CONTROL SYSTEMS

(Modeling, Analysis and Design)

Swami Karunamoorthy

DESIRED ACTUAL
KNOWLEDGE KNOWLEDGE
OF CONTROLS OF CONTROLS
(INPUT) (OUTPUT)
CONTROL SYSTEM
EDUCATION
+
_

ASSESSMENT
(FEEDBACK)
 E 37 - MEMS 4301
Modeling, Simulation
and Control

Dr. Swami Karunamoorthy

Washington University in St. Louis
Spring 2020
Thank you for purchasing this book! Cognella Custom is an imprint of Cognella, Inc., a
student-founded company with a mission to support instructors in creating engaging
learning materials at the lowest possible cost to students. While many textbooks include
chapters you’ll never read, every piece of content in this book has been hand-picked by your
instructor to help you get the most out of their course.

The cost of this book includes the copyright fees required by the original rightsholders. By
purchasing this text, you are protecting yourself and your school from copyright
infringement, as well as supporting the scholars and authors who spent time developing
this content.

If you have any questions regarding this text, please don’t hesitate to contact us at
[email protected].

Best wishes for a successful academic term,

The Cognella Custom Team

Copyright © 2019 by original copyright holders. Cognella gratefully acknowledges

permission to reprint previously-published materials. No part of this publication may be
reprinted, reproduced, transmitted, or utilized in any form or by any electronic, mechanical,
or other means, now known or hereafter invented, including photocopying, microfilming,
and recording, or in any information retrieval system without the written permission of the
original rights holders. For inquiries regarding permissions, or any other form of
reproduction, please contact the Cognella Licensing Department at [email protected].
 CONTROL SYSTEMS
(Modeling, Analysis, and Design)

1.0 Introduction 1

2.0 Modeling of Dynamic Systems 13

3.0 Modeling of Control Systems 25

4.0 Characteristics of Control Systems 45

5.0 Performance of Control Systems 63

6.0 Stability of Control Systems 77

7.0 Root-Locus method 93

8.0 Bode Plot and Nyquist Plot 115

9.0 Design of Control Systems 133

10.0 State Variable Model and Design 149

Appendix: MATLAB & SIMULINK 173

 1.0 Introduction

Control system is a study of methods for modeling, analysis and design to accomplish the
actual output that is equal to or very close to desired output used as input to the system.
Learning Objectives
1. To understand various types of control system.
2. To identify various elements of a control system
3. To learn about Transfer Functions.

1.1 Types of Control System

The types of control system are in general classified as (i) open loop and (ii) closed loop.
The open loop control system has no feedback and it does not depend on output. Since the
information is forwarded from start to finish, it is also called as feed-forward system. In a
closed loop system sensor are used to measure the actual output and feedback to input that
alters the desired output, until the output matches with input. Such a system is also called
as feed-back control system.

Open-loop control system (or feed-forward system)

For example, microwave oven, bread toaster and space heater are typical devices with
open loop control system.

These systems function as a typical on-off switch. They do not control how hot the food
should be heated in a microwave or how well the bread should be toasted or how warm the
room should be heated.

1
1
Closed loop system (or feed Back system)

A -> Summing point

B -> Pick off point

Temperature controlled oven, those heating with thermostat, cruise control in a car are
typical example of closed loop system. The heating turns on or off automatically when the
actual heat is below or above the desired heat level in order to maintain the set level of heat
in both oven or house heating. In a car, the speed is controlled by cruise control such that
the car travels at set speed level. The cruise control, controls the fuel flow into engine.
When the car is at uphill, the speed drops and the cruise control increases the fuel
(equivalent to increasing accelerator by driver) to engine. When the car is at downhill, the
speed increases and the cruise control decreases the fuel flow to engine to maintain the set
level of speed.

The type of feedback can be further classified as positive feedback and negative feedback.
Positive Feedback
Let △= (Desired output - Actual output)
The value of △, is used as feedback. If its value decreases, the actual output converges
closer to desired output. The decreased value of △ is attributed to negative feedback while
increased value of △ is attributed to positive feedback. The feedback from actual output is
added to input, making it larger in positive feedback.
For example, if a person speak on a microphone, the voice is amplified to certain level in
a speaker. If the microphone is close to speaker, the output from speaker fed into
microphone continuously, thus increasing the desired output to very high level. Such a
positive feedback result in loud squealing or howling noise. So, the positive feedback
makes the input larger instead of minimizing the difference between input and output. Thus,
in general the positive feedback lead to system instability and it is not desirable in the
control of engineering systems.

2
2
 Control Diagram

Negative Feedback
In negative feedback system, the feedback from actual output is subtracted at input to
control the difference between them.

3
3
For example, in a centralized heating system in a house, the thermostat controls the
temperature to a set level value.

If the actual temperature is less than the desired temperature, the feedback to thermostat
activates the heating system to increase the temperature in the house. It turns off the heating
system when the feedback indicated the actual temperature is greater or equal to the desired
temperature. So, the negative feedback system leads to system stability and hence
commonly used in control used in control of engineering systems.

An emphasis is given here to understand, model analyze and design negative feedback
automatic control systems and they are also called as modern control systems.

1.2 Elements of control system

The various elements of a typical control system are

(i) Controller
(ii) Actuator
(iii) Plant or Process
(iv) Sensor

4
4
The above control diagram represents closed loop system and without the feedback loop it
becomes open loop system as shown below.

If the actual output is equal to or very close to the desired output, open loop system is
adequate to use. When the actual output is not equal to the desired output, it is not only
necessary but also required to use a closed loop system with negative feedback. It ensures
both system control and system stability.

Let us look at the house heating control system with all the control elements.

5
5
Controller: A device used to activate the actuator.
In the above example, thermostat is used as controller to turn-on or turn-off the
heating furnace. If a person is used for this operation, it would be manual
control. Since a device is used in this case, it is automatic control.

Actuator: A device in the control system used to alter or adjust the environment. In the
above example, the heating furnace is the actuator and it adjusts the house
heating by switching it on or off.

Plant: A system (a device or an industrial plant or a process) under control. In the above
example, house is the plant and its environment or temperature is being controlled
to a set level of comfort.

Sensor: A device that provides measurement of actual output. In the above example,
temperature sensing device within the thermostat is the sensor element in the
control system.

One of the typical temperature sensing devices is bi-metallic strip. It is used to make or
break contact with an electric circuit that turns on or turns off the heating system.

Cruise control in a car

6
6
Another example is the cruise control in a car. Here, the controller is a throttle that regulates
the fuel flow.
Engine is the activator that increase or decrease the speed. Car is the plant being controlled
to travel at set speed. Speed sensor gives feedback on any change in actual speed with
respect to desired speed. Based on this feedback, throttle will adjust the fuel flow into
engine.

Learning process Control

A student’s learning process can be represented as a negative feedback control system.

In a classroom learning, the course outline sets the goal for desired knowledge. Motivation
and time management are the controller in this process. Studying the book or class notes
and practicing problems are the actuators in this system. The plant in this system is the
“Learning Process” that is controlled to gain the desired knowledge. Homework,
examinations and assessment of course outcomes are the sensors used to measure the actual
knowledge and used as feedback for better learning.

1.3 Transfer Function

Transfer function is widely used in the study of control system. It is designed as the ratio
of output to input in Laplace Transform domain with all initial conditions assumed to be
zero.

7
7
Open Loop

𝑌𝑌(𝑠𝑠) = 𝐺𝐺(𝑠𝑠)𝑅𝑅(𝑠𝑠)
𝑌𝑌(𝑠𝑠) 𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂
𝐺𝐺(𝑠𝑠) = =
𝑅𝑅(𝑠𝑠) 𝐼𝐼𝐼𝐼𝑂𝑂𝑂𝑂𝑂𝑂
Where
S = Frequency domain in Laplace Transform (L.T.)
R(s) = L.T. of input
Y(s) = L.T. of output
G(s) = Transfer function

Closed loop

Negative feedback

Where
G(s) = Feed Forward T.F.
H(s) = Feed Back T.F.
E/R = Error ratio
Y/R = Control ratio
𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑂𝑂𝑁𝑁𝑁𝑁𝑁𝑁 → 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑂𝑂𝑁𝑁𝑁𝑁𝑁𝑁 𝑓𝑓𝑁𝑁𝑁𝑁𝑓𝑓𝑓𝑓𝑁𝑁𝑓𝑓𝑓𝑓
B/R = Feedback ratio = �
𝑃𝑃𝑃𝑃𝑠𝑠𝑁𝑁𝑂𝑂𝑁𝑁𝑁𝑁𝑁𝑁 → 𝑃𝑃𝑃𝑃𝑠𝑠𝑁𝑁𝑂𝑂𝑁𝑁𝑁𝑁𝑁𝑁 𝑓𝑓𝑁𝑁𝑁𝑁𝑓𝑓𝑓𝑓𝑁𝑁𝑓𝑓𝑓𝑓

8
8
 E = R-B
G = Y/E
H = B/Y
Y = GE = G(R-B) = G(R-HY) = GR-GHY
Y(1+GH) = GR
Transfer function:
Y/R = G/(1+GH) = Output/Input

Positive feedback

E = R+B ; B = HY
Y = GE = G(R+B) = GR+GHY
Y(1-GH) = GR
Transfer function:
Y/R = G/(1-GH) = Output/Input
If GH ≅ 1, output may become infinity making the system unstable.

9
9
 Assessment of learning objective
1. Control system is a study
(a) To control the failure of a system
(b) To control the operation of a function
(c) To control the actual output to match with desired output
(d) To control the input

2. Open loop control system is the one that

(a) Depend on output
(b) Depend on input
(c) Does not depend on input
(d) Does not depend on output

3. Closed loop control system is the one that

(a) Depend on output
(b) Provides feedback
(c) Provides stability
(d) All of the above

4. Positive feedback system

(a) Makes the system unstable
(b) Makes the system stable
(c) Makes the output positive
(d) None of the above

5. Controller is defined as
(a) A device that activates the plant
(b) A device that activates the actuator
(c) A device that controls the sensor
(d) All of the above

6. Actuator is defined as
(a) A device that activates the plant
(b) A device that activates the controller
(c) A device that activates the sensor
(d) All of the above

10
10
7. A Plant or Process is defined as
(a) A system that controls the activator
(b) A system that controls the sensor
(c) A system that activates the controller
(d) A system to be controlled

8. A sensor is defined as
(a) A device to measure the output from plant
(b) A device to give feedback to controller
(c) A device to function as sensing element in the system
(d) All of the above

9. Transfer function is defined as

(a) Ratio of input to output
(b) Ratio of output to input
(c) Sum of output and input
(d) Difference between output and input

10. G/(1+GH) is the transfer function for

(a) Open loop system
(b) Positive feedback system
(c) Negative feedback system
(d) None of the above

11. G/(1-GH) is the transfer function for

(a) Open loop system
(b) Positive feedback system
(c) Negative feedback system
(d) None of the above

12. The transfer function of a dynamic system requires

(a) The initial conditions to be zero
(b) The initial conditions to be not zero
(c) The initial conditions to be infinity
(d) None of the above

11
11
 2.0 Modeling of Dynamic System
Learning Objectives:
(1) To understand the transfer function for a dynamic system with single degree of freedom
or multi degrees of freedom.
(2) To model the given dynamic system with equations of motion in matrix format.
(3) To understand Poles, Zeros and Final Value Theorem.

2.1 Transfer function for dynamic systems

A typical single degree of freedom system is given below:

m – mass
b – damping coefficient
k – stiffness
y(t) = Output (response); r(t) = Input (Forcing function)

Equation of motion:
𝑚𝑚𝑦𝑦̈ + 𝑏𝑏𝑦𝑦̇ + 𝑘𝑘𝑦𝑦 = 𝑟𝑟(𝑡𝑡)
Assume the initial conditions are zero since the transfer function is not a function of them.
Taking Laplace Transform (L.T)
(𝑚𝑚𝑠𝑠 2 + 𝑏𝑏𝑠𝑠 + 𝑘𝑘)𝑌𝑌(𝑠𝑠) = 𝑅𝑅(𝑠𝑠)

Z(s)Y(s) = R(s); Z(s) = 𝑚𝑚𝑠𝑠 2 + 𝑏𝑏𝑠𝑠 + 𝑘𝑘

1
Y(s) = [ ] R(s)
Z(s)

= G(s) R(s)
1 1
Where the Transfer function, G(s) = =
Z(s) 𝑚𝑚𝑠𝑠 2 +bs+k

13
Multi degrees of freedom:
[𝑀𝑀]{𝑦𝑦̈ } + [𝐶𝐶 }{𝑦𝑦̇ } + [𝑘𝑘]{𝑦𝑦} = {𝑟𝑟(𝑡𝑡)}

[M]—Mass matrix
[C]—Damping matrix
[K]—Stiffness matrix

Taking L.T. 〈𝑆𝑆 2 [𝑀𝑀] + 𝑆𝑆[𝐶𝐶 ] + [𝐾𝐾]〉{Y(s)} = {R(s)}

[Z(s)] {Y(s)} = {R(s)}

{Y(s)} = [𝑍𝑍(𝑠𝑠)]−1 {R(s)}

=[G(s)] {R(s)}

Where the Transfer function matrix,

[G(s)] = [𝑍𝑍(𝑠𝑠)]−1 =〈𝑆𝑆 2 [𝑀𝑀] + 𝑆𝑆[𝐶𝐶 ] + [𝐾𝐾]〉−1

Two degrees of freedom

𝑌𝑌1 (𝑠𝑠) 𝐺𝐺 𝐺𝐺12 𝑅𝑅1 (𝑠𝑠)

� � = � 11 �� �
𝑌𝑌2 (𝑠𝑠) 𝐺𝐺21 𝐺𝐺22 𝑅𝑅2 (𝑠𝑠)
𝑌𝑌1 (𝑠𝑠)
Let R2=0 => Y1(s)=G11(s) R1(s) => G11(s) =
𝑅𝑅1 (𝑆𝑆)

𝑌𝑌2 (𝑠𝑠)
Y2(s)=G21(s) R1(s) => G21(s) =
𝑅𝑅1 (𝑆𝑆)

𝑌𝑌1 (𝑠𝑠)
Let R1=0 => Y1(s)=G12(s) R2(s) => G12(s) =
𝑅𝑅2 (𝑆𝑆)

𝑌𝑌2 (𝑠𝑠)
Y2(s)=G22(s) R2(s) => G22(s) =
𝑅𝑅2 (𝑆𝑆)

14
 𝑌𝑌𝑖𝑖 (𝑆𝑆)
In general, 𝐺𝐺𝚤𝚤̇𝚥𝚥̇ (𝑠𝑠) =
𝑅𝑅𝑗𝑗 (𝑆𝑆)

2.2 Modeling of dynamic system

A multi-degrees of freedom system when it is dynamically uncoupled, “Rule based method” can
be applied to obtain the Mass, Damping, and Stiffness matrices.

Rule-Based method
Rule #1: Degree of freedom, (N)
Each mass has a degree of freedom. If there is a degree of freedom not associated with a mass,
assume a virtual mass (or dummy mass) with zero value for that degree of freedom.
It helps to apply this method. The size of matrices, [M], [C] and [K] are N x N.
Rule #2: Mass matrix, [M]
Since the system is assumed as dynamically uncoupled, the mass matrix is always diagonal.
Rule #3: Stiffness Matrix:

Diagonal Elements 𝐾𝐾𝑖𝑖𝑖𝑖 = ∑𝑁𝑁

𝑖𝑖=1(𝑠𝑠𝑡𝑡𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠𝑠𝑠𝑐𝑐𝑡𝑡𝑠𝑠𝑐𝑐 𝑡𝑡𝑐𝑐 𝑀𝑀𝑀𝑀𝑠𝑠𝑠𝑠, 𝑀𝑀𝑖𝑖 )

Off-diagonal Elements 𝐾𝐾𝑖𝑖𝑖𝑖 = − ∑𝑁𝑁

𝑖𝑖=1(𝑠𝑠𝑡𝑡𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠𝑠𝑠𝑐𝑐𝑡𝑡𝑠𝑠𝑐𝑐 𝑏𝑏𝑠𝑠𝑡𝑡𝑏𝑏𝑠𝑠𝑠𝑠𝑠𝑠 𝑀𝑀𝑖𝑖 𝑀𝑀𝑠𝑠𝑐𝑐 𝑀𝑀𝑖𝑖 )

𝐾𝐾𝑖𝑖𝑖𝑖 = 𝐾𝐾𝑖𝑖𝑖𝑖 due to symmetry

Rule #4: Damping Matrix:

Diagonal Elements, 𝐶𝐶𝑖𝑖𝑖𝑖 = ∑𝑁𝑁

𝑖𝑖=1(𝐷𝐷𝑀𝑀𝑚𝑚𝐷𝐷𝑠𝑠𝑠𝑠𝐷𝐷 𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑐𝑐𝑠𝑠𝑠𝑠𝑠𝑠𝑡𝑡𝑠𝑠 𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠𝑠𝑠𝑐𝑐𝑡𝑡𝑠𝑠𝑐𝑐 𝑡𝑡𝑐𝑐 𝑀𝑀𝑀𝑀𝑠𝑠𝑠𝑠, 𝑀𝑀𝑖𝑖 )

Off-damping Elements,
𝑁𝑁

𝐶𝐶𝑖𝑖𝑖𝑖 = − �(𝐷𝐷𝑀𝑀𝑚𝑚𝐷𝐷𝑠𝑠𝑠𝑠𝐷𝐷 𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑐𝑐𝑠𝑠𝑠𝑠𝑠𝑠𝑡𝑡𝑠𝑠 𝑐𝑐𝑐𝑐𝑠𝑠𝑠𝑠𝑠𝑠𝑐𝑐𝑡𝑡𝑠𝑠𝑐𝑐 𝑏𝑏𝑠𝑠𝑡𝑡𝑏𝑏𝑠𝑠𝑠𝑠𝑠𝑠 𝑀𝑀𝑖𝑖 𝑀𝑀𝑠𝑠𝑐𝑐 𝑀𝑀𝑖𝑖 )

𝑖𝑖=1

𝐶𝐶𝑖𝑖𝑖𝑖 = 𝐶𝐶𝑖𝑖𝑖𝑖 due to symmetry

15
Example-1

Find the Transfer Function.

𝑚𝑚 0 𝑐𝑐 + 𝑐𝑐 −𝑐𝑐3 𝑘𝑘 −𝑘𝑘
[𝑀𝑀] = � 1 � [𝐶𝐶] = � 1−𝑐𝑐 3 𝑐𝑐2 + 𝑐𝑐3
� [𝐾𝐾] = � �
0 𝑚𝑚2 3 −𝑘𝑘 𝑘𝑘

Let m1 = 1 kg, m2 = 2 kg, c1 = 1 kg/s, c2 = 2 kg/s, c3 = 3 kg/s, k = 1 N/m

1 0 4 −3 1 −1
[𝑀𝑀] = � � [𝐶𝐶] = � � [𝐾𝐾] = � �
0 2 −3 5 −1 1

(𝑠𝑠 2 + 4𝑠𝑠 + 1) −(3𝑠𝑠 + 1)

Z(s) = S2[M]+S[C]+[K] = � �
−(3𝑠𝑠 + 1) (2𝑠𝑠 2 + 5𝑠𝑠 + 1)

Transfer Function
G(s) = [Z(s)]-1

1 (2𝑠𝑠 2 + 5𝑠𝑠 + 1) 3𝑠𝑠 + 1

= � 2 �
𝐷𝐷𝐷𝐷𝐷𝐷 3𝑠𝑠 + 1 (𝑠𝑠 + 4𝑠𝑠 + 1)

Det = (s2+4s+1) (2s2+5s+1) −(3s+1)2 = s(2s3+13s2+14s+3)

16
Example-2

Introduce a virtual (dummy) mass m0 at the degree of freedom, y1

Let m = 1kg, c = 1kg/s, k = 1N/m, m0 = 0

0 0 0 0 1 −1
[𝑀𝑀] = � � [𝐶𝐶] = � � [𝐾𝐾] = � �
0 1 0 1 −1 1

1 −1
Z(s) = S2[M]+S[C]+[K] = � �
−1 (𝑠𝑠 2 + 𝑠𝑠 + 1)

Transfer Function
G(s) = [Z(s)]-1
2
�(𝑠𝑠 + 𝑠𝑠 + 1) 1�
1
=
𝐷𝐷𝐷𝐷𝐷𝐷 1 1
Det = (1) (𝑠𝑠 2 + 𝑠𝑠 + 1) −1 = s(s+1)

17
Example-3

1 0 3 −3 1 −1
[𝑀𝑀] = � � [𝐶𝐶] = � � [𝐾𝐾] = � �
0 2 −3 3 −1 1

(𝑠𝑠 2 + 3𝑠𝑠 + 1) −(3𝑠𝑠 + 1)

Z(s) = S2[M]+S[C]+[K] = � �
−(3𝑠𝑠 + 1) (2𝑠𝑠 2 + 3𝑠𝑠 + 1)

Transfer Function
G(s) = [Z(s)]-1

1 (2𝑠𝑠 2 + 3𝑠𝑠 + 1) 3𝑠𝑠 + 1

= � 2 �
𝐷𝐷𝐷𝐷𝐷𝐷 3𝑠𝑠 + 1 (𝑠𝑠 + 3𝑠𝑠 + 1)

Det = (2𝑠𝑠 4 + 9𝑠𝑠 3 + 3𝑠𝑠 2 − 8)

2.3 Poles, Zeros and Final Value theorem

Single Degree of Freedom (SDOF)
𝑚𝑚𝑦𝑦̈ + 𝑏𝑏𝑦𝑦̇ + 𝑘𝑘𝑦𝑦 = 0
𝑏𝑏 𝑘𝑘
𝑦𝑦̈ + 𝑦𝑦̇ + 𝑦𝑦 = 0
𝑚𝑚 𝑚𝑚
𝑦𝑦̈ + 2ζ𝜔𝜔𝑛𝑛 𝑦𝑦̇ + 𝜔𝜔𝑛𝑛2 𝑦𝑦 = 0

Initial Conditions:
y(0) = y0 𝑦𝑦̇ (0) = 0

L.T. => [𝑠𝑠 2 𝑌𝑌(𝑠𝑠) − 𝑠𝑠𝑦𝑦(0) − 𝑦𝑦̇ (0)] + (2ζ𝜔𝜔𝑛𝑛 )[𝑠𝑠𝑌𝑌(𝑠𝑠) − 𝑦𝑦(0)] + (𝜔𝜔𝑛𝑛 2 )[𝑌𝑌(𝑠𝑠)] = 0

(𝑠𝑠 2 + 2ζ𝜔𝜔𝑛𝑛 𝑠𝑠 + 𝜔𝜔𝑛𝑛 2 )𝑌𝑌(𝑠𝑠) = (𝑠𝑠 + 2ζ𝜔𝜔𝑛𝑛 )𝑦𝑦0

(𝑠𝑠+2ζ𝜔𝜔𝑛𝑛 )𝑦𝑦0 𝑁𝑁(𝑠𝑠)
Y(s) = = 𝐷𝐷(𝑠𝑠)
(𝑠𝑠 +2ζ𝜔𝜔𝑛𝑛 𝑠𝑠+𝜔𝜔𝑛𝑛 2 )
2

Characteristic equation => D(s) = 0

The roots of this equation are called Poles.

18
Numerator equation => N(s) = 0
The roots of this equation are called Zeros.
At Poles, D(s) = 0 => Y(s) = ∞
At Zeros, N(s) = 0 => Y(s) = 0
The plot of Poles and Zeros is called S-plane plot.
In this case, Y(s) has one zero and two poles
Zero: S = −2ζ𝜔𝜔𝑛𝑛

Poles: S1,2 = −ζ𝜔𝜔𝑛𝑛 ± 𝜔𝜔𝑛𝑛 �𝜁𝜁 2 − 1

S-plane Plot
S1,2 = −𝜁𝜁𝜔𝜔𝑛𝑛 ± 𝜔𝜔𝑛𝑛 �𝜁𝜁 2 − 1

Undamped: 𝜁𝜁 = 0; S1,2 = ±𝜔𝜔𝑛𝑛 √−1 = ±𝑗𝑗𝜔𝜔𝑛𝑛

Critically damped: 𝜁𝜁 = 1; S1,2 = −𝜔𝜔𝑛𝑛

Underdamped: 𝜁𝜁 < 1; S1 𝑀𝑀𝑠𝑠𝑐𝑐 S2 are complex

Overdamped: 𝜁𝜁 > 1; S1 𝑀𝑀𝑠𝑠𝑐𝑐 S2 are real

For 𝜁𝜁 < 1

S1,2 = −𝜁𝜁𝜔𝜔𝑛𝑛 ± 𝑗𝑗𝜔𝜔𝑛𝑛 �1 − 𝜁𝜁 2 = 𝐴𝐴 ± 𝑗𝑗𝑗𝑗

|𝑆𝑆| = √𝐴𝐴2 + 𝑗𝑗2 = �𝜁𝜁 2 𝜔𝜔𝑛𝑛2 + 𝜔𝜔𝑛𝑛2 (1 − 𝜁𝜁 2 ) = 𝜔𝜔𝑛𝑛 = 𝐶𝐶𝐴𝐴

19
 𝑗𝑗 𝜔𝜔𝑛𝑛 �1 − 𝜁𝜁 2 �1 − 𝜁𝜁 2
tan 𝜃𝜃 = = =−
𝐴𝐴 −𝜁𝜁𝜔𝜔𝑛𝑛 𝜁𝜁

For 0 < 𝜁𝜁 < 1 => 0 < 𝜃𝜃 < 90𝑜𝑜

𝜁𝜁 0 1
𝜃𝜃 90 0
(1 − 𝜁𝜁 2 ) 1
𝑠𝑠𝑠𝑠𝑐𝑐 2 𝜃𝜃 = 1 + 𝑡𝑡𝑀𝑀𝑠𝑠2 𝜃𝜃 = 1 + 2
= 2
𝜁𝜁 𝜁𝜁
1
𝑠𝑠𝑠𝑠𝑐𝑐𝜃𝜃 = ; 𝑐𝑐𝑐𝑐𝑠𝑠𝜃𝜃 = 𝜁𝜁; 𝜃𝜃 = cos −1 𝜁𝜁
𝜁𝜁

20
 Final Value Theorem
Final value (or Steady State) of y(t) is given by
lim 𝑦𝑦(𝑡𝑡) = lim 𝑠𝑠𝑠𝑠(𝑠𝑠)
𝑡𝑡→∞ 𝑠𝑠→0

Example:
𝑦𝑦̈ + 4𝑦𝑦̇ + 3𝑦𝑦 = 2𝑟𝑟(𝑡𝑡)
𝑦𝑦(0)=1, 𝑦𝑦̇ (0) = 0, r(t)=1, 𝑡𝑡 ≥ 0 (unit step function)

L.T.
2
[𝑠𝑠 2 𝑦𝑦(𝑠𝑠) − 𝑠𝑠𝑠𝑠(0) − 𝑦𝑦̇ (0)] + 4[𝑠𝑠𝑠𝑠(𝑠𝑠) − 𝑦𝑦(0)] + 3𝑦𝑦(𝑠𝑠) =
𝑠𝑠
2
𝑦𝑦(𝑠𝑠)(𝑠𝑠 2 + 4𝑠𝑠 + 3) = + (𝑠𝑠 + 4)
𝑠𝑠
2 (𝑠𝑠 + 4)
𝑦𝑦(𝑠𝑠) = +
𝑠𝑠(𝑠𝑠 + 3)(𝑠𝑠 + 1) (𝑠𝑠 + 3)(𝑠𝑠 + 1)
2 + 4𝑠𝑠 + 𝑠𝑠 2
=
𝑠𝑠(𝑠𝑠 + 3)(𝑠𝑠 + 1)

Steady State value,

lim 𝑦𝑦(𝑡𝑡) = 𝑠𝑠𝑠𝑠(𝑠𝑠)|𝑠𝑠=0
𝑡𝑡→∞

2 + 4𝑠𝑠 + 𝑠𝑠 2 2
= |𝑠𝑠=0 =
(𝑠𝑠 + 3)(𝑠𝑠 + 1) 3

21
 Example:
𝑦𝑦̈ + 3𝑦𝑦̇ + 2𝑦𝑦 = 1
𝑦𝑦(0) = 1, 𝑦𝑦̈ (0) = 0

L.T.
1
[𝑠𝑠 2 𝑦𝑦(𝑠𝑠) − 𝑠𝑠𝑠𝑠(0) − 𝑦𝑦̇ (0)] + 3[𝑠𝑠𝑠𝑠(𝑠𝑠) − 𝑦𝑦(0)] + 2𝑦𝑦(𝑠𝑠) =
𝑠𝑠
1
𝑦𝑦(𝑠𝑠)[𝑠𝑠 2 + 3𝑠𝑠 + 2] = + (𝑠𝑠 + 3)
𝑠𝑠
1 𝑠𝑠 + 3
𝑦𝑦(𝑠𝑠) = + 2
𝑠𝑠(𝑠𝑠 2 + 3𝑠𝑠 + 2) 𝑠𝑠 + 3𝑠𝑠 + 2

𝑠𝑠 2 + 3𝑠𝑠 + 1
=
𝑠𝑠(𝑠𝑠 2 + 3𝑠𝑠 + 2)
Final value,
lim 𝑦𝑦(𝑡𝑡) = 𝑠𝑠𝑠𝑠(𝑠𝑠)|𝑠𝑠=0
𝑡𝑡→∞

𝑠𝑠 2 + 3𝑠𝑠 + 1 1
= 2
|𝑠𝑠=0 =
𝑠𝑠 + 3𝑠𝑠 + 2 2

Initial value theorem

lim[ 𝑦𝑦(𝑡𝑡)] = lim [𝑠𝑠𝑠𝑠(𝑠𝑠)] = 𝑦𝑦(0)
𝑡𝑡→0 𝑠𝑠→∞

However, 𝑦𝑦(0) is the initial condition

In the above example,
1 3
1 + 3𝑠𝑠 + 𝑠𝑠 2 𝑠𝑠 2 + 𝑠𝑠 + 1
lim � 2 �=� � = 1 = 𝑦𝑦(0)
𝑠𝑠→∞ 𝑠𝑠 + 3𝑠𝑠 + 2 3 2
1+ + 2
𝑠𝑠 𝑠𝑠 𝑠𝑠=∞

22
 Assessment
1. [Z(s)] matrix is obtained from
A. 𝑆𝑆 2 [𝑀𝑀] + 𝑆𝑆 2 [𝐶𝐶] + 𝑆𝑆 2 [𝐾𝐾]
B. 𝑆𝑆[𝑀𝑀] + 𝑆𝑆 2 [𝐶𝐶] + 𝑆𝑆 3 [𝐾𝐾]
C. [𝑀𝑀] + 𝑆𝑆[𝐶𝐶] + 𝑆𝑆 2 [𝐾𝐾]
D. 𝑆𝑆 2 [𝑀𝑀] + 𝑆𝑆[𝐶𝐶] + [𝐾𝐾]

2. Transfer function G(s) can be obtained from

A. [𝑍𝑍(𝑠𝑠)]
B. [𝑍𝑍(𝑠𝑠)]−1
C. [𝑍𝑍(𝑠𝑠)][𝑍𝑍(𝑠𝑠)]
D. none of the above

3. Rule based method is applicable when,

A. the mass matrix is dynamically uncoupled
B. the mass matrix is diagonal
C. the stiffness and damping matrix are symmetric
D. all of the above

4. Poles are defined as

A. the roots of the characteristic equation
B. the square root of characteristic equation
C. the roots of numerator equation of transfer function
D. the square root of numerator equation

5. Zeros are defined as,

A. square root of characteristic equation
B. square root of numerator equation
C. roots of numerator equation
D. roots of denominator equation

6. At Poles, the output y(s) is equal to

A. zero
B. infinity
C. one
D. none of the above

7. At Zeros, the output y(s) is equal to

A. zero
B. infinity
C. one
D. none of the above

23
8. If 𝑺𝑺𝟏𝟏,𝟐𝟐 = −𝜻𝜻𝝎𝝎𝒏𝒏 ± 𝒋𝒋𝝎𝝎𝒅𝒅 , the magnitude is equal to
A. Natural frequency
B. Damped natural frequency
C. Damping ratio
D. None of the above

9. If 𝑺𝑺𝟏𝟏,𝟐𝟐 = −𝜻𝜻𝝎𝝎𝒏𝒏 ± 𝒋𝒋𝝎𝝎𝒅𝒅 , the angle is equal to

A. cos −1 𝜁𝜁
B. cos 𝜁𝜁
C. tan−1 𝜁𝜁
D. sin 𝜁𝜁

10. The final value theorem is useful to find

A. the value of output at t=0
B. the value of output at t=∞
C. the value of input at t=0
D. the value of input at t=∞

24
 3.0 Modeling of Control Systems

Learning Objectives:
1. To model the control system with block diagram and determine the transfer
function for the total system
2. To reduce the given block diagram with multiple blocks into single equivalent
block
3. To model the control system with signal flow graph and determine the transfer
function for the whole system
4. To find path transfer functions and loop transfer functions for applying Mason’s
formula

Block Diagram model

A control system would have several blocks and each block has a transfer function.
However, the transfer function for the whole system can be achieved by minimizing all the
blocks into a single equivalent block. The resulting transfer function would give the
required transfer function for the whole system. Minimizing the blocks is based on a set of
rules and they are given below.

3.1 Rules for minimizing the number of Blocks

1. Blocks in Series

2. Feedback Loop

25
3. Moving Summing Point

a) Ahead to Behind

b) Behind to Ahead

4. Moving Pick-off Point

a) Ahead to Behind

26
 b) Behind to Ahead

3.2 Equivalent Block diagram

Example: Minimize the given Block diagram to a single equivalent Block.

