10 operated under indirect field oriented control: torque at 5 Hz. Compare this value with the voltage corresponding to constant
voltsihz operation.
a) evaluate the torque vs. torque command current (iqs) characteristic if the
machine is operated with a flux command current (ids) which yields rated flux. c) Find the value of the torque if IST is held constant at the value in part (a)
Plot the curve from zero to 150% of rated torque. and the new value of IST needed to bring the torque back to the value of part (a)
if the slip gain is too small by a factor of 2. The value of IS# (command value)
b) find the torque vs. torque command current (iqs) characteristics which is held constant at the value in part (a). Find the rotor and stator fluxes for the
result under the conditions of part a) but for the cases where the slip calculator original value of IST (express the fluxes as a multiple of the values for rated
gain is 50% too large and 50% too small. operation).

d) Find and plot (using simulation) the torque transient following a step
Problem 8-2 Voltslhz for Maximum Torquelampere Operation change in IST from zero to the value in part (a) for the case of correct slip gain
Use the results of problem 5-2 to find the required voltsihz to cause opera- and for the slip gain error situation in part (c). Compare the predicted steady
tion at maximum torquelampere (peak of curve of part 5-2a) for the given state torque with the result in part (c). The value of Is# (command value) is
machine at 60 Hz and at 5 Hz. Comment on how main flux saturation will held constant at the value in part (a).
affect these answers. Explain how these results illustrate the difficulty of
obtaining good starting torque (maximum torquelampere) in an open loop
adjustable speed drive.

Problem 8-3 Detuned Induction Machine Field Orientation

Consider a 100 hp induction machine with parameters and performance
data as follows:
460 volts (line to line) 100 hp 60 hz 4 pole
ZB = 2.84 R VB= 266 volts IB = 93.6 amp TB= 395.8
rs = r, = 0.015 pu = 0.0425 Q xls = XI, = 0.100 pu = 0.284 R
x, = 3.0 pu = 8.51 Q J = 2.0 kg - m2
Rated Operation SR = 0.0177 pu IR = 1.17 pu cos OR = 0.880
The machine is operated from a current regulated variable frequency supply.
For this problem assume an ideal current regulator.
I

9 Field Weakening Operation

9.1 Introduction
So far we have examined carefully how torque can be accurately controlled in
an induction motor drive. However, we have made little mention of the
demands of the load. Clearly the electromagnetic torque and load torque must
be equal in magnitude and opposing in direction during steady state operation.
In general we can separate loads into four categories, 1) constant speed, vari-
able or constant torque loads, 2) adjustable speed, constant torque loads, 3)
adjustable speed, variable torque loads and 4) variable speed, variable torque
loads.
In category #1 are included those typical applications which require only a
fixed frequency supply such as constant speed pumps and fans, and compres-
sors. This category is serviced by conventional induction motors operating
from the utility 50 or 60 Hz supply in which the pole number is chosen to be
compatible with the speed requirement of the application. (Note that constant
speed is, in this case, slightly violated since the rotor speed changes by a few
per cent with changes in load from zero to rated value.)
In category #2 applications such as elevators, fiber spinning, hoists, etc. are
found. Note that here, constant torque is meant to imply that the torque
changes only very slowly, such as when an elevator stops at each floor. By the
term adjustable in contrast to the term variable is meant that the speed is
adjusted externally by an operator, typically a human operator. In these appli-
cations, the management of acceleration or deceleration is also typically
important.
In category #3 are found applications such as fans, pumps and compressors
which are called upon to operate with changing speed. Because of the inherent
flow versus pressure (or head) characteristics of such devices, the torque load
follows square law characteristic as a function of speed. As a result the power
requirement varies as the cube of speed. All three of the above applications are
adequately served by the open loop speed controller of Figure 1.7 or Figure
1.13. With a correctly tuned slip frequency compensator (the gain box f s l i p / I i
402 9 Field Weakemg Operation Torque Demand and Capability of High Performance Drives 403

in these figures), a speed regulation capability of 1% or less can be obtained at

rated speed.
Finally, category #4 contains those applications in which the speed is vari-
able rather than merely adjustable implying that dynamic requirements are
attached to the speed andlor torque response. Category #4 contains those high
Constant Torque Constant Power
performance applications addressed by this book, namely, traction motors, ser- Limit Region Limit Region
vos and the like.
Constant Power*Speed
Limit Region
9.2 Torque Demand and Capability of High Performance I

Drives
Prior to the modem ac inverter, variable speed, variable torque applications
were the province solely of dc motor drives. The torque available from a dc
motor was limited by three factors 1) the maximum armature and field currents
which could be safely handled by the dc motor and dc power converter, 2) the
maximum available voltage obtainable from the dc armature power supply and
3) the maximum current that could be handled by the brushlcommutator mech-
Rotor Speed (Per Unit)
anism at a particular speed [I].
If the field and armature currents are set to their maximum permissible val-
ues then, from eqn (5.24), Figure 9.1 Torque vs. speed capability curve for dc motor drive