Step-1 Move the Pick-off point from ahead of G4 to behind G4

27
Step-2 Minimize the feedback loop of G3G4H1

Step-3

𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯
Let 𝑲𝑲𝟏𝟏 = � �, 𝑲𝑲𝟐𝟐 = � 𝟐𝟐 � and minimize the feedback loop of G2K1K2
𝟏𝟏−𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟏𝟏 𝑮𝑮𝟒𝟒

𝑮𝑮𝟐𝟐 𝑲𝑲𝟏𝟏
𝑷𝑷 =
𝟏𝟏 + 𝑮𝑮𝟐𝟐 𝑲𝑲𝟏𝟏 𝑲𝑲𝟐𝟐

Step-4 Minimize the feedback G1, P, H3

𝑮𝑮𝟏𝟏 𝑷𝑷
𝑸𝑸 =
𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑯𝑯𝟑𝟑 𝑷𝑷

𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒

𝑷𝑷 = �𝟏𝟏 + 𝑮𝑮𝟐𝟐
𝟏𝟏 − 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟏𝟏 𝑮𝑮𝟒𝟒 (𝟏𝟏 − 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟏𝟏 )

𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝟏𝟏 − 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟏𝟏 + 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑯𝑯𝟐𝟐

= �� �
𝟏𝟏 − 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟏𝟏 (𝟏𝟏 − 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟏𝟏 )

28
 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 (𝟏𝟏 − 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟏𝟏 )
=
(𝟏𝟏 − 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟏𝟏 ) (𝟏𝟏 − 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟏𝟏 + 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑯𝑯𝟐𝟐 )

𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒

= ; 𝑷𝑷 =
𝟏𝟏 − 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟏𝟏 + 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑯𝑯𝟐𝟐 𝑫𝑫

𝑫𝑫 = 𝟏𝟏 − 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟏𝟏 + 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑯𝑯𝟐𝟐

𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒

𝑮𝑮𝟏𝟏 𝑷𝑷 =
𝑫𝑫

𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟑𝟑

𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑷𝑷𝑯𝑯𝟑𝟑 = 𝟏𝟏 +
𝑫𝑫

𝑫𝑫 + 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟑𝟑

=
𝑫𝑫

𝑮𝑮𝟏𝟏 𝑷𝑷 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒

𝑸𝑸 = = =
𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑷𝑷𝑯𝑯𝟑𝟑 𝑫𝑫 + 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟑𝟑 (𝟏𝟏 − 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟏𝟏 + 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑯𝑯𝟐𝟐 + 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟑𝟑 )

Example:

Step-1 Minimize the feedback loop G1, H1

𝑮𝑮𝟏𝟏
𝑷𝑷 =
𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑯𝑯𝟏𝟏

29
Step-2 Minimize the feedback loop P, G2, H2

𝑷𝑷𝑮𝑮𝟐𝟐 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐

𝑸𝑸 = =
𝟏𝟏 + 𝑷𝑷𝑮𝑮𝟐𝟐 𝑯𝑯𝟐𝟐 𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑯𝑯𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑯𝑯𝟐𝟐

Step-3 Minimize Q, G3 blocks in series

𝒀𝒀 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑

𝑻𝑻 = = 𝑮𝑮𝟑𝟑 𝑸𝑸 =
𝑹𝑹 (𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑯𝑯𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑯𝑯𝟐𝟐 )

30
 Alternate Solution

This method is alternate to applying to the rules to minimize the blocks. In this method, the
output for each block is determined and then consolidated to find the transfer function for
a single equivalent block.

Step-1 Label the output from each block and summing points

Step-2 Write the equation, Output = (Input) (Transfer Function) for each block and simplify
to find Y = (T)(R) where T(s) is the Total or Equivalent transfer function.

E = R-C-D; C = AH1; D = BH2; E = R- AH1- BH2

A = EG1 = (R-AH1-BH2)G1; A(1+G1H1) = (R-BH2)G1

(𝑹𝑹 − 𝑩𝑩𝑯𝑯𝟐𝟐 )𝑮𝑮𝟏𝟏

𝑨𝑨 =
𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑯𝑯𝟏𝟏

(𝑹𝑹 − 𝑩𝑩𝑯𝑯𝟐𝟐 )𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐

𝑩𝑩 = 𝑨𝑨𝑮𝑮𝟐𝟐 = ; 𝑩𝑩(𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑯𝑯𝟏𝟏 ) = 𝑹𝑹𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 − 𝑩𝑩𝑯𝑯𝟐𝟐 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐
𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑯𝑯𝟏𝟏

𝑹𝑹𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐
𝑩𝑩(𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑯𝑯𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑯𝑯𝟐𝟐 ) = 𝑹𝑹𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 ; 𝑩𝑩 =
𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑯𝑯𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑯𝑯𝟐𝟐

𝑹𝑹𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑

𝒀𝒀 = 𝑩𝑩𝑮𝑮𝟑𝟑 =
𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑯𝑯𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑯𝑯𝟐𝟐

𝒀𝒀 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑

𝑻𝑻 = =
𝑹𝑹 𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑯𝑯𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑯𝑯𝟐𝟐

31
 3.3 Signal Flow Graph Models

“Signal flow graph” is also a viable method to model the control system. In this method
the input and output for each block is represented by “Nodes” designated by a circle and
the transfer function is represented by “Branches” designated by a line. The graph consists
of Nodes connected by Branches.

Block Diagram Signal Flow Graph

In signal flow graph the description of parameters are given below.

Nodes: Input and output points or junctions.

Path (or Branch): A line connecting one signal node with another Signal node. It represents
the transfer function (or the ratio of output to input)

Loop: A closed path that starts from output node and end at input node. It represents
feedback in the control system.

𝒀𝒀(𝒔𝒔)
Signal flow gain: 𝑻𝑻(𝒔𝒔) =
𝑹𝑹(𝒔𝒔)

The relation between paths and Signal flow gain (or transfer function) is given by Mason’s
∑(𝑷𝑷𝒌𝒌 △𝒌𝒌 )
formula, 𝑻𝑻 =
△

Where 𝑷𝑷𝒌𝒌 = 𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷 𝑷𝑷𝒐𝒐 𝒈𝒈𝒈𝒈𝒈𝒈𝒈𝒈𝒔𝒔 𝑷𝑷𝒈𝒈 𝒌𝒌𝑷𝑷𝒕𝒕 path from input, R(s) to output, Y(s)

Δ = determinant = 𝟏𝟏 − ∑𝑵𝑵
𝒈𝒈=𝟏𝟏 𝑳𝑳𝒈𝒈 + ∑ 𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎 − ∑ 𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎 𝑳𝑳𝒑𝒑 +⋅⋅⋅⋅⋅⋅

32
𝑳𝑳𝒈𝒈 = Loop gain, N = Number of Loops

𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎 = Product of gains of all combinations of two loops that are not touching each other.

𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎 𝑳𝑳𝒑𝒑 = Product of gains of all combinations of three loops that are not touching each
other.

𝚫𝚫𝒌𝒌 = Cofactor of the path 𝑷𝑷𝒌𝒌 .

= the value of Δ, after removing the loops, that are touching the kth path. (or the value of Δ
reduced by setting the loops touching the kth path to be zero)

Single Path: There is only one path connecting input(R) of output(Y).

N = 2 loops = L1 & L2

𝟐𝟐 𝟐𝟐

△= 𝟏𝟏 − � 𝑳𝑳𝒈𝒈 + � 𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎

𝒈𝒈=𝟏𝟏 𝒈𝒈=𝟏𝟏

= 𝟏𝟏 − (𝑳𝑳𝟏𝟏 + 𝑳𝑳𝟐𝟐 ) + 𝑳𝑳𝟏𝟏 𝑳𝑳𝟐𝟐

L1 of L2 are non-touching loops, they do not touch each other. So L1L2 ≠ 0.

Since there is only one path, k = 1. So ∆𝒌𝒌 = ∆𝟏𝟏 = 𝟏𝟏, since the loops L1 & L2 touch the path.
L1 = 0 = L2

Overlapping Loop
Since L1 and L2 overlap, these loops are
assumed touching each other. So L1L2 = 0.

Δ=1-(L1+L2) ∆𝟏𝟏 = 𝟏𝟏

In general, for all single paths, the loops are

touching each other. So, the product 𝑳𝑳𝟏𝟏 𝑳𝑳𝟐𝟐 = 𝟎𝟎

33
Sample for Touching loops:

Δ=1-(L1+L2) ∆𝟏𝟏 = 𝟏𝟏 Δ=1-(L1+L2) ∆𝟏𝟏 = 𝟏𝟏

L1, L2, L3  do not touch each

other. L4 overlaps L1, L2, L3
and hence it touches all other
loops.

∆= 𝟏𝟏 − (𝑳𝑳𝟏𝟏 + 𝑳𝑳𝟐𝟐 + 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟒𝟒 ) + (𝑳𝑳𝟏𝟏 𝑳𝑳𝟐𝟐 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟏𝟏 𝑳𝑳𝟑𝟑 ) − 𝑳𝑳𝟏𝟏 𝑳𝑳𝟐𝟐 𝑳𝑳𝟑𝟑

∆𝟏𝟏 = 𝟏𝟏 (Since L1, L2, L3, L4 touch the path)

Multiple Path
2 Path:
Path-1 = R-A-B-C-Y

Path-2 = R-D-E-F-Y

k = 1,2 (2 paths)

N = 4 Loops

L1 & L2 are touching and L3 & L4 are

touching; L1L2 = 0 = L3L4

L1 & L2 does not touch L3 & L4

34
So, ∑ 𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎 = 𝑳𝑳𝟏𝟏 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟏𝟏 𝑳𝑳𝟒𝟒 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟒𝟒

∑ 𝑳𝑳𝒈𝒈 = 𝑳𝑳𝟏𝟏 + 𝑳𝑳𝟐𝟐 + 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟒𝟒 ; ∑ 𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎 𝑳𝑳𝑷𝑷 = 𝟎𝟎

∆= 𝟏𝟏 − � 𝑳𝑳𝒈𝒈 + � 𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎

= 𝟏𝟏 − (𝑳𝑳𝟏𝟏 + 𝑳𝑳𝟐𝟐 + 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟒𝟒 ) + (𝑳𝑳𝟏𝟏 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟏𝟏 𝑳𝑳𝟒𝟒 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟒𝟒 )

k=1 ∆𝟏𝟏 = ∆|𝑳𝑳𝟏𝟏 =𝟎𝟎,𝑳𝑳𝟐𝟐 =𝟎𝟎 = 𝟏𝟏 − (𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟒𝟒 ) + (𝟎𝟎)

k=2 ∆𝟐𝟐 = ∆|𝑳𝑳𝟑𝟑 =𝟎𝟎,𝑳𝑳𝟒𝟒 =𝟎𝟎 = 𝟏𝟏 − (𝑳𝑳𝟏𝟏 + 𝑳𝑳𝟐𝟐 ) + (𝟎𝟎)

3-Path:
N=6

k = 1, 2, 3

Path-1: L1 = 0 = L2

Path-2: L3 = 0 = L4

Path-3: L5 = 0 = L6

Touching Loops: L1L2=0, L3L4=0, L5L6=0

∆= 𝟏𝟏 − � 𝑳𝑳𝒈𝒈 + � 𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎 − � 𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎 𝑳𝑳𝒑𝒑

� 𝑳𝑳𝒈𝒈 = 𝑳𝑳𝟏𝟏 + 𝑳𝑳𝟐𝟐 + 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟒𝟒 + 𝑳𝑳𝟓𝟓 + 𝑳𝑳𝟔𝟔

� 𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎 = 𝑳𝑳𝟏𝟏 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟏𝟏 𝑳𝑳𝟒𝟒 + 𝑳𝑳𝟏𝟏 𝑳𝑳𝟓𝟓 + 𝑳𝑳𝟏𝟏 𝑳𝑳𝟔𝟔 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟒𝟒 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟓𝟓 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟔𝟔 + 𝑳𝑳𝟑𝟑 𝑳𝑳𝟓𝟓

+ 𝑳𝑳𝟑𝟑 𝑳𝑳𝟔𝟔 + 𝑳𝑳𝟒𝟒 𝑳𝑳𝟓𝟓 + 𝑳𝑳𝟒𝟒 𝑳𝑳𝟔𝟔

� 𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎 𝑳𝑳𝒑𝒑 = 𝑳𝑳𝟏𝟏 𝑳𝑳𝟑𝟑 𝑳𝑳𝟓𝟓 + 𝑳𝑳𝟏𝟏 𝑳𝑳𝟑𝟑 𝑳𝑳𝟔𝟔 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟒𝟒 𝑳𝑳𝟓𝟓 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟒𝟒 𝑳𝑳𝟔𝟔 + 𝑳𝑳𝟏𝟏 𝑳𝑳𝟒𝟒 𝑳𝑳𝟓𝟓 + 𝑳𝑳𝟏𝟏 𝑳𝑳𝟒𝟒 𝑳𝑳𝟔𝟔 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟑𝟑 𝑳𝑳𝟓𝟓

+ 𝑳𝑳𝟐𝟐 𝑳𝑳𝟑𝟑 𝑳𝑳𝟔𝟔

35
k = 1: ∆𝟏𝟏 = ∆|𝑳𝑳𝟏𝟏 =𝟎𝟎,𝑳𝑳𝟐𝟐 =𝟎𝟎 = 𝟏𝟏 − (𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟒𝟒 + 𝑳𝑳𝟓𝟓 + 𝑳𝑳𝟔𝟔 ) + (𝑳𝑳𝟑𝟑 𝑳𝑳𝟓𝟓 + 𝑳𝑳𝟑𝟑 𝑳𝑳𝟔𝟔 + 𝑳𝑳𝟒𝟒 𝑳𝑳𝟓𝟓 + 𝑳𝑳𝟒𝟒 𝑳𝑳𝟔𝟔 )

k = 2: ∆𝟐𝟐 = ∆|𝑳𝑳𝟑𝟑 =𝟎𝟎,𝑳𝑳𝟒𝟒 =𝟎𝟎 = 𝟏𝟏 − (𝑳𝑳𝟏𝟏 + 𝑳𝑳𝟐𝟐 + 𝑳𝑳𝟓𝟓 + 𝑳𝑳𝟔𝟔 ) + (𝑳𝑳𝟏𝟏 𝑳𝑳𝟓𝟓 + 𝑳𝑳𝟏𝟏 𝑳𝑳𝟔𝟔 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟓𝟓 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟔𝟔 )

k = 3: ∆𝟑𝟑 = ∆|𝑳𝑳𝟓𝟓 =𝟎𝟎,𝑳𝑳𝟔𝟔 =𝟎𝟎 = 𝟏𝟏 − (𝑳𝑳𝟏𝟏 + 𝑳𝑳𝟐𝟐 + 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟒𝟒 ) + (𝑳𝑳𝟏𝟏 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟏𝟏 𝑳𝑳𝟒𝟒 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟒𝟒 )

Example:
For the given block diagram model, determine the signal flow graph model and find the
total transfer function.

Equivalent Signal Flow Graph Model

Number of Loops, N = 3 (All are

touching each other)

Number of Path, k = 1 (Only one

path from R(s) to Y(s))

𝒀𝒀(𝒔𝒔) 𝑷𝑷𝟏𝟏 ∆𝟏𝟏

𝑻𝑻(𝒔𝒔) = = (Mason’s formula)
𝑹𝑹(𝒔𝒔) ∆

Path gain, P1 = G1G2G3G4 = Product of T.F. along the path from input to output.

∆= 𝟏𝟏 − � 𝑳𝑳𝒈𝒈 + � 𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎

� 𝑳𝑳𝒈𝒈 = 𝑳𝑳𝟏𝟏 + 𝑳𝑳𝟐𝟐 + 𝑳𝑳𝟑𝟑

� 𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎 = 𝑳𝑳𝟏𝟏 𝑳𝑳𝟐𝟐 + 𝑳𝑳𝟏𝟏 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟑𝟑 = 𝟎𝟎

36
The loops L1, L2, and L3 overlap and hence they touch each other. So, the product (LnLm)
are zero.

∆= 𝟏𝟏 − (𝑳𝑳𝟏𝟏 + 𝑳𝑳𝟐𝟐 + 𝑳𝑳𝟑𝟑 ); ∆𝟏𝟏 = 𝟏𝟏

For a single path, all the loops touch the path and hence set L1= 0, L2 = 0, L3 = 0 in Δ.

Loop Transfer function = Product of T.F along the loop.

L1 = −G2G3H2

L2 = G3G4H1

L3 = −G1G2G3G4H3

𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒

𝑻𝑻(𝒔𝒔) =
(𝟏𝟏 + 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑯𝑯𝟐𝟐 − 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 𝑯𝑯𝟑𝟑 )

Example:
For the given signal flow Graph, determine the transfer function.

𝒑𝒑𝟏𝟏 ∆𝟏𝟏
𝑻𝑻(𝒔𝒔) =
∆

Number of Path = 1; k = 1

Path gain, 𝑷𝑷𝟏𝟏 = 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑

Number of loops = 2; L1 and L2

Loop gain: 𝑳𝑳𝟏𝟏 = −𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑯𝑯𝟏𝟏 and 𝑳𝑳𝟐𝟐 = −𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑯𝑯𝟐𝟐

∆= 𝟏𝟏 − � 𝑳𝑳𝒈𝒈 + � 𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎

� 𝑳𝑳𝒈𝒈 = 𝑳𝑳𝟏𝟏 + 𝑳𝑳𝟐𝟐

� 𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎 = 𝑳𝑳𝟏𝟏 𝑳𝑳𝟐𝟐 = 𝟎𝟎

Since the loops are touching.

37
 ∆= 𝟏𝟏 − (𝑳𝑳𝟏𝟏 + 𝑳𝑳𝟐𝟐 )

∆𝟏𝟏 = ∆|𝑳𝑳𝟏𝟏 =𝟎𝟎,𝑳𝑳𝟐𝟐 =𝟎𝟎 = 𝟏𝟏 (𝑺𝑺𝒈𝒈𝒈𝒈𝑷𝑷𝑺𝑺 𝒃𝒃𝑷𝑷𝑷𝑷𝒕𝒕 𝑷𝑷𝒕𝒕𝑺𝑺 𝒍𝒍𝑷𝑷𝑷𝑷𝒑𝒑𝒔𝒔 𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝑷𝒕𝒕 𝑷𝑷𝒕𝒕𝑺𝑺 𝒑𝒑𝒈𝒈𝑷𝑷𝒕𝒕)

𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑

𝑻𝑻(𝒔𝒔) =
(𝟏𝟏 + 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑯𝑯𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑯𝑯𝟐𝟐 )

This signal flow graph model can be represented as block diagram model as given below.

Step-1 Move the pick off point from ahead of G3 to behind G3

38
Step-2 Minimize 𝑮𝑮𝟐𝟐 , 𝑮𝑮𝟑𝟑 , 𝑯𝑯𝟏𝟏 feedback loop

𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑
𝑷𝑷 =
𝟏𝟏+𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑯𝑯𝟏𝟏

Step-3 Minimize 𝑮𝑮𝟏𝟏 , 𝑷𝑷, 𝑯𝑯𝟐𝟐 /𝑮𝑮𝟑𝟑 feedback loop

𝒀𝒀 𝑮𝑮𝟏𝟏 𝑷𝑷 𝑮𝑮𝟏𝟏 𝑮𝑮𝟑𝟑 𝑷𝑷

𝑻𝑻 = = =
𝑹𝑹 𝟏𝟏 + (𝑮𝑮𝟏𝟏 𝑷𝑷𝑯𝑯𝟐𝟐 /𝑮𝑮𝟑𝟑 ) 𝑮𝑮𝟑𝟑 + 𝑮𝑮𝟏𝟏 𝑯𝑯𝟐𝟐 𝑷𝑷

𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑
𝑮𝑮𝟏𝟏 𝑮𝑮𝟑𝟑 ( )
𝟏𝟏 + 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑯𝑯𝟏𝟏
= � 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑯𝑯𝟐𝟐
𝑮𝑮𝟑𝟑 �𝟏𝟏 + �
𝟏𝟏 + 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑯𝑯𝟏𝟏

𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑

𝑻𝑻 =
(𝟏𝟏 + 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑯𝑯𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑯𝑯𝟐𝟐 )

Example:

𝒑𝒑𝟏𝟏 ∆𝟏𝟏
𝑻𝑻(𝒔𝒔) =
∆

𝑷𝑷𝟏𝟏 = 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑

𝑳𝑳𝟏𝟏 = −𝑮𝑮𝟐𝟐 𝑯𝑯𝟏𝟏 and 𝑳𝑳𝟐𝟐 = −𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑯𝑯𝟐𝟐

∆= 𝟏𝟏 − � 𝑳𝑳𝒈𝒈 = 𝟏𝟏 − (𝑳𝑳𝟏𝟏 + 𝑳𝑳𝟐𝟐 )

∆𝟏𝟏 = 𝟏𝟏

39
 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑
𝑻𝑻(𝒔𝒔) =
(𝟏𝟏 + 𝑮𝑮𝟐𝟐 𝑯𝑯𝟏𝟏 + 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑯𝑯𝟐𝟐 )

Example:

∑ 𝑷𝑷𝒌𝒌 ∆𝒌𝒌 𝟏𝟏
𝑻𝑻(𝒔𝒔) = = (𝑷𝑷𝟏𝟏 ∆𝟏𝟏 + 𝑷𝑷𝟐𝟐 ∆𝟐𝟐 )
∆ ∆

Number of Paths = 2; k = 1, 2

Path gain,

𝑷𝑷𝟏𝟏 = 𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 (upper path)

𝑷𝑷𝟐𝟐 = 𝑮𝑮𝟓𝟓 𝑮𝑮𝟔𝟔 𝑮𝑮𝟕𝟕 𝑮𝑮𝟖𝟖 (lower path)

Number of loops = 4

Loop gain: 𝑳𝑳𝟏𝟏 = −𝑮𝑮𝟐𝟐 𝑯𝑯𝟐𝟐 ; 𝑳𝑳𝟐𝟐 = −𝑮𝑮𝟑𝟑 𝑯𝑯𝟑𝟑 ; 𝑳𝑳𝟑𝟑 = −𝑮𝑮𝟔𝟔 𝑯𝑯𝟔𝟔 ; 𝑳𝑳𝟒𝟒 = −𝑮𝑮𝟕𝟕 𝑯𝑯𝟕𝟕

Non-touching Loops: Loops L1 and L2 do not touch L3 and L4

∆= 𝟏𝟏 − � 𝑳𝑳𝒈𝒈 + � 𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎

� 𝑳𝑳𝒈𝒈 = 𝑳𝑳𝟏𝟏 + 𝑳𝑳𝟐𝟐 + 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟒𝟒 = −(𝑮𝑮𝟐𝟐 𝑯𝑯𝟐𝟐 + 𝑮𝑮𝟑𝟑 𝑯𝑯𝟑𝟑 + 𝑮𝑮𝟔𝟔 𝑯𝑯𝟔𝟔 + 𝑮𝑮𝟕𝟕 𝑯𝑯𝟕𝟕 )

� 𝑳𝑳𝒈𝒈 𝑳𝑳𝒎𝒎 = 𝑳𝑳𝟏𝟏 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟏𝟏 𝑳𝑳𝟒𝟒 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟒𝟒

(𝑳𝑳𝟏𝟏 𝑳𝑳𝟐𝟐 = 𝟎𝟎 = 𝑳𝑳𝟑𝟑 𝑳𝑳𝟒𝟒 , since there are touching loops)

∆= 𝟏𝟏 − (𝑳𝑳𝟏𝟏 + 𝑳𝑳𝟐𝟐 + 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟒𝟒 ) + (𝑳𝑳𝟏𝟏 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟏𝟏 𝑳𝑳𝟒𝟒 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟐𝟐 𝑳𝑳𝟒𝟒 )

∆𝟏𝟏 = ∆|𝑳𝑳𝟏𝟏 =𝟎𝟎=𝑳𝑳𝟐𝟐 = 𝟏𝟏 − (𝑳𝑳𝟑𝟑 + 𝑳𝑳𝟒𝟒 ) = 𝟏𝟏 + 𝑮𝑮𝟔𝟔 𝑯𝑯𝟔𝟔 + 𝑮𝑮𝟕𝟕 𝑯𝑯𝟕𝟕

∆𝟐𝟐 = ∆|𝑳𝑳𝟑𝟑 =𝟎𝟎=𝑳𝑳𝟒𝟒 = 𝟏𝟏 − (𝑳𝑳𝟏𝟏 + 𝑳𝑳𝟐𝟐 ) = 𝟏𝟏 + 𝑮𝑮𝟐𝟐 𝑯𝑯𝟐𝟐 + 𝑮𝑮𝟑𝟑 𝑯𝑯𝟑𝟑

𝑻𝑻(𝒔𝒔)
(𝑮𝑮𝟏𝟏 𝑮𝑮𝟐𝟐 𝑮𝑮𝟑𝟑 𝑮𝑮𝟒𝟒 )(𝟏𝟏 + 𝑮𝑮𝟔𝟔 𝑯𝑯𝟔𝟔 + 𝑮𝑮𝟕𝟕 𝑯𝑯𝟕𝟕 ) + (𝑮𝑮𝟓𝟓 𝑮𝑮𝟔𝟔 𝑮𝑮𝟕𝟕 𝑮𝑮𝟖𝟖 )(𝟏𝟏 + 𝑮𝑮𝟐𝟐 𝑯𝑯𝟐𝟐 + 𝑮𝑮𝟑𝟑 𝑯𝑯𝟑𝟑 )
=
[𝟏𝟏 + (𝑮𝑮𝟐𝟐 𝑯𝑯𝟐𝟐 + 𝑮𝑮𝟑𝟑 𝑯𝑯𝟑𝟑 + 𝑮𝑮𝟔𝟔 𝑯𝑯𝟔𝟔 + 𝑮𝑮𝟕𝟕 𝑯𝑯𝟕𝟕 ) + (𝑮𝑮𝟐𝟐 𝑮𝑮𝟔𝟔 𝑯𝑯𝟐𝟐 𝑯𝑯𝟔𝟔 + 𝑮𝑮𝟐𝟐 𝑮𝑮𝟕𝟕 𝑯𝑯𝟐𝟐 𝑯𝑯𝟕𝟕 + 𝑮𝑮𝟑𝟑 𝑮𝑮𝟔𝟔 𝑯𝑯𝟑𝟑 𝑯𝑯𝟔𝟔 + 𝑮𝑮𝟑𝟑 𝑮𝑮𝟕𝟕 𝑯𝑯𝟑𝟑 𝑯𝑯𝟕𝟕 )]

40
 Assessment

1. If G1 and G2 are in series, the equivalent transfer function is equal to

(a) G1/G2

(b) G1+G2

(c) G1-G2

(d) G1*G2

2. When you move the summing point from behind to ahead, the input ‘R2’ becomes
(a) R2+G

(b) R2/G

(c) R2-G

(d) R2*G

3. When you move the Pick-off point from ahead to behind, the Pick-off value is
(a) Divided by G

(b) Multiplied by G

(c) All of the above

(d) None of the above

4. In SFG (signal flow Graph), the Nodes represent

(a) Input

(b) Output

(c) Junction point in a path

(d) All of the above

41
5. In SFG, the lines (Branches) represent,
(a) The total transfer function

(b) The transfer function of a block

(c) The loop transfer function

(d) The path transfer function

6. The path transfer function is equal to

(a) Sum of all transfer function in a path

(b) Ratio of all transfer function in a path

(c) Product of all transfer function in a path

(d) None of the above

7. Mason’s formula is a function of

(a) Path transfer function

(b) Loop transfer function

(c) Determinant function

(d) All of the above

8. For single path,

(a) ∆1=1

(b) All the loop touches the path

(c) T(s)=P1∆1/∆

(d) All of the above

42
9. For two paths,
(a) T(s)= (P1+P2) ( ∆1+∆2)/ ∆

(b) T(s)= (P1∆1+P2∆2)/ ∆

(c) T(s)= (P1-P2) ( ∆1-∆2)/ ∆

(d) T(s)= (P1∆1+P2∆2)/ ∆

10. The loop are identified as touching, if

(a) They are adjacent

(b) They across

(c) They overlap

(d) All of the above

43
 4.0 Characteristics of Control System

Learning Objectives:
1) To study the role of error signals in negative feedback control system
2) To minimize the System Sensitivity to parameter changes and unwanted
disturbance or noise.
3) To apply the final value theorem to determine the Steady State error.

4.1 Error Signal Analysis

The error is defined as the difference between input (desired output) and output (actual
output). E(s) = R(s) -Y(s)
N =Noise
Td = Disturbance
H = 1 (unit feedback)
G = Plant T.F.
Gc = Controller T.F.

Y1 = Y+N
A1 = R – Y2; Y2 = Y1H = Y1
A1 = R – Y1 = R – Y – N
A2 = A1Gc = Gc (R – Y – N)
A3 = A2 + Td = Td + GcR –GcY –GcN
Y = A3G = GTd + GGcR – GGcY –GGcN
Y(1 + GGc) = GTd + GGcR – GGcN
𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺 𝐺𝐺𝐺𝐺𝑐𝑐
𝑌𝑌 = ( )𝑅𝑅 + ( )𝑇𝑇𝑑𝑑 − ( )𝑁𝑁
1 + 𝐺𝐺𝐺𝐺𝑐𝑐 1 + 𝐺𝐺𝐺𝐺𝑐𝑐 1 + 𝐺𝐺𝐺𝐺𝑐𝑐
Let L =𝐺𝐺𝐺𝐺𝑐𝑐 , the Loop Transfer function (forward Loop)
𝐿𝐿 𝐺𝐺 𝐿𝐿
𝑌𝑌 = ( )𝑅𝑅 + ( )𝑇𝑇𝑑𝑑 − ( )𝑁𝑁
1 + 𝐿𝐿 1 + 𝐿𝐿 1 + 𝐿𝐿

45
 1 𝐺𝐺 𝐿𝐿
Tracking Error, 𝐸𝐸(𝑠𝑠) = 𝑅𝑅(𝑠𝑠) − 𝑌𝑌(𝑠𝑠) = 𝑅𝑅( )−( )𝑇𝑇𝑑𝑑 + ( )𝑁𝑁
1+𝐿𝐿 1+𝐿𝐿 1+𝐿𝐿
1
Let 𝑆𝑆 = = 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓
1+𝐿𝐿

𝐿𝐿
𝐶𝐶 = = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓
1 + 𝐿𝐿
= 1 − 𝑆𝑆 ; 𝑆𝑆 + 𝐶𝐶 = 1
So, 𝑌𝑌 = 𝐶𝐶𝐶𝐶 + 𝐺𝐺𝐺𝐺𝑇𝑇𝑑𝑑 − 𝐶𝐶𝐶𝐶;
𝐸𝐸 = 𝑆𝑆𝑆𝑆 − 𝐺𝐺𝐺𝐺𝑇𝑇𝑑𝑑 + 𝐶𝐶𝐶𝐶

Disturbance
To minimize the disturbance, the Sensitivity ‘S’ should be small.
1
Since 𝑆𝑆 = ; ‘L’ should be large. Since 𝐿𝐿 = 𝐺𝐺𝐺𝐺𝑐𝑐 , ‘𝐺𝐺𝑐𝑐 ’ should be large for given ‘G’.
1+𝐿𝐿
So, we need to design a controller with large gain ‘𝐺𝐺𝑐𝑐 ’ over the required range of
frequencies. A typical frequency for disturbance is at low frequency range.
Noise
To minimize the noise, the complementary sensitivity function, ‘C’ should be small.
𝐿𝐿
Since 𝐶𝐶 = ; ‘L’ should be small.
1+𝐿𝐿

Since 𝐿𝐿 = 𝐺𝐺𝐺𝐺𝑐𝑐 , “𝐺𝐺𝑐𝑐 ” should be small over the required range of frequencies. So, the
controller gain should be large to reject the disturbance and it should be small to attenuate
the noise. It is difficult to achieve over the same frequency range. However, the typical
frequency range for disturbance is low and for noise is high. Therefore, the control
system can be designed such that the loop gain, “L” is large at low frequencies (to
minimize the disturbance) and small at high frequencies (to minimize the noise).