As speed continues to increase the minimum value of field current needed

to supply rated armature current decreases in accordance with eqn 5.2-1 which
where hf, -- L max and I f , is the maximum field current. This con- specifies that as speed increases, the field flux linkage must decrease inversely
straint sets the limit curve #1 on F~gure9.1. with speed if Ea is to be maintained constant. However, as the field current is
When the speed of the dc motor increases, the voltage applied to the motor decreased, the distortion of the airgap flux caused by the armature current
must increase to match the back emf of the dc motor as defined by eqn 5.2-1. increases creating a very non-uniform distribution of bar to bar voltage on the
Eventually, at some speed, for any practical power supply, the voltage can not commutator. This distortion and the high speed eventually cause bar to bar
be further increased to maintain the current at the limit value Ia, If the commutator voltages to reach unacceptable levels. Sparking begins to occur
field current is not changed at this point the current will rapidly drop to zero, usually terminated by spectacular fireworks, a flashover of the brush and com-
i.e. at the speed where Vdc(max) = Ea ,where Vdc(max) is the dc output volt- mutator rigging. Hence, above a certain speed the reduction of the field current
age of the power supply. Torque can continue to be maintained however, if the to maintain constant armature current must be curtailed. The armature current
emf of the motor is held at a value lower than Vdc( m a x ) . The minimum value must now begin to decrease. Since the need for commutating ability is propor-
of field current at any speed is set by the maximum permissible value of I , . tional to the current to be cornmutated, the field current (as well as interpole
Hence, another boundary curve, curve #2 in Figure 9.1, can be identified which current) now becomes proportional to the armature current resulting in a third
is set by the saturation of the dc power supply. This curve is usually labelled as mode of operation termed series field operation, curve #3. The comer point at
the constant power region. which the machine reaches crossover between curve #2 and curve #3 is gener-
 404 9 Field Weakening Operation Induction Machine Control Strategy for Constant Power Operation 405

ally considered to be rated maximum speed. Operation along curve #3 is rarely 9.3 Induction Machine Control Strategy for Constant Power
used since commutating ability is marginal and the motor enters a speed region Operation
where physical integrity of the commutator bars becomes difficult to maintain. Referring back to Figure 5.21 it is quite easy to see that a maximum torque
In addition the torque capability in this region is not much more than windage requirement imposed by curve #1 on Figure 9.1 is readily satisfied in the case
and friction losses making the additional complexity of operating in this region of a vector controlled induction motor. In particular, when the slip frequency is
difficult to justify. The variation of the motor electrical variables needed to held constant then both the reactance j ( L , / L , ) X, and the resistance
obtain Figure 9.1 is shown in Figure 9.2. 2
( L , / L r ) [ r r / (SO,) ] O, are proportional to the applied frequency. Hence if
the stator current amplitude Is is also held constant by means of current regu-
lator, then the division of Is between the magnetizing component Ise and the
Armature Current torque producing component IsT remains unchanged and thus the torque is

1.0- , \ \Armature Voltage constant, as demonstrated by eqn 5.9-9. Except for the slightly different effect
of the stator I r drop the curves of the electrical variables are remarkably simi-
lar to the dc machine.
In the high speed range, the output torque capability of the induction
machine depends on the field weakening strategy. The conventional method for
the field weakening is to vary the rotor flux reference in proportion to the
inverse of the rotor speed [2] as shown in Figure 9.3. The torque therefore

Induction
Machine
with Field
Figure 9.2 Electrical variables corresponding to Figure 9.1 as a function of Oriented
speed Controller

It should be noted that since the armature current and field current are con-
stant the constraint set by curve #1 on Figure 9.1 inherently produces a con-
stant torque limit on the speed-torque plane while constant armature current
and constant back emf constraint of curve #2 imposes inherently a constant
power limit. Neglecting armature resistance, Curve #3 imposes an inverse Figure 9.3 Block diagram of field weakening using rotor flux weakening
. square law torque characteristic. Popular acceptance of these capability limita-
tions of a dc machine has caused users of electrical equipment to define these will decrease inversely in accordance with eqn (6.2-14). A plot of the corre-
same characteristics from the point of view of torque requirements not only to sponding electrical variables is shown in Figure 9.4. The parameters used to
be imposed on a dc machine but any type of motor drive. It remains to be seen calculate these curves correspond to the parameters of the example in Section
whether the maximum capability curves of an induction machine are consistent 5.10.1. However in this case, one per unit voltage is defined to be that voltage
with torque requirements, as adopted from the maximum capability curves of a which produces 3 four to one field weakening range. The maximum torque in
dc machine. the torque limited region was selected to be 0.719 per unit.
406 9 Field Weakening Operation Inverter Imposed Voltage and Current Limits
407

voltage which the inverter can apply to the machine. Therefore, considering the
limited voltage and current capacities, it is desirable to consider a control
scheme which yields the best possible torque per ampere over the entire speed
range. In the high speed range, the output torque capability of the induction
machine depends on the field weakening strategy. The conventional method for
field weakening, discussed in the previous section, is to vary the rotor flux ref-
erence in proportion to the inverse of the rotor speed. In this method, if maxi-
mum inverter voltage is reached at one per unit speed, the voltage margin
required to regulate current is not maintained. The problem can be overcome if
we turn our attention to the stator flux linkage.
Assume that the voltage applied to the motor reaches its limit. The d,q axis
model of the machine in the synchronous reference frame is, from eqns (6.2-1)
and (6.2-2),

Figure 9.4 Electrical variables of induction machine satisfying the torque v:s = rs& + p ~ : s-
vs. speed requirements of Figure 9.1
In the steady state the p becomes zero. In field orientation
Note that the constant power limit region can apparently be extended with-
out limit. However, it is important to note that, in contrast to the dc machine, A:, = ~ ~ i : ,
the voltage of the power supply must continue to increase during the power
and
limited region to counteract the increased voltage drop across the stator leak-
age inductance. It should be noted that the price to be paid for an extended con-
stant power region is a sacrifice in the torque producing capability. For A;, = LSiiS+ ~ , , , i ; ~
example, from Figure 9.4 note that one per unit speed is reached at only 95% where .
of the voltage capability of the inverter. Hence, 5% of the torque production
capability is being sacrificed to meet the constant horsepower limit over the 4
to 1 speed range. In essence, the torque capability curve of an induction
machine is, in fact, somewhat greater than the torque requirements imposed by Combining eqns (9.4-4) and (9.4-5),
the dc motor torque-speed characteristics. In the next section we will explore
exactly how this additional torque capability of the induction machine can be
extracted.

9.4 Inverter Imposed Voltage and Current Limits

The maximum output torque and power developed by the machine is ulti-
mately dependent on the allowable inverter current rating and the maximum
408 9 Field Weakening Operation Maximum Torque Capability Curve Using the Conventional Method
409

where L,' is the stator transient inductance.

Equations (9.4-1) and (9.4-2) become:

viS= r, I:, + o e ~ s l ~ s (9.4-8)

(9.4-9)
'is= rsd-we's",
The maximum stator voltage, V,, max is determined by the available dc link
voltage and the PWM strategy and must satisfy,
Current Limited Circle
2 2 2
(vi,) + (G)5 vS,max (9.4-10)
.e
Hence, the voltage vector must remain on the locus of a circle of radius V, . 'ds
Neglecting the stator Ir drop for high speed operation, the limit for the
steady state currents is
2 2
Voltage Limited Ellipse
( o e ~ s+~( to)e ~ , ' ~5bv:,s ) max
(9.4-1 1 )

Equation (9.4-1 1 ) can be rearranged to form

Figure 9.5 Voltage limited ellipse and current limited circle in the field
weakening region

e 2 e 2 2
Hence, the locus of the current vector is constrained to the locus of a voltage ('p) + (Ids) -< Is, max (9.4-14)
limited ellipse with major axis b = V,/ ( o e L S 1 )and minor axis a = This expression represents a current limited circle the radius of which is the
(V,/ ( o e L s )) .The eccentricity of the ellipse is defined by, maximum allowable stator current as shown in Figure 9.5. Note that the cur-
rent limited circle remains constant for any speed while the voltage limited
ellipse begins to decrease as field weakening operation is entered.

The shape of the voltage limited ellipse depends upon the leakage of the 9.5 Maximum Torque Capability Curve Using the
motor as a per unit of the stator self inductance. Figure 9.5 shows the voltage Conventional Method
limited ellipse for several speeds in the field weakening region. Note that when Consider now the commonly used means for field weakening operation in
the voltage is fixed the ellipse shrinks inversely with w e . which the rotor flux reference is varied inversely proportional to a,.The d-
In general, the maximum stator current is also limited by the inverter cur-
rent rating as well as the thermal rating of the induction motor. Hence, the d-
axis reference current c: is decreased so as to reduce the rotor flux linkage as
in Figure 9.4. However, in this case, the q-axis reference current I,e* can be
and q-axis currents must also satisfy a limit in the stator current amplitude, i.e. increased in order to fully utilize the current rating of the power converter. The
two current components are related to the maximum stator current I,, max by,
410 9 Field Weakening Operation Scheme for Achieving Maximum Torque Capability [4],[5] 41 1

where Idsrrated is the rated d-axis current producing rated rotor flux and
is the mtor speed in per unit and equals unity at the transition point. By
combming (9.5-1) and (9.5-2), we have,

Equation (9.5-3) indicates that when the rotor flux is reduced inversely
with increasing speed, the trajectory of the reference current vector moves
along a current limited circle in a counterclockwise direction. As long as the
current vector for a given frequency is enclosed by the voltage limited ellipse
for that frequency, field weakening operation can be maintained. However, as
the voltage limited ellipse continues to shrink, a point is ultimately reached
where the current vector lies on both the current limited and voltage limited
curves shown as w3 in Figure 9.6. At this point the inverter is completely satu- Figure 9.6 Locus, of current vector for I/or field weakening method
rated and no voltage margin is available to regulate the current. Operation at showing operation reaching the voltage limit ellipse at m3= q , d
any higher speed requires that the current vector now remain on the voltage
responds to the allowable values of current at a certain speed. Clearly, the max-
limited ellipse.
imum torque is reached where the idsiqs hyperbola just touches the current
limited curves at point A in Figure 9.7 which becomes the transition point from
9.6 Scheme for Achieving Maximum Torque Capability [4],[5] constant torque (region #1) to constant power operation (region #2).
When expressed in terms of stator currents, the torque of a field oriented induc-
tion machine can be expressed as, 9.6.1 Transition from Constant Torque to Constant Power Operation
When the maximum values of voltage and current are defined, the fre-
quency at which the motor enters the field weakening region (region #2) is not
an independent variable. If we combine eqns (9.5-l), (9.5-2) with (9.4-ll),
this results in an expression for the transition frequency, cobose,is obtained
hence, the torque can be maximized by maximizing the idsiqs product. It
should be mentioned here that this statement is-true - only- if. saturation is
3 l e c t e d . However, since this discussion pertains to _field weakening the Obase -- 2 e 2
"s, max
2 2 e 2
(9.6-2)
assumption is generally valid. 'dwated + Ll (Is, max - Ids,rated)
Figure 9.7 shows a sketch of this product superimposed on the voltage and Above this frequency the trajectory of the optimal current vector for maxi-
current limited curves. The area within the intersection of these two curves cor- mum torque subsequently moves along the current limited boundary in the
counter clockwise direction as the speed increases. As the speed continues to
412 9 Field Weakening Operation Scheme for Achieving Maximum Torque Capability [4],[5] 413

At the transition from region #2 to region $3 the maximum torque is lim-

ited by both the voltage limited ellipse and the current limited circle. Combin-
ing eqns (9.4-11) describing the voltage limit and (9.4-14) describing the
current limit and assuming the equality sign holds, then solving for the d- and
q-axis currents results in,

B TransitionfromRegion #2 to Region #3

Torque Hyperbolas

-
The maximum value of slip is derived from eqn (9.6-4) and (9.6-5) subject to
the current limit and voltage limit constraints, eqns (9.4-11) and (9.4-14),
whereupon,

This value of slip frequency corresponds .to the slip for maximum torque for a
voltage driven machine. As the machine moves through region #2, the slip fre-
quency increases from the normal value in region #1 to the breakdown value of
Obd
eqn (9.6-6). Increasing beyond this point produces less torque even though the
Figure 9.7 Illustrating intersection of the constant torque hyperbolas current increases. This result indicates that a wide constant horsepower range
with the voltage limit curves and showing Region #3 is obtained by selecting a motor with as small a leakage inductance as possible.
operation The frequency at the transition to region #3 is found to be
increase beyond abase, the slip frequency increases in order to keep the stator
current amplitude at its prescribed maximum value.