46
4.2 Sensitivity Analysis
𝐿𝐿 𝐺𝐺 𝐿𝐿
𝑌𝑌(𝑠𝑠) = ( )𝑅𝑅 + ( )𝑇𝑇𝑑𝑑 − ( )𝑁𝑁
1 + 𝐿𝐿 1 + 𝐿𝐿 1 + 𝐿𝐿
𝐿𝐿
𝐹𝐹𝑓𝑓𝐶𝐶 𝑇𝑇𝑑𝑑 = 0 = 𝑁𝑁, 𝑌𝑌 = 𝑅𝑅
1 + 𝐿𝐿
𝐼𝐼𝑓𝑓 𝐿𝐿 ≫ 1, 1 + 𝐿𝐿 ≈ 𝐿𝐿, 𝑌𝑌 ≈ 𝑅𝑅
1 𝐺𝐺 𝐿𝐿
𝐸𝐸(𝑠𝑠) = ( )𝑅𝑅 − ( )𝑇𝑇𝑑𝑑 + ( )𝑁𝑁
1 + 𝐿𝐿 1 + 𝐿𝐿 1 + 𝐿𝐿
1
𝐹𝐹𝑓𝑓𝐶𝐶 𝑇𝑇𝑑𝑑 = 0 = 𝑁𝑁, 𝐸𝐸 = ( )𝑅𝑅
1 + 𝐿𝐿
1
𝐼𝐼𝑓𝑓 𝐿𝐿 ≫ 1, 𝐸𝐸 ≈ ( )𝑅𝑅
𝐿𝐿
Since 𝐿𝐿 = 𝐺𝐺𝐺𝐺𝑐𝑐 and if 𝐺𝐺𝑐𝑐 is large, any small change in plant transfer function, G does not
affect sensitivity.
System Sensitivity
𝑌𝑌(𝑠𝑠)
System transfer function 𝑇𝑇(𝑠𝑠) =
𝑅𝑅(𝑠𝑠)

Sensitivity of T with respect to G, the plant transfer function,

𝜕𝜕𝑇𝑇/𝑇𝑇 𝜕𝜕𝑇𝑇 𝐺𝐺 𝜕𝜕(𝐶𝐶𝑆𝑆𝑇𝑇)
𝑆𝑆𝐺𝐺𝑇𝑇 = = ( ) ; 𝑆𝑆𝐺𝐺𝑇𝑇 =
𝜕𝜕𝐺𝐺/𝐺𝐺 𝜕𝜕𝐺𝐺 𝑇𝑇 𝜕𝜕(𝐶𝐶𝑆𝑆𝐺𝐺)

𝐺𝐺𝑐𝑐 𝐺𝐺
𝑌𝑌(𝑠𝑠) = ( )𝑅𝑅(𝑠𝑠)
1 + 𝐺𝐺𝑐𝑐 𝐺𝐺
𝑌𝑌(𝑠𝑠) 𝐺𝐺𝑐𝑐 𝐺𝐺
𝑇𝑇(𝑠𝑠) = =
𝑅𝑅(𝑠𝑠) 1 + 𝐺𝐺𝑐𝑐 𝐺𝐺
𝜕𝜕𝑇𝑇 (1 + 𝐺𝐺𝑐𝑐 𝐺𝐺)𝐺𝐺𝑐𝑐 − 𝐺𝐺𝑐𝑐 𝐺𝐺(𝐺𝐺𝑐𝑐 ) 𝐺𝐺𝑐𝑐
= 2
=
𝜕𝜕𝐺𝐺 (1 + 𝐺𝐺𝑐𝑐 𝐺𝐺) (1 + 𝐺𝐺𝑐𝑐 𝐺𝐺)2
𝜕𝜕𝑇𝑇 𝐺𝐺 𝐺𝐺𝑐𝑐 𝐺𝐺(1 + 𝐺𝐺𝑐𝑐 𝐺𝐺) 1 1
𝑆𝑆𝐺𝐺𝑇𝑇 = ( ) = 2
= =
𝜕𝜕𝐺𝐺 𝑇𝑇 (1 + 𝐺𝐺𝑐𝑐 𝐺𝐺) 𝐺𝐺𝑐𝑐 𝐺𝐺 1 + 𝐺𝐺𝑐𝑐 𝐺𝐺 1 + 𝐿𝐿
If 𝐺𝐺 = 𝑓𝑓(𝛼𝛼); sensitivity of T with respect to 𝛼𝛼,
𝑆𝑆𝛼𝛼𝑇𝑇 = 𝑆𝑆𝐺𝐺𝑇𝑇 𝑆𝑆𝛼𝛼𝐺𝐺

47
Error and System Sensitivity

𝑌𝑌 𝐺𝐺
Transfer Function, 𝑇𝑇 = = ;
𝑅𝑅 1+𝐺𝐺𝐺𝐺

𝜕𝜕𝑇𝑇 (1 + 𝐺𝐺𝐺𝐺)(1) − 𝐺𝐺(𝐺𝐺) 1

= 2
=
𝜕𝜕𝐺𝐺 (1 + 𝐺𝐺𝐺𝐺) (1 + 𝐺𝐺𝐺𝐺)2
𝜕𝜕𝑇𝑇 𝐺𝐺 1 𝐺𝐺(1 + 𝐺𝐺𝐺𝐺) 1
𝑆𝑆𝐺𝐺𝑇𝑇 = ( ) = =
𝜕𝜕𝐺𝐺 𝑇𝑇 (1 + 𝐺𝐺𝐺𝐺)2 𝐺𝐺 1 + 𝐺𝐺𝐺𝐺

𝑇𝑇 1
For unity feedback, H=1; 𝑆𝑆𝐺𝐺 =
1+𝐺𝐺
𝑌𝑌 𝐺𝐺
Error, 𝐸𝐸 = 𝑅𝑅 − 𝑌𝑌 = 𝑅𝑅(1 − ) = 𝑅𝑅(1 − 𝑇𝑇) = 𝑅𝑅 �1 − �
𝑅𝑅 1+𝐺𝐺𝐺𝐺

𝐸𝐸 1 + 𝐺𝐺(𝐺𝐺 − 1)
=
𝑅𝑅 1 + 𝐺𝐺𝐺𝐺
For H=1,
𝐸𝐸 1
= = 𝑆𝑆𝐺𝐺𝑇𝑇
𝑅𝑅 1 + 𝐺𝐺

Example.
𝐾𝐾 𝐺𝐺
Given, 𝐺𝐺 = ; 𝑇𝑇 =
𝜏𝜏𝑠𝑠+1 1+𝐺𝐺

Find the sensitivity of T with respect to 𝜏𝜏.

𝑆𝑆𝜏𝜏𝑇𝑇 = 𝑆𝑆𝐺𝐺𝑇𝑇 𝑆𝑆𝜏𝜏𝐺𝐺
𝐺𝐺 𝜕𝜕𝑇𝑇 (1 + 𝐺𝐺) − 𝐺𝐺 1
𝑇𝑇 = ; = =
1 + 𝐺𝐺 𝜕𝜕𝐺𝐺 (1 + 𝐺𝐺)2 (1 + 𝐺𝐺)2
𝜕𝜕𝑇𝑇 𝐺𝐺 1 𝐺𝐺(1 + 𝐺𝐺) 1 𝜏𝜏𝑠𝑠 + 1
𝑆𝑆𝐺𝐺𝑇𝑇 = ( ) = = =
𝜕𝜕𝐺𝐺 𝑇𝑇 (1 + 𝐺𝐺)2 𝐺𝐺 1 + 𝐺𝐺 𝜏𝜏𝑠𝑠 + 1 + 𝐾𝐾
𝐾𝐾 𝜕𝜕𝐺𝐺 𝐾𝐾𝑠𝑠
𝐺𝐺 = ; =−
𝜏𝜏𝑠𝑠 + 1 𝜕𝜕𝜏𝜏 (𝜏𝜏𝑠𝑠 + 1)2

48
 𝜕𝜕𝐺𝐺 𝜏𝜏 𝐾𝐾𝑠𝑠 𝜏𝜏 𝜏𝜏𝑠𝑠
𝑆𝑆𝜏𝜏𝐺𝐺 = ( ) =− 2
(𝜏𝜏𝑠𝑠 + 1) = −
𝜕𝜕𝜏𝜏 𝐺𝐺 (𝜏𝜏𝑠𝑠 + 1) 𝐾𝐾 (𝜏𝜏𝑠𝑠 + 1)
(𝜏𝜏𝑠𝑠 + 1) 𝜏𝜏𝑠𝑠 𝜏𝜏𝑠𝑠
𝑆𝑆𝜏𝜏𝑇𝑇 = 𝑆𝑆𝐺𝐺𝑇𝑇 𝑆𝑆𝜏𝜏𝐺𝐺 = (− )=−
(𝜏𝜏𝑠𝑠 + 1 + 𝐾𝐾) 𝜏𝜏𝑠𝑠 + 1 (𝜏𝜏𝑠𝑠 + 1 + 𝐾𝐾)

Disturbance Signal
It is an unwanted input signal that affect the output. For example, a car running over a
speed bump experiences an impulse as disturbance. It is an unwanted input to the cruise
control system of the car.

𝐴𝐴1 = 𝑅𝑅 − 𝑌𝑌1 ; 𝑌𝑌1 = 𝐺𝐺𝑌𝑌

𝐴𝐴1 = 𝑅𝑅 − 𝐺𝐺𝑌𝑌
𝐴𝐴2 = 𝐴𝐴1 𝐺𝐺1 = 𝐺𝐺1 𝑅𝑅 − 𝐺𝐺1 𝐺𝐺𝑌𝑌
𝐴𝐴3 = 𝐴𝐴2 + 𝑇𝑇𝑑𝑑 = 𝑇𝑇𝑑𝑑 + 𝐺𝐺1 𝑅𝑅 − 𝐺𝐺1 𝐺𝐺𝑌𝑌
𝑌𝑌 = 𝐴𝐴3 𝐺𝐺2 = 𝐺𝐺2 𝑇𝑇𝑑𝑑 + 𝐺𝐺1 𝐺𝐺2 𝑅𝑅 − 𝐺𝐺1 𝐺𝐺2 𝐺𝐺𝑌𝑌
𝑌𝑌(1 + 𝐺𝐺1 𝐺𝐺2 𝐺𝐺) = 𝐺𝐺2 𝑇𝑇𝑑𝑑 + 𝐺𝐺1 𝐺𝐺2 𝑅𝑅
𝐺𝐺2 𝐺𝐺1 𝐺𝐺2
𝑌𝑌 = � � 𝑇𝑇𝑑𝑑 + � � 𝑅𝑅
1 + 𝐺𝐺1 𝐺𝐺2 𝐺𝐺 1 + 𝐺𝐺1 𝐺𝐺2 𝐺𝐺

𝐺𝐺1 𝐺𝐺2 𝐺𝐺2

Error, 𝐸𝐸 = 𝑅𝑅 − 𝑌𝑌 = 𝑅𝑅 �1 − �−� � 𝑇𝑇𝑑𝑑
1+𝐺𝐺1 𝐺𝐺2 𝐺𝐺 1+𝐺𝐺1 𝐺𝐺2 𝐺𝐺

For unity feedback (H = 1);

1 𝐺𝐺2
𝐸𝐸 = 𝑅𝑅 � �−� � 𝑇𝑇
1 + 𝐺𝐺1 𝐺𝐺2 1 + 𝐺𝐺1 𝐺𝐺2 𝑑𝑑
The error due to disturbance alone can be obtained by assuming R = 0

49
 𝐺𝐺2
(𝐸𝐸)𝑑𝑑 = − � � 𝑇𝑇
1 + 𝐺𝐺1 𝐺𝐺2 𝑑𝑑
If 𝐺𝐺1 𝐺𝐺2 ≫ 1; 1 + 𝐺𝐺1 𝐺𝐺2 ≈ 𝐺𝐺1 𝐺𝐺2
𝐺𝐺2 1
(𝐸𝐸)𝑑𝑑 = − � � 𝑇𝑇𝑑𝑑 = −( )𝑇𝑇𝑑𝑑
𝐺𝐺1 𝐺𝐺2 𝐺𝐺1
To minimize the error due to disturbance, the gain of G1 should be large. Here ‘G1’
represents the controller gain. It agrees with previous conclusion that the controller
should be designed with large gain to minimize the effect of disturbance.

Noise attenuation
The error function in general,
𝐸𝐸 = (𝑠𝑠)𝑅𝑅 − (𝐺𝐺𝑠𝑠)𝑇𝑇𝑑𝑑 + (𝐶𝐶)𝑁𝑁
The error due to Noise can be obtained by assuming, R = 0 = Td
Then,
𝐿𝐿
(𝐸𝐸)𝑁𝑁 = (𝐶𝐶)𝑁𝑁 = � � 𝑁𝑁
1 + 𝐿𝐿
The error due to noise is low if the loop gain, L is low. Since 𝐿𝐿 = 𝐺𝐺𝐺𝐺𝑐𝑐 , it leads to the
controller gain 𝐺𝐺𝑐𝑐 to be low. Small loop gain ensures good noise attenuation while large
loop gain ensures rejection of disturbance. Therefore, the controller should have high
gain at low frequencies to minimize the effect of disturbance and low gain at high
frequencies to minimize the effect of noise.

Example.

Find the Error function.

𝑌𝑌1 = 𝑌𝑌 + 𝑁𝑁
𝑌𝑌2 = 𝐺𝐺𝑌𝑌1 = 𝐺𝐺𝑌𝑌 + 𝐺𝐺𝑁𝑁

50
 𝐴𝐴1 = 𝑅𝑅 − 𝑌𝑌2 = 𝑅𝑅 − 𝐺𝐺𝑌𝑌 − 𝐺𝐺𝑁𝑁
𝐴𝐴2 = 𝐴𝐴1 𝐺𝐺𝑐𝑐 = 𝐺𝐺𝑐𝑐 𝑅𝑅 − 𝐺𝐺𝑐𝑐 𝐺𝐺𝑌𝑌 − 𝐺𝐺𝑐𝑐 𝐺𝐺𝑁𝑁
𝑌𝑌 = 𝐴𝐴2 𝐺𝐺 = 𝐺𝐺𝐺𝐺𝑐𝑐 𝑅𝑅 − 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺𝑌𝑌 − 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺𝑁𝑁
𝑌𝑌(1 + 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺) = 𝐺𝐺𝐺𝐺𝑐𝑐 𝑅𝑅 − 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺𝑁𝑁
𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺
𝑌𝑌 = ( )𝑅𝑅 − ( )𝑁𝑁
1 + 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺 1 + 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺
𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺
𝐸𝐸 = 𝑅𝑅 − 𝑌𝑌 = �1 − � 𝑅𝑅 + � �N
1+𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺 1+𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺

Steady State Error (𝒆𝒆𝒔𝒔𝒔𝒔 )

The steady state error is the value of error function when time, ‘t’ is infinity. So,the final
value theorem can be applied to find the steady state value of the error function.
lim 𝑆𝑆(𝑆𝑆) = lim 𝑠𝑠𝐸𝐸(𝑠𝑠) = 𝑆𝑆𝑠𝑠𝑠𝑠
𝑡𝑡→∞ 𝑠𝑠→0

Or
𝑆𝑆𝑠𝑠𝑠𝑠 = [𝑠𝑠𝐸𝐸(𝑠𝑠)]𝑠𝑠=0

Example:
𝐾𝐾
If 𝐺𝐺(𝑠𝑠) = 𝐶𝐶𝑆𝑆𝑑𝑑 𝑅𝑅(𝑠𝑠) = 1/𝑠𝑠 ; find the value of K for the steady state error to be zero.
𝜏𝜏𝑠𝑠+1

Open Loop:

𝑌𝑌 = 𝑅𝑅𝐺𝐺
1 𝐾𝐾
𝐸𝐸 = 𝑅𝑅 − 𝑌𝑌 = 𝑅𝑅(1 − 𝐺𝐺) = �1 − �
𝑠𝑠 𝜏𝜏𝑠𝑠 + 1
𝐾𝐾
𝑆𝑆𝑠𝑠𝑠𝑠 = [𝑠𝑠𝐸𝐸(𝑠𝑠)]𝑠𝑠=0 = �1 − � = 1 − 𝐾𝐾
𝜏𝜏𝑠𝑠 + 1 𝑠𝑠=0
The steady state error, 𝑆𝑆𝑠𝑠𝑠𝑠 = 0 at K = 1

51
Closed Loop:

𝑌𝑌 𝐺𝐺
𝑇𝑇(𝑠𝑠) = =
𝑅𝑅 1 + 𝐺𝐺
𝑌𝑌 𝐺𝐺 1
𝐸𝐸(𝑠𝑠) = 𝑅𝑅 − 𝑌𝑌 = 𝑅𝑅 �1 − � = 𝑅𝑅(1 − 𝑇𝑇) = 𝑅𝑅 �1 − � = 𝑅𝑅 � �
𝑅𝑅 1 + 𝐺𝐺 1 + 𝐺𝐺
1 𝜏𝜏𝑠𝑠 + 1
𝐸𝐸(𝑠𝑠) = =
𝐾𝐾 𝑠𝑠[𝜏𝜏𝑠𝑠 + 1 + 𝐾𝐾]
𝑠𝑠 �1 + �
𝜏𝜏𝑠𝑠 + 1
𝜏𝜏𝑠𝑠 + 1 1
𝑆𝑆𝑠𝑠𝑠𝑠 = [𝑠𝑠𝐸𝐸(𝑠𝑠)]𝑠𝑠=0 = � � =
𝜏𝜏𝑠𝑠 + 1 + 𝐾𝐾 𝑆𝑆=0 1 + 𝐾𝐾
The steady state error will be minimum at large value of ‘K’. It will be zero when the
value of ‘K’ is infinity.
Example.
We derived the error function for a system with noise.
𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺
𝐸𝐸 = 𝑅𝑅 �1 − �+� � 𝑁𝑁
1 + 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺 1 + 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺
100 1 1
If 𝐺𝐺(𝑆𝑆) = ; 𝐺𝐺𝑐𝑐 = 𝐾𝐾; 𝐺𝐺 = ; 𝑅𝑅 = = 𝑁𝑁;
𝑆𝑆+100 𝑠𝑠+5 𝑠𝑠

Determine the steady state error for the cases (a) N = 0 and (b) R = 0
(a) N = 0, R = 1/s
1 + 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺 − 𝐺𝐺𝐺𝐺𝑐𝑐 1
𝐸𝐸(𝑠𝑠) = � �� �
1 + 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺 𝑠𝑠
100𝐾𝐾 100𝐾𝐾
𝐺𝐺𝐺𝐺𝑐𝑐 = ; 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺 =
𝑠𝑠 + 100 (𝑠𝑠 + 100)(𝑠𝑠 + 5)
100𝐾𝐾 100𝐾𝐾
1 + 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺 − 𝐺𝐺𝐺𝐺𝑐𝑐 = 1 + −
(𝑠𝑠 + 100)(𝑠𝑠 + 5) 𝑠𝑠 + 100
(𝑠𝑠 + 100)(𝑠𝑠 + 5) + 100𝐾𝐾 − 100𝐾𝐾(𝑠𝑠 + 5)
=
(𝑠𝑠 + 100)(𝑠𝑠 + 5)
(𝑠𝑠 + 100)(𝑠𝑠 + 5) + 100𝐾𝐾
1 + 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺 =
(𝑠𝑠 + 100)(𝑠𝑠 + 5)

52
 (𝑠𝑠 + 100)(𝑠𝑠 + 5) + 100𝐾𝐾 − 100𝐾𝐾(𝑠𝑠 + 5)
𝑆𝑆𝑠𝑠𝑠𝑠 = [𝑠𝑠𝐸𝐸(𝑠𝑠)]𝑠𝑠=0 = � �
(𝑠𝑠 + 100)(𝑠𝑠 + 5) + 100𝐾𝐾 𝑠𝑠=0
500 − 400𝐾𝐾 5 − 4𝐾𝐾
𝑆𝑆𝑠𝑠𝑠𝑠 = =
500 + 100𝐾𝐾 5 + 𝐾𝐾
For 𝑆𝑆𝑠𝑠𝑠𝑠 = 0, 𝐾𝐾 = 5/4 = 1.25
(b) R = 0, N = 1/s
𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺 1
𝐸𝐸(𝑠𝑠) = � �� �
1 + 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺 𝑠𝑠
100𝐾𝐾
𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺 =
(𝑠𝑠 + 100)(𝑠𝑠 + 5)
(𝑠𝑠 + 100)(𝑠𝑠 + 5) + 100𝐾𝐾
1 + 𝐺𝐺𝐺𝐺𝑐𝑐 𝐺𝐺 =
(𝑠𝑠 + 100)(𝑠𝑠 + 5)
100𝐾𝐾 100𝐾𝐾 𝐾𝐾
𝑆𝑆𝑠𝑠𝑠𝑠 = [𝑠𝑠𝐸𝐸(𝑠𝑠)]𝑠𝑠=0 = � � = =
(𝑠𝑠 + 100)(𝑠𝑠 + 5) + 100𝐾𝐾 𝑠𝑠=0 500 + 100𝐾𝐾 5 + 𝐾𝐾

The steady state error will be minimum for lower value of controller gain.
For 𝐾𝐾 = 1, 𝑆𝑆𝑠𝑠𝑠𝑠 = 1/6

Example.

10
Given: 𝐺𝐺𝑐𝑐 = 𝐾𝐾 & 𝐺𝐺(𝑠𝑠) =
𝑠𝑠(𝑠𝑠+2)

Determine the controller gain, K for ramp input such that the steady state error is 1%.
1
For ramp input, (𝑠𝑠) = ; steady state error, 𝑆𝑆𝑠𝑠𝑠𝑠 = 0.01
𝑠𝑠 2
𝑌𝑌 𝐺𝐺𝐺𝐺𝑐𝑐
System Transfer function, 𝑇𝑇(𝑠𝑠) = =
𝑅𝑅 1+𝐺𝐺𝐺𝐺𝑐𝑐

Error function,
1 𝐺𝐺𝐺𝐺𝑐𝑐 1 1
𝐸𝐸 = 𝑅𝑅(1 − 𝑇𝑇) = �1 − � = � �
𝑠𝑠 2 1 + 𝐺𝐺𝐺𝐺𝑐𝑐 𝑠𝑠 2 1 + 𝐺𝐺𝐺𝐺𝑐𝑐

53
 10𝐾𝐾 𝑠𝑠(𝑠𝑠 + 2) + 10𝐾𝐾
1 + 𝐺𝐺𝐺𝐺𝑐𝑐 = 1 + =
𝑠𝑠(𝑠𝑠 + 2) 𝑠𝑠(𝑠𝑠 + 2)
1 𝑠𝑠(𝑠𝑠 + 2) 1 𝑠𝑠 + 2
𝐸𝐸(𝑠𝑠) = 2
� �= � �
𝑠𝑠 𝑠𝑠(𝑠𝑠 + 2) + 10𝐾𝐾 𝑠𝑠 𝑠𝑠(𝑠𝑠 + 2) + 10𝐾𝐾
𝑠𝑠 + 2 2
𝑆𝑆𝑠𝑠𝑠𝑠 = [𝑠𝑠𝐸𝐸(𝑠𝑠)]𝑠𝑠=0 = � � =
𝑠𝑠(𝑠𝑠 + 2) + 10𝐾𝐾 𝑠𝑠=0 10𝐾𝐾
1
0.01 = , 𝐾𝐾 = 20
5𝐾𝐾

Time Constant

Time constant is defined as time taken for the output to change due to applied step input.
Standard format for first order system transfer function is
1
𝐺𝐺(𝑠𝑠) =
𝜏𝜏𝑐𝑐 𝑠𝑠 + 1
The coefficient of s, “𝜏𝜏𝑐𝑐 ” is the time constant.
𝑌𝑌 1 1
Let = 𝐺𝐺 = ; 𝑅𝑅 =
𝑅𝑅 𝜏𝜏𝑐𝑐 𝑠𝑠+1 𝑠𝑠

1 1 𝜏𝜏𝑐𝑐
𝑌𝑌 = = −
𝑠𝑠(𝜏𝜏𝑐𝑐 𝑠𝑠 + 1) 𝑠𝑠 𝜏𝜏𝑐𝑐 𝑠𝑠 + 1
1 1
𝑌𝑌(𝑠𝑠) = −
𝑠𝑠 𝑠𝑠 + 1
𝜏𝜏𝑐𝑐
𝑡𝑡 1 @ 𝜏𝜏𝑐𝑐 = 0
𝑆𝑆(𝑆𝑆) = 1 − 𝑆𝑆 − �𝜏𝜏𝑐𝑐 = �
0 @𝜏𝜏𝑐𝑐 = ∞
Example:
𝐾𝐾
Let 𝐺𝐺(𝑠𝑠) =
𝑎𝑎𝑠𝑠+𝑏𝑏

Open Loop

54
 𝑌𝑌 𝐾𝐾
𝑇𝑇 = = 𝐺𝐺 =
𝑅𝑅 𝐶𝐶𝑠𝑠 + 𝑏𝑏
𝐾𝐾 1 𝐾𝐾1
In standard format, 𝑇𝑇 = � �=
𝑏𝑏 (𝑎𝑎/𝑏𝑏)𝑠𝑠+1 𝜏𝜏𝑐𝑐 𝑠𝑠+1

Time constant, 𝜏𝜏𝑐𝑐 = 𝐶𝐶/𝑏𝑏; 𝐾𝐾1 = 𝐾𝐾/𝑏𝑏

Closed Loop

𝑌𝑌 𝐺𝐺 𝐾𝐾 𝐾𝐾
𝑇𝑇 = = =� ���1 + �
𝑅𝑅 1 + 𝐺𝐺 𝐶𝐶𝑠𝑠 + 𝑏𝑏 𝐶𝐶𝑠𝑠 + 𝑏𝑏

𝐾𝐾 𝐾𝐾 1 𝐾𝐾
𝑇𝑇(𝑠𝑠) = = � �=
𝐶𝐶𝑠𝑠 + (𝑏𝑏 + 𝐾𝐾) (𝑏𝑏 + 𝐾𝐾) �𝐶𝐶� 𝜏𝜏𝑐𝑐 𝑠𝑠 + 1
(𝑏𝑏 + 𝑘𝑘)� 𝑠𝑠 + 1
Time constant, 𝜏𝜏𝑐𝑐 = 𝐶𝐶/(𝑏𝑏 + 𝐾𝐾); 𝑘𝑘1 = 𝐾𝐾/(𝑏𝑏 + 𝐾𝐾)

Let K = 150, a = 5, b = 10; 𝐺𝐺(𝑠𝑠) = 150�(5𝑠𝑠 + 10)

For open loop, 𝜏𝜏𝑐𝑐 = 𝐶𝐶�𝑏𝑏 = 0.5 𝑠𝑠𝑆𝑆𝑓𝑓; 𝐾𝐾1 = 15

For closed loop, 𝜏𝜏𝑐𝑐 = 𝐶𝐶�(𝑏𝑏 + 𝐾𝐾) = 1�32 = 0.03 𝑠𝑠𝑆𝑆𝑓𝑓; 𝐾𝐾1 = 15�16

55
 Blood Pressure Control during Anesthesia

Objective:
To develop an automated system to regulate the depth of anesthesia. If the depth is low,
the patient will feel greater pain and if the depth is high, the patient will die or become
comatose. To measure the depth of anesthesia, the parameter used in general is Mean
Arterial Pressure (MAP). The level of MAP serves as a guide for the delivery of
anesthesia.
The control system is given below:

1
Let 𝐺𝐺𝑐𝑐 (𝑠𝑠) = [𝑘𝑘𝐷𝐷 𝑠𝑠 2 + 𝑘𝑘𝑃𝑃 𝑠𝑠 + 𝑘𝑘𝐼𝐼 ] 𝑤𝑤ℎ𝑆𝑆𝐶𝐶𝑆𝑆 𝑘𝑘𝐷𝐷 , 𝑘𝑘𝑃𝑃 , 𝑘𝑘𝐼𝐼 𝐶𝐶𝐶𝐶𝑆𝑆 𝑓𝑓𝑓𝑓𝑆𝑆𝑆𝑆𝐶𝐶𝑓𝑓𝐶𝐶𝐶𝐶𝑆𝑆𝐶𝐶 𝑔𝑔𝐶𝐶𝑆𝑆𝑆𝑆𝑠𝑠
𝑠𝑠

1 1
𝐺𝐺𝑝𝑝 (𝑠𝑠) = ; 𝐺𝐺(𝑠𝑠) = ; 𝐺𝐺(𝑠𝑠) = 1
𝑠𝑠 (𝑠𝑠 + 𝐶𝐶)2
𝑅𝑅(𝑠𝑠) = 𝑅𝑅0 ⁄𝑠𝑠 (𝑠𝑠𝑆𝑆𝑆𝑆𝐶𝐶 𝑆𝑆𝑆𝑆𝐶𝐶𝑓𝑓𝑆𝑆); 𝑇𝑇𝑑𝑑 (𝑠𝑠) = 𝐷𝐷0 ⁄𝑠𝑠 (𝑠𝑠𝑆𝑆𝑆𝑆𝐶𝐶 𝑑𝑑𝑆𝑆𝑠𝑠𝑆𝑆𝑓𝑓𝐶𝐶𝑏𝑏𝐶𝐶𝑆𝑆𝑓𝑓𝑆𝑆)
𝐴𝐴1 = 𝑅𝑅 − 𝑌𝑌, 𝑠𝑠𝑆𝑆𝑆𝑆𝑓𝑓𝑆𝑆 𝐺𝐺 = 1
𝐴𝐴2 = 𝐺𝐺𝑐𝑐 𝐴𝐴1 = 𝐺𝐺𝑐𝑐 𝑅𝑅 − 𝐺𝐺𝑐𝑐 𝑌𝑌
𝐴𝐴3 = 𝐺𝐺𝑝𝑝 𝐴𝐴2 = 𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 𝑅𝑅 − 𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 𝑌𝑌

𝐴𝐴4 = −𝑇𝑇𝑑𝑑 + 𝐴𝐴3 = −𝑇𝑇𝑑𝑑 + 𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 𝑅𝑅 − 𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 𝑌𝑌

𝑌𝑌 = 𝐺𝐺𝐴𝐴4 = −𝐺𝐺𝑇𝑇𝑑𝑑 + 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 𝑅𝑅 − 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 𝑌𝑌

𝑌𝑌(1 + 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 ) = (𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 )𝑅𝑅 − 𝐺𝐺𝑇𝑇𝑑𝑑

𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 𝐺𝐺
𝑌𝑌(𝑠𝑠) = � � 𝑅𝑅 − � � 𝑇𝑇
1 + 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 1 + 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 𝑑𝑑

1 𝐺𝐺
𝐸𝐸(𝑠𝑠) = 𝑅𝑅(𝑠𝑠) − 𝑌𝑌(𝑠𝑠) = � � 𝑅𝑅 + � � 𝑇𝑇
1 + 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 1 + 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 𝑑𝑑

56
Steady state error when 𝐓𝐓𝐝𝐝 = 𝟎𝟎
𝑘𝑘𝐷𝐷 𝑠𝑠 2 + 𝑘𝑘𝑃𝑃 𝑠𝑠 + 𝑘𝑘𝐼𝐼 𝑘𝑘𝐷𝐷 𝑠𝑠 2 + 𝑘𝑘𝑃𝑃 𝑠𝑠 + 𝑘𝑘𝐼𝐼
𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 = =
𝑠𝑠 2 (𝑠𝑠 + 𝑝𝑝)2 𝑠𝑠 4 + 2𝑝𝑝𝑝𝑝 3 + 𝑝𝑝2 𝑠𝑠 2
1 𝑅𝑅0 𝑠𝑠 4 + 2𝑝𝑝𝑝𝑝 3 + 𝑝𝑝2 𝑠𝑠 2 𝑅𝑅0
𝐸𝐸(𝑠𝑠) = � �� � = � 4 � � �
1 + 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 𝑠𝑠 𝑠𝑠 + 2𝑝𝑝𝑝𝑝 3 + (𝑝𝑝2 + 𝑘𝑘𝐷𝐷 )𝑠𝑠 2 + 𝑘𝑘𝑃𝑃 𝑠𝑠 + 𝑘𝑘𝐼𝐼 𝑠𝑠

𝑅𝑅0 (𝑠𝑠 4 + 2𝑝𝑝𝑝𝑝 3 + 𝑝𝑝2 𝑠𝑠 2 )

𝑒𝑒𝑠𝑠𝑠𝑠 = [𝑠𝑠𝑠𝑠(𝑠𝑠)]𝑠𝑠=0 = = 0; 𝑎𝑎𝑎𝑎 𝑠𝑠 = 0
𝑠𝑠 4 + 2𝑝𝑝𝑝𝑝 3 + (𝑝𝑝2 + 𝑘𝑘𝐷𝐷 )𝑠𝑠 2 + 𝑘𝑘𝑃𝑃 𝑠𝑠 + 𝑘𝑘𝐼𝐼
Steady state error due to step input of magnitude, 𝑅𝑅0 is zero.

Steady state output to step disturbance (R=0)

𝐷𝐷0
𝑇𝑇𝑑𝑑 =
𝑠𝑠
𝐺𝐺 −𝑠𝑠 2 𝐷𝐷0
𝑌𝑌(𝑠𝑠) = − � � 𝑇𝑇𝑑𝑑 = 4 � �
1 + 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 [𝑠𝑠 + 2𝑝𝑝𝑝𝑝 + (𝑝𝑝 + 𝑘𝑘𝐷𝐷 )𝑠𝑠 + 𝑘𝑘𝑃𝑃 𝑠𝑠 + 𝑘𝑘𝐼𝐼 ] 𝑠𝑠
3 2 2

−𝐷𝐷0 𝑠𝑠 2
𝑙𝑙𝑙𝑙𝑙𝑙 𝑦𝑦(𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙[𝑠𝑠𝑠𝑠(𝑠𝑠)] = 4 = 0 @ 𝑠𝑠 = 0
𝑡𝑡→∞ 𝑠𝑠→0 [𝑠𝑠 + 2𝑝𝑝𝑝𝑝 3 + (𝑝𝑝2 + 𝑘𝑘𝐷𝐷 )𝑠𝑠 2 + 𝑘𝑘𝑃𝑃 𝑠𝑠 + 𝑘𝑘𝐼𝐼 ]
The step disturbance (surgical disturbance) of magnitude, D0 does not affect the steady
state output.

Sensitivity of closed loop T.F. to changes in ‘p’, the MAP. (𝐓𝐓𝐝𝐝 = 𝟎𝟎)
𝑆𝑆𝑝𝑝𝑇𝑇 = 𝑆𝑆𝐺𝐺𝑇𝑇 𝑆𝑆𝑝𝑝𝐺𝐺
𝑌𝑌 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐
𝑇𝑇 = =
𝑅𝑅 1 + 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐

𝜕𝜕𝜕𝜕 �(1 + 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 )(𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 ) − (𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 )(𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 )� 𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐
= 2
=
𝜕𝜕𝜕𝜕 (1 + 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 ) (1 + 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 )2
𝜕𝜕𝜕𝜕 𝐺𝐺 𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 (1 + 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 ) 1
𝑆𝑆𝐺𝐺𝑇𝑇 = � �� � = =
𝜕𝜕𝜕𝜕 𝑇𝑇 (1 + 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 )2 𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 1 + 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐
1
𝐺𝐺 =
(𝑠𝑠 + 𝑝𝑝)2
𝜕𝜕𝜕𝜕 (𝑠𝑠 + 𝑝𝑝)2 (0) − 2(𝑠𝑠 + 𝑝𝑝) −2
= =
𝜕𝜕𝜕𝜕 (𝑠𝑠 + 𝑝𝑝)4 (𝑠𝑠 + 𝑝𝑝)3

57
 𝜕𝜕𝐺𝐺 𝐶𝐶 −2 2𝐶𝐶
𝑆𝑆𝑝𝑝𝐺𝐺 = � �� � = 3
𝐶𝐶(𝑠𝑠 + 𝐶𝐶)2 = −
𝜕𝜕𝜕𝜕 𝐺𝐺 (𝑠𝑠 + 𝐶𝐶) 𝑠𝑠 + 𝐶𝐶
1 𝑠𝑠 2 (𝑠𝑠 + 𝐶𝐶)2
𝑆𝑆𝐺𝐺𝑇𝑇 = = 4
1 + 𝐺𝐺𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 [𝑠𝑠 + 2𝐶𝐶𝑠𝑠 3 + (𝐶𝐶2 + 𝑘𝑘𝐷𝐷 )𝑠𝑠 2 + 𝑘𝑘𝑃𝑃 𝑠𝑠 + 𝑘𝑘𝐼𝐼 ]

−2𝐶𝐶𝑠𝑠 2 (𝑠𝑠 + 𝐶𝐶)

𝑆𝑆𝑝𝑝𝑇𝑇 = 𝑆𝑆𝐺𝐺𝑇𝑇 𝑆𝑆𝑝𝑝𝐺𝐺 =
[𝑠𝑠 4 + 2𝐶𝐶𝑠𝑠 3 + (𝐶𝐶2 + 𝑘𝑘𝐷𝐷 )𝑠𝑠 2 + 𝑘𝑘𝑃𝑃 𝑠𝑠 + 𝑘𝑘𝐼𝐼 ]
Let 𝑘𝑘𝑃𝑃 = 6, 𝑘𝑘𝐷𝐷 = 4, 𝑘𝑘𝐼𝐼 = 1, 𝐶𝐶 = 2
(−4𝑠𝑠 3 − 8𝑠𝑠 2 )
𝑆𝑆𝑝𝑝𝑇𝑇 = 4
(𝑠𝑠 + 4𝑠𝑠 3 + 8𝑠𝑠 2 + 6𝑠𝑠 + 1)

Mars Rover Control

The goal is to operate the Rover with minimum disturbances such as rocks, with low
sensitivity to changes in controller gain, ‘K’.

1 1
𝐺𝐺(𝑠𝑠) = = 2
(𝑠𝑠 + 1)(𝑠𝑠 + 3) (𝑠𝑠 + 4𝑠𝑠 + 3)
𝐴𝐴1 = 𝑅𝑅 − 𝑌𝑌
𝐴𝐴2 = 𝐴𝐴1 𝐾𝐾 = 𝐾𝐾𝑅𝑅 − 𝐾𝐾𝑌𝑌
𝐴𝐴3 = 𝑇𝑇𝑑𝑑 + 𝐴𝐴2 = 𝑇𝑇𝑑𝑑 + 𝐾𝐾𝑅𝑅 − 𝐾𝐾𝑌𝑌
𝑌𝑌 = 𝐺𝐺𝐴𝐴3 = 𝐺𝐺𝑇𝑇𝑑𝑑 + 𝐺𝐺𝐾𝐾𝑅𝑅 − 𝐺𝐺𝐾𝐾𝑌𝑌
𝑌𝑌(1 + 𝐺𝐺𝐾𝐾) = (𝐺𝐺𝐾𝐾)𝑅𝑅 − 𝐺𝐺𝑇𝑇𝑑𝑑
𝐺𝐺𝐾𝐾 𝐺𝐺
𝑌𝑌(𝑠𝑠) = � � 𝑅𝑅 − � � 𝑇𝑇
1 + 𝐺𝐺𝐾𝐾 1 + 𝐺𝐺𝐾𝐾 𝑑𝑑
Sensitivity of T(s) when 𝑇𝑇𝑑𝑑 = 0
𝑌𝑌 𝐺𝐺𝐾𝐾
𝑇𝑇 = =
𝑅𝑅 1 + 𝐺𝐺𝐾𝐾

58
 𝜕𝜕𝜕𝜕 𝐾𝐾
𝑆𝑆𝐾𝐾𝑇𝑇 = � �� �
𝜕𝜕𝜕𝜕 𝑇𝑇
𝜕𝜕𝜕𝜕 [(1 + 𝐺𝐺𝐺𝐺)𝐺𝐺 − 𝐺𝐺𝐺𝐺(𝐺𝐺)] 𝐺𝐺
= =
𝜕𝜕𝜕𝜕 (1 + 𝐺𝐺𝐺𝐺)2 (1 + 𝐺𝐺𝐺𝐺)2
𝐺𝐺 𝐾𝐾(1 + 𝐺𝐺𝐺𝐺) 1 𝑠𝑠 2 + 4𝑠𝑠 + 3
𝑆𝑆𝐾𝐾𝑇𝑇 = = =
(1 + 𝐺𝐺𝐺𝐺)2 𝐺𝐺𝐺𝐺 1 + 𝐺𝐺𝐺𝐺 𝑠𝑠 2 + 4𝑠𝑠 + (3 + 𝐾𝐾)
Let 𝑠𝑠 = 𝑗𝑗𝑗𝑗
−𝜔𝜔2 + 4𝑗𝑗𝑗𝑗 + 3
𝑆𝑆𝐾𝐾𝑇𝑇 =
−𝜔𝜔 2 + 4𝑗𝑗𝑗𝑗 + 3 + 𝐾𝐾
(3 − 𝜔𝜔2 ) + 𝑗𝑗(4𝜔𝜔)
=
(𝐾𝐾 + 3 − 𝜔𝜔 2 ) + 𝑗𝑗(4𝜔𝜔)

�(3 − 𝜔𝜔 2 )2 + 16𝜔𝜔 2
|𝑆𝑆𝐾𝐾𝑇𝑇 | =
�(𝐾𝐾 + 3 − 𝜔𝜔 2 )2 + 16𝜔𝜔 2
At low frequencies, 𝜔𝜔2 is negligibly small.
3
Therefore, |𝑆𝑆𝐾𝐾𝑇𝑇 | =
𝐾𝐾+3
3
For K = 2, |𝑆𝑆𝐾𝐾𝑇𝑇 | = = 0.6
5

√22 +16 20
For 𝜔𝜔 = 1, |𝑆𝑆𝐾𝐾𝑇𝑇 | = =� = 0.79
√42 +16 32

For 0 ≤ 𝜔𝜔 ≤ 1, the sensitivity varies from 0.6 to 0.8.

Effect of Disturbance (R = 0)
𝐺𝐺 1
𝑌𝑌 = � � 𝑇𝑇𝑑𝑑 ; 𝑙𝑙𝑙𝑙𝑙𝑙 𝑇𝑇𝑑𝑑 =
1 + 𝐺𝐺𝐺𝐺 𝑠𝑠
1 1
𝑌𝑌 = � �
𝑠𝑠 2 + 4𝑠𝑠 + (3 + 𝐾𝐾) 𝑠𝑠
Steady state response, 𝑙𝑙𝑙𝑙𝑙𝑙 𝑦𝑦(𝑡𝑡) = 𝑙𝑙𝑙𝑙𝑙𝑙 𝑠𝑠𝑠𝑠(𝑠𝑠)
𝑡𝑡→∞ 𝑠𝑠→0

1 1
𝑦𝑦(𝑡𝑡) = � � =
𝑠𝑠 2 + 4𝑠𝑠 + 3 + 𝐾𝐾 𝑠𝑠=0 3 + 𝐾𝐾
For K = 2, 𝑦𝑦(𝑡𝑡) = 1/5 = 0.2

59
To minimize the effect of disturbance, choose a higher value for ‘K’.
Let K = 100; 𝑦𝑦(𝑡𝑡) = 1/103 = 0.0097
Effect of disturbance is negligibly small on steady state response at larger value of ‘K’.

60
 Assessment

1. The error is given as

a) E=R–Y
b) E = R (1− T)
c) E = R − R(Y/R)
d) All of the above

2. To minimize the disturbance, the controller should have,

a) Large gain at high frequencies
b) Small gain at high frequencies
c) Large gain at low frequencies
d) Small gain at low frequencies

3. To minimize the noise, the controller should have,

a) Small gain at high frequencies
b) Small gain at low frequencies
c) Large gain at low frequencies
d) Large gain at high frequencies

4. Sensitivity of a system T.F. (T) with respect to Plant T.F. (G) is given by
𝜕𝜕𝜕𝜕 𝑇𝑇
a) 𝑆𝑆𝐺𝐺𝑇𝑇 = � � � �
𝜕𝜕𝜕𝜕 𝐺𝐺
𝜕𝜕𝜕𝜕 𝑇𝑇
b) 𝑆𝑆𝐺𝐺𝑇𝑇 = � ��� �
𝜕𝜕𝜕𝜕 𝐺𝐺
𝜕𝜕𝜕𝜕 𝑇𝑇
c) 𝑆𝑆𝑇𝑇𝐺𝐺 = � �� �
𝜕𝜕𝜕𝜕 𝐺𝐺
d) None of the above

5. Sensitivity of T(s) with respect to a parameter (𝛼𝛼) in G(s) is given by

𝜕𝜕𝜕𝜕/𝑇𝑇
a) 𝑆𝑆𝛼𝛼𝑇𝑇 =
𝜕𝜕𝜕𝜕/𝛼𝛼
b) 𝑆𝑆𝛼𝛼𝑇𝑇 = 𝑆𝑆𝐺𝐺𝑇𝑇 𝑆𝑆𝛼𝛼𝐺𝐺
𝜕𝜕𝜕𝜕 𝐺𝐺 𝜕𝜕𝜕𝜕 𝛼𝛼
c) 𝑆𝑆𝛼𝛼𝑇𝑇 = � � � � � � � �
𝜕𝜕𝜕𝜕 𝑇𝑇 𝜕𝜕𝜕𝜕 𝐺𝐺
d) All of the above

61
6. The steady state error is the value of error function, e(t) evaluated at
a) 𝑡𝑡 = 0
b) 𝑡𝑡 = ∞
c) 𝑠𝑠 = ∞
d) All of the above

7. Gain setting is defined as

a) The effect of change in gain on error function
b) The effect of change in gain on steady state error
c) The effect of error on gain
d) The effect of gain on sensitivity

(𝑠𝑠+4)(𝑠𝑠+5)
8. If 𝐸𝐸(𝑠𝑠) = ; 𝑡𝑡ℎ𝑒𝑒 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑒𝑒𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑖𝑖𝑖𝑖,
𝑠𝑠(𝑠𝑠 2 +3𝑠𝑠+2)
a) 10
b) 0
c) 0.1
d) ∞

1 𝑠𝑠 3 +2𝑠𝑠 2
9. If 𝐸𝐸(𝑠𝑠) = � 3 � � � ; 𝑡𝑡ℎ𝑒𝑒 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑖𝑖𝑖𝑖,
𝑠𝑠 𝑠𝑠 3 +4𝑠𝑠 2 +5𝑠𝑠+1
a) 0.5
b) ∞
c) 0
d) 2

𝑠𝑠 2 +3𝑠𝑠+4
10. If 𝐸𝐸(𝑠𝑠) = 𝑎𝑎𝑎𝑎𝑎𝑎 𝑒𝑒𝑠𝑠𝑠𝑠 = 0.05, 𝑡𝑡ℎ𝑒𝑒 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 𝑜𝑜𝑜𝑜 ′𝐾𝐾′ 𝑖𝑖𝑖𝑖,
𝑠𝑠 3 +4𝑠𝑠 2 +(5+2𝐾𝐾)𝑠𝑠
a) 375
b) 3.75
c) 37.5
d) 0.375

62
 5.0 Performance of control systems
Learning Objectives
1. Able to identify performance parameters for the control of transient and steady
state responses of a control system.
2. Able to apply time domain performance specifications to a second order system
response.
3. To study the steady state error for step, ramp and parabolic input functions.

A control system in general represents a dynamic system, and hence its performance is
given in terms of transient response and steady state response. Performance parameters
serves as a measure to study how well a control system perform and how to change the
parameters for a desired performance.

5.1 Transient and steady state responses

First order system
9
Let 𝑇𝑇(𝑠𝑠) = = T.F. for first order system
𝑠𝑠+10

𝑅𝑅(𝑠𝑠) = 1/𝑠𝑠 (Step input)

9 0.9 0.9
𝑌𝑌(𝑠𝑠) = 𝑇𝑇(𝑠𝑠)𝑅𝑅(𝑠𝑠) = = − (𝐵𝐵𝐵𝐵 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑝𝑝𝑝𝑝𝑓𝑓𝑓𝑓)
𝑠𝑠(𝑠𝑠 + 10) 𝑠𝑠 𝑠𝑠 + 10
Taking inverse Laplace Transform
𝐵𝐵(𝑝𝑝) = 0.9(1 − 𝑒𝑒 −10𝑡𝑡 ) = 𝑇𝑇𝑝𝑝𝑝𝑝𝑓𝑓𝑠𝑠𝑝𝑝𝑒𝑒𝑓𝑓𝑝𝑝 𝑝𝑝𝑒𝑒𝑠𝑠𝑝𝑝𝑓𝑓𝑓𝑓𝑠𝑠𝑒𝑒
Steady state response (at “t” equal to infinity),
𝐵𝐵(∞) = 0.9(1 − 0) = 0.9
Also, by final value theorem,
9
𝑝𝑝𝑝𝑝𝑙𝑙 𝑌𝑌(𝑝𝑝) = 𝑝𝑝𝑝𝑝𝑙𝑙[𝑠𝑠𝑌𝑌(𝑠𝑠)]; 𝑌𝑌(𝑠𝑠) =
𝑡𝑡→∞ 𝑠𝑠→0 𝑠𝑠(𝑠𝑠 + 10)
9
𝑌𝑌(∞) = [𝑆𝑆𝑌𝑌(𝑠𝑠)]𝑠𝑠=0 = � � = 9/10 = 0.9
𝑠𝑠 + 10 𝑠𝑠=0
Error, 𝐸𝐸 = 𝑅𝑅(1 − 𝑇𝑇)
1 9 1 𝑠𝑠 + 1
𝐸𝐸(𝑠𝑠) = �1 − �= � �
𝑠𝑠 𝑠𝑠 + 10 𝑠𝑠 𝑠𝑠 + 10

63
 𝑠𝑠+1
Steady state error, 𝑒𝑒𝑠𝑠𝑠𝑠 = [𝑆𝑆𝐸𝐸(𝑠𝑠)]𝑠𝑠=0 = � � = 1/10 = 0.1
𝑠𝑠+10 𝑠𝑠=0

= [1 − 𝐵𝐵(∞)] = 1 − (𝑠𝑠𝑝𝑝𝑒𝑒𝑝𝑝𝑠𝑠𝐵𝐵 𝑠𝑠𝑝𝑝𝑝𝑝𝑝𝑝𝑒𝑒 𝑝𝑝𝑒𝑒𝑠𝑠𝑝𝑝𝑓𝑓𝑓𝑓𝑠𝑠𝑒𝑒)

Second order system

𝐺𝐺
𝑇𝑇(𝑠𝑠) =
1 + 𝐺𝐺

2
𝜔𝜔𝑛𝑛
Let 𝐺𝐺(𝑠𝑠) = = 𝑇𝑇. 𝐹𝐹. 𝑓𝑓𝑓𝑓 𝑝𝑝 𝑠𝑠𝑒𝑒𝑓𝑓𝑓𝑓𝑓𝑓𝑠𝑠 𝑓𝑓𝑝𝑝𝑠𝑠𝑒𝑒𝑝𝑝 𝑠𝑠𝐵𝐵𝑠𝑠𝑝𝑝𝑒𝑒𝑙𝑙
𝑠𝑠 2 +(2𝜁𝜁𝜔𝜔𝑛𝑛 )𝑠𝑠

𝜔𝜔𝑛𝑛2
𝑇𝑇(𝑠𝑠) = ; 𝜁𝜁 < 1(𝑓𝑓𝑓𝑓𝑠𝑠𝑒𝑒𝑝𝑝𝑠𝑠𝑝𝑝𝑙𝑙𝑝𝑝𝑒𝑒𝑠𝑠)
𝑠𝑠 2 + (2𝜁𝜁𝜔𝜔𝑛𝑛 )𝑠𝑠 + 𝜔𝜔𝑛𝑛2
Let 𝑅𝑅(𝑠𝑠) = 1/𝑠𝑠 (𝑠𝑠𝑝𝑝𝑒𝑒𝑝𝑝 𝑝𝑝𝑓𝑓𝑝𝑝𝑓𝑓𝑝𝑝)
𝑌𝑌(𝑠𝑠) = 𝑇𝑇(𝑠𝑠)𝑅𝑅(𝑠𝑠)
𝜔𝜔𝑛𝑛2
=
𝑠𝑠[𝑠𝑠 2 + (2𝜁𝜁𝜔𝜔𝑛𝑛 )𝑠𝑠 + 𝜔𝜔𝑛𝑛2 ]
Taking inverse L.T. (from L.T. Table)
𝑒𝑒 −𝜁𝜁𝜔𝜔𝑛𝑛 𝑡𝑡
𝐵𝐵(𝑝𝑝) = 1 − 𝑠𝑠𝑝𝑝𝑓𝑓(𝜔𝜔𝑑𝑑 𝑝𝑝 + 𝜙𝜙)
�1 − 𝜁𝜁 2

𝜙𝜙 = 𝑓𝑓𝑓𝑓𝑠𝑠 −1 𝜁𝜁 ; 𝜔𝜔𝑑𝑑 = 𝜔𝜔𝑛𝑛 �1 − 𝜁𝜁 2

Let 𝑅𝑅(𝑠𝑠) = 1 (𝑝𝑝𝑙𝑙𝑝𝑝𝑓𝑓𝑝𝑝𝑠𝑠𝑒𝑒 𝑝𝑝𝑓𝑓𝑝𝑝𝑓𝑓𝑝𝑝)
𝜔𝜔𝑛𝑛2
𝑌𝑌(𝑠𝑠) = = 𝑠𝑠[𝑌𝑌(𝑠𝑠)]𝑢𝑢𝑛𝑛𝑢𝑢𝑡𝑡 𝑠𝑠𝑡𝑡𝑠𝑠𝑠𝑠
𝑠𝑠 2 + (2𝜁𝜁𝜔𝜔𝑛𝑛 )𝑠𝑠 + 𝜔𝜔𝑛𝑛2
𝜔𝜔𝑛𝑛 𝑠𝑠
𝐵𝐵(𝑝𝑝) = 𝑒𝑒 −𝜁𝜁𝜔𝜔𝑛𝑛 𝑡𝑡 𝑠𝑠𝑝𝑝𝑓𝑓𝜔𝜔𝑑𝑑 𝑝𝑝 = [𝐵𝐵(𝑝𝑝)]𝑢𝑢𝑛𝑛𝑢𝑢𝑡𝑡 𝑠𝑠𝑡𝑡𝑠𝑠𝑠𝑠
�1 − 𝜁𝜁 2 𝑠𝑠𝑝𝑝

64
5.2 Performance Parameters
Performance parameters are defined in terms of the step response of the closed loop
system. Typical parameters used for performance measure are
i. Rise time
ii. Peak time
iii. Settling time
iv. Final value
v. Percent overshoot
The unit step response is given below.

∗ 𝑇𝑇𝑟𝑟 : Rise time (Measures Swiftness of response)

∗ 𝑇𝑇𝑠𝑠 : Peak time (The time at peak value of response)

∗ 𝑇𝑇𝑠𝑠 : Settling time (Time required to settle with in some 𝛿𝛿% of final value)
𝑀𝑀𝑠𝑠𝑡𝑡 : Peak value at 𝑝𝑝 = 𝑇𝑇𝑠𝑠 , (𝑀𝑀𝑠𝑠𝑡𝑡 = 𝐵𝐵(𝑇𝑇𝑠𝑠 )); 𝑀𝑀𝑠𝑠𝑡𝑡 = 1 + 𝑓𝑓𝑜𝑜𝑒𝑒𝑝𝑝𝑠𝑠ℎ𝑓𝑓𝑓𝑓𝑝𝑝

∗ 𝐹𝐹𝐹𝐹: Final value -> steady state value at 𝑝𝑝 = ∞ [𝐵𝐵(∞) = 𝑠𝑠𝑌𝑌(𝑠𝑠)𝑠𝑠=0 ] = 1 for step input
in the figure above.
∗ 𝑃𝑃. 𝑂𝑂.: Percent overshoot -> Difference between Peak value and final value expressed as
percentage with respect to Final Value.
𝑀𝑀𝑠𝑠𝑡𝑡 − 𝐹𝐹𝐹𝐹
𝑃𝑃. 𝑂𝑂 = � � 100
𝐹𝐹𝐹𝐹
𝛿𝛿: Percent of FV used as reference to find Ts (Typical value of 𝛿𝛿 is equal to 2% of
FV; For FV = 1, 𝛿𝛿 = 2% = 0.02)
The exponential decay in the response is 𝑒𝑒 −𝜁𝜁𝜔𝜔𝑛𝑛 𝑡𝑡 .

65
 @ 𝑡𝑡 = 𝑇𝑇𝑠𝑠 , 𝑒𝑒 −𝜁𝜁𝜔𝜔𝑛𝑛 𝑇𝑇𝑠𝑠 = 𝛿𝛿 = 0.02
1
𝑖𝑖. 𝑒𝑒. = 0.02 ⟹ 𝑒𝑒 𝜁𝜁𝜔𝜔𝑛𝑛 𝑇𝑇𝑠𝑠 = 1/0.02 = 50
𝑒𝑒 𝜁𝜁𝜔𝜔𝑛𝑛 𝑇𝑇𝑠𝑠
𝜁𝜁𝜔𝜔𝑛𝑛 𝑇𝑇𝑠𝑠 = 𝑙𝑙𝑙𝑙 50 = 3.91 ≈ 4
4 1
𝑇𝑇𝑠𝑠 = ; Let Time constant, = ; Then 𝑇𝑇𝑠𝑠 = 4𝜏𝜏
𝜁𝜁𝜔𝜔𝑛𝑛 𝜁𝜁𝜔𝜔𝑛𝑛

For arbitrary value of 𝛿𝛿, 𝑒𝑒 −𝜁𝜁𝜔𝜔𝑛𝑛 𝑇𝑇𝑠𝑠 = 𝛿𝛿

−𝜁𝜁𝜔𝜔𝑛𝑛 𝑇𝑇𝑠𝑠 = 𝑙𝑙𝑙𝑙𝑙𝑙
𝑙𝑙𝑙𝑙𝑙𝑙
𝑇𝑇𝑠𝑠 = −
𝜁𝜁𝜔𝜔𝑛𝑛
4
For 𝛿𝛿 = 2%, 𝑇𝑇𝑠𝑠 =
𝜁𝜁𝜔𝜔𝑛𝑛

3
For 𝛿𝛿 = 5%, 𝑇𝑇𝑠𝑠 =
𝜁𝜁𝜔𝜔𝑛𝑛

Transient Performance Measures

Swiftness: Rise time, Tr and Peak time, Tp
Closeness to desired response: Settling time, Ts and Percent Overshoot, P.O.
To find Peak Value, 𝑀𝑀𝑝𝑝𝑝𝑝 :

𝑀𝑀𝑝𝑝𝑝𝑝 = 𝑦𝑦(𝑇𝑇𝑝𝑝 ) = 𝑦𝑦𝑚𝑚𝑚𝑚𝑚𝑚

1
𝑦𝑦(𝑡𝑡)|𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 = 1 − 𝑒𝑒 −𝜁𝜁𝜔𝜔𝑛𝑛 𝑡𝑡 𝑠𝑠𝑠𝑠𝑠𝑠 (𝜔𝜔𝑑𝑑 𝑡𝑡 + 𝜙𝜙); 𝛽𝛽 = �1 − 𝜁𝜁 2 ; 𝜔𝜔𝑑𝑑 = 𝜔𝜔𝑛𝑛 𝛽𝛽
𝛽𝛽

𝑑𝑑
To find 𝑦𝑦𝑚𝑚𝑚𝑚𝑚𝑚 , �𝑦𝑦𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 � = 0
𝑑𝑑𝑑𝑑
𝑑𝑑
Since �𝑦𝑦𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 � = 𝑦𝑦|𝑖𝑖𝑖𝑖𝑖𝑖
𝑑𝑑𝑑𝑑
𝜔𝜔𝑛𝑛 −𝜁𝜁𝜔𝜔 𝑡𝑡
𝑦𝑦𝑚𝑚𝑚𝑚𝑚𝑚 = 𝑒𝑒 𝑛𝑛 𝑠𝑠𝑠𝑠𝑠𝑠 (𝜔𝜔 𝑡𝑡) = 0 ⟹ 𝑠𝑠𝑠𝑠𝑠𝑠 (𝜔𝜔 𝑡𝑡) = 0
𝑑𝑑 𝑑𝑑
𝛽𝛽
𝜋𝜋
𝜔𝜔𝑑𝑑 𝑡𝑡 = 𝑛𝑛𝑛𝑛; 𝐹𝐹𝐹𝐹𝐹𝐹 𝑛𝑛 = 1; 𝑡𝑡 = = 𝑇𝑇𝑝𝑝
𝜔𝜔𝑑𝑑
1
So 𝑀𝑀𝑝𝑝𝑝𝑝 = 1 − 𝑒𝑒 −𝜁𝜁𝜔𝜔𝑛𝑛 𝑇𝑇𝑝𝑝 𝑠𝑠𝑠𝑠𝑠𝑠 (𝜔𝜔𝑑𝑑 𝑇𝑇𝑝𝑝 + 𝜙𝜙)
𝛽𝛽

𝜔𝜔𝑑𝑑 𝑇𝑇𝑝𝑝 = 𝜋𝜋, 𝜔𝜔𝑛𝑛 𝑇𝑇𝑝𝑝 = 𝜋𝜋�𝛽𝛽

66
 1 −�𝜁𝜁𝜁𝜁�
𝑀𝑀𝑝𝑝𝑝𝑝 = 1 − 𝑒𝑒 𝛽𝛽 𝑠𝑠𝑠𝑠𝑠𝑠 (𝜋𝜋 + 𝜙𝜙)
𝛽𝛽
−�
𝜁𝜁𝜁𝜁
� 𝑠𝑠𝑠𝑠𝑠𝑠 𝜙𝜙
= 1 + 𝑒𝑒 𝛽𝛽
𝛽𝛽
𝜁𝜁𝜁𝜁
−� �
= 1 + 𝑒𝑒 𝛽𝛽 = 1 + 𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜ℎ𝑜𝑜𝑜𝑜𝑜𝑜

𝜁𝜁 = 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐; 𝛽𝛽 = �1 − 𝜁𝜁 2 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝜙𝜙
𝜁𝜁𝜁𝜁
−� �
𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜ℎ𝑜𝑜𝑜𝑜𝑜𝑜 = 𝑒𝑒 𝛽𝛽

𝜁𝜁𝜁𝜁
−� �
Percent overshoot, P.O. =100𝑒𝑒 𝛽𝛽

Also, from the definition of P.O.

𝑀𝑀𝑝𝑝𝑝𝑝 − 𝐹𝐹𝐹𝐹
𝑃𝑃. 𝑂𝑂. = � � 100
𝐹𝐹𝐹𝐹
𝜁𝜁𝜁𝜁
−� �
For 𝐹𝐹𝐹𝐹 = 1, 𝑃𝑃. 𝑂𝑂. = 100(𝑀𝑀𝑝𝑝𝑝𝑝 − 1) = 100𝑒𝑒 𝛽𝛽

To summarize, the performance parameters are:

4 1
1. Settling Time, 𝑇𝑇𝑠𝑠 = = 4𝜏𝜏; 𝜏𝜏 = = Time constant
𝜁𝜁𝜔𝜔𝑛𝑛 𝜁𝜁𝜔𝜔𝑛𝑛
𝜁𝜁𝜁𝜁
−� �
2. Percent overshoot, P.O. =100𝑒𝑒 ; 𝛽𝛽 = �1 − 𝜁𝜁 2 𝛽𝛽

3. Peak Time,𝑇𝑇𝑝𝑝 = 𝜋𝜋�𝜔𝜔𝑑𝑑 ; 𝜔𝜔𝑑𝑑 = 𝜔𝜔𝑛𝑛 �1 − 𝜁𝜁 2

1
4. Rise Time, 𝑇𝑇𝑟𝑟 = [2.16𝜁𝜁 + 0.6]; 0.3 ≤ 𝜁𝜁 ≤ 0.8. Rise time is defined as the
𝜔𝜔𝑛𝑛
swiftness of step response to rise from 10% to 90% of the magnitude of the step
input. A linear approximation given above is commonly used to determine the rise
time.
5. Peak value,
𝜁𝜁𝜁𝜁 𝑃𝑃. 𝑂𝑂.
−� �
𝑀𝑀𝑝𝑝𝑝𝑝 = 1 + 𝑒𝑒 𝛽𝛽 =1+
100
6. Final value, 𝐹𝐹𝐹𝐹 = [𝑆𝑆𝑆𝑆(𝑠𝑠)]𝑠𝑠=0 = 1, for step input
Percent overshoot should be lower for closeness to desired response. Peak time should be
low for swiftness to reach the peak value. With increase in damping ratio, Percent
Overshoot decreases while the peak time increases.
At 𝜁𝜁 = 0, 𝑃𝑃. 𝑂𝑂. = 100 𝑎𝑎𝑎𝑎𝑎𝑎 𝜔𝜔𝑛𝑛 𝑇𝑇𝑝𝑝 = 𝜋𝜋

At 𝜁𝜁 = 1, 𝑃𝑃. 𝑂𝑂. = 0 𝑎𝑎𝑎𝑎𝑎𝑎 𝜔𝜔𝑛𝑛 𝑇𝑇𝑝𝑝 = ∞

67
The settling time decreases with increase in damping ratio. Rise time increases with
increase in damping ratio. Peak value is directly proportional to percent overshoot.
All these performance parameters are functions of natural frequency and damping ratio.
For a second order system, the value of 𝜔𝜔𝑛𝑛 and 𝜁𝜁 can be determined from its transfer
function. Then all the performance parameters can be calculated using the relations given.
Similarly, 𝜔𝜔𝑛𝑛 and 𝜁𝜁 can be evaluated from given values of P.O. and Ts.
1 𝑃𝑃. 𝑂𝑂.
𝜁𝜁𝜁𝜁
=
−� � 100
𝑒𝑒 𝛽𝛽

𝜁𝜁𝑛𝑛 100
= 𝑝𝑝𝑓𝑓( ) = 𝛼𝛼
𝛽𝛽 𝑃𝑃. 𝑂𝑂.
𝜁𝜁𝑛𝑛
= 𝛼𝛼
�1 − 𝜁𝜁 2

𝜁𝜁 2 𝑛𝑛 2 = 𝛼𝛼 2 (1 − 𝜁𝜁 2 )
𝜁𝜁 2 (𝛼𝛼 2 + 𝑛𝑛 2 ) = 𝛼𝛼 2

𝜁𝜁 = 𝛼𝛼� ; 𝜔𝜔𝑛𝑛 = 4�(𝜁𝜁𝑇𝑇 )

�√𝛼𝛼 2 + 𝑛𝑛 2 � 𝑠𝑠

Then the other parameters, 𝑀𝑀𝑠𝑠𝑡𝑡 , 𝑇𝑇𝑠𝑠 , 𝑇𝑇𝑟𝑟 can be easily calculated.

Example:
K
Let =
S(S+P)

Select the gain K and pole, p

such that P.O. = 4.3% and Ts = 4 sec
Let us first find 𝜔𝜔𝑛𝑛 and 𝜁𝜁.
100 100
𝛼𝛼 = 𝑝𝑝𝑓𝑓 = 𝑝𝑝𝑓𝑓 = 3.146
𝑃𝑃. 𝑂𝑂. 4.3
𝜁𝜁 = 𝛼𝛼� = 3.146� = 0.707
�√𝛼𝛼 2 + 𝑛𝑛 2 � √3.1462 + 𝑛𝑛 2

𝜔𝜔𝑛𝑛 = 4�𝜁𝜁𝑇𝑇 = 4�[0.707 × 4] = 1.414

𝑠𝑠

𝐺𝐺 𝐾𝐾 𝐾𝐾
System T.F., T = = =
1+𝐺𝐺 𝑆𝑆(𝑆𝑆+𝑠𝑠)+𝐾𝐾 𝑆𝑆 2 +𝑠𝑠𝑆𝑆+𝐾𝐾

68
Comparing this with standard T.F. for a second order system,
𝜔𝜔𝑛𝑛2
𝑇𝑇 =
𝑠𝑠 2 + (2𝜁𝜁𝜔𝜔𝑛𝑛 )𝑠𝑠 + 𝜔𝜔𝑛𝑛2
𝑝𝑝 = 2𝜁𝜁𝜔𝜔𝑛𝑛 = 2(0.707)(1.414) = 2
K = 𝜔𝜔𝑛𝑛2 = (1.414)2 = 2

Steady State Error

𝑌𝑌 𝐺𝐺𝑐𝑐 𝐺𝐺
T.F. 𝑇𝑇(𝑠𝑠) = =
𝑅𝑅 1+𝐺𝐺𝑐𝑐 𝐺𝐺

Error 𝐸𝐸(𝑠𝑠) = 𝑅𝑅(1 − 𝑇𝑇)

1
= 𝑅𝑅(𝑠𝑠) � �
1 + 𝐺𝐺𝑐𝑐 𝐺𝐺
Steady state error 𝑒𝑒𝑠𝑠𝑠𝑠 = [𝑆𝑆𝐸𝐸(𝑠𝑠)]𝑠𝑠=0
Let us study "𝑒𝑒𝑠𝑠𝑠𝑠 " for different input functions.
Test input functions,

Step: r(t) = A; 𝑅𝑅(𝑠𝑠) = 𝐴𝐴�𝑠𝑠

Ramp: r(t) = (At); 𝑅𝑅(𝑠𝑠) = 𝐴𝐴�𝑠𝑠 2

Quadratic: r(t) = (At2/2); 𝑅𝑅(𝑠𝑠) = 𝐴𝐴�𝑠𝑠 3

Step input
R(s) = A/s where ‘A’ is the magnitude of step function.
𝐴𝐴 1
𝐸𝐸(𝑠𝑠) = � �
𝑠𝑠 1 + 𝐺𝐺𝑐𝑐 (𝑠𝑠)𝐺𝐺(𝑠𝑠)
𝐴𝐴
𝑒𝑒𝑠𝑠𝑠𝑠 =
1 + 𝐺𝐺𝑐𝑐 (0)𝐺𝐺(0)
𝐾𝐾 ∏𝑀𝑀
𝑖𝑖=1(𝑆𝑆+𝑍𝑍𝑖𝑖 )
In general, 𝐺𝐺𝑐𝑐 (𝑠𝑠)𝐺𝐺(𝑠𝑠) = 𝑄𝑄 ; 𝑍𝑍𝑢𝑢 → 𝑍𝑍𝑒𝑒𝑝𝑝𝑓𝑓𝑠𝑠 𝑃𝑃𝑘𝑘 → 𝑃𝑃𝑓𝑓𝑝𝑝𝑒𝑒𝑠𝑠
𝑆𝑆 𝑁𝑁 ∏𝑘𝑘=1(𝑆𝑆+𝑃𝑃𝑘𝑘 )

N → system type number

N 0 1 2
1⁄𝑆𝑆 𝑁𝑁 1 1⁄𝑆𝑆 1⁄𝑆𝑆 2

69
Type zero (N = 0)
𝐾𝐾 ∏(𝑆𝑆 + 𝑍𝑍𝑖𝑖 )
𝐺𝐺𝑐𝑐 (𝑠𝑠)𝐺𝐺(𝑠𝑠) =
∏(𝑆𝑆 + 𝑃𝑃𝑘𝑘 )
𝐾𝐾 ∏(𝑍𝑍𝑖𝑖 )
𝐺𝐺𝑐𝑐 (0)𝐺𝐺(0) = = 𝐾𝐾𝑝𝑝 = 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐
∏(𝑃𝑃𝑘𝑘 )
Therefore,
𝐴𝐴
𝑒𝑒𝑠𝑠𝑠𝑠 = ; 𝐾𝐾 = [𝐺𝐺𝑐𝑐 (𝑠𝑠)𝐺𝐺(𝑠𝑠)]𝑠𝑠=0
1 + 𝐾𝐾𝑝𝑝 𝑝𝑝
𝐴𝐴
For 𝑁𝑁 ≥ 1, 𝑒𝑒𝑠𝑠𝑠𝑠 = 𝐾𝐾 ∏(𝑆𝑆+𝑍𝑍 )
𝑖𝑖 �
1+ � 𝑁𝑁
𝑆𝑆 ∏(𝑆𝑆+𝑃𝑃 ) 𝑘𝑘 𝑠𝑠=0

𝐴𝐴𝑆𝑆 𝑁𝑁 ∏(𝑆𝑆 + 𝑃𝑃𝑘𝑘 )

=� � =0
𝑆𝑆𝑁𝑁 ∏(𝑆𝑆 + 𝑃𝑃𝑘𝑘 ) + 𝐾𝐾 ∏(𝑆𝑆 + 𝑍𝑍𝑖𝑖 ) 𝑠𝑠=0

The steady state error is zero for all systems except type – zero system.