9.6.2 Region #2 - Field Weakening with Constant Output Power

In region #2 the slip angular frequency Sw, must be increased to maintain 9.6.3 Region #3 - Field Weakening with Constant Speed*Power
constant stator current. The slip frequency relationship, eqn (6.2-17) reduces
to
,
As the frequency is increased beyond we, , the ellipse continues to be reduced
and eventually is included within the circle as shown in Figure 9.7. Thus the
.e output torque is limited only by the voltage limited ellipse. The optimal cur-
r r lqs
SO, = --
L .e
rents for this case can be obtained from (9.6-1) and (9.4-11). The optimal d-
"ds and q- axis currents are, in this case,
414 9 Field Weakening Operation Control System Implementation 415

e
iqs
Region #2
I %a~ed'~e<%d Wr ---) Regulator Rotor Flux
Oriented Induction
Region #I
Controller Machine
and
(Single Point) CRP WM

Is

Z r Rotor Flux
Estimator
0,

Figure 9.9 Block diagram of the overall control system for optimized
operation in the field weakening range

with voltage space vector PWM is readily adopted as the current regulation
scheme. The rotor flux is estimated from the terminal quantities as described
by eqns (9.4-1) and (9.4-2) together with eqns (6.8-1) and (6.8-2). Feedfor-
Figure 9.8 Overall trajectory of optimal stator current vector in the field. ward compensation is added in order to decouple the d,q axes interactions as in
weakening r e ~ i q t
Figure 7.23.
In practice eqn (9.4-1) is difficult to implement because it requires a pure
integrator. which has initial value and drift problems. To avoid these problems,
a pure integrator must be replaced by a low pass filter which results in good
performance in the high speed region [3]. Figure 9.10 shows the flux reference
and resulting torque vs. speed for the machine of Figure 9.4. From the compar-
1e =- Vs max
9s &we~; ison of the two methods it is apparent that the optimized method provides sig-
nificantly improved torque production when compared with the simple
The torque developed by the machine drops rapidly as roughly the square of algorithm which decreases rotor flux inversely with rotor speed.
the speed reproducing, to a remarkable extent the series field mode of opera-
tion of the dc machine. The trajectory moves along the constraint curve during
regions#l, #2 and #3 as illustrated in Figure 9.8.

9.7 Control System Implementation

Figure 9.9 shows a block diagram of a possible implementation scheme [4].
The overall system consists of speed regulator, flux regulator, current regulator
and indirect rotor flux oriented controller. A synchronous frame PI regulator
416 9 Field Weakening Operation

Appendix
1 I

-/--
C__________-----------
Per Unit Representation
/ a / Electric machine parameters and performance characteristics are often
i y p ids (per unit) p
i 4s expressed in per unit terms rather than in actual units. This approach facilitates

0.5 -
-11 or Methud comparisons between machines and is helpful in clarifying the relative sizes of
-- Maximum Capability Method the machine parameters.
The most appropriate and commonly used per unit system is one based on
the rated output power of the machine (shaft power if a motor, electrical tenni-
nal output power if a generator). The primary base quantities are:
0 -
V, = VR = rated rms phase voltage

P B = P R = rated output power

a, = oR= rated electrical radian frequency

from which the following secondary base quantities are derived

Figure 9.10 Flux reference and torque command currents vs. speed for
llo, method and optimized capability method

References
D. W. Novotny and R D. Lorenz, "Introduction to Field Orientation and High Perfor-
mance AC Drives", in Tutorial Course Rec.. IEEE-IAS Annual Meeting Conf. Rec.,
1985, Sections 1 and 6.
R. Jotten and H. Schierling. "Control of the Induction Machine in the Field Weakening where q is the number of phases, P i s the number of pole pairs and omB repre-
Range", in Proc. IFAC 1983, pp. 297-304. sents the base mechanical angular velocity of the rotor shaft. The subscripts
X. Xu, R. De Doncker and D. W. Novotny, "Stator Flux Orientation Control of Induction 'Byand 'R' denote 'base' and 'rated' respectively. The base quantities are typ-
Machines in the Field Weakening Region", IEEE-IAS Annual Meeting Conf. Rec., ically taken to be equal to the corresponding rated quantity although this is not
1988, pp. 4 3 7 4 3 . necessary. All electrical and mechanical variables can now be expressed as a
S.-H. Kim, S.-K. Sul, and M.-H. Park, "Maximum Torque Control of an Induction Ma-
per unit quantity by taking the ratio of the physical quantity and dividing by
chine in the Field Weakening Region", IEEE IAS Annual Meeting Conf. Rec., 1993, pp.
the base.'~orexample the per unit stator voltage, stator current and resistance
401407.
are:
S.-H. Kim and S.-K. Sul, "Voltage Control Strategy for Maximum Torque Operation of
an Induction Machine in the Field Weakening Region", in IECON Conf. Rec., 1994, pp.
599-604.
418 Appendix