Ramp Input
𝑟𝑟(𝑡𝑡) = 𝐴𝐴𝐴𝐴; 𝑅𝑅(𝑠𝑠) = 𝐴𝐴�𝑠𝑠 2

1 𝐴𝐴
𝐸𝐸(𝑠𝑠) = � � � 2�
1 + 𝐺𝐺𝑐𝑐 𝐺𝐺 𝑆𝑆
𝐴𝐴
𝑒𝑒𝑠𝑠𝑠𝑠 = [𝑆𝑆𝑆𝑆(𝑠𝑠)]𝑠𝑠=0 = |
𝑆𝑆[1 + 𝐺𝐺𝑐𝑐 (𝑠𝑠)𝐺𝐺(𝑠𝑠)] 𝑠𝑠=0
If 𝐺𝐺𝑐𝑐 𝐺𝐺 ≫ 1, 1 + 𝐺𝐺𝑐𝑐 𝐺𝐺 ≈ 𝐺𝐺𝑐𝑐 𝐺𝐺
𝐴𝐴
𝑒𝑒𝑠𝑠𝑠𝑠 = |
𝑆𝑆𝐺𝐺𝑐𝑐 (𝑠𝑠)𝐺𝐺(𝑠𝑠) 𝑠𝑠=0
Type-1 system (N = 1)
𝐾𝐾 ∏(𝑆𝑆 + 𝑍𝑍𝑖𝑖 )
𝐺𝐺𝑐𝑐 𝐺𝐺 =
𝑆𝑆 ∏(𝑆𝑆 + 𝑃𝑃𝑘𝑘 )
𝐴𝐴 ∏(𝑆𝑆 + 𝑃𝑃𝑘𝑘 ) 𝐴𝐴 ∏(𝑃𝑃𝑘𝑘 ) 𝐴𝐴
𝑒𝑒𝑠𝑠𝑠𝑠 = |𝑠𝑠=0 = =
𝐾𝐾 ∏(𝑆𝑆 + 𝑍𝑍𝑖𝑖 ) 𝐾𝐾 ∏(𝑍𝑍𝑖𝑖 ) 𝐾𝐾𝑣𝑣
𝐾𝐾 ∏(𝑍𝑍𝑖𝑖 )
𝐾𝐾𝑣𝑣 = �∏(𝑃𝑃 ) = [𝑆𝑆𝐺𝐺𝑐𝑐 (𝑠𝑠)𝐺𝐺(𝑠𝑠)]𝑠𝑠=0 = velocity error constant
𝑘𝑘

70
For N > 1, 𝑒𝑒𝑠𝑠𝑠𝑠 = 0 and for N < 1, 𝑒𝑒𝑠𝑠𝑠𝑠 = ∞
For ramp input, the steady state error is finite only for Type-1 system.

Quadratic Input
𝐴𝐴𝑝𝑝 2 𝐴𝐴 2 𝐴𝐴
𝑝𝑝(𝑝𝑝) = ; 𝑅𝑅(𝑠𝑠) = � 3 � = 3
2 2 𝑆𝑆 𝑆𝑆
1 𝐴𝐴 𝐴𝐴
𝐸𝐸(𝑠𝑠) = � � � 3� ≈ 3
1 + 𝐺𝐺𝑐𝑐 𝐺𝐺 𝑆𝑆 𝑆𝑆 [𝐺𝐺𝑐𝑐 (𝑠𝑠)𝐺𝐺(𝑠𝑠)]
𝐴𝐴
𝑒𝑒𝑠𝑠𝑠𝑠 = [𝑆𝑆𝐸𝐸(𝑠𝑠)]𝑠𝑠=0 = |
𝑆𝑆 2 [𝐺𝐺𝑐𝑐 (𝑠𝑠)𝐺𝐺(𝑠𝑠)] 𝑠𝑠=0
Type-2 System (N=2)
𝐾𝐾 ∏(𝑆𝑆 + 𝑍𝑍𝑢𝑢 )
𝐺𝐺𝑐𝑐 𝐺𝐺 =
𝑆𝑆 2 ∏(𝑆𝑆 + 𝑃𝑃𝑘𝑘 )
𝐴𝐴 ∏(𝑆𝑆 + 𝑃𝑃𝑘𝑘 ) 𝐴𝐴 ∏(𝑃𝑃𝑘𝑘 ) 𝐴𝐴
𝑒𝑒𝑠𝑠𝑠𝑠 = |𝑠𝑠=0 = =
𝐾𝐾 ∏(𝑆𝑆 + 𝑍𝑍𝑢𝑢 ) 𝐾𝐾 ∏(𝑍𝑍𝑢𝑢 ) 𝐾𝐾𝑚𝑚
𝐾𝐾 ∏(𝑍𝑍𝑢𝑢 )
𝐾𝐾𝑚𝑚 = �∏(𝑃𝑃 ) = [𝑆𝑆 2 𝐺𝐺𝑐𝑐 (𝑠𝑠)𝐺𝐺(𝑠𝑠)]𝑠𝑠=0 = Acceleration error constant.
𝑘𝑘

For N > 2, 𝑒𝑒𝑠𝑠𝑠𝑠 = 0 and for N < 2, 𝑒𝑒𝑠𝑠𝑠𝑠 = ∞

For quadratic input, the steady state error is finite only for Type-2 system.

The steady state error for different input are summarized as below
System Type Step input Ramp input Quadratic input
0 𝐴𝐴⁄(1 + 𝐾𝐾𝑠𝑠 ) ∞ ∞
1 0 𝐴𝐴⁄𝐾𝐾𝑣𝑣 ∞
2 0 0 𝐴𝐴⁄𝐾𝐾𝑚𝑚

Example:
1
𝐺𝐺(𝑠𝑠) =
𝑠𝑠+2

2
𝐻𝐻(𝑠𝑠) =
𝑠𝑠 + 4
1
𝑅𝑅(𝑠𝑠) =
𝑠𝑠

71
Find K, such that steady state error is zero.
𝐾𝐾𝐺𝐺
T.F; 𝑇𝑇(𝑠𝑠) =
1+𝐾𝐾𝐺𝐺𝐾𝐾
1 1 + 𝐾𝐾𝐺𝐺𝐻𝐻 − 𝐾𝐾𝐺𝐺
𝐸𝐸(𝑠𝑠) = 𝑅𝑅(1 − 𝑇𝑇) = � �
𝑆𝑆 1 + 𝐾𝐾𝐺𝐺𝐻𝐻
1 + 𝐾𝐾𝐺𝐺(0)𝐻𝐻(0) − 𝐾𝐾𝐺𝐺(0)
𝑒𝑒𝑠𝑠𝑠𝑠 = [𝑆𝑆𝐸𝐸(𝑠𝑠)]𝑠𝑠=0 = � �
1 + 𝐾𝐾𝐺𝐺(0)𝐻𝐻(0)
G(0) = 1/2, H(0) = 2/4=1/2

1 + 𝐾𝐾�1�4 − 1�2�
𝑒𝑒𝑠𝑠𝑠𝑠 = =0
1 + �𝐾𝐾�4�
𝐾𝐾
1− =0 𝐾𝐾 = 4
4

Hubble Space Telescope Control

The control system for telescope pointing system is given below.
a. Choose K and K1 so
that percent overshoot is
less than 10% for step
input.
b. The steady state error
due to ramp input is
minimum.
c. The effect of step
disturbance is reduced.

72
Reduced Block Diagram

�1�𝑆𝑆 2 � 1
𝐺𝐺(𝑠𝑠) = =
1 + �1�𝑆𝑆 2 � (𝐾𝐾1 𝑆𝑆) 𝑆𝑆 2 + 𝐾𝐾1 𝑆𝑆
(a) P.O. = 10%
100
𝛼𝛼 = ln � � = 𝑝𝑝𝑓𝑓10 = 2.3
𝑃𝑃. 𝑂𝑂.
𝜁𝜁 = 𝛼𝛼� = 0.59
�√𝛼𝛼 2 + 𝑛𝑛 2 �
1 𝐾𝐾𝐺𝐺 𝑌𝑌 𝐾𝐾
𝑇𝑇𝑑𝑑 = 0, 𝑅𝑅(𝑠𝑠) = , 𝑇𝑇 = = = 2
𝑠𝑠 1 + 𝐾𝐾𝐺𝐺 𝑅𝑅 (𝑆𝑆 + 𝐾𝐾1 𝑆𝑆 + 𝐾𝐾)
1 𝐾𝐾
𝑌𝑌(𝑠𝑠) = � � � 2 �
𝑆𝑆 (𝑆𝑆 + 𝐾𝐾1 𝑆𝑆 + 𝐾𝐾)
𝐾𝐾
𝐵𝐵(∞) = 𝑆𝑆𝑌𝑌(𝑠𝑠)|𝑠𝑠=0 = =1
𝐾𝐾
2
𝜔𝜔𝑛𝑛
Comparing T(s) with standard form of T.F., 2
𝑠𝑠 2 +(2𝜁𝜁𝜔𝜔𝑛𝑛 )𝑠𝑠+𝜔𝜔𝑛𝑛

𝐾𝐾 = 𝜔𝜔𝑛𝑛2
𝐾𝐾1 = 2𝜁𝜁𝜔𝜔𝑛𝑛 = 1.2𝜔𝜔𝑛𝑛

(b) Ramp input

1
𝑅𝑅(𝑠𝑠) =
𝑆𝑆 2
1 1 1 𝑆𝑆 2 + 𝐾𝐾1 𝑆𝑆 1 𝑆𝑆 + 𝐾𝐾1
𝐸𝐸(𝑠𝑠) = 𝑅𝑅(1 − 𝑇𝑇) = 2 � � = 2� 2 �= � 2 �
𝑆𝑆 1 + 𝐾𝐾𝐺𝐺 𝑆𝑆 𝑆𝑆 + 𝐾𝐾1 𝑆𝑆 + 𝐾𝐾 𝑆𝑆 𝑆𝑆 + 𝐾𝐾1 𝑆𝑆 + 𝐾𝐾
𝐾𝐾1 1
𝑒𝑒𝑠𝑠𝑠𝑠 = [𝑆𝑆𝐸𝐸(𝑠𝑠)]𝑠𝑠=0 = =
𝐾𝐾 𝐾𝐾 ⁄𝐾𝐾1
For minimum steady state error; (𝐾𝐾 ⁄𝐾𝐾1 ) should be large, or ‘K’ should be large.

73
From (a),

𝐾𝐾 𝜔𝜔𝑛𝑛 √𝐾𝐾
= =
𝐾𝐾1 1.2 1.2
If 𝐾𝐾 = 100, 𝐾𝐾1 = 1.2√𝐾𝐾 = 12; 𝐾𝐾 ⁄𝐾𝐾1 = 100/12 = 8.33
𝐾𝐾1
𝑒𝑒𝑠𝑠𝑠𝑠 = = 0.12
𝐾𝐾

(c) Step Disturbance

𝑇𝑇𝑑𝑑 = 1/𝑆𝑆, 𝑅𝑅 = 0

𝑌𝑌 𝐺𝐺
=
𝑇𝑇𝑑𝑑 1 + 𝐾𝐾𝐺𝐺
1
𝑌𝑌(𝑠𝑠) = 𝑇𝑇𝑑𝑑 � �
𝑆𝑆 2 + 𝐾𝐾1 𝑆𝑆 + 𝐾𝐾
1 1
= � 2 �
𝑆𝑆 𝑆𝑆 + 𝐾𝐾1 𝑆𝑆 + 𝐾𝐾
1
𝐵𝐵(∞) = 𝑆𝑆𝑌𝑌(𝑠𝑠)|𝑠𝑠=0 =
𝐾𝐾

The effect of 𝑇𝑇𝑑𝑑 is less for large value of K. For K = 100, 𝐵𝐵(∞) = 0.01

74
 Assessment
9𝑆𝑆
1. 𝑇𝑇 = represents
𝑆𝑆 2 +10𝑆𝑆
a) First order system
b) Second order system
c) Third order system
d) All of the above

2. Find value theorem is useful to find,

a) Steady state error
b) Steady state response
c) The final value of output
d) All of the above

8𝑠𝑠 2 +5𝑠𝑠
3. If 𝐸𝐸(𝑠𝑠) = , the steady state error is
9𝑠𝑠 3 +4𝑠𝑠 2
a) 8/9
b) 9/8
c) 5/4
d) 4/5

4. If Ts = K�𝜁𝜁𝜔𝜔 the value of K for settling time within 2% of final value is

n

a) 4.6
b) 3.9
c) 3.0
d) 2.3

5. For the above problem, the value of K for settling time within 5% of final value
is,
a) 4.6
b) 3.9
c) 3.0
d) 2.3

75
6. For step input, the steady state error is zero for
a) Type-1 system
b) Type-2 system
c) Type-3 system
d) All of the above

7. For ramp input, the steady state error is NOT zero for
a) Type-1 system
b) Type-2 system
c) Type-3 system
d) All of the above

𝐴𝐴
8. For quadratic input, the steady state error, 𝑒𝑒𝑠𝑠𝑠𝑠 = for
𝐾𝐾𝑎𝑎
a) Type-1 system
b) Type-2 system
c) Type-3 system
d) All of the above

9. The error function is defined for

a) 𝐸𝐸(𝑠𝑠) = 𝑅𝑅(𝑠𝑠) − 𝑌𝑌(𝑠𝑠)
b) 𝐸𝐸(𝑠𝑠) = 𝑅𝑅(𝑠𝑠)[1 − 𝑇𝑇(𝑠𝑠)]
𝑌𝑌(𝑠𝑠)−𝑅𝑅(𝑠𝑠)
c) 𝐸𝐸(𝑠𝑠) = −𝑅𝑅(𝑠𝑠) � �
𝑅𝑅(𝑠𝑠)
d) All of the above

10. If percent overshoot and settling time are given, we can find
a) Damping ratio
b) Natural frequency
c) Peak time and Peak value
d) All of the above

76
 6. Stability of control system
The stability of control system is a function of poles, the roots of the characteristic
equation given by the denominator polynomial of the system transfer function. Routh –
Hurwitz method is used to determine the stability of the system without computing the
roots of the characteristic equation.

Learning Objectives:
1) To understand the stability of dynamic systems.
2) To understand the role of poles and their location in finding the system stability.
3) To apply Routh – Hurwitz method to determine the system stability.

Stability of Dynamic System

Let the transfer function of a dynamic system is,
1
𝑇𝑇(𝑠𝑠) =
𝑠𝑠(𝑠𝑠 + 𝑎𝑎)(𝑠𝑠 − 𝑎𝑎)
Characteristic equation is, 𝑠𝑠(𝑠𝑠 + 𝑎𝑎)(𝑠𝑠 − 𝑎𝑎) = 0
The roots are, s = 0, s = −𝑎𝑎, s = +a
These roots are also called as the poles of the system and they can be represented on a S-
plane stability plot.

In general, the complex poles are −𝑎𝑎 + 𝑗𝑗𝑦𝑦1 or 𝑎𝑎 + 𝑗𝑗𝑦𝑦2

If 𝑦𝑦1 = 0 = 𝑦𝑦2 the roots are real. The real poles, 0, −𝑎𝑎, 𝑎𝑎 are shown in the figure below.

Here 𝑠𝑠 = −𝑎𝑎 is stable and s = a is unstable.

The point s = 0 is neutrally stable.

77
The response for each case is given below.

If 𝑦𝑦1 ≠ 0 ≠ 𝑦𝑦2 , the roots are complex; the respective responses are,

Characteristic equation
Let the system transfer function, 𝑇𝑇(𝑠𝑠) = 𝑝𝑝(𝑠𝑠)/𝑞𝑞(𝑠𝑠). The polynomial in the numerator is
‘p(s)’ and its roots are called ZEROS of the system. The polynomial in the denominator
is ‘q(s)’ and its roots are called POLES of the system. The negative poles (real or
complex) contribute to stability of the system, while the positive poles (real or complex)
contribute to instability in the system. The characteristic equation is given as q(s) = 0.
Let 𝑞𝑞(𝑠𝑠) = 𝑎𝑎𝑛𝑛 𝑠𝑠 𝑛𝑛 + 𝑎𝑎𝑛𝑛−1 𝑠𝑠 𝑛𝑛−1 + 𝑎𝑎𝑛𝑛−2 𝑠𝑠 𝑛𝑛−2 + 𝑎𝑎𝑛𝑛−3 𝑠𝑠 𝑛𝑛−3 + ⋯ + 𝑎𝑎1 𝑠𝑠 + 𝑎𝑎0 = 0
The highest power of ‘s’ gives the order of control system. The general equation given
above represents a nth order control system. The characteristic equation is the basis to
apply the Routh – Hurwitz criterion.

78
Routh – Hurwitz criterion
This criterion states that the number of roots of q(s) with positive real parts is equal to the
number of changes in sign in the first column of Routh array. So, a stable system requires
all the coefficients in the first column of Routh array to be positive. An unstable system
will have at least one sign change in the first column indicating at least there is one
positive real root. A typical Routh array is given below.
𝑠𝑠 𝑛𝑛 𝑎𝑎𝑛𝑛 𝑎𝑎𝑛𝑛−2 𝑎𝑎𝑛𝑛−4 ⋯ ⋯ 𝑎𝑎2 𝑎𝑎0
𝑛𝑛−1
𝑠𝑠 𝑎𝑎𝑛𝑛−1 𝑎𝑎𝑛𝑛−3 𝑎𝑎𝑛𝑛−5 ⋯ ⋯ 𝑎𝑎3 𝑎𝑎1
𝑠𝑠 𝑛𝑛−2
𝑠𝑠 𝑛𝑛−3
⋮
𝑠𝑠 2
𝑠𝑠1
𝑠𝑠 0
The first row in the array begins with the coefficient of sn and continues with successive
alternate term coefficients. The second row in the array begins with the coefficient of sn-1
and continues with successive alternate term coefficients. All other rows in the array need
to be computed.

Example:
𝑞𝑞(𝑠𝑠) = 𝑎𝑎3 𝑠𝑠 3 + 𝑎𝑎2 𝑠𝑠 2 + 𝑎𝑎1 𝑠𝑠 + 𝑎𝑎0

Routh array
𝑠𝑠 3 𝑎𝑎3 𝑎𝑎1 0
𝑠𝑠 2 𝑎𝑎2 𝑎𝑎0 0
𝑠𝑠1 𝑏𝑏1 𝑏𝑏2
𝑠𝑠 0 𝑐𝑐1 𝑐𝑐2

To find b1
Consider two elements in the same column, just above b1 and other two elements in the
immediate next column. They are

𝑎𝑎0 1 1 𝑎𝑎3 𝑎𝑎1

𝑏𝑏1 = − � � 𝑎𝑎3 + 𝑎𝑎1 = − [𝑎𝑎3 𝑎𝑎0 − 𝑎𝑎2 𝑎𝑎1 ] = − �𝑎𝑎 𝑎𝑎0 �
𝑎𝑎2 𝑎𝑎2 𝑎𝑎2 2

79
To find b2
Skip the elements in the same column, just above b2. Consider the elements in a column
to the left and a column to the right above b2.
0 1 1 𝑎𝑎 0
𝑏𝑏2 = − � � 𝑎𝑎3 + 0 = − [𝑎𝑎3 (0) − 𝑎𝑎2 (0)] = − � 3 �=0
𝑎𝑎2 𝑎𝑎2 𝑎𝑎2 𝑎𝑎2 0

To find c1
Similar to b1, the procedure is to consider four elements in the columns just above c1.
𝑏𝑏
𝑐𝑐1 = − � 2 � 𝑎𝑎2 + 𝑎𝑎0 = 𝑎𝑎0 ; since b2 = 0
𝑏𝑏1

To find c2
Similar to b2, the elements to be considered are
0
𝑐𝑐2 = − � � 𝑎𝑎2 + 0 = 0
𝑏𝑏1

Completed Routh Array

𝑠𝑠 3 𝑎𝑎3 𝑎𝑎1 0
𝑠𝑠 2 𝑎𝑎2 𝑎𝑎0 0
𝑠𝑠1 𝑏𝑏1 0
𝑠𝑠 0 𝑎𝑎0 0

1
𝑏𝑏1 = (𝑎𝑎 𝑎𝑎 − 𝑎𝑎3 𝑎𝑎0 )
𝑎𝑎2 1 2

80
For the system to be stable all the elements in the first column of Routh Array should be
positive. In this case, 𝑎𝑎3 , 𝑎𝑎2 , 𝑏𝑏1 and 𝑎𝑎0 should be positive. If 𝑎𝑎3 and 𝑎𝑎2 are positive, 𝑎𝑎0
should be positive. For 𝑏𝑏1 to be positive, 𝑎𝑎1 𝑎𝑎2 > 𝑎𝑎3 𝑎𝑎0 . It requires 𝑎𝑎1 also to be positive
since all other coefficients are positive. So, all the coefficients are required to be positive.
In stability analysis, it is necessary that all the coefficients of a characteristic equation
should be positive. If there is any sign change, it is certain that the system is likely to be
unstable. The characteristic equation can be inspected for any sign change among
coefficients as initial check. If all the coefficients are positive, then Routh – Hurwitz
method can be applied for further investigation of system stability.
The procedure for Routh – Hurwitz method can be classified into four categories. They
are (1) No zero element in the first column, (2) There is a zero element in the first
column, (3) All elements in a row are zero, (4) Repeated roots on the vertical axis of s-
Plane plot.

(1) No zero element in the first column

Characteristic equation is,
𝑞𝑞(𝑠𝑠) = 𝑠𝑠 3 + 𝑠𝑠 2 + 2𝑠𝑠 + 24
The polynomial is complete. All coefficients exist, and they are positive. Let us apply the
Routh – Hurwitz criteria.
𝑠𝑠 3 1 2 0
𝑠𝑠 2 1 24 0
𝑠𝑠1 −22 0
𝑠𝑠 0 24

In the first column we see two sign changes, one from 𝑠𝑠 2 to 𝑠𝑠1 and other from 𝑠𝑠1 to 𝑠𝑠 0 .
There will be two positive roots and they will be on the right half of s-plane. So, the
system will be unstable.
Roots of the characteristic equation:
𝑞𝑞(𝑠𝑠) = 𝑠𝑠 3 + 𝑠𝑠 2 + 2𝑠𝑠 + 24 = (𝑠𝑠 + 3)(𝑠𝑠 2 − 2𝑠𝑠 + 8)

The roots are, 𝑠𝑠1 = −3, 𝑠𝑠2,3 = 1 ± 𝑗𝑗√7 = 𝜎𝜎 ± 𝑗𝑗𝑗𝑗

81
 Although it satisfies the necessary condition
in the initial check (polynomial is complete
and all coefficients are positive) the Routh
array has indicated that the system is
unstable.

(2) There is a zero element in the first column.

If there is zero in the first column of the Routh array while other elements in the
respective row are nonzero, replace the zero by a small positive number 𝜖𝜖 for computing
the elements in the first column of the array.
Let the characteristic equation is
𝑞𝑞(𝑠𝑠) = 𝑠𝑠 5 + 2𝑠𝑠 4 + 2𝑠𝑠 3 + 4𝑠𝑠 2 + 11𝑠𝑠 + 10
It satisfies the necessary condition in the initial check that the polynomial is complete and
all the coefficients are positive.

Routh Array
𝑠𝑠 5 1 2 11 0
𝑠𝑠 4 2 4 10 0
𝑠𝑠 3 𝑏𝑏1 𝑏𝑏2 0
𝑠𝑠 2 𝑐𝑐1 𝑐𝑐2 0
𝑠𝑠1 𝑑𝑑1 𝑑𝑑2 0
𝑠𝑠 0 𝑒𝑒1 𝑒𝑒2

𝑏𝑏1 = −(4/2)(1) + 2 = 0
Let 𝑏𝑏1 = 𝜖𝜖
𝑏𝑏2 = −(10/2)(1) + 11 = 6
𝑐𝑐1 = −(𝑏𝑏2 /𝑏𝑏1 )(2) + 4 == −(6/𝜖𝜖)(2) + 4 = 4 − 12/𝜖𝜖
𝑐𝑐1 = 1/𝜖𝜖(4𝜖𝜖 − 12) ≈ −12/𝜖𝜖
𝑐𝑐2 = −(0/𝑏𝑏1 )(2) + 10 = 10

82
 𝑑𝑑1 = −(𝑐𝑐2 /𝑐𝑐1 )𝑏𝑏1 + 𝑏𝑏2 = 10𝜖𝜖 2 /12 + 6 ≈ 6
𝑑𝑑2 = 0
𝑒𝑒1 = −(𝑑𝑑2 /𝑑𝑑1 )𝑐𝑐1 + 𝑐𝑐2 = 𝑐𝑐2 = 10
𝑒𝑒2 = 0

Completed Routh Array

𝑠𝑠 5 1 2 11
𝑠𝑠 4 2 4 10
𝑠𝑠 3
∈ 6
− 12�∈ 10
2
𝑠𝑠
𝑠𝑠1 6 0
𝑠𝑠 0 10 0
We see that there are two sign changes in the first column. So, the system is unstable with
two roots in the right half plane.

(3) ALL elements in a row are zero

In this case, the elements in a row that includes zero in the first column are also zero.
When the roots of characteristic equation are symmetric pairs such as (𝑠𝑠 + 𝜎𝜎) and
(𝑠𝑠 − 𝜎𝜎) or (𝑠𝑠 + 𝑗𝑗𝑗𝑗) and (𝑠𝑠 − 𝑗𝑗𝑗𝑗), the Routh array has a row with all zero elements. In
such situation, Auxiliary Polynomial is used to determine the stability. Auxiliary
Polynomial, U(s) is a polynomial that results from a row which immediately precedes the
row of zeros in the Routh array. The order of the auxiliary polynomial is always even
indicating the number of symmetrical pair of roots.
For example, let the characteristic equation is,
𝑞𝑞(𝑠𝑠) = 𝑠𝑠 3 + 2𝑠𝑠 2 + 4𝑠𝑠 + 8

Routh Array:
𝑠𝑠 3 1 4 0
𝑠𝑠 2 2 8 0
𝑠𝑠1 ∈ 0
𝑠𝑠 0 8 0

83
The row s1 has all zero elements and the row s2 just above the row of zeros is used to
form the auxiliary polynomial. Therefore, the auxiliary polynomial is given as,
U(s) = 2𝑠𝑠 2 + 8 = 0
= 𝑠𝑠 2 + 4 = 0
𝑠𝑠 2 = −4; 𝑠𝑠 = ±𝑗𝑗2
The roots are 𝑠𝑠1 = (𝑠𝑠 + 𝑗𝑗2), 𝑠𝑠2 = (𝑠𝑠 − 𝑗𝑗2)
To find the other root of characteristic equation, divide the polynomial by the auxiliary
polynomial.
𝑞𝑞(𝑠𝑠) 𝑠𝑠 3 + 2𝑠𝑠 2 + 4𝑠𝑠 + 8
=
𝑈𝑈(𝑠𝑠) 𝑠𝑠 2 + 4
𝑠𝑠 + 2
𝑠𝑠 2 + 4 �𝑠𝑠 3+ 2𝑠𝑠 2 + 4𝑠𝑠 + 8
𝑠𝑠 3 + 4𝑠𝑠
(−) 2𝑠𝑠 2 +8
2
2𝑠𝑠 +8
(−) 0
Therefore, 𝑞𝑞(𝑠𝑠) = (𝑠𝑠 + 2)(𝑠𝑠 + 𝑗𝑗2)(𝑠𝑠 − 𝑗𝑗2)
The roots (or Poles) of q(s) are, −2, 𝑗𝑗2, −𝑗𝑗2

S – Plane Plot

Since the poles ±𝑗𝑗2 are on the vertical axis that divides the stable and unstable region;
the system is neither stable nor unstable. It is classified as marginally stable. For all
practical purpose, it is considered as instability.

84
(4) Repeated roots on the vertical axis of s-plane plot.
In the previous case, we had one row of zeros on the Routh array resulting into one
auxiliary polynomial with one pair of symmetric roots on the vertical axis of s-plane plot.
In this case, we will see multiple rows of zeros on the Routh Array resulting into more
than one auxiliary polynomial with several repeated roots on the vertical axis of s-plane
plot.
Let the characteristic equation is,
𝑞𝑞(𝑠𝑠) = 𝑠𝑠 5 + 𝑠𝑠 4 + 2𝑠𝑠 3 + 2𝑠𝑠 2 + 𝑠𝑠 + 1

Routh Array

𝑠𝑠 5 1 2 1 0
𝑠𝑠 4 1 2 1 0
𝑠𝑠 3 ∈ ∈ 0
𝑠𝑠 2 1 1 0
𝑠𝑠1 ∈ 0
𝑠𝑠 0 1

Here the rows of s3 and s1 contains all zero elements. The value ∈ is used at appropriate
places in order to complete the following rows on the Routh array.
Auxiliary Polynomial for the row s1 is,
𝑠𝑠 2 + 1 = 0; 𝑠𝑠 2 = −1; 𝑠𝑠 = ±𝑗𝑗; 𝑠𝑠1 = +𝑗𝑗; 𝑠𝑠2 = −𝑗𝑗
Auxiliary Polynomial for the row s3 is,
𝑠𝑠 4 + 2𝑠𝑠 2 + 1 = 0
(𝑠𝑠 2 + 1)(𝑠𝑠 2 + 1) = 0
(𝑠𝑠 2 + 1) = 0; 𝑠𝑠3 = +𝑗𝑗; 𝑠𝑠4 = −𝑗𝑗
(𝑠𝑠 2 + 1) = 0; 𝑠𝑠5 = +𝑗𝑗; 𝑠𝑠6 = −𝑗𝑗
There are three pairs of repeated symmetric roots on the vertical axis leading to marginal
instability. The system is unstable.

85
Example
Given, 𝑞𝑞(𝑠𝑠) = 𝑠𝑠 5 + 𝑠𝑠 4 + 15𝑠𝑠 3 + 10𝑠𝑠 2 + 44𝑠𝑠 + 24
Find all the roots and check for stability.

Routh Array
𝑠𝑠 5 1 15 44
𝑠𝑠 4 1 10 24
𝑠𝑠 3 5 20 0
𝑠𝑠 2
6 24 0
𝑠𝑠 1 ∈ 0
𝑠𝑠 0 24
Auxiliary polynomial is, 6𝑠𝑠 2 + 24 = 0
𝑈𝑈(𝑠𝑠) = 𝑠𝑠 2 + 4 = 0; 𝑠𝑠 = ±𝑗𝑗2
To find other roots,
𝑞𝑞(𝑠𝑠) 𝑠𝑠 5 + 𝑠𝑠 4 + 15𝑠𝑠 3 + 10𝑠𝑠 2 + 44𝑠𝑠 + 24
=
𝑈𝑈(𝑠𝑠) 𝑠𝑠 2 + 4

𝑠𝑠 3 + 𝑠𝑠 2 + 11𝑠𝑠 + 6
2
𝑠𝑠 + 4 �𝑠𝑠 5 + 𝑠𝑠 4 + 15𝑠𝑠 3 + 10𝑠𝑠 2 + 44𝑠𝑠 + 24
𝑠𝑠 5 + 4𝑠𝑠 3
(−) 𝑠𝑠 4 + 11𝑠𝑠 3 + 10𝑠𝑠 2
𝑠𝑠 4 + + 4𝑠𝑠 2
(−) 11 𝑠𝑠 3 + 6𝑠𝑠 2 + 44𝑠𝑠
11 𝑠𝑠 3 + 44𝑠𝑠
(−) 2
6𝑠𝑠 + 24
6𝑠𝑠 2 + 24
(−) 0

The reduced polynomial is, 𝑠𝑠 3 + 𝑠𝑠 2 + 11𝑠𝑠 + 6 = 0

Applying Routh criterion for this polynomial,

Routh Array:
𝑠𝑠 3 1 11
𝑠𝑠 2 1 6
𝑠𝑠1 5 0
𝑠𝑠 0 6

86
There are no sign changes in the first column and hence there are no positive real roots.
The roots of this cubic polynomial are −0.56, −0.22 ± 𝑗𝑗6.54. So, the system has one
negative real pole, two negative complex poles, and a pair of poles on vertical axis. It can
be interpreted that the system is marginally stable.