Answers to Chapter 2 Problems

Problem 2 4
3
Ps = -Re [ yqdsi;ds
2
I + 3vosiOs
In this per unit system, one per unit voltage denotes rated rms voltage while
Problem 2-5
1.41 per unit voltage would designate the corresponding peak voltage when the
waveform is sinusoidal. It is interesting to note that the tendency to use rms Stator Frame
voltage as the base loses its meaning when the supply voltage is non-sinusoi-
dal, e.g. an inverter. In this case peak rated (sinusoidal) voltage is frequently qs = 375.6~0s(wet - 30')
vS vis = -375.6sin (wet - 30')
used as the base. In this book we have employed only rms values as base.
j (met - 30')
Per unit quantities can also be employed in a system in which the fre- yids = 375.6e
quency varies. The per unit reactance corresponding to an inductance L is then
defined to be,
Rotor Frame
vrqs = 375.6~0s( (we - o r ) r - 30")

where o, is the angular radian frequency of the supply. This equation can also
be written as

Synchronous Frame
where X is the inductive reactance at the base frequency.
vis = 375.6 =0 yids = 375.6 + j 0
It is important to note that in this per unit system:
1)All computations use the same mathematical expressions as used with actual Problem 2-6
units except that the number of phases and the number of poles no longer
appear anywhere in the computations. Stator Frame

2) For a motor, the rated current is considerably greater than 1.0 because I B vis = 433.7cosaet vis = 0 yids = 216.8eim,t + 216.8e - j a r !
does not incorporate the influence of the power factor or efficiency.
Rotor Frame
3) For an induction motor rated torque is slightly greater than 1.0 because T B
is based on synchronous speed and not on rated speed. Rated torque is equal to vLs = 433.7coswer c o s q t v& = 433.7cosqr sinwrr
in per unit.
1 -S,

Synchronous Frame
420 Answers to Problems Answers to Problems 42 1

ve'7s = 216.8 (1 + cos20et) v;, = 216.8sin2wet Problem 2-10

-j2o,t
yids=216.8(1+e )

Problem 2-7
3P
a) T , = --lm [htqdmiqdrl
22

Answers to Chapter 3 Problems

Problem 2-8
Problem 3-1
1 1
b)viS = i i ~ [ c o s o r +jcos50r--cos70t-
7 I
... ...

:
Problem 2-9
[
23 jot
~ ) y i ~ ~ = - - ~+-e
V e
-jhr
- -lej7mt ......
7 - I
Problem 3-5
422 Answers to Problems Answers to Problems 423

a) c)

ii = 6.67 A (constant)
 1.
424 Answers to Problems Answers to Problems 425

Problem 3-7

Torque characteristic similar to dc series machine.

Problem 3- 13
ii {dc) (A)= Torque characteristic similar to dc shunt machine.

ii {6) (A)=

s (o,L,) - -rsrr
Problem 3-8

vs9s { 1 ) (V)=
Answers to Chapter 4 Problems
vs9s ( 5 ) (V)= Problem 4 1

v;$ { 1 ) (V)=

vis (5) (V)=

vi {dc) (V)=

vi ( 6 ) (V)=

Problem 3- 10
426 1
I
Answers to Problems Answers to Problems 427

Problem 4 4
-
- -
c) Vx = j- vbc
A
-Ix = jz
Ib

Problem 4-5
a) jb = 1.77 L (-37.7)
b) T, =0.78 T, =0.0074 T, =0.77
c) Te2d . 8 0 2

Problem 4-6

I
whe e 9 = tan- 1 -
'-r

%Lr
C)V; (SS) = -I r sinar0t
1 $3 dC

d) s h e as problem 4-1 d

1.2 s to decay to 10%

Problem 4-7
'ram Lmzm (phasor quantity) -j (0, 1-90') +j(qr+9W)
a) iqds (t) = 5.08e + 4.77e + 10.2e-j9'
fir: + (a:o - a:) L: + j2rroeLr

1 p5~,~,S
d) T = J 3 p [ c o s t + cos ( 2 q r + 9,) ]
b) Te = 4.76sin20,t - 9.55 sin wet

Tm,,= 1 2 . 9 ~ ~

Problem 4-8
i(Ss ( 1) = 0.865 pu, $:i (5) = 0.0072 pu, i:$ ( 1) = 0.629 pu,
428 Answers to Problems Answers to Problems 429

ijs ( 5 ) = 0.205, ii ( d c ) = 0.831 pu, ii = 0.306.

.S
iqds ( 1 ) = 1.07 L -36", iids ( 5 )= 0.205 L -88"

Problem 5-5
a) o=1.25 pu
b) 1=0.410 pu
c) y = 90"
Answers to Chapter 5 Problems d) y = 20.2"

Problem 5-1 Problem 5-6

a) I I = 1 . 2 6 ~ ~ a) ~ ~ = 0 . 1 sec.
03

d) f T I V
No Load same same same same
Loaded same increase same increase

Problem 5-3 Problem 5-7

All answers in pu with q -axis as reference for phase angles.
a) or = UrB

Gds Iqds lLqds Iqdr lLqdr

MachineICond. Mag Ang Mag Ang Mag Ang Mag Ang Mag Ang
Problem 5-4 IM/T = T R I .OOLI 3. 1' 1.31~-20.0' 0.97 L-75.7' I . 17 L180° 0.89L-90'
a) Is, = 1.186 Is@ = 0.455 S o , = O.00952IST/Is4 SM/T = T R 0.93 L23.2' 1.22 LO.OO 0.901 -65.9' 5.78 1-90' 0.85 1-73.4"
IM/T =0 0.9 L-0.8' 0.45 1-90' O.94L-90" 0.0 - 0.89 1-90'
SM/T =0 0.82 LO.OO O.OOL0.0 0.82 L-90" 5.78 L-90" 0.82 L-90"
430 Answers to Problems Answers to Problems 43 1