Example

For a given control system, find the range of ‘k’ and ‘a’ for which the system is stable.

𝑘𝑘(𝑠𝑠 + 𝑎𝑎)
𝐺𝐺1 (𝑠𝑠) =
𝑠𝑠 + 1
1
𝐺𝐺2 (𝑠𝑠) =
𝑠𝑠(𝑠𝑠 + 2)(𝑠𝑠 + 3)
𝐺𝐺1 𝐺𝐺2 𝑘𝑘(𝑠𝑠 + 𝑎𝑎)
𝑇𝑇(𝑠𝑠) = =
1 + 𝐺𝐺1 𝐺𝐺2 [𝑠𝑠(𝑠𝑠 + 1)(𝑠𝑠 + 2)(𝑠𝑠 + 3) + 𝑘𝑘(𝑠𝑠 + 𝑎𝑎)]
The characteristic equation is given by the denominator.
𝑞𝑞(𝑠𝑠) = 𝑠𝑠(𝑠𝑠 + 1)(𝑠𝑠 + 2)(𝑠𝑠 + 3) + 𝑘𝑘(𝑠𝑠 + 𝑎𝑎) = 𝑠𝑠 4 + 6𝑠𝑠 3 + 11𝑠𝑠 2 + (𝑘𝑘 + 6)𝑠𝑠 + 𝑘𝑘𝑎𝑎

Routh Array
𝑠𝑠 4 1 11 ka
𝑠𝑠 3 6 (k+6) 0
𝑠𝑠 2 𝑏𝑏1 𝑏𝑏2 0
𝑠𝑠1 𝑐𝑐1 0 0
𝑠𝑠 0 𝑑𝑑1 0
(𝑘𝑘 + 6) 60 − 𝑘𝑘
𝑏𝑏1 = − + 11 =
6 6
𝑏𝑏2 = 𝑘𝑘𝑎𝑎
𝑏𝑏2 1
𝑐𝑐1 = − (6) + (𝑘𝑘 + 6) = [−𝑘𝑘 2 + (54 − 36𝑎𝑎)𝑘𝑘 + 360]
𝑏𝑏1 (60 − 𝑘𝑘)
𝑑𝑑1 = 0 + 𝑏𝑏2 = 𝑘𝑘𝑎𝑎

87
For the system to be stable, the elements b1, c1, d1 in the first column should be positive.
For b1 to be positive, k < 60
For d1 to be positive, k > 0 and a > 0
The range of ‘a’ depends on the value of ‘k’ chosen.
Let k = 40; then
1
𝑐𝑐1 = [920 − 1440𝑎𝑎]
20
For c1 > 0, a < 0.639
For k = 1, a < 11.47
k = 59, a < 0.03
The range of ‘k’ = 1 ≤ 𝑘𝑘 ≤ 59
𝑘𝑘 = 1; 0 ≤ 𝑎𝑎 ≤ 11.47
The range of ‘a’ = �
𝑘𝑘 = 59; 0 ≤ 𝑎𝑎 ≤ 0.03

88
 Assessment
1) The characteristic equation is,
A. The number of controller transfer function
B. The denominator of controller transfer function
C. The numerator of total(system) transfer function
D. The denominator of total(system) transfer function

2) The roots of characteristic equation are called as

A. Zeros
B. Poles
C. Complex zeros
D. None of the above

3) For stability, the roots of characteristic equation should be on,

A. Left half plane of S-Plot
B. Right half plane of S-Plot
C. On the Real Axis of S-Plot
D. On the vertical Axis of S-Plot

4) For stability, the coefficients of s in the characteristic equation should be

A. All positive
B. All negative
C. At least one positive and one negative
D. Equal number of positive and negative

5) For stability, the Routh array should be

A. All values in the first column be positive
B. All values in the second column be positive
C. At least one negative value in the first column
D. None of the above

6) Regarding first two rows on Routh array,

A. One row is with odd powers and the other with even powers
B. The coefficients of “s” are used to complete these two rows
C. First row begins with highest power of characteristic equation
D. All of the above

89
7) On the given Routh array, the value b1 is,
𝑠𝑠 3 1 8 0
𝑠𝑠 2 2 4 0
𝑠𝑠1 𝑏𝑏1 𝑏𝑏2
𝑠𝑠 0 𝑐𝑐1

A. -12
B. 6
C. 0
D. 3

8) The value of c1 on the Routh array is,

A. 4
B. 0
C. 8
D. -16

9) Auxiliary polynomial is used for stability if

A. One of the elements on the first column of Routh array is zero
B. One of the columns on the Routh array is zero
C. One of the rows on the Routh array has all zero
D. One of the elements on the first row of Routh array is zero

10) Auxiliary polynomial,

A. Is the row preceding the row of zeros.
B. Is an even order polynomial.
C. Has symmetric roots
D. All of the above

11) If the roots are on vertical axis, the system is

A. Stable
B. Unstable
C. Marginally stable
D. None of the above

90
12) A marginally stable system in general considered as,
A. Stable system
B. Unstable system
C. Neutral system
D. None of the above

91
 7. Root – Locus Method

Learning Objectives:
1) To understand the roots (Poles and Zeros) of a transfer function and how they
travel on the S-Plane plot with respect to changes in controller gain, K.
2) Learn to sketch the root locus diagram that shows the path or loci from a pole to a
zero at a finite distance or at infinity.
3) To apply the three term PID parameters in controller design to achieve the desired
performance.

Root locus method has been developed by Walter R. Evans who received his engineering
education in Electrical Engineering at Washington University in St. louis in 1941. He
invented the root locus method in 1948 before he completed his master’s degree in
electrical engineering at University of California, Los Angeles in 1951.
Loci of the Roots of Transfer function are the paths they travel in a journey on the S-
Plane. Let us consider a closed loop control system with ‘K’ as controller parameter and
‘P(s)’ as plant transfer function.

𝐾𝐾𝐾𝐾
The transfer function, 𝑇𝑇(𝑆𝑆) =
1+𝐾𝐾𝐾𝐾

The characteristic equation 𝑞𝑞(𝑠𝑠) = 1 + 𝐾𝐾𝐾𝐾 = 0 and its roots are the poles.
𝑁𝑁(𝑠𝑠) 𝐾𝐾𝑁𝑁(𝑠𝑠)
If 𝐾𝐾(𝑠𝑠) = ; 𝑇𝑇(𝑆𝑆) =
𝐷𝐷(𝑠𝑠) 𝐷𝐷(𝑠𝑠)+𝐾𝐾𝑁𝑁(𝑠𝑠)

𝑞𝑞(𝑠𝑠) = 𝐷𝐷(𝑠𝑠) + 𝐾𝐾𝐾𝐾(𝑠𝑠) = 0

for K = 0, it reduces to D(s) = 0

93
Open Loop

𝐾𝐾𝐾𝐾(𝑠𝑠)
𝑇𝑇(𝑠𝑠) = 𝐾𝐾𝐾𝐾(𝑠𝑠) =
𝐷𝐷(𝑠𝑠)
The characteristic equation is 𝐷𝐷(𝑠𝑠) = 0 and its roots are the poles of open loop system.
These poles are equal to the poles of the closed loop system with 𝐾𝐾 = 0.
Open loop zeros are the roots of 𝐾𝐾(𝑠𝑠) = 0
In closed loop, if K is very large and tend to infinity, 𝐷𝐷(𝑠𝑠) + 𝐾𝐾𝐾𝐾(𝑠𝑠) ≈ 𝐾𝐾𝐾𝐾(𝑠𝑠) = 0; So,
N(s)=0 and its roots are the poles of closed loop system. Comparing with open loop
system, the poles of the closed loop system with 𝐾𝐾 → ∞ is equal to zeros of the open
loop system.
To summarize, the poles of closed loop system are equal to open loop poles at 𝐾𝐾 = 0 and
equal to open loop zeros, at 𝐾𝐾 = ∞.

In other words, when the control parameter ‘K’ varies from zero to infinity, the locus of
roots starts from a pole and ends at a zero. The travel from pole to zero is defined as Root
locus or journey of roots. If the path taken by the roots in such a journey remains
completely in the left half plane, the system is said to be stable for all values of ‘𝐾𝐾’ and
the system is not sensitive to the variation of ‘𝐾𝐾’. However, if some of the path is on the
right half plane, the system will be unstable, and we can determine the range of ‘𝐾𝐾’ for
instability and a range of ‘𝐾𝐾’ for the system to be stable.
Also, ‘𝐾𝐾’ should be positive for the path or loci to exist for travel. The journey of the
roots takes place as ‘𝐾𝐾’ varies from zero to infinity. If ‘𝐾𝐾’ is negative, no path or loci
exists for the roots to travel.

Example
𝐾𝐾𝐾𝐾(𝑠𝑠) 𝑠𝑠+1
Given, 𝑇𝑇(𝑠𝑠) = ; 𝐾𝐾(𝑠𝑠) = ;
𝐾𝐾𝐾𝐾(𝑠𝑠)+1 𝑠𝑠(𝑠𝑠+2)

𝐾𝐾(𝑠𝑠 + 1)
𝑇𝑇(𝑠𝑠) =
𝑠𝑠(𝑠𝑠 + 2) + 𝐾𝐾(𝑠𝑠 + 1)
Characteristic equation, 𝑞𝑞(𝑠𝑠) = 𝑠𝑠(𝑠𝑠 + 2) + 𝐾𝐾(𝑠𝑠 + 1) = 0
𝑠𝑠 2 + (𝐾𝐾 + 2)𝑠𝑠 + 𝐾𝐾 = 0

94
The poles, 𝑠𝑠1,2 = (1/2)[−(𝐾𝐾 + 2) ± �(𝐾𝐾 + 2)2 − 4𝐾𝐾]

The zeros 𝐾𝐾(𝑠𝑠 + 1) = 0 ⟶ (𝑠𝑠 + 1) = 0 ⟶ 𝑠𝑠 = −1

Since 𝐾𝐾 = 0 at poles, we can find the poles by setting 𝐾𝐾 = 0 as,

𝑠𝑠1,2 = (1/2)�−2 ± √4 − 0� = −1 ± 1 = 0, −2

Now, let us plot the poles and zeros on the S-Plane plot.

In this plot, zero at 𝑠𝑠 = −1 is represented by a small circle on the real axis, while
poles at 𝑠𝑠 = 0 𝑎𝑎𝑎𝑎𝑎𝑎 𝑠𝑠 = −2 are represented by ‘x’ on the real axis. The first path of travel
starts from the pole at origin and end at zero at 𝑠𝑠 = −1. The second path of travel starts
from the pole at 𝑠𝑠 = −2 and travels to the zero at infinity. For the second path, a zero is
assumed at infinity since there is no known zero in its path. In the region 𝑠𝑠 = −1 to 𝑠𝑠 =
−2, there is no path and ‘𝐾𝐾’ is negative.

For example, let K = 1.5, then 𝑠𝑠1,2 = −0.5, −3. The first root is on first path of travel
while the second root is on second path of travel. Let 𝐾𝐾 = −1.5; then the roots are
𝑠𝑠1,2 = −1.5, 1. The first root is in the region where there is no path exists and the other
root is on the positive half plane indicating instability. This example highlights that K
should be positive for stability and it is the required condition for the journey of roots
starting from a pole (𝐾𝐾 = 0) and end at a zero (𝐾𝐾 = ∞).

In this example, the poles and zeros are real values. They can be complex values in
general. Also, it will be challenging to find all the poles and zeros for each value of ‘𝐾𝐾’.
To predict the journey of roots without finding them, a set of rules are provided by
Walter R. Evans, the inventor of this method.

95
Process for Root Locus Plot
For real values of poles and zeros, the root locus diagram requires the knowledge of
number of loci (or paths) for travel on real axis. For complex value of poles and zeros,
more information is required, and they are Asymptotes, Cross–over point, Break–away
point and Angle of departure. The process to determine each step in this journey of the
roots is given below.

Step-1 Zeros, Poles and Loci

Given the plant transfer function 𝐾𝐾(𝑠𝑠), the closed loop system transfer function is
𝐾𝐾𝐾𝐾(𝑠𝑠) 𝑁𝑁(𝑠𝑠)
𝑇𝑇(𝑠𝑠) = Let 𝐾𝐾(𝑠𝑠) =
𝐾𝐾𝐾𝐾(𝑠𝑠)+1 𝐷𝐷(𝑠𝑠)

𝐾𝐾(𝑠𝑠) = Numerator polynomial in ‘s’ and its roots are the Zeros.
𝐷𝐷(𝑠𝑠) = Denominator polynomial in ‘s’ and its roots are the Poles.
Let number of zeros = 𝑚𝑚 and number of Poles = 𝑎𝑎.
Then, number of Loci = 𝑎𝑎; Number of loci travel to infinity = 𝑎𝑎 − 𝑚𝑚.
Each loci travel from a pole to a zero located at a finite distance. If there is no zero in its
path, a zero is assumed at infinity and the loci will travel to infinity.

Step – 2 Real Axis Loci (left half plane in S-Plot)

A loci exist in a region (between two poles or between a pole and a zero) if the number of
poles and zeros to the right of an arbitrary point in the region is an Odd number. The
controller gain, K is positive in this region (𝐾𝐾 > 0)
A loci does not exist in a region if the number of poles and zeros to the right of an
arbitrary point is an Even number. The controller gain is negative in this region (𝐾𝐾 < 0)

96
In figure (i), the loci travel is simple. In figure (ii), the loci between the two poles meet at
a point and break away towards the zero. Since there is no path exists in the region
between the pole and zero, the loci takes a detour of going over that region and enter or
break-in on the real axis with one branch traveling to zero in the vicinity and other branch
traveling to zero at infinity.

Step – 3 Loci travel to a zero at infinity

If a zero at infinity is not on real axis, but located at some point on S-plane, the loci needs
guidance to travel. The required parameters for guidance are (a) Asymptotes, (b) Cross-
over point for both loci and asymptotes, (c) Break-away or Break-in point on the real
axis.

(a) Asymptotes
∑(𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝)−∑(𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧)
Center of Asymptote, 𝜎𝜎𝐴𝐴 =
𝑛𝑛−𝑚𝑚

2𝑘𝑘+1
Angles of Asymptotes, (𝜙𝜙𝐴𝐴 )𝑘𝑘 = � � (180)
𝑛𝑛−𝑚𝑚

0 ≤ 𝑘𝑘 ≤ (𝑛𝑛 − 𝑚𝑚 − 1)

(b) Cross-over points

Cross-over points are of two types – (i) Asymptote cross over point and (ii) Loci cross
over point. The cross over points are defined as points where the asymptote and loci
crosses the vertical axis of the diagram.
Asymptote cross-over point = ±𝑗𝑗|𝜎𝜎𝐴𝐴 |𝑡𝑡𝑡𝑡𝑡𝑡(𝜙𝜙𝐴𝐴 )1
Loci cross-over point on the vertical axis represent the condition of marginal stability. So,
these points need to be determined by Routh – Hurwitz criterion.
In the Routh array, find ‘𝐾𝐾’ such that one of the row has all zeros. Then find the roots of
auxiliary polynomial and they are the Loci cross-over points.

(c) Break-away or Break-in point

Break-away point is the point on real axis where two loci meet and depart (break-away)
from real axis. Break-in point is the point on real axis where two loci from the S-plane
arrive (Break-in) to the real axis. 𝐾𝐾 = 𝐾𝐾𝑚𝑚𝑚𝑚𝑚𝑚 at this point.

Characteristic equation is, 1 + 𝐾𝐾𝐾𝐾(𝑠𝑠) = 0

1
𝐾𝐾 = − � �
𝑃𝑃(𝑠𝑠)

97
 𝑑𝑑𝑑𝑑
For 𝐾𝐾 = 𝐾𝐾𝑚𝑚𝑚𝑚𝑚𝑚 , =0
𝑑𝑑𝑑𝑑

Solving for roots of the resulting polynomial, the root in appropriate region of loci gives
the break-away or break-in points.

Step-4 Complex Poles and Zeros

When a loci depart from a complex pole, the angle of departure needs to be determined.
Similarly, when a loci arrive at a complex zero, the angle of arrival needs to be
determined.
(a) Angle of departure (𝜽𝜽𝒅𝒅 ):
At a complex pole, 𝑠𝑠 = −𝑎𝑎 + 𝑗𝑗𝑗𝑗
𝑁𝑁′(𝑠𝑠)
𝑃𝑃′(𝑠𝑠) = [(𝑠𝑠 + 𝑎𝑎 − 𝑗𝑗𝑗𝑗)𝑃𝑃(𝑠𝑠)]𝑠𝑠=(−𝑎𝑎+𝑗𝑗𝑗𝑗) =
𝐷𝐷′(𝑠𝑠)
∠𝑃𝑃′(𝑠𝑠) = ∠𝑁𝑁′ − ∠𝐷𝐷′
𝑏𝑏 0 𝑓𝑓𝑓𝑓𝑓𝑓 𝑏𝑏 = 0
∠𝑠𝑠 = − tan−1 � � = �
𝑎𝑎 90° 𝑓𝑓𝑓𝑓𝑓𝑓 𝑎𝑎 = 0
Angle of departure, 𝜃𝜃𝑑𝑑 = ±[∠𝑃𝑃′(𝑠𝑠) + 360]

(b) Angle of arrival (𝜽𝜽𝒂𝒂 ):

At a complex zero, 𝑠𝑠 = −𝑐𝑐 + 𝑗𝑗𝑗𝑗
𝑃𝑃′′(𝑠𝑠) = 𝑃𝑃(𝑠𝑠)/(𝑠𝑠 + 𝑐𝑐 − 𝑗𝑗𝑗𝑗) = 𝑁𝑁′′/𝐷𝐷′′
The angle,
∠𝑃𝑃′′(𝑠𝑠) = ∠𝑁𝑁′′ − ∠𝐷𝐷′′
Angle of arrival, 𝜃𝜃𝑎𝑎 = ±[∠𝑃𝑃′′(𝑠𝑠) − 360]

98
Example
1
Given, 𝐾𝐾(𝑠𝑠) = , draw the Root Locus diagram
𝑠𝑠(𝑠𝑠+2)(𝑠𝑠+4)

1) Open loop T.F., 𝑲𝑲𝑲𝑲(𝒔𝒔) = 𝑲𝑲/[𝒔𝒔(𝒔𝒔 + 𝟐𝟐)(𝒔𝒔 + 𝟒𝟒)]

Number of Zeros, 𝑚𝑚 = 0
Number of Poles, 𝑎𝑎 = 3; 𝑠𝑠 = 0, −2, −4
Number of loci = 𝑎𝑎 = 3
Number of loci travel to infinity = 𝑎𝑎– 𝑚𝑚 = 3

2)
Existence of Loci on Real Axis
𝑠𝑠1 𝑡𝑡𝑓𝑓 𝑠𝑠2 : Sum of poles and zeros to
the right of an arbitrary point chosen
in this region is odd (1 pole)
So, the path exists in this region and K
is positive.

𝑠𝑠2 𝑡𝑡𝑓𝑓 𝑠𝑠3 : Sum of poles and zeros to the right of an arbitrary point chosen in this region
is even (2 poles). So, no path exist in this region and K is negative.

𝑠𝑠3 𝑡𝑡𝑓𝑓 ∞: Sum of poles and zeros to the right of an arbitrary point chosen in this region is
odd (3 poles). So, the path exists in this region and K is positive.

99
 3) Loci Travel

The loci from 𝑠𝑠 = 0 and 𝑠𝑠 = −2 come together and break away to a zero at infinity. To
guide the path of this travel asymptotes are used. The loci follows these asymptotes in its
journey to a zero at infinity.
∑ 𝐾𝐾−∑ 𝑍𝑍 (0−2−4)−(0)
Center of Asymptotes, 𝜎𝜎𝐴𝐴 = = = −2
𝑛𝑛−𝑚𝑚 3

2𝑘𝑘+1
Angle of Asymptotes, 𝜙𝜙𝐴𝐴 = � � (180); 0 ≤ 𝑘𝑘 ≤ (𝑎𝑎 − 𝑚𝑚 − 1)
𝑛𝑛−𝑚𝑚

2𝑘𝑘 + 1
𝜙𝜙𝐴𝐴 = � � (180) = (2𝑘𝑘 + 1)(60); 0 ≤ 𝑘𝑘 ≤ 2
3
𝑘𝑘 = 0; 𝜙𝜙1 = 60°
𝑘𝑘 = 1; 𝜙𝜙2 = 180°
𝑘𝑘 = 2; 𝜙𝜙3 = 300°

100
Cross-over Point
Characteristic equation 1 + 𝐾𝐾𝐾𝐾(𝑠𝑠) = 0
𝑠𝑠(𝑠𝑠 + 2)(𝑠𝑠 + 4) + 𝐾𝐾 = 0
𝑠𝑠 3 + 6𝑠𝑠 2 + 8𝑠𝑠 + 𝐾𝐾 = 0
𝑠𝑠 3 1 8 0
𝑠𝑠 2 6 𝐾𝐾 0
𝑠𝑠1 𝑏𝑏1 0
𝑠𝑠 0 𝑐𝑐1

𝐾𝐾
𝑏𝑏1 = − + 8 = (1/6)(48 − 𝐾𝐾) 𝑐𝑐1 = 𝐾𝐾
6
For stability, 𝑏𝑏1 and 𝑐𝑐1 should be positive. So, K should be positive but less than 48;
0 < 𝐾𝐾 < 48. If 𝑏𝑏1 = 0, the row 𝑆𝑆’ will have all zero; the system will be marginally
stable.
For 𝐾𝐾 = 48, 𝑏𝑏1 = 0
The auxiliary polynomial is, 6𝑠𝑠 2 + 𝐾𝐾 = 0; 6𝑠𝑠 2 + 48 = 0

48
𝑆𝑆 = �− = ±𝑗𝑗2.828; These are the loci cross-over points.
6

Asymptote cross-over points are ±𝑗𝑗|𝜎𝜎𝐴𝐴 |𝑡𝑡𝑡𝑡𝑡𝑡𝜙𝜙1 = ±𝑗𝑗2𝑡𝑡𝑡𝑡𝑡𝑡60 = ±𝑗𝑗3.46

Break-away point
𝐾𝐾 = −1/𝑃𝑃(𝑠𝑠) = −(𝑠𝑠 3 + 6𝑠𝑠 2 + 8𝑠𝑠) = 0
𝑑𝑑𝑑𝑑/𝑑𝑑𝑑𝑑 = 0; 3𝑠𝑠 2 + 12𝑠𝑠 + 8 = 0

𝑠𝑠 = (1/6)�−12 ± √144 − 96 � = −0.845, −3.155

Since −3.155 lies on the region where there is no loci, 𝑠𝑠 = −0.845 is the viable break
away point. So, the loci from 𝑠𝑠 = 0 𝑎𝑎𝑎𝑎𝑎𝑎 𝑠𝑠 = −2 travel towards each other and break
away at 𝑠𝑠 = −0.845 and follow the asymptotes to the zero at infinity. In this process, the
loci crosses the vertical axis at 𝑠𝑠 = ±𝑗𝑗2.828 and the asymptote crosses the vertical axis at
±𝑗𝑗3.46.
Since the pole is not a complex pole, it is not required to find the angle of departure.

101
Example:
Given, 𝐾𝐾(𝑠𝑠) = (𝑠𝑠 + 2)/𝑠𝑠(𝑠𝑠 + 1), draw the Root Locus diagram.
Number of Zeros, 𝑚𝑚 = 1; 𝑠𝑠 = −2
Number of Poles, 𝑎𝑎 = 2; 𝑠𝑠 = 0, −1
Number of loci = 𝑎𝑎 = 2
Number of loci travel to infinity = 𝑎𝑎– 𝑚𝑚 = 1

Real Axis loci

There is a path between P1 and P2. There is no path between P2 and Z1. There is a path
between Z1 and infinity. The loci from P1 and P2 travel towards each other and break
away. Then they take a detour over the zero and break-in on the real axis in the region Z1
to ∞. One branch goes to zero at 𝑠𝑠 = −2 and the other branch goes to zero at infinity as
shown in the diagram.

102
Break-away and Break-in points
𝐾𝐾 = −1/𝑃𝑃(𝑠𝑠) = −𝑠𝑠(𝑠𝑠 + 1)/(𝑠𝑠 + 2) = −(𝑠𝑠 2 + 𝑠𝑠)/(𝑠𝑠 + 2)
(𝑠𝑠 + 2)(2𝑠𝑠 + 1)−(𝑠𝑠 2 + 𝑠𝑠)(1)
𝑑𝑑𝑑𝑑/𝑑𝑑𝑑𝑑 = 0 = − � �
(𝑠𝑠 + 2)2

𝑠𝑠 2 + 4𝑠𝑠 + 2 = 0
𝑠𝑠 = −0.585, −3.415
𝑠𝑠 = −0.585 is the Break-away point while 𝑠𝑠 = −3.415 is the Break-in point.
Let the maximum height of the detour path is −2 ± 𝑗𝑗𝑗𝑗
Characteristic equation is,
1 + 𝐾𝐾𝐾𝐾(𝑠𝑠) = 0
𝑠𝑠(𝑠𝑠 + 1) + 𝐾𝐾(𝑠𝑠 + 2) = 0; 𝑠𝑠 2 + (𝐾𝐾 + 1)𝑠𝑠 + 2𝐾𝐾 = 0 ⟶ (1)
For the roots, 𝑠𝑠 = −2 ± 𝑗𝑗𝑗𝑗. the desired characteristic equation is,
(𝑠𝑠 + 2 − 𝑎𝑎𝑎𝑎)(𝑠𝑠 + 2 + 𝑎𝑎𝑎𝑎) = 0
𝑠𝑠 2 + 4𝑠𝑠 + (4 + 𝑎𝑎2 ) = 0 ⟶ (2)
Comparing the equations (1) and (2),
𝐾𝐾 + 1 = 4; 𝐾𝐾 = 3
4 + 𝑎𝑎2 = 2𝐾𝐾 = 6

So, 𝑎𝑎2 = 2; 𝑎𝑎 = √2

The coordinates of loci at 𝑠𝑠 = −2 are, 𝑠𝑠 = −2 ± 𝑗𝑗√2

On the root locus diagram, the points A and B are at equal distance of 1.415 from the
zero at 𝑠𝑠 = −2. The points, C and D are at equal distance of √2 from the pole at 𝑠𝑠 = −2.
So, the shape of the root locus is a circle.

K-value at break-away and break-in points

Since 𝐾𝐾 = −(𝑠𝑠 2 + 𝑠𝑠)/(𝑠𝑠 + 2)
At break-away point, 𝑠𝑠 = −0.585; 𝐾𝐾 = 0.17
At break-in point, 𝑠𝑠 = −3.415; 𝐾𝐾 = 5.83

103
Example
Given 𝐾𝐾(𝑠𝑠) = 1�[𝑠𝑠(𝑠𝑠 + 4)(𝑠𝑠 2 + 8𝑠𝑠 + 32)]

Draw the Root Locus diagram

Zeros, Poles and Loci

Number of Zeros, 𝑚𝑚 = 0

Number of Poles, 𝑎𝑎 = 4; 𝑠𝑠1 = 0, 𝑠𝑠2 = −4; 𝑠𝑠3,4 = 1/2�−8 ± √64 − 128� = −4 ± 𝑗𝑗4

Number of loci = 𝑎𝑎 = 4
Number of loci travel to infinity = 𝑎𝑎– 𝑚𝑚 = 4

Real Axis Loci

There is a loci between 𝑠𝑠 = 0 and 𝑠𝑠 = −4. There is no loci in the region 𝑠𝑠 = −4 and
𝑠𝑠 = ∞. The loci between the poles 𝑠𝑠 = 0 and 𝑠𝑠 = −4 travel towards each other and break
away to a zero at infinity. The loci from complex poles begins their journey with a
departure angle and travel to a zero at infinity. Since the path to infinity is not along the
real axis, they need the guidance of asymptotes.

104
Loci Travel
Asymptotes:
∑ 𝐾𝐾−∑ 𝑍𝑍
Center of asymptotes, 𝜎𝜎𝐴𝐴 = = (1/4)[0 − 4 − 4 + 𝑗𝑗4 − 4 − 𝑗𝑗4] = −3
𝑛𝑛−𝑚𝑚

2𝑘𝑘+1 2𝑘𝑘+1
Angle of asymptotes, 𝜙𝜙𝐴𝐴 = � � (180°) = � � (180°) = (2𝑘𝑘 + 1)(45°)
𝑛𝑛−𝑚𝑚 4

0 ≤ 𝑘𝑘 ≤ (𝑎𝑎 − 𝑚𝑚 − 1); 0 ≤ 𝑘𝑘 ≤ 3
k 0 1 2 3
𝜙𝜙𝐴𝐴 45° 135° 225° 315°

Cross-over point
The characteristic equation is given by
1 + 𝐾𝐾𝐾𝐾(𝑠𝑠) = 0
𝑠𝑠(𝑠𝑠 + 4)(𝑠𝑠 2 + 8𝑠𝑠 + 32) + 𝐾𝐾 = 0; 𝑠𝑠 4 + 12𝑠𝑠 3 + 64𝑠𝑠 2 + 128𝑠𝑠 + 𝐾𝐾 = 0

Routh Array
𝑠𝑠 4 1 64 K
𝑠𝑠 3 12 128 0
𝑠𝑠 2 𝑗𝑗1 𝑗𝑗2
𝑠𝑠1 𝑐𝑐1 𝑐𝑐2
𝑠𝑠 0 𝑎𝑎1

105
 𝑏𝑏1 = −(128/12)(1) + 64 = 160/3
𝑏𝑏2 = 𝐾𝐾
𝑐𝑐1 = 128 − (9𝐾𝐾/40)
𝑐𝑐2 = 0
𝑑𝑑1 = 𝐾𝐾
For stability, 𝑏𝑏1 , 𝑐𝑐1 , 𝑑𝑑1 should be positive.
9𝐾𝐾
So, 𝐾𝐾 > 0 𝑎𝑎𝑎𝑎𝑎𝑎 < 128; 𝑜𝑜𝑜𝑜 𝐾𝐾 < 568.89
40

For 𝐾𝐾 = 568.89, 𝑐𝑐1 = 0 and hence the row of 𝑠𝑠1 has all zeros.
Then, the auxiliary polynomial is given by, 𝑏𝑏1 𝑠𝑠 2 + 𝑏𝑏2 = 0

𝒔𝒔 = �−(𝑏𝑏2 /𝑏𝑏1 ) = �−3𝐾𝐾/160 = ±𝑗𝑗(3.27)

These are the crossover points of loci.

The crossover point of asymptotes = ±|𝜎𝜎𝐴𝐴 |𝑡𝑡𝑡𝑡𝑡𝑡𝜙𝜙1 = ±3𝑡𝑡𝑡𝑡𝑡𝑡45 = ±3

Break-away point:
Characteristic equation is 1 + 𝐾𝐾𝐾𝐾(𝑠𝑠) = 0
𝐾𝐾 = −1/𝑃𝑃(𝑠𝑠) = −(𝑠𝑠 4 + 12𝑠𝑠 3 + 64𝑠𝑠 2 + 128𝑠𝑠)
The value of s at 𝐾𝐾 = 𝐾𝐾𝑚𝑚𝑚𝑚𝑚𝑚 gives the break-away point.
So, 𝑑𝑑𝑑𝑑/𝑑𝑑𝑑𝑑 = 0; 4𝑠𝑠 3 + 36𝑠𝑠 2 + 128𝑠𝑠 + 128 = 0;
Let F(s) = 𝑑𝑑𝑑𝑑/𝑑𝑑𝑑𝑑; Since 𝐹𝐹(𝑠𝑠) is cubic, we can solve for a root by graphical method.
𝑠𝑠 0 −1 −2
𝐹𝐹(𝑠𝑠) 128 32 −16

106
 Break-away point is, 𝜎𝜎𝑗𝑗 = −1.58
Alternately, one can find the roots
of this cubic polynomial by
MATLAB and choose the real
value of root that is within the
region of loci.

Departure Angle
Let us find the departure angle at the complex pole, −4 + 𝑗𝑗4

𝐾𝐾(𝑠𝑠) = 1�[𝑠𝑠(𝑠𝑠 + 4)(𝑠𝑠 + 4 + 𝑗𝑗4)(𝑠𝑠 + 4 − 𝑗𝑗4)]

At 𝑠𝑠 = −4 + 𝑗𝑗4,
𝐾𝐾’(𝑠𝑠) = [(𝑠𝑠 + 4 − 𝑗𝑗4)𝐾𝐾(𝑠𝑠)]𝑠𝑠=−4+𝑗𝑗4

= �1�𝑠𝑠(𝑠𝑠 + 4)(𝑠𝑠 + 4 + 𝑗𝑗4)�

𝑠𝑠=−4+𝑗𝑗4

= 1�[(−4 + 𝑗𝑗4)(𝑗𝑗4)(𝑗𝑗8)]

𝐾𝐾’(𝑠𝑠) = 𝐾𝐾’/𝐷𝐷’;
∠𝐾𝐾’ = ∠𝐾𝐾’ − ∠𝐷𝐷’
= (0) − (−45 + 90 + 90) = −135°
𝜃𝜃𝑑𝑑 = ±(−135 + 360) = ±225°
For the pole, −4 + 𝑗𝑗4; 𝜃𝜃𝑑𝑑 = 225° and for the pole, −4 − 𝑗𝑗4; 𝜃𝜃𝑑𝑑 = −225°

107
Root-Locus Diagram

PID Controllers
It is a three terms controller commonly used in industrial process control. The three terms
are (1) Proportional, (2) Integral and (3) Derivative.
𝐾𝐾
𝐺𝐺𝑐𝑐 (𝑠𝑠) = 𝐾𝐾𝐾𝐾 + � 𝐼𝐼�𝑆𝑆� + 𝐾𝐾𝐷𝐷 𝑆𝑆

Where 𝐾𝐾𝐾𝐾 , 𝐾𝐾𝐼𝐼 , 𝐾𝐾𝐷𝐷 are gains of proportional, integral and derivative terms respectively. It
is used to improve stability and steady state error.
𝐾𝐾
PI-Controller: 𝐺𝐺𝑐𝑐 (𝑠𝑠) = 𝐾𝐾𝐾𝐾 + � 𝐼𝐼�𝑆𝑆�; it is used to improve the steady state error.

PD-Controller: 𝐺𝐺𝑐𝑐 (𝑠𝑠) = 𝐾𝐾𝐾𝐾 + 𝐾𝐾𝐷𝐷 𝑆𝑆; it is used to improve the stability of system.
A PI and PD controller placed in series is equivalent to a PID controller.

108
 =
𝐾𝐾
Let 𝐺𝐺1 = PI Controller = 𝐾𝐾𝐾𝐾1 + � 𝐼𝐼�𝑆𝑆�

Let 𝐺𝐺2 = PD Controller = 𝐾𝐾𝐾𝐾2 + 𝐾𝐾𝐷𝐷 𝑆𝑆

𝐾𝐾𝐼𝐼�
𝐺𝐺𝑐𝑐 = 𝐺𝐺1 𝐺𝐺2 = �𝐾𝐾𝐾𝐾1 + 𝑆𝑆� �𝐾𝐾𝐾𝐾2 + 𝐾𝐾𝐷𝐷 𝑆𝑆�
𝐾𝐾𝐼𝐼 𝐾𝐾𝐾𝐾2
= (𝐾𝐾𝐾𝐾1 𝐾𝐾𝐾𝐾2 + 𝐾𝐾𝐼𝐼 𝐾𝐾𝐷𝐷 ) + � � � + 𝐾𝐾𝐾𝐾1 𝐾𝐾𝐷𝐷 𝑆𝑆
𝑆𝑆
(𝐾𝐾𝐼𝐼 )𝑝𝑝�
= (𝐾𝐾𝐾𝐾 )𝑝𝑝 + � 𝑆𝑆� + (𝐾𝐾𝐷𝐷 )𝑝𝑝 𝑆𝑆
where the subscript ‘𝑒𝑒’ represents equivalent value.

Controller gains, 𝑲𝑲𝑲𝑲 , 𝑲𝑲𝑰𝑰 , 𝑲𝑲𝑫𝑫

𝐺𝐺𝑐𝑐 𝐺𝐺
𝑇𝑇 =
1 + 𝐺𝐺𝑐𝑐 𝐺𝐺

𝐾𝐾
𝐺𝐺𝑐𝑐 = 𝐾𝐾𝐾𝐾 + � 𝐼𝐼�𝑠𝑠� + 𝐾𝐾𝐷𝐷 𝑠𝑠

= �1�𝑠𝑠� [𝐾𝐾𝐾𝐾 𝑠𝑠 + 𝐾𝐾𝐼𝐼 + 𝐾𝐾𝐷𝐷 𝑠𝑠 2 ]

The controller gains can be calculated by Zeigler-Nichols method.

1.2𝐾𝐾𝑈𝑈
𝐾𝐾𝐾𝐾 = 0.6𝐾𝐾𝑈𝑈 , 𝐾𝐾𝐼𝐼 = , 𝐾𝐾𝐷𝐷 = 0.075𝐾𝐾𝑈𝑈 𝑇𝑇𝑈𝑈
𝑇𝑇𝑈𝑈
For PI controller, 𝐾𝐾𝐾𝐾 = 0.45𝐾𝐾𝑈𝑈 , 𝐾𝐾𝐼𝐼 = 0.54𝐾𝐾𝑈𝑈 /𝑇𝑇𝑈𝑈
For PD controller, 𝐾𝐾𝐾𝐾 = 0.6𝐾𝐾𝑈𝑈 , 𝐾𝐾𝐷𝐷 = 0.075𝐾𝐾𝑈𝑈 𝑇𝑇𝑈𝑈
For proportional term controller, 𝐾𝐾𝐾𝐾 = 0.5𝐾𝐾𝑈𝑈
where Ultimate gain, 𝐾𝐾𝑈𝑈 = 𝐾𝐾 for the system to be marginally stable and it can be
determined by Routh-Hurwitz method. The Ultimate period, 𝑇𝑇𝑈𝑈 = 2𝜋𝜋/𝜔𝜔; 𝜔𝜔 = real root
of characteristic equation, 𝑞𝑞(𝑠𝑠) with 𝑠𝑠 = 𝑗𝑗𝜔𝜔, evaluated at 𝐾𝐾 = 𝐾𝐾𝑈𝑈 .

109
Example
Given, 𝐺𝐺(𝑠𝑠) = 1�[𝑠𝑠(𝑠𝑠 + 1)(𝑠𝑠 + 5)]

Find the PID controller transfer function and the system transfer function.

To find 𝐾𝐾𝑈𝑈 𝑎𝑎𝑎𝑎𝑎𝑎 𝑇𝑇𝑈𝑈 . It is assumed that 𝐺𝐺𝑐𝑐 = 𝐾𝐾𝑃𝑃 = 𝐾𝐾

Characteristic equation, 𝑞𝑞(𝑠𝑠) = 1 + 𝐺𝐺𝑐𝑐 𝐺𝐺 = 0; 1 + 𝐾𝐾𝐾𝐾 = 0
𝐾𝐾
1+ =0
𝑆𝑆(𝑆𝑆 + 1)(𝑆𝑆 + 5)
𝑆𝑆(𝑆𝑆 + 1)(𝑆𝑆 + 5) + 𝐾𝐾 = 0
𝑆𝑆 3 + 6𝑆𝑆 2 + 5𝑆𝑆 + 𝐾𝐾 = 0

Routh Array:
𝑠𝑠 3 1 5 0
𝑠𝑠 2 6 𝐾𝐾 0
𝑠𝑠1 𝑏𝑏1 0
𝑠𝑠 0 𝐾𝐾

𝑏𝑏1 = (−𝐾𝐾/6) + 5
For stability, 𝑏𝑏1 and 𝐾𝐾 should be positive.
For marginal stability, 𝑏𝑏1 = 0; (𝐾𝐾/6) = 5
𝐾𝐾 = 30 = 𝐾𝐾𝑈𝑈
𝑞𝑞(𝑠𝑠) = 𝑠𝑠 3 + 6𝑠𝑠 2 + 5𝑠𝑠 + 𝐾𝐾 = 0
𝑞𝑞(𝑗𝑗𝑗𝑗) = −𝜔𝜔2 (𝑗𝑗𝑗𝑗) − 6𝜔𝜔2 + 5(𝑗𝑗𝑗𝑗) + 30 = 0
6(5 − 𝜔𝜔2 ) + 𝑗𝑗𝑗𝑗(5 − 𝜔𝜔2 ) = 0; (5 − 𝜔𝜔2 )(6 + 𝑗𝑗𝑗𝑗) = 0

𝜔𝜔2 = 5; 𝜔𝜔 = √5 (Real value of 𝜔𝜔)

𝑇𝑇𝑈𝑈 = 2𝜋𝜋/𝜔𝜔 = 2𝜋𝜋/√5 = 2.81

𝐾𝐾𝑃𝑃 , 𝐾𝐾𝐼𝐼 , 𝐾𝐾𝐷𝐷 for this example:
𝐾𝐾𝑃𝑃 = 0.6𝐾𝐾𝑈𝑈 = 18
𝐾𝐾𝐼𝐼 = 1.2𝐾𝐾𝑈𝑈 /𝑇𝑇𝑈𝑈 = 12.81

110
 𝐾𝐾𝐷𝐷 = 0.075𝐾𝐾𝑈𝑈 𝑇𝑇𝑈𝑈 = 6.32
1
𝐺𝐺𝑐𝑐 (𝑠𝑠) = � � [18𝑆𝑆 + 12.81 + 6.32𝑆𝑆 2 ]
𝑆𝑆
6.32
=� � [𝑆𝑆 2 + 2.85𝑆𝑆 + 2.03]
𝑆𝑆
6.32
=� � [(𝑆𝑆 + 1.425)2 ]
𝑆𝑆
ONE Pole: 𝑠𝑠 = 0
TWO Zeros: 𝑠𝑠 = −1.425, 𝑠𝑠 = −1.425

Transfer function
6.32
Controller Transfer Function: 𝐺𝐺𝑐𝑐 (𝑠𝑠) = � � (𝑆𝑆 2 + 2.85𝑆𝑆 + 2.03)
𝑆𝑆

1
𝐺𝐺 =
𝑆𝑆(𝑆𝑆 + 1)(𝑆𝑆 + 5)
6.32(𝑆𝑆 2 + 2.85𝑆𝑆 + 2.03)
𝐺𝐺𝑐𝑐 𝐺𝐺 =
𝑆𝑆 2 (𝑆𝑆 + 1)(𝑆𝑆 + 5)
System Transfer Fuction:
𝐺𝐺𝑐𝑐 𝐺𝐺 6.32(𝑆𝑆 2 + 2.85𝑆𝑆 + 2.03)
𝑇𝑇(𝑆𝑆) = = 2
1 + 𝐺𝐺𝑐𝑐 𝐺𝐺 𝑆𝑆 (𝑆𝑆 + 1)(𝑆𝑆 + 5) + 6.32(𝑆𝑆 2 + 2.85𝑆𝑆 + 2.03)
6.3𝑆𝑆 2 + 18𝑆𝑆 + 12.8
𝑇𝑇(𝑆𝑆) =
𝑆𝑆 4 + 6𝑆𝑆 3 + 11.3𝑆𝑆 2 + 18𝑆𝑆 + 12.8

111
 Assessment

1. For 𝑇𝑇 = 𝐾𝐾𝐾𝐾/(1 + 𝐾𝐾𝐾𝐾), the poles of a closed loop system at K = 0 are equal to
a) Open loop poles
b) Open loop zeros
c) Roots of the characteristic equation
d) All of the above

2. The path or loci exist in a region where

a) 𝐾𝐾 > 0
b) 𝐾𝐾 = 0
c) 𝐾𝐾 < 0
d) None of the above

3. 𝐾𝐾 > 0 in a region if the sum of poles and zeros to the right of an arbitrary point in
this region is
a) Zero
b) Odd
c) Even
d) None of the above

4. The value of 𝐾𝐾 is zero at

a) A zero
b) On the horizontal axis of Root Locus
c) On the vertical axis of Root Locus
d) A pole

5. At a given zero, the value of ‘K’ is

a) Zero
b) Infinity
c) Positive
d) Negative

112
6. if there is no path or loci in a region, then
a) 𝐾𝐾 = 0
b) 𝐾𝐾 = ∞
c) 𝐾𝐾 > 0
d) 𝐾𝐾 < 0

7. If a loci travels to a zero at infinity, it needs

a) Cross-over points
b) Break-away point
c) Asymptotes
d) Break-in point

8. Angle of departure is required for a loci

a) To begin the travel from a real pole
b) To begin the travel from a complex pole
c) To end the travel at complex pole
d) To end the travel at real pole

9. Cross-over points for a loci are the points where

a) The system is marginally stable
b) The loci intersects the vertical axis
c) They represent the roots of the auxiliary polynomial
d) All of the above

10. PID controllers are used to improve

a) Stability
b) Steady state error
c) Industrial process control
d) All of the above

113
 8. Bode Plot and Nyquist Plot

Learning Objectives:
1) To express the magnitude of system transfer function in frequency domain in
terms of decibel (dB).
2) To draw the Bode plots of magnitude (dB) and phase angle in separate plots as a
function of frequency.
3) To draw the Nyquist plot of magnitude and phase angle in a same plot (polar plot)
4) To determine Gain margin, Phase margin and bandwidth frequency

Hendrik Wade Bode (1905 – 1982) invented Bode – Plots to represent the magnitude and
phase angle of system transfer function in two separate logarithmic graphs. Also, he
invented gain margin, phase margin and bandwidth in the stability analysis of linear
system. Harry Theodor Nyquist (1889 – 1976) represented the magnitude and phase
angle together in one plot using polar graph. These graphs are called Nyquist Plots and
they provide alternate method in stability analysis.

Transfer function as a function of frequency:

Let the system transfer function = 𝑇𝑇(𝑠𝑠). It can be expressed as a function of frequency
by substituting 𝑠𝑠 = 𝑗𝑗𝑗𝑗. Then, the transfer function 𝑇𝑇(𝑗𝑗𝑗𝑗) would have a ratio of complex
values. The magnitude of transfer function is expressed in terms of decibels and the phase
angle is expressed in terms of degrees.

Decibel:
“Bel” is defined as logarithmic ratio of two power levels. A decibel is equal to one tenth
of a Bel, and it is commonly used in the measurement of noise in acoustic analysis.
𝑃𝑃 𝑃𝑃
Sound level, 𝐿𝐿 = 𝑙𝑙𝑙𝑙𝑙𝑙10 � 1 � 𝐵𝐵𝐵𝐵𝐵𝐵; 1𝐵𝐵𝐵𝐵𝐵𝐵 = 10𝑑𝑑𝑑𝑑; 𝐿𝐿 = 10[𝑙𝑙𝑙𝑙𝑙𝑙10 � 1 �]𝑑𝑑𝑑𝑑
𝑃𝑃2 𝑃𝑃2

Where 𝑃𝑃1 and 𝑃𝑃2 are power levels.

Power 𝑃𝑃 = 𝜔𝜔𝜔𝜔; Torque, 𝑇𝑇 = 𝐼𝐼𝐼𝐼 where ‘𝐼𝐼’ is the mass moment of inertia and 𝛼𝛼 = �𝜔𝜔�𝑡𝑡�
𝐼𝐼𝐼𝐼 𝐼𝐼
So, 𝑃𝑃 = 𝜔𝜔 � � = 𝜔𝜔2
𝑡𝑡 𝑡𝑡

115
 𝑃𝑃1 𝜔𝜔12
=
𝑃𝑃2 𝜔𝜔22
𝜔𝜔1 2 𝜔𝜔1
𝐿𝐿 = 10𝑙𝑙𝑙𝑙𝑙𝑙10 � � = 20𝑙𝑙𝑙𝑙𝑙𝑙10 � � 𝑑𝑑𝑑𝑑
𝜔𝜔2 𝜔𝜔2
For an Octave, 𝜔𝜔2 = 2𝜔𝜔1 ; 𝐿𝐿 = |20𝑙𝑙𝑙𝑙𝑙𝑙(1⁄2)| = 6𝑑𝑑𝑑𝑑
For a Decade, 𝜔𝜔2 = 10𝜔𝜔1 ; 𝐿𝐿 = |20𝑙𝑙𝑙𝑙𝑙𝑙(1⁄10)| = 20𝑑𝑑𝑑𝑑
For comparison of “𝑑𝑑𝑑𝑑” level, the auditory threshold is (0)𝑑𝑑𝑑𝑑, whisper is (30)𝑑𝑑𝑑𝑑, loud
thunder is (110)𝑑𝑑𝑑𝑑, Jet engine noise during take-off of an airplane is 140𝑑𝑑𝑑𝑑, a rocket
engine noise during take-off is 180𝑑𝑑𝑑𝑑.

Magnitude and Phase Angle

𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪– (𝒊𝒊)
𝑇𝑇(𝑗𝑗𝑗𝑗) = (𝐴𝐴 + 𝑗𝑗𝑗𝑗)𝑁𝑁 ; |𝑇𝑇| = (𝐴𝐴2 + 𝐵𝐵2 )𝑁𝑁⁄2
Magnitude, 𝑑𝑑𝑑𝑑 = 20𝑙𝑙𝑙𝑙𝑙𝑙|𝑇𝑇| = 20(𝑁𝑁⁄2)𝑙𝑙𝑙𝑙𝑙𝑙(𝐴𝐴2 + 𝐵𝐵2 )
= (10𝑁𝑁)𝑙𝑙𝑙𝑙𝑙𝑙(𝐴𝐴2 + 𝐵𝐵2 )
Phase Angle, 𝜙𝜙 = (𝑁𝑁) 𝑡𝑡𝑡𝑡𝑡𝑡−1 (𝐵𝐵⁄𝐴𝐴)
𝐴𝐴 = 0; 𝑑𝑑𝑑𝑑 = 20𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(𝐵𝐵), 𝜙𝜙 = (𝑁𝑁)(𝜋𝜋⁄2)
𝐵𝐵 = 0; 𝑑𝑑𝑑𝑑 = 20𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁(𝐴𝐴), 𝜙𝜙 = 0

Log Algebra: 𝑙𝑙𝑙𝑙𝑙𝑙(𝑎𝑎𝑎𝑎) = 𝑙𝑙𝑙𝑙𝑙𝑙(𝑎𝑎) + 𝑙𝑙𝑙𝑙𝑙𝑙(𝑏𝑏)

𝑙𝑙𝑙𝑙𝑙𝑙(𝑎𝑎/𝑏𝑏) = 𝑙𝑙𝑙𝑙𝑙𝑙(𝑎𝑎) − 𝑙𝑙𝑙𝑙𝑙𝑙(𝑏𝑏)

𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪 − (𝒊𝒊𝒊𝒊)
𝐾𝐾(𝑧𝑧 + 𝑗𝑗𝑗𝑗)
𝑇𝑇(𝑗𝑗𝑗𝑗) = = 𝑁𝑁/𝐷𝐷
(𝑗𝑗𝑗𝑗)2 (𝑎𝑎+ 𝑗𝑗𝑗𝑗)(𝑏𝑏 + 𝑗𝑗𝑗𝑗)
Magnitude, 𝑑𝑑𝑑𝑑 = 20𝑙𝑙𝑙𝑙𝑙𝑙|𝑇𝑇| = 20𝑙𝑙𝑙𝑙𝑙𝑙|𝑁𝑁| − 20𝑙𝑙𝑙𝑙𝑙𝑙|𝐷𝐷|
= [20𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 + 10𝑙𝑙𝑙𝑙𝑙𝑙(𝑧𝑧 2 + 𝜔𝜔2 ) − 20(2)𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙] − 10𝑙𝑙𝑙𝑙𝑙𝑙(𝑎𝑎2 + 𝜔𝜔2 ) − 10𝑙𝑙𝑙𝑙𝑙𝑙(𝑏𝑏 2 + 𝜔𝜔2 )
Phase Angle,
𝜙𝜙 = ∠𝑇𝑇 = ∠𝑁𝑁 − ∠𝐷𝐷 = 0 + 𝑡𝑡𝑡𝑡𝑡𝑡−1 (𝜔𝜔⁄𝑧𝑧) − 2(𝜋𝜋⁄2) − 𝑡𝑡𝑡𝑡𝑡𝑡−1 (𝜔𝜔⁄𝑎𝑎) − 𝑡𝑡𝑡𝑡𝑡𝑡−1 (𝜔𝜔⁄𝑏𝑏)

116
𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪 − (𝒊𝒊𝒊𝒊𝒊𝒊)
A typical transfer function, 𝑇𝑇(𝑠𝑠) can be expressed as,
𝐾𝐾(1 + 𝜏𝜏1 𝑠𝑠)
𝑇𝑇(𝑠𝑠) =
𝑠𝑠 𝑁𝑁 (1 + 𝜏𝜏2 𝑠𝑠)(𝑠𝑠 2 + 2𝜁𝜁𝜔𝜔𝑛𝑛 𝑠𝑠 + 𝜔𝜔𝑛𝑛2 )
𝐾𝐾(1 + 𝑗𝑗𝑗𝑗𝜏𝜏1 )
𝑇𝑇(𝑗𝑗𝑗𝑗) =
(𝑗𝑗𝑗𝑗)𝑁𝑁 (1 + 𝑗𝑗𝑗𝑗𝜏𝜏2 )(−𝜔𝜔 2 + 𝑗𝑗(2𝜁𝜁𝜔𝜔𝑛𝑛 )𝜔𝜔 + 𝜔𝜔𝑛𝑛2 )
𝐾𝐾1 (1 + 𝑗𝑗𝑗𝑗𝜏𝜏1 )
=
(𝑗𝑗𝑗𝑗)𝑁𝑁 (1 + 𝑗𝑗𝑗𝑗𝜏𝜏2 ){(1 − 𝑢𝑢2 ) + 𝑗𝑗(2𝜁𝜁𝜁𝜁)}
Where 𝐾𝐾1 = (𝐾𝐾/𝜔𝜔𝑛𝑛2 ) and 𝑢𝑢 = (𝜔𝜔/𝜔𝜔𝑛𝑛 )
𝑑𝑑𝑑𝑑 = 20𝑙𝑙𝑙𝑙𝑙𝑙|𝑇𝑇(𝑗𝑗𝑗𝑗)|
= [20𝑙𝑙𝑙𝑙𝑙𝑙𝐾𝐾1 + 10𝑙𝑙𝑙𝑙𝑙𝑙(1 + 𝜔𝜔2 𝜏𝜏1 2 )] − [20𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 + 10𝑙𝑙𝑙𝑙𝑙𝑙(1 + 𝜔𝜔2 𝜏𝜏2 2 ) + 10𝑙𝑙𝑙𝑙𝑙𝑙{(1 − 𝑢𝑢2 )2 + (2𝜁𝜁𝜁𝜁)2 }]

𝜙𝜙 = 𝑡𝑡𝑡𝑡𝑡𝑡−1 (𝜔𝜔𝜏𝜏1 ) − 𝑁𝑁(𝜋𝜋⁄2) − 𝑡𝑡𝑡𝑡𝑡𝑡−1 (𝜔𝜔𝜏𝜏2 ) − 𝑡𝑡𝑡𝑡𝑡𝑡−1 [2𝜁𝜁𝜁𝜁/(1 − 𝑢𝑢2 )]

Example
𝑇𝑇(𝑠𝑠) = 1000/[(𝑠𝑠 + 10)(𝑠𝑠 + 100)]
Find the magnitude in dB and phase angle in degrees, for 𝜔𝜔 = 0,10,200 𝑟𝑟𝑟𝑟𝑟𝑟/𝑠𝑠
𝑇𝑇(𝑗𝑗𝑗𝑗) = 1000/[(10 + 𝑗𝑗𝑗𝑗)(100 + 𝑗𝑗𝑗𝑗)]
Magnitude: 𝑑𝑑𝑑𝑑 = 20𝑙𝑙𝑙𝑙𝑙𝑙|𝑇𝑇(𝑗𝑗𝑗𝑗)|
= 20𝑙𝑙𝑙𝑙𝑙𝑙(1000) − 10𝑙𝑙𝑙𝑙𝑙𝑙(100 + 𝜔𝜔2 ) − 10𝑙𝑙𝑙𝑙𝑙𝑙(10000 + 𝜔𝜔2 )
= 60 − 10[𝑙𝑙𝑙𝑙𝑙𝑙(100 + 𝜔𝜔2 ) + 10𝑙𝑙𝑙𝑙𝑙𝑙(10000 + 𝜔𝜔2 )]
Phase angle:
𝜙𝜙 = 0 − 𝑡𝑡𝑡𝑡𝑡𝑡−1 (𝜔𝜔⁄10) − 𝑡𝑡𝑡𝑡𝑡𝑡−1 (𝜔𝜔⁄100)

𝜔𝜔 𝑑𝑑𝑑𝑑 𝜙𝜙(𝑑𝑑𝑑𝑑𝑑𝑑)

0 0 0
10 −3.05 −50.71
200 −33.02 −150.57

117
Peak value of Magnitude and Frequency Ratio for Complex poles:
For, 𝐺𝐺(𝑗𝑗𝑗𝑗) = �1�[(1 − 𝑢𝑢2 ) 𝑢𝑢 = �𝜔𝜔�𝜔𝜔𝑛𝑛 �
+ 𝑗𝑗(2𝜁𝜁𝜁𝜁)]� ;

|𝐺𝐺(𝑗𝑗𝑗𝑗)| = �1� � = �1� � = �1� �

�(1 − 𝑢𝑢2 )2 + (2𝜁𝜁𝜁𝜁)2 �{𝑢𝑢4 + 2𝑢𝑢2 (2𝜁𝜁 2 − 1) + 1} �𝛽𝛽

𝛽𝛽 = 𝑢𝑢4 + 2𝑢𝑢2 (2𝜁𝜁 2 − 1) + 1

Peak value of magnitude, 𝑀𝑀𝑝𝑝𝑝𝑝 = |𝐺𝐺|𝑚𝑚𝑚𝑚𝑚𝑚 ; (𝑑𝑑𝑑𝑑)𝑚𝑚𝑚𝑚𝑚𝑚 = 20𝑙𝑙𝑙𝑙𝑙𝑙|𝐺𝐺|𝑚𝑚𝑚𝑚𝑚𝑚 = 20𝑙𝑙𝑙𝑙𝑙𝑙𝑀𝑀𝑝𝑝𝑝𝑝
𝑑𝑑|𝐺𝐺|
For |𝐺𝐺| = |𝐺𝐺|𝑚𝑚𝑚𝑚𝑚𝑚 , =0
𝑑𝑑𝑑𝑑

𝑑𝑑|𝐺𝐺| −{4𝑢𝑢3 + 4𝑢𝑢(2𝜁𝜁 2 − 1)}

= 4 =0
𝑑𝑑𝑑𝑑 {𝑢𝑢 + 2𝑢𝑢2 (2𝜁𝜁 2 − 1) + 1}
4𝑢𝑢3 + 4𝑢𝑢(2𝜁𝜁 2 − 1) = 0
𝑢𝑢2 = 1 − 2𝜁𝜁 2
𝜔𝜔𝑟𝑟
𝑢𝑢 = �1 − 2𝜁𝜁 2 = ; 𝜔𝜔 = 𝜔𝜔𝑟𝑟 @ |𝐺𝐺| = |𝐺𝐺|𝑚𝑚𝑚𝑚𝑚𝑚
𝜔𝜔𝑛𝑛
Substitute this value of ‘𝑢𝑢’ in |𝐺𝐺|;

|𝐺𝐺|𝑚𝑚𝑚𝑚𝑚𝑚 = �1� � = 𝑀𝑀𝑝𝑝𝑝𝑝

�𝛽𝛽𝑝𝑝
𝛽𝛽𝑝𝑝 = (1 − 2𝜁𝜁 2 )2 − 2(1 − 2𝜁𝜁 2 )2 + 1 = 1 − (1 − 2𝜁𝜁 2 )2 = 4𝜁𝜁 2 (1 − 𝜁𝜁 2 )

Peak value:

Magnitude, 𝑀𝑀𝑝𝑝𝑝𝑝 = �1� � = �1� �

�𝛽𝛽𝑏𝑏 �2𝜁𝜁�1 − 𝜁𝜁 2 �
𝜔𝜔𝑟𝑟
𝑢𝑢𝑟𝑟 = = �1 − 2𝜁𝜁 2 ; 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹, 𝜔𝜔𝑟𝑟 = 𝜔𝜔𝑛𝑛 �1 − 2𝜁𝜁 2
𝜔𝜔𝑛𝑛

For 𝜻𝜻 < 𝟏𝟏
𝜁𝜁 𝑀𝑀𝑝𝑝𝑝𝑝 𝜔𝜔𝑟𝑟 ⁄𝜔𝜔𝑛𝑛

0 ∞ 1
0.1 5.025 0.99
0.2 2.55 0.96
0.707 1.0 0

118
Example

1
For 𝐺𝐺 = and 𝐾𝐾 = 2 draw the Bode Plot.
𝑠𝑠(𝑠𝑠+1)(𝑠𝑠+2)

𝑇𝑇(𝑠𝑠) = 𝐾𝐾𝐺𝐺�(1 + 𝐾𝐾𝐺𝐺)

2 2 2
𝐾𝐾𝐺𝐺 = = = 3
𝑠𝑠(𝑠𝑠 + 1)(𝑠𝑠 + 2) 𝑠𝑠(𝑠𝑠 + 1)(𝑠𝑠 + 2) + 2 (𝑠𝑠 + 3𝑠𝑠 + 2𝑠𝑠 + 2)
2

2 2
𝑇𝑇(𝑗𝑗𝑗𝑗) = =
[(−𝑗𝑗 2 )𝑗𝑗𝑗𝑗 − 3𝑗𝑗 + 2(𝑗𝑗𝑗𝑗) + 2] [(2 − 3𝑗𝑗 ) + 𝑗𝑗𝑗𝑗(2 − 𝑗𝑗 2 )]
2 2

Magnitude, 𝑑𝑑𝐵𝐵 = 20𝑙𝑙𝑙𝑙𝑙𝑙2 − 10𝑙𝑙𝑙𝑙𝑙𝑙{(2 − 3𝑗𝑗2 )2 + 𝑗𝑗2 (2 − 𝑗𝑗2 )2 }

𝐼𝐼�2−𝐼𝐼2 �
Phase Angle, 𝜙𝜙 = 0 − 𝑡𝑡𝑡𝑡𝑡𝑡−1 (2−3𝐼𝐼2 )

𝑗𝑗 dB 𝜙𝜙(𝑑𝑑𝐵𝐵𝑙𝑙)

0.1 0.09 −6
0.8 5 −86
1.0 3 −135
2.0 −14.6 −202

Example:
For the same example above, find 𝑀𝑀𝑝𝑝𝐼𝐼 , 𝜁𝜁, 𝑃𝑃. 𝑂𝑂. , 𝑇𝑇𝑠𝑠

(𝑑𝑑𝐵𝐵)𝑚𝑚𝑚𝑚𝑚𝑚 = 5; 𝑗𝑗𝑟𝑟 = 0.8 (frequency at 𝑑𝑑𝐵𝐵 = (𝑑𝑑𝐵𝐵)𝑚𝑚𝑚𝑚𝑚𝑚 )

(𝑑𝑑𝐵𝐵)𝑚𝑚𝑚𝑚𝑚𝑚 = 20𝑙𝑙𝑙𝑙𝑙𝑙|𝑇𝑇|𝑚𝑚𝑚𝑚𝑚𝑚 = 20𝑙𝑙𝑙𝑙𝑙𝑙𝑀𝑀𝑝𝑝𝐼𝐼 = 5𝑑𝑑𝐵𝐵; 𝑙𝑙𝑙𝑙𝑙𝑙𝑀𝑀𝑝𝑝𝐼𝐼 = 5/20 = 0.25

𝑀𝑀𝑝𝑝𝐼𝐼 = (10)0.25 = 1.8

119
 1
𝑀𝑀𝑝𝑝𝑝𝑝 = = 1.8
2𝜁𝜁�1 − 𝜁𝜁 2

𝜁𝜁 4 − 𝜁𝜁 2 + 0.077 = 0
𝜁𝜁 = 0.29, 0.96
Choose 𝜁𝜁 = 0.29 since 𝜁𝜁 < 1; the other value is close to one.
Maximum value of 5 𝑑𝑑𝑑𝑑 occurs at 𝜔𝜔 = 0.8; 𝜔𝜔𝑟𝑟 = 0.8.
𝜔𝜔𝑟𝑟 0.8
𝜔𝜔𝑛𝑛 = = = 0.88
�1 − 2𝜁𝜁 2 �1 − 2(0.29)2
−𝜁𝜁𝜁𝜁
� � �
𝑃𝑃. 𝑂𝑂. = 100𝑒𝑒 �1−𝜁𝜁 2 = 38.6%
4
𝑇𝑇𝑠𝑠 = = 15.67𝑠𝑠𝑠𝑠𝑠𝑠
𝜁𝜁𝜔𝜔𝑛𝑛

System Bandwidth Frequency (𝝎𝝎𝑩𝑩 )

Bandwidth frequency is defined as the frequency at which the magnitude is equal to
−3 𝑑𝑑𝐵𝐵. Since one octave is 6 𝑑𝑑𝑑𝑑, the bandwidth frequency is also defined as the
frequency at which the frequency response is declined one half octave from zero 𝑑𝑑𝑑𝑑.

Example
120
𝑇𝑇(𝑠𝑠) =
𝑠𝑠(𝑠𝑠 + 1)(𝑠𝑠 + 2)
Find the bandwidth frequency, 𝜔𝜔𝐵𝐵 and respective phase angle 𝜙𝜙𝐵𝐵

120
𝑇𝑇(𝑗𝑗𝑗𝑗) =
(𝑗𝑗𝑗𝑗)(𝑗𝑗𝑗𝑗 + 1)(𝑗𝑗𝑗𝑗 + 2)

𝑑𝑑𝑑𝑑 = 20𝑙𝑙𝑙𝑙𝑙𝑙|𝑇𝑇| = 20𝑙𝑙𝑙𝑙𝑙𝑙(120) − 20𝑙𝑙𝑙𝑙𝑙𝑙(𝜔𝜔) − 10𝑙𝑙𝑙𝑙𝑙𝑙(𝜔𝜔2 + 1) − 10𝑙𝑙𝑙𝑙𝑙𝑙(𝜔𝜔2 + 4)

𝜙𝜙 = 0 − �𝜋𝜋�2� − 𝑡𝑡𝑡𝑡𝑡𝑡−1 (𝜔𝜔) − 𝑡𝑡𝑡𝑡𝑡𝑡−1 �𝜔𝜔�2�

120
 𝑗𝑗 𝑑𝑑𝐵𝐵 𝜙𝜙(𝑑𝑑𝐵𝐵𝑙𝑙)

3 10.9 −217.9
4 4.25 −229.4
5 −1.15 −236.9
6 −5.66 −242
7 −9.53 −246

𝑑𝑑𝐵𝐵 𝑗𝑗

−1.15 5
−3 ?
−5.66 6

Linear Interpolation
𝑌𝑌 = 𝑌𝑌1 + [(𝑋𝑋 − 𝑋𝑋1 )(𝑌𝑌2 − 𝑌𝑌1 )/(𝑋𝑋2 − 𝑋𝑋1 )]
= 5 + [(−1.85)(1)/(−4.51)] = 5.4

The bandwidth frequency 𝑗𝑗𝐵𝐵 = 5.4

The phase angle 𝜙𝜙𝐵𝐵 = 𝜙𝜙 at 𝑗𝑗 = 𝑗𝑗𝐵𝐵
= −90 − 𝑡𝑡𝑡𝑡𝑡𝑡−1 (5.4) − 𝑡𝑡𝑡𝑡𝑡𝑡−1 (2.7) = −90 − 79.5 − 69.7
𝜙𝜙𝐵𝐵 = −239°

121
Second order system – Bandwidth Frequency and Damping ratio
For a second order system,
2
𝜔𝜔𝑛𝑛 2
𝜔𝜔𝑛𝑛
𝑇𝑇(𝑠𝑠) = 2 ; 𝜁𝜁 < 1; 𝑇𝑇(𝑗𝑗𝜔𝜔𝐵𝐵 ) = 2 2
𝑠𝑠 2 +(2𝜁𝜁𝜔𝜔𝑛𝑛 )𝑠𝑠+𝜔𝜔𝑛𝑛 �𝜔𝜔𝑛𝑛 −𝜔𝜔𝐵𝐵 �+𝑗𝑗2𝜁𝜁𝜔𝜔𝑛𝑛 𝜔𝜔𝐵𝐵

1 𝜔𝜔𝐵𝐵
= ; where 𝑢𝑢 = �𝜔𝜔𝑛𝑛
(1 − 𝑢𝑢2 )+ 𝑖𝑖(2𝜁𝜁𝜁𝜁)
1
|𝑇𝑇| =
�(1 − 𝑢𝑢2 )2 + (2𝜁𝜁𝜁𝜁)2
3
At 20𝑙𝑙𝑙𝑙𝑙𝑙|𝑇𝑇| = −3 𝑑𝑑𝑑𝑑, 𝑙𝑙𝑙𝑙𝑙𝑙|𝑇𝑇| = − � � ; |𝑇𝑇| = 0.7079
20

(1 − 𝑢𝑢2 )2 + 4𝜁𝜁 2 𝑢𝑢2 = 2

𝑢𝑢4 + (4𝜁𝜁 2 − 2)𝑢𝑢2 − 1 = 0

𝑢𝑢2 = (1 − 2𝜁𝜁 2 ) + �(2𝜁𝜁 2 − 1)2 + 1

(1⁄2) 𝜔𝜔𝐵𝐵
𝑢𝑢 = �(1 − 2𝜁𝜁 2 ) + �4𝜁𝜁 4 − 4𝜁𝜁 2 + 2� =
𝜔𝜔𝑛𝑛
(1⁄2)
𝜔𝜔𝐵𝐵 = 𝜔𝜔𝑛𝑛 �(1 − 2𝜁𝜁 2 ) + �4𝜁𝜁 4 − 4𝜁𝜁 2 + 2� = 𝑓𝑓(𝜔𝜔𝑛𝑛 , 𝜁𝜁)

For given Percent Overshoot and Settling time, the value of 𝜁𝜁 can be determined from
Percent Overshoot while the value of 𝜔𝜔𝑛𝑛 can be determined from Settling time. Then the
bandwidth frequency (𝜔𝜔𝐵𝐵 ) can be easily determined from the above equation.