b) or = 0 b) E= 1.38 pu, I=0.856 pu, i*9s = 0.785 pu, i& = 0.346 pu, y = 23.8"
Vqds l9ds &qds lqdr &qdr c) Part a, Te(0) = 0, Te(final)=0.835 pu, Instantaneous response
MachineICond. Mag Ang Mag Ang Mag Ang Mag Ang Mag Ang Part b, T,(O)=0.703, Te(final)=0.835pu, Time Constant =
IM/T = TR 0.074 L-3.1' 1.31 L-20" 0.97 L-75.7' 1.17 L180° 0.89 L-90'
Lqr/rqr = 0.0243 sec.
SM/T = T, 0.037 LO.OO 1.22 LO.OO 0.90 -65.9' 5.78 L -90' 0.85L-73.4'
IM/T = 0 0.013 L-90' 0.45 L-90' 0.94 L-90' 0.0 - 0.89 L-90' Answers to Chapter 7 Problems
SM/T=O 0.0- 0.0- 0.82 L-90" 5.78 L-90' 0.82 L-90'
Problem 7-1

C) or = 2orB (a) K=16.1 z =O.O345 sec

Vqds lqds &qds lqdr &9dr (b) f(H4 IHI L H (deg)
MachineICond. Mag Ang Mag Ang Mag Ang Mag Ang Mag Ang 0.0 1.o 0.0"

IM/T = 1.08 L23.8 1.24 L-10.3 0.53 L-63.0' 1.17 L180° 0.45 L-90'
10 0.963 -3.01"
20 0.910 -5.06"
SM/T = not possible (flux constrained by magnet)
40 0.821 -8.21 "
IM/T = 0 0.9 L-0.4" 0.22 L0.90° 0.47 L-90' 0.0 - 0.47 L-90'
60 0.747 -10.7"
SM/T = 0 not possible (flux constrained by magnet)

Answers to Chapter 8 Problems

Answers to Chapter 6 Problems Problem 8-1
a) Tuned operation
Problem 6- 1
Linear Te vs. iqscharacteristic T , = 0.849iqs
a) I:s = 60.6 A (peak) = 0.457 pu b) Detuned operation
1; = 157A(peak)= 1 . 1 8 6 ~ ~ Non-linear T, vs. iqscharacteristic
So, = 9.35 rlsec
b) T, = 204N-m
Slip Gain Too Slip Gain Too
So, = 18.7 rlsec Small Large
we = 754rlsec
= 496 volts line to line (rms)

Problem 6-2 Svnchronous Machine Field Orientation

a) Te = 0.835 pu, E = 0,835 pu, cose = 0.865
432 Answers to Problems Answers to Problems 433

1.O 1.13 1.0 0.625 d) T, (steady state) = 1.67 pu

1.2 1.49 1.2 0.73
1.4 1.86 1.4 0.83

Problem 8-2
f=60 hz., Vhz=2.0lpu
f=5 hz., V/hz=2.21 pu
Saturation will substantially reduce the Vhz values which yield maximum
torque ampere.

a) IST = 1-13pu, IS# =O.309 PU

b) Vs = 0.0995 pu
C) T , = 1.67 pu, IST = 0.63 1 pu, 4, = 1.88 PU, 4, = 1.78 pu.

0 2 4 6 8
Time - Sec.
A hysteresis limit cycles ..............327
hysteresis regulators ............... 322
air gap predictive control .................349
field intensity ......................39 PWM inverter .................... 321
flux ...............................39
ramp comparison .................. 331
flux linkage .....................40.75 stationary frame ................... 335
non-uniform .......................88
packet of flux ......................40 synchronous frame .................340
total air gap MMF .................49 current source inverter
commutation delay ................ 119
current regulated .................. 318
B d.q model .........................116
BJT(bipolar junction tiansistor) .......11 dc link inductor ................... 129
brushless dc machine ................ 224 duality with VSI ...................144
sine wave machines ................223 switching functions ................ 119
synchronous frame switching functions
...................................'23
complex vector with synchronous machine .........208
defmition ..........................51
equivalent circuit ..................287 D
equivalent circuit in operational form d.q model
................................... I53 air gap flux referred form ...........283
induction machine equivalent circuit modified forms ....................155
................................152 of synchronous machine ........... 290
interpretation ......................61 ramp comparison controller ........ 335
operational impedances ............171 rotor flux referred equations ........ 258
voltage equations ...................52 stator flux oriented induction machime
complex vector diagram ...................................310
air gap flux reference ...............167 statorvoltage equations in terms of rotor
for positive sequence .............. 166 flux ...........................304
rotor flux linkage reference .........168 synchronous frame ramp comparison
round rotor synchronous machine ...215 current regulator ..............340
salient pole synchronous machine ...215 voltages for sinusoidal excitation ....159
controlled rectifier ................... 26 dc machine
current limit circle .................. 409 brushless dc machine ..............222
current regulation capability curve ....................403
augmented f e e d f o m d compensation commutating capability ..............6
...................................347 commutation ..................... 3 3 4
augmented feedforward compensation constant power limit .............. 404
with decoupling ....................348 constant torque limit ............. 404
block diagram of series compensated dc drives ............................2
.................................. 346 efficiency ...........................5
block diagram with feedfonvard com- field weakening operation ..........401
pensation ..................... 346 requirements for torque control ... 206
CSI ...............................318 size and weight ......................7
CSI limitations ....................321 speed'power product ................5
dc link ............................318 torque ..............................7
feedforward compensation ........344 torque control .....................203
hysteresis controller decoupling 331 detuning in indirect field orientation
hysteresis controller switching diagram detuned field orientation lines ......360
................................... 324 eigenvalues ........................379
A? h INDEX
I INDEX 437

*
field orientation torqueYcharacteristics invariance of the steady state ....... 164 wide constant horsepower range ....413 mutual between phases ............. 46
...................................362 negative sequence .................163 field oriented control ................32 mutual from stator to rotor ..........47
flux characteristics .................363 positive sequence ..................160 flux linkage rotor leakage .......................48
influence of saturation ............364 referred variables ..................154 magnetizing component ............4 0
stator leakage .....................4 6
loss trends ......................... 369 rotor flux reference ................169 self component of flux linkages .....4 0 stator magnetizing .................
normalized torque command .......361 salient pole synchronous machine ...291 four quadrant drive ...................2 46
stator transient ................'9% 337
phasor diagram ....................353 stator flux reference ...............170 fundamental component model total self ..........................4 6
power loss ......................... 365 stator referred .....................158 CSI steady state current ............I37 induction machine
representation of slip gain error .....383 steady state .......................156 CSI steady state voltage ........... 140 block diagram for current fed .......263
rotor copper loss characteristics ....370 transient ......................191, 196 d. q equivalent circuit of CSI drive in syn-
cogging torques ....................25
slip gain error multipliers ..........360 turns ratio ........................I53 chronous frame ................136 complex vector equivalent circuit ... 152
small signal transfer function .......379 d.q equivalent circuit of VSI drive in syn-
complex vector model ............. 151
torque response with slip gain error F chronous frame ................'34 constant power limit ..............406
................................... 377 equivalent circuit ......... , .141. 143
field ............................217. 410 direct field orientation .............247
torque step response ...............382 field orientation
equivalent circuit of PWM-VSI inverter dynamics of field orientation ....... 257
transient response .............3791 384 achieving maximum torque capability
................................I43 equivalent circuit ...................61
PWM steady state voltage ..........I39 field weakening operation ..........401
direct field orientation ...................................410 three phase six step VSI ............142
controller for air gap flux ...........289 block diagram for induction machine idealized three phase ................43
decoupler for stator flux ............3" ....................................263 VSI steady state current ............I39 in rotor flux rotating frame ..........87
implementation of ................a80 VSI steady state voltage ............I37 in rotor reference frame .............3f-'
constant flux voltage decoupler .....307
measurement of air gap flux ........276 direct .............................247 in stationary axis d,q,o variables .....81
stator flux based ...................312 direct controllers ..................276 G in synchronously rotating reference
using flux sensors ..................277 dynamic response ..................261 general purpose drives .................31 frame ..........................86
voltage and current sensing .........278 dynamic response of air gap flux ....284 GTO (gate turn off thyristor) ..........11 indirect field orientation .......232, 241
dynamics of .......................257 indirect field orientation using a CSI
E dynamics of synchronous machine . .293 H ..........................244
equivalent circuit for induction machine indirect field orientation using a
eigenvalues high performance drives ............... 31 CRPWM .....................242
constant speed ....................177 ...................................230
limiting cases ......................184 field orientation lines .......... 357, 389 inrush current ......................78
loci in the rotor reference frame ....182 field weakening operation ..........399 I magnetizing inductance .............55
loci in the stationary reference frame indirect .......................232, 241 IGBT (insulated gate bipolar transistor) negative sequence equivalent circuit 163
...................................180 optimized operation in field weakening ......................................I1 open loop speed controller ..........17
loci in the synchronously rotating refer- range .........................415 indirect field orientation optimized operation in the field weaken-
ence frame ....................181 phase correction for CSI system ....246 compensated flux response .........267 ing range ......................415
ranges of machine parameten for normal rotor flux referred equations .......258 controller for air gap flux ...........288 positive sequence equivalent circuit 160
machines ...................... 186 slip expression .....................232 detuning ..........................353 pulsation torque ............. y . 8 0 . 162
rotor poles ........................ I79 stator flux based ...................309 field orientation lines ..............357 rotor reference frame ...............84
stator poles ........................I79 steady state d,q currents ........... 236 indirect control using a CSI ....... 268 short circuit of unloaded machine 187
steady state d,q model ............. 234 slip gain error ......................384 simulated with phase variables .......78
typical values ......................187
equwalent circuit synchronous and induction machine slip relation .......................260 size and weight ......................7
comparison ...................249 start up transient .................269 squirrel cage winding ................36
complex vector operational .........153
torque characteristics ..............355 stationary frame d.q model with CSI
constant rotor flux linkage model .. 190 synchronous machine withconstant field
for air gap flux control .............287 excitation .....................293 using a CRPWM .................24" ...................................126
fundamental component circuit of three synchronous machine with variable field using a CSI .......................244 stationary frame d,q model with VSI
phase six step CSI .............143 excitation .....................295 using voltage controlled inverter ... 308 ...................................124
fundamental component circuit of three terminal behavior ..................233 with d,q synchronous frame currents as steady state complex vector equivalent
phase six step VSI .............142 torque characteristics ..............355 Inputs .........................265 circuit ........................."87
fundamental component model of torque control requirements .......-247 with synchronous frame rotor flux and q- steady state equivalent circuit derivation
PWM-VSI .................... 143 torque production .................261 axis stator current as inputs .....265 ...................................156
illustrating induction machine field ori- using air gap flux ..................283 steady state torque .................161
entation .......................23O using voltage as the controlled variable inductance synchronous frame d.q model with CSI
induction machine complex vector form ...................................
304 coupling factor ....................178 ...................................127
voltage decoupler .................. 306 leakage ............................ synchronous frame d.q model with VSI
magnetizing .................... 412I54
45
438 INDEX INDEX e
.
439

N synchronously rotating ............. 63 field orientation with a CSI ........217

torque capability curve ............406
turns ratio .........................I53 zero sequence component .......... 56 flux linkage equation ................93
negative sequence voltage ............ I57 rotor cogging .......................320
vector control in steady state .......226 Gorges Effect .................... roo
voltage decoupler .................306 induction motor torque ........... loo
0 kVA requirement ....................8
wide constant horsepower range ....413 one quadrant converter ................2
wound rotor .................... 36, 61 selection of flux level line start of permanent magnet machine
inertia best optimal .......................392 ...................................104
ac machine .........................6 P changes in magnetizing inductance and mutual inductance between stator
dc motor ........................... 6 positive sequence voltage ............157 rotor time constant ............365 phases ........................ 90
inverter power effect of detuning ..................365 Park's Equations ....................98
commutation ...................... 10 electromechanical ..................74 field orientation lines ..............389 permanent magnet .............102. 223
CSI ................................25 in d.q circuits ......................
69 maximum drive efficiency .........390 power input equation ...............99
commutation ...................27 three phase ........................68 maximum torque pei stator ampere reaction torque ........... roo. 214. 216
inddction motor ................28 power factor ...................................389 reluctance torque ..............214. 216
turn off devices ................ 30 input hndamental component .......5 optimal ...........................388 rotor flux linkage equations .........96
CSI (current stiff inverter) ...........9 pulse width modulation optimal utilization of the converter saliency torque ................... loo
current controlled PWM ...........321 current regulated .................. 321 ..................................394 self inductance .....................89
current regulated ..................317 hysteresis regulator ................322 peak torque effects .................365 steady state d.q model ............. 213
current stiff ........................ 9 ramp comparison ..................331 reduced sensitivity to controller detun- three phase stator voltages .......... 91
d q models in synchronous reference ing .......................... .393 torque control .....................212
frame .........................120 R switching transients torque control requirements ........ zzr
feedback diodes ....................12 ac component magnitude ...........189 torque control using a CRPWM ....218
ramp comparison PWM torque output equation .............99
imposed voltage and current limits . 405 augmented feedforward compensation complex eigenvalue ................172
PWM constant speed eigenvalues ........ .I77 torque pulsation ...................103
.................................. .347 vector control ................ 208. 292
block modulation ...............19 augmented feedforward compensation coupling factor ....................178
changeover to six step mode .... 24 with decoupling ............... 348 dc component magnitude ...........189 vector diagram for field orientation .217
gear changing ..................$22 general solution ....................I77 wound field ......................... 8
basic operation ....................331
natural modulation ..............22 block diagram .....................338 ranges of machine parameters for normal
phase locking ...................22 block diagram of series compensated machines ......................186
pulse dropping .................24 ...................................346 rotor poles ....................... J79 time constant
sinusoidal modulation ..........20 block diagram with feedforward com- stator poles ........................179 rotor open circuit .................. 178
PWM inverter ................... .9, 18 pensation .................... .346 stator transient inductance .........191 rotor short circuit .................178
resonant reversal ................... 10 dc current controller ............... 335 sudden application of stator current stator open circuit .................178
return current ...................... 12 describing function ................ 333 source excitation ...............172 stator short circuit .................178
ripple current ....................... 9 feed forward compensation .........344 sudden short circuit ................187 torque ..............................103
six step inverter ..................... 9 modulation index .................. 333 transient equivalent circuit ........ 196 adjustable .........................401
six step voltage inverter .............12 models of PI controller ............342 voltage behind transient reactance . .192 characteristics for field orientation .355
snubbing ........................... II predictive control .................349 synchronous machine complex vector expressions ..........77
torque pulsations ...................25 simpler synchronous regulator ......344 angle control ......................292 constant ..........................401
voltage controlled oscillator (VCO) . .18 stationary frame d,q variables .......335 as commutatorless dc machine ..... 208 cross product form .................76
voltage source ...................... 9 braking torque .....................103 electromagnetic .................73, 74
voltage stiff .........................9 choice of gamma ...................212
stationary frame regulator cross coupling response with slip gain error ........377
VoltsMertz control .................15 ...................................341 complex space vector ...............95 variable ...........................
VSI (voltage stiff inverter) ........... 9
401
rectifier .............................9 converter ........................... 8 transformation
inverters reference frame CSI driven .......................208 general complex form ...............65
losses caused by .................... 24 arbitrary ...........................56 CSI torque control ................220 general phase .......................65
attached to vector ..................56 d,q variable vector diagram .........214 general rotating complexvector .....58
M d,q axes ............................56 damping torque ...................IOO graphical interpretation of inverse ...66
magnetic saturation ...................393 freely rotating ......................56 dynamic response with gamma not zero inverse phase .......................66
MCT (MOS controlled thyristor) ......11 normal component .................57 ................................... 295 inverse rotation ....................66
MOSFET (metal oxide semiconductor rotor ..............................56 dynamics of field orientation .......293 power invariant .....................56
field effect transistor) ...........11 s t a t i o n ~.......................f6, 63 excitation torque .................loo turns ratio transformation .............54
stator .............................. 56 field orientation ................... 2x9
440 INDEX

vector control ........................32

voltage behind transient reactance ....192
voltage controlled oscillator (VCO) ....17
voltage limit ellipse .................408
voltage source inverter
complex vector d.q form ...........114
d.q model .........................115
duality with CSI ..................J4-4
PWM switching functions .........115
switching functions ................111
synchronous frame switching functions
...................................122
.................1x5
zero voltage states

winding
concentrated full pitch ..............39
effective number of turns ............38
flux l i e s ....................44346
flux linkages of rotor ...............48
leakage flux ..........................45
rotating ............................ 43
rotor ............................... 43
rotor flux linkages .................47
stationary .......................... 43
stator ..............................43
voltage equations ...................43
winding density distribution .........37
winding distribution ................31

z
zero sequence .......................66
circuits ............................60
equation ...........................59