Example
81
Given 𝑇𝑇(𝑠𝑠) =
𝑠𝑠 2 +9𝑠𝑠+81

Find 𝜔𝜔𝐵𝐵 , 𝑀𝑀𝑝𝑝𝑝𝑝 and 𝜔𝜔𝑟𝑟

Characteristic equation: 𝑠𝑠 2 + 9𝑠𝑠 + 81 = 0

Compare with 𝑠𝑠 2 + (2𝜁𝜁𝜔𝜔𝑛𝑛 )𝑠𝑠 + 𝜔𝜔𝑛𝑛2 = 0
𝜔𝜔𝑛𝑛2 = 81; 𝜔𝜔𝑛𝑛 = 9
2𝜁𝜁𝜔𝜔𝑛𝑛 = 9; 𝜁𝜁 = 0.5
(1⁄2)
Bandwidth frequency, 𝜔𝜔𝐵𝐵 = 𝜔𝜔𝑛𝑛 �(1 − 2𝜁𝜁 2 ) + �4𝜁𝜁 4 − 4𝜁𝜁 2 + 2� = 11.45
1
Peak value of Magnitude, 𝑀𝑀𝑝𝑝𝑝𝑝 = = 1.15
2𝜁𝜁�1−𝜁𝜁 2

Peak value frequency, 𝜔𝜔𝑟𝑟 = 𝜔𝜔𝑛𝑛 �1 − 2𝜁𝜁 2 = 6.36

122
Nyquist Stability Criterion
It is based on mapping contours from S – plane to 𝐹𝐹(𝑠𝑠) plane.

𝐺𝐺
𝑇𝑇(𝑠𝑠) =
1 + 𝐺𝐺𝐺𝐺
𝑁𝑁(𝑠𝑠)
Characteristic equation, 𝐹𝐹(𝑠𝑠) = 1 + 𝐺𝐺𝐺𝐺 = 1 + 𝐿𝐿(𝑠𝑠); 𝐿𝐿(𝑠𝑠) = 𝐺𝐺(𝑠𝑠)𝐺𝐺(𝑠𝑠) =
𝐷𝐷(𝑠𝑠)

𝐷𝐷 + 𝑁𝑁
𝐹𝐹(𝑠𝑠) =
𝐷𝐷
The poles of 𝐹𝐹(𝑠𝑠) are same as the poles of 𝐿𝐿(𝑠𝑠). Nyquist plot is a polar plot of magnitude
and angle of 𝐿𝐿(𝑗𝑗𝑗𝑗). 𝐿𝐿(𝑠𝑠) = 𝐹𝐹(𝑠𝑠) − 1; 𝐿𝐿(𝑠𝑠) = −1 for 𝐹𝐹(𝑠𝑠) = 0.
The Nyquist Stability criterion states that a feedback system is stable if and only if the
contour of the plot does not encircle the (−1,0) point.

Example
1000
Sketch a Nyquist plot for given 𝐿𝐿(𝑠𝑠) = (𝑠𝑠+1)(𝑠𝑠+10)

1000
𝐿𝐿(𝑠𝑠) =
𝑠𝑠 2 + 11𝑠𝑠 + 10
1000
𝐿𝐿(𝑗𝑗𝑗𝑗) =
[(10 − 𝑗𝑗 2 ) + 𝑗𝑗(11𝑗𝑗)]
1000
|𝐿𝐿(𝑗𝑗𝑗𝑗)| =
�(10 − 𝑗𝑗 2 )2 + 121𝑗𝑗 2
11𝑗𝑗
∠𝐿𝐿(𝑗𝑗𝑗𝑗) = 0 − 𝑡𝑡𝑡𝑡𝑡𝑡−1 � �
10 − 𝑗𝑗 2
Polar Plot of |𝑳𝑳| and ∠𝑳𝑳
𝐿𝐿(𝑗𝑗𝑗𝑗) = 𝑓𝑓(𝑟𝑟, 𝜃𝜃) where 𝑟𝑟 = |𝐿𝐿|𝑡𝑡𝑡𝑡𝑑𝑑𝜃𝜃 = ∠𝐿𝐿

123
 𝑗𝑗 𝑟𝑟 𝜃𝜃(𝑑𝑑𝐵𝐵𝑙𝑙)

0 100 0
1 70.35 −50.71
10 6.8 −129.3
100 0.1 −173.7
∞ 0 −180

The negative sign in the angle represent they are measured in clockwise direction. The
plot is symmetric. MATLAB can be used to generate the plot. The procedure for manual
plot is given below.

The contour does not encircle the point (−1,0) and hence the system is stable.

124
Example
𝐾𝐾 𝐾𝐾
𝐿𝐿(𝑠𝑠) = 2
= 3
𝑠𝑠(𝑠𝑠 + 1) 𝑠𝑠 + 2𝑠𝑠 2 + 𝑠𝑠
𝐾𝐾
𝐿𝐿(𝑗𝑗𝑗𝑗) =
(−2𝑗𝑗 2 ) + 𝑗𝑗(𝑗𝑗 − 𝑗𝑗 3 )
𝐾𝐾
|𝐿𝐿| =
�4𝑗𝑗 2 + (𝑗𝑗 − 𝑗𝑗 3 )2

𝑗𝑗(1 − 𝑗𝑗2 ) 1 − 𝑗𝑗2

∠𝐿𝐿 = 0 − � � = tan−1 � �
−2𝑗𝑗 2𝑗𝑗

Nyquist Plot for different value of ‘𝐾𝐾’ are given below.

The magnitude of 𝐿𝐿(𝑗𝑗𝑗𝑗) at ∠𝐿𝐿 = −180° can be determined by equating

−1
1 − 𝑗𝑗2
𝑡𝑡𝑡𝑡𝑡𝑡 � � = −180°
2𝑗𝑗
(1 − 𝑗𝑗2 )
= 𝑡𝑡𝑡𝑡𝑡𝑡(−180) = 0; 𝑗𝑗 = 1
2𝑗𝑗
Then 𝐿𝐿 = 𝐾𝐾/(−2)
At 𝐾𝐾 = 1, 𝐿𝐿 = −(1/2); the contour does not encircle the point (−1,0) and the system is
stable.
𝐾𝐾 = 2, 𝐿𝐿 = −1; the contour cut through the point (−1, 0) and the system is marginally
stable.
𝐾𝐾 = 3, 𝐿𝐿 = −1.5; the contour encircle the point (−1,0) and the system is unstable.

125
Relative Stability

In time domain, a system with lower value of settling time (𝑇𝑇𝑠𝑠 ) is considered relatively
more stable than a system with higher value of settling time. In frequency domain, the
parameters used to measure relative stability are Gain Margin (𝐺𝐺𝐺𝐺) and Phase Margin
(𝑃𝑃𝑃𝑃).

Gain Margin is defined as the increase in system gain at a frequency when the phase
angle, 𝜙𝜙 = −180° that will result in a marginally stable system with intersection of
(−1,0) point on the Nyquist diagram.

∠𝐿𝐿 = tan−1 𝜙𝜙 = 0; 𝜙𝜙 = ±180° or 0°

However, (−1,0) point lies on the redial line, 𝜙𝜙 = −180°.
So, from Bode plot the gain margin can be obtained as magnitude in 𝑑𝑑𝑑𝑑
when 𝜙𝜙 = −180°.
𝐺𝐺𝐺𝐺 = 20𝑙𝑙𝑙𝑙𝑙𝑙|𝐿𝐿| at the frequency 𝜔𝜔 , when 𝜙𝜙 = −180°.

Phase margin is defined as the amount of phase shift at the cross over frequency when the
magnitude is 𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧 𝑑𝑑𝑑𝑑 (20𝑙𝑙𝑙𝑙𝑙𝑙|𝐿𝐿| = 0) that will result in a marginally stable system with
intersection of (−1,0) point on the Nyquist diagram.
At |𝐿𝐿| = 1, 20𝑙𝑙𝑙𝑙𝑙𝑙|𝐿𝐿| = 0 𝑑𝑑𝑑𝑑
The gain cross-over frequency, 𝜔𝜔𝑐𝑐 = 𝜔𝜔 at |𝐿𝐿| = 1. So, from Bode plot the phase margin
can be obtained as the phase angle when 20𝑙𝑙𝑙𝑙𝑙𝑙|𝐿𝐿| = 0.
𝑃𝑃𝑃𝑃 = 𝜙𝜙 at the gain cross-over frequency, 𝜔𝜔𝑐𝑐 .
A typical Bode Plot with 𝑃𝑃𝑃𝑃 and 𝐺𝐺𝐺𝐺 is given below.

126
Computation of 𝑮𝑮𝑮𝑮 and 𝑷𝑷𝑮𝑮
Given the transfer function, 𝐿𝐿(𝑠𝑠); determine 𝐿𝐿(𝑗𝑗𝑗𝑗) and then find magnitude (20𝑙𝑙𝑙𝑙𝑙𝑙|𝐿𝐿|)
and the angle ∠𝐿𝐿.
Gain Margin (𝑑𝑑𝐵𝐵): Set ∠𝐿𝐿 = −180 and find 𝑗𝑗 = 𝑗𝑗0 , the phase cross-over frequency.
Then 𝐺𝐺𝑀𝑀 = 20𝑙𝑙𝑙𝑙𝑙𝑙|𝐿𝐿| evaluated at 𝑗𝑗0 .
Phase Margin (𝑑𝑑𝐵𝐵g): Set |𝐿𝐿| = 1 and find 𝑗𝑗 = 𝑗𝑗𝑐𝑐 , the gain cross-over frequency. Then
𝑃𝑃𝑀𝑀 = ∠𝐿𝐿 evaluated at 𝑗𝑗𝑐𝑐 .

Example
𝐾𝐾
𝐿𝐿(𝑠𝑠) =
(𝑠𝑠 + 2)(𝑠𝑠 + 3)
a) Find the 𝐺𝐺𝑀𝑀 for 𝐾𝐾 = 1
b) Find ‘𝐾𝐾’ for given 𝐺𝐺𝑀𝑀 of −3 𝑑𝑑𝐵𝐵
c) Find 𝑃𝑃𝑀𝑀 for 𝐾𝐾 = 10
𝐾𝐾
𝐿𝐿(𝑠𝑠) =
𝑠𝑠 2 + 5𝑠𝑠 + 6

127
 𝐾𝐾
𝐿𝐿(𝑗𝑗𝑗𝑗) =
[(6 − 𝜔𝜔 2 )+ 𝑗𝑗(5𝜔𝜔)]
𝐾𝐾 5𝜔𝜔
|𝐿𝐿| = ; ∠𝐿𝐿 = − 𝑡𝑡𝑡𝑡𝑡𝑡−1
�(6 − 𝜔𝜔 2 )2 + 25𝜔𝜔 2 (6 − 𝜔𝜔 2 )

(6 − 𝜔𝜔2 )2 + 25𝜔𝜔2 = 𝜔𝜔4 + 13𝜔𝜔2 + 36 = (𝜔𝜔2 + 4)(𝜔𝜔2 + 9)

𝐾𝐾
|𝐿𝐿| =
[(𝜔𝜔 2 + 4)(𝜔𝜔 2 + 9)](1⁄2)
𝑑𝑑𝑑𝑑 = 20𝑙𝑙𝑙𝑙𝑙𝑙|𝐿𝐿| = 20𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 − 10𝑙𝑙𝑙𝑙𝑙𝑙(𝜔𝜔2 + 4) − 10𝑙𝑙𝑙𝑙𝑙𝑙(𝜔𝜔2 + 9)
(a) 𝐾𝐾 = 1
5𝜔𝜔
∠𝐿𝐿 = 0; 𝑡𝑡𝑡𝑡𝑡𝑡−1 = 0 ; 𝜔𝜔0 = 0
(6 − 𝜔𝜔 2 )
𝑑𝑑𝑑𝑑 = 0 − 10𝑙𝑙𝑙𝑙𝑙𝑙4 − 10𝑙𝑙𝑙𝑙𝑙𝑙9 = −15.56
𝐺𝐺𝐺𝐺 = −15.56 𝑑𝑑𝑑𝑑
(b) 𝐺𝐺𝐺𝐺 = −3 𝑑𝑑𝑑𝑑
𝐺𝐺𝐺𝐺 = 20𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 − 10𝑙𝑙𝑙𝑙𝑙𝑙4 − 10𝑙𝑙𝑙𝑙𝑙𝑙9
−3 = 20𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 − 15.56
20𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 = 12.56; 𝐾𝐾 = 4.246

(c) K=10
𝐾𝐾
|𝐿𝐿| = =1
[𝜔𝜔 4 + 13𝜔𝜔 2 + 36](1⁄2)
𝐾𝐾 2 = 𝜔𝜔4 + 13𝜔𝜔2 + 36 = 100
𝜔𝜔4 + 13𝜔𝜔2 − 64 = 0
1
𝜔𝜔2 = � � �−13 ± √425� = 3.8, −16.8
2
Since 𝜔𝜔 is positive, 𝜔𝜔𝑐𝑐 = √3.8 = 1.95
5𝜔𝜔𝑐𝑐
∠𝐿𝐿 = − 𝑡𝑡𝑡𝑡𝑡𝑡−1 = −77°
(6 − 𝜔𝜔𝑐𝑐 2 )
𝑃𝑃𝑃𝑃 = −77°

128
Phase Margin and Damping
2
𝜔𝜔𝑛𝑛
For a second order system, 𝐿𝐿(𝑠𝑠) =
𝑠𝑠(𝑠𝑠+2𝜁𝜁𝜔𝜔𝑛𝑛 )

𝜔𝜔𝑛𝑛2
𝐿𝐿(𝑗𝑗𝑗𝑗) =
(−𝜔𝜔 2 ) + 𝑗𝑗(2𝜁𝜁𝜔𝜔𝑛𝑛 𝜔𝜔)
𝜔𝜔𝑛𝑛2
|𝐿𝐿| = 4
[𝜔𝜔 + (4𝜁𝜁 2 𝜔𝜔𝑛𝑛2 )𝜔𝜔 2 ](1⁄2)
2𝜁𝜁𝜔𝜔𝑛𝑛 𝜔𝜔 2𝜁𝜁𝜔𝜔𝑛𝑛
∠𝐿𝐿 = − 𝑡𝑡𝑡𝑡𝑡𝑡−1 � � = 𝑡𝑡𝑡𝑡𝑡𝑡 −1
� �
−𝜔𝜔 2 𝜔𝜔
𝜔𝜔 = 𝜔𝜔𝑐𝑐 at |𝐿𝐿| = 1
𝜔𝜔4 + (4𝜁𝜁 2 𝜔𝜔𝑛𝑛2 )𝜔𝜔2 − 𝜔𝜔𝑛𝑛4 = 0
Solving,

𝜔𝜔2 = 𝜔𝜔𝑛𝑛2 �−2𝜁𝜁 2 ± �4𝜁𝜁 4 + 1�

(1/2)
𝜔𝜔𝑐𝑐 = 𝜔𝜔𝑛𝑛 �−2𝜁𝜁 2 + �4𝜁𝜁 4 + 1�

2𝜁𝜁𝜔𝜔𝑛𝑛 2𝜁𝜁
𝑃𝑃𝑃𝑃 = 𝑡𝑡𝑡𝑡𝑡𝑡−1 � � = 𝑡𝑡𝑡𝑡𝑡𝑡−1 � (1/2)
�
𝜔𝜔𝑐𝑐
�−2𝜁𝜁 2 + �4𝜁𝜁 4 + 1�

Linear approximation, 𝑃𝑃𝑃𝑃 = 100𝜁𝜁

𝜁𝜁 = 0.01(𝑃𝑃𝑃𝑃) = 0.01𝜙𝜙
This approximation is good for 10° ≤ 𝜙𝜙 ≤ 70°
It results in a damping range of 0.1 ≤ 𝜁𝜁 ≤ 0.707

129
 Assessment
1. Bode plot is,
a) A logarithmic plot
b) A polar plot
c) A linear plot
d) None of the above

2. Nyquist plot is
a) A logarithmic plot
b) A polar plot
c) A linear plot
d) None of the above

1
3. If 𝑇𝑇 = , the magnitude |𝑇𝑇| is
3+𝑗𝑗4

a) 1/5
b) 1/7
c) 1/25
d) 25

4. The angle, ∠𝑇𝑇 is

a) 37°
b) −37°
c) −53°
d) 53°

130
5. If 𝑇𝑇 = 4, ∠𝑇𝑇 is
a) 90°
b) 0°
c) 76°
d) 0.07°

6. If 𝑇𝑇 = 𝑗𝑗4, ∠𝑇𝑇 is
a) 90°
b) 0°
c) 76°
d) 0.07°

7. Bandwidth frequency is the frequency at

a) Magnitude = −3 𝑑𝑑𝑑𝑑
b) Magnitude = −(1/2)(𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜)
c) |𝑇𝑇| = 0.71
d) All of the above

8. If |𝑇𝑇| = −1𝑑𝑑𝑑𝑑 @ 𝜔𝜔 = 5𝑎𝑎𝑎𝑎𝑎𝑎 − 5𝑑𝑑𝑑𝑑 @ 𝜔𝜔 = 6, the bandwidth frequency is,

a) 11
b) 5/6
c) 5.5
d) 6/5

131
9. Phase margin is the phase angle evaluated at
a) Crossover frequency
b) Magnitude is 𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧 𝑑𝑑𝑑𝑑
c) Magnitude = 1
d) All of the above

10. Gain margin is the magnitude in 𝑑𝑑𝑑𝑑 evaluated at

a) Frequency at 𝜙𝜙 = −180°
b) Frequency at 𝜙𝜙 = 90°
c) Frequency at 𝜙𝜙 = −90°
d) Frequency at 𝜙𝜙 = 180°

132
 9. Design of control system
Learning Objectives:
1) To design a lead compensator.
2) To design a pre-filter.
3) To design a Lag compensator.

Compensator Design:
Compensator (𝐺𝐺𝑐𝑐 ) is an additional component added to the control system to compensate
for deficiency in performance. There are four types of compensators based on where it is
added in the loop of control system.
(i) Cascade Type: The compensator is added in the Feed-Forward loop as shown
below.
𝐺𝐺𝑐𝑐 ⟶ Compensator T.F
𝐺𝐺 ⟶Plant T.F.

(ii) Feedback Type: The compensator is added in the Feed-Back loop as shown
below.

(iii) Output Type:

The compensator is added in the output part of the loop, after the pick-off point as shown
below.

133
 (iv) Input Type:
The compensator is added in the input part of the loop, before the summing point as
shown below.

However, cascade type is commonly used in the design of a compensator.

𝐾𝐾(𝑠𝑠+𝑧𝑧)
Let 𝐺𝐺𝑐𝑐 (𝑠𝑠) = (𝑠𝑠+𝑝𝑝)
for lead compensator.

𝐾𝐾(𝑗𝑗𝑗𝑗 + 𝑧𝑧)
𝐺𝐺𝑐𝑐 (𝑗𝑗𝑗𝑗) =
(𝑗𝑗𝑗𝑗 + 𝑝𝑝)
𝑗𝑗
𝐾𝐾 �1 + 𝑗𝑗 � (𝑧𝑧)
= 𝑧𝑧
𝑗𝑗
�1 + 𝑗𝑗 � (𝑝𝑝)
𝑝𝑝
1 1
Let = 𝜏𝜏, and = 𝛼𝛼𝜏𝜏
𝑝𝑝 𝑧𝑧

𝑧𝑧 1 𝑧𝑧 𝐾𝐾
= ; 𝐾𝐾 � � =
𝑝𝑝 𝛼𝛼 𝑝𝑝 𝛼𝛼
𝐾𝐾
Let 𝐾𝐾1 =
𝛼𝛼

𝐾𝐾1 (1+𝑗𝑗𝑗𝑗𝛼𝛼𝑗𝑗)
So, 𝐺𝐺𝐶𝐶 (𝑗𝑗𝑗𝑗) =
(1+𝑗𝑗𝑗𝑗𝑗𝑗)

134
Lead Compensator Design
Design Parameters (𝜶𝜶, 𝝉𝝉)
Let 𝐾𝐾1 = 1
1 1
[1 + 𝑗𝑗𝑗𝑗(𝛼𝛼𝛼𝛼)] 1 + (𝛼𝛼𝛼𝛼)𝑠𝑠 𝛼𝛼𝛼𝛼(𝑠𝑠 + 𝛼𝛼𝛼𝛼) 𝛼𝛼(𝑠𝑠 + 𝛼𝛼𝛼𝛼)
𝐺𝐺𝐶𝐶 (𝑗𝑗𝑗𝑗) = ; 𝐺𝐺𝐶𝐶 (𝑠𝑠) = = =
[1 + 𝑗𝑗𝑗𝑗(𝜏𝜏)] 1 + (𝜏𝜏)𝑠𝑠 1 1
𝜏𝜏(𝑠𝑠 + ) (𝑠𝑠 + )
𝜏𝜏 𝜏𝜏
√1 + 𝛼𝛼 2 𝜔𝜔 2 𝜏𝜏 2
|𝐺𝐺𝑐𝑐 | =
√1 + 𝜔𝜔 2 𝜏𝜏 2
�1+𝛼𝛼 2 𝜔𝜔2 𝜏𝜏2 �
Magnitude, 𝑀𝑀 𝑖𝑖𝑖𝑖 𝑑𝑑𝑑𝑑 = 20𝑙𝑙𝑙𝑙𝑙𝑙|𝐺𝐺𝑐𝑐 | = 10𝑙𝑙𝑙𝑙𝑙𝑙 � (1+𝜔𝜔2 𝜏𝜏2 )
�

1 + 𝛼𝛼 2 𝜔𝜔2 𝜏𝜏 2
𝑀𝑀 = 10𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙; 𝑤𝑤ℎ𝑒𝑒𝑒𝑒𝑒𝑒 𝐶𝐶 =
1 + 𝜔𝜔 2 𝜏𝜏 2
𝑀𝑀� �
𝐶𝐶 = (10)� 10

1 + 𝛼𝛼 2 𝜔𝜔2 𝜏𝜏 2 = 𝐶𝐶(1 + 𝜔𝜔2 𝜏𝜏 2 )

𝜏𝜏 2 {𝜔𝜔2 (𝛼𝛼 2 − 𝐶𝐶)} = 𝐶𝐶 − 1

1 𝐶𝐶−1
At 𝜔𝜔 = 𝜔𝜔𝑐𝑐 , 𝜏𝜏 = �
𝜔𝜔𝑐𝑐 𝛼𝛼2 −𝐶𝐶

Phase, 𝜙𝜙 = ∠𝐺𝐺𝐶𝐶 = 𝑡𝑡𝑡𝑡𝑡𝑡−1 (𝛼𝛼𝛼𝛼𝛼𝛼) − 𝑡𝑡𝑡𝑡𝑡𝑡−1 (𝜔𝜔𝜔𝜔)

𝐴𝐴−𝐵𝐵
In general, 𝑡𝑡𝑡𝑡𝑡𝑡−1 (𝐴𝐴) − 𝑡𝑡𝑡𝑡𝑡𝑡−1 (𝐵𝐵) = 𝑡𝑡𝑡𝑡𝑡𝑡−1 � �
1+𝐴𝐴𝐴𝐴

𝜔𝜔𝜔𝜔(𝛼𝛼−1)
So, 𝜙𝜙 = 𝑡𝑡𝑡𝑡𝑡𝑡−1 � �
1+𝛼𝛼𝜔𝜔2 𝜏𝜏2

𝜔𝜔𝜔𝜔(𝛼𝛼 − 1)
𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 =
1 + 𝛼𝛼𝜔𝜔 2 𝜏𝜏 2
1
At 𝜔𝜔 = 𝜔𝜔𝑚𝑚𝑚𝑚𝑚𝑚 , 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 = ∞ ⇒ 1 + 𝛼𝛼𝜔𝜔2 𝜏𝜏 2 = 0; |𝜔𝜔𝜔𝜔| =
√𝛼𝛼
𝛼𝛼 − 1 1 (𝛼𝛼 − 1)
𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 = � �� �=
√𝛼𝛼 1+1 2√𝛼𝛼
𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 (𝛼𝛼 − 1) 2√𝛼𝛼 𝛼𝛼 − 1
𝑠𝑠𝑠𝑠𝑠𝑠 𝜙𝜙 = = =
�1 + 𝑡𝑡𝑡𝑡𝑡𝑡2 (𝜙𝜙) 2√𝛼𝛼 (𝛼𝛼 + 1) 𝛼𝛼 + 1
1+𝑠𝑠𝑠𝑠𝑠𝑠 𝜙𝜙
So, we can find, 𝛼𝛼 =
1−𝑠𝑠𝑠𝑠𝑠𝑠 𝜙𝜙

135
 1 + 𝛼𝛼 2 𝜔𝜔2 𝜏𝜏 2
𝑐𝑐 =
1 + 𝜔𝜔 2 𝜏𝜏 2
1
Substituting 𝜔𝜔2 𝜏𝜏 2 =
𝛼𝛼

1 + 𝛼𝛼
𝐶𝐶 = = 𝛼𝛼
1
1+� �
𝛼𝛼
So, 𝑀𝑀 = 10𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙; it is rounded off to next higher integer.

Design Process
Given the percent Overshoot (P.O.) and settling time (𝑇𝑇𝑠𝑠 ), we can find the values of 𝜁𝜁,
and 𝜔𝜔𝑛𝑛 .
𝛿𝛿 100
𝜁𝜁 = ; 𝛿𝛿 = 𝑙𝑙𝑙𝑙 � �
√𝛿𝛿 2 + 𝜋𝜋 2 𝑃𝑃. 𝑂𝑂.
4
𝜔𝜔𝑛𝑛 =
𝜁𝜁𝑇𝑇𝑠𝑠
𝜙𝜙 = 100𝜁𝜁 = Phase margin
1 + 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
𝛼𝛼 =
1 − 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
𝑀𝑀 = 10𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 (Round-off to next higher integer)
𝑀𝑀� �
𝐶𝐶 = (10)� 10 ; it is not equal to 𝛼𝛼 since ‘𝑀𝑀’ is rounded off to an integer.

1 𝐶𝐶 − 1
𝜏𝜏 = �
𝜔𝜔𝑐𝑐 𝛼𝛼 2 − 𝐶𝐶

𝜔𝜔𝑐𝑐 = Cross-over frequency and either it is given or can be found from Bode Plot
1
𝛼𝛼�𝑠𝑠+ �
Compensator, 𝐺𝐺𝑐𝑐 = 𝛼𝛼𝛼𝛼
1
�𝑠𝑠+ �
𝜏𝜏

1
Zero: 𝑠𝑠 = −
𝛼𝛼𝛼𝛼
1
Polo: 𝑠𝑠 = −
𝜏𝜏

136
Design check
𝜙𝜙 > 0; 𝑀𝑀 > 0; 𝑙𝑙 > 𝑝𝑝2 + 1; 𝑝𝑝 = 𝑡𝑡𝑡𝑡𝑖𝑖𝜙𝜙
Pole should be greater than (−𝜁𝜁𝑗𝑗𝑠𝑠 ) and it should be in the desired region.
If the pole requirement is not satisfied, change the frequency (ωc ) and repeat the process.
For a second order system, characteristic equation is 𝑠𝑠 2 + (2𝜁𝜁𝑗𝑗𝑠𝑠 )𝑠𝑠 + 𝑗𝑗𝑠𝑠2 = 0

For 𝜁𝜁 < 1, Roots (Poles) are 𝑠𝑠1,2 = −𝜁𝜁𝑗𝑗𝑠𝑠 ± 𝑗𝑗𝑗𝑗𝑠𝑠 �1 − 𝜁𝜁 2 = −𝑡𝑡 ± 𝑗𝑗𝑗𝑗
|𝑡𝑡| 𝜁𝜁𝑗𝑗𝑠𝑠
𝑙𝑙𝑙𝑙𝑠𝑠𝐶𝐶 = = = 𝜁𝜁
𝑙𝑙𝐴𝐴 𝑗𝑗𝑠𝑠

𝑙𝑙𝐴𝐴 = �𝑡𝑡2 + 𝑗𝑗 2 = 𝑗𝑗𝑠𝑠

𝐶𝐶 = 𝑐𝑐𝑙𝑙𝑠𝑠 −1 𝜁𝜁

Shaded area represents desired region for performance of stability

Example
Design a compensator for given 𝜁𝜁 = 0.45, 𝑇𝑇𝑠𝑠 = 4 𝑠𝑠𝑒𝑒𝑐𝑐 and 𝑗𝑗𝑐𝑐 = 5 𝑒𝑒𝑡𝑡𝑑𝑑/𝑠𝑠

𝜙𝜙 = 100𝜁𝜁 = 45°
4 4
𝑗𝑗𝑠𝑠 = = = 2.22
𝜁𝜁𝑇𝑇𝑠𝑠 0.45 × 4

137
 1 + 𝑠𝑠𝑖𝑖𝑖𝑖𝜙𝜙
𝛼𝛼 = = 5.83
1 − 𝑠𝑠𝑖𝑖𝑖𝑖𝜙𝜙
𝑀𝑀 = 10𝑙𝑙𝑙𝑙𝑙𝑙𝛼𝛼 = 10𝑙𝑙𝑙𝑙𝑙𝑙(5.83) = 7.66𝑑𝑑𝑑𝑑 ≈ 8𝑑𝑑𝑑𝑑 (rounded off to next higher value)
𝑀𝑀� �
𝑙𝑙 = (10)� 10 = (10)0.8 = 6.31

1 𝑙𝑙 − 1 1 5.31
𝜏𝜏 = � 2 = � = 0.087
𝑗𝑗𝑐𝑐 𝛼𝛼 − 𝑙𝑙 5 (5.832 − 6.31)

1
𝛼𝛼 �𝑠𝑠 + � 5.83(𝑠𝑠 + 1.97)
𝐺𝐺𝑐𝑐 = 𝛼𝛼𝜏𝜏 =
1 (𝑠𝑠 + 11.5)
�𝑠𝑠 + �
𝜏𝜏
Zero: 𝑠𝑠 = −1.97
Pole: 𝑠𝑠 = −11.5

Check:
𝜙𝜙 > 0, 𝑀𝑀 > 0; 𝑝𝑝 = 𝑡𝑡𝑡𝑡𝑖𝑖45 = 1; 𝑝𝑝2 + 1 = 2
𝑙𝑙 > (𝑝𝑝2 + 1); 𝜁𝜁𝑗𝑗𝑠𝑠 = 1; Pole > (−𝜁𝜁𝑗𝑗𝑠𝑠 )
𝐶𝐶 = 𝑐𝑐𝑙𝑙𝑠𝑠 −1 0.45 = 63.26°

−𝜁𝜁𝑗𝑗𝑠𝑠 ± 𝑗𝑗𝑗𝑗𝑠𝑠 �1 − 𝜁𝜁 2 = −1 ± 𝑗𝑗1.98

The pole is in the desired region.

Example
10
𝐺𝐺(𝑠𝑠) =
𝑠𝑠(𝑠𝑠+1)

Desired value of 𝜁𝜁 = 0.5 and 𝑇𝑇𝑠𝑠 = 2.67𝑠𝑠𝑒𝑒𝑐𝑐

(a) Check if a compensator is required

138
 (b) If required, design a compensator
𝐺𝐺 10
(a) 𝑇𝑇(𝑠𝑠) = =
1+𝐺𝐺 𝑠𝑠 2 +𝑠𝑠+10

Characteristic equation, 𝑠𝑠 2 + 𝑠𝑠 + 10 = 0
Comparing with standard form, 𝑠𝑠 2 + (2𝜁𝜁𝜔𝜔𝑛𝑛 )𝑠𝑠 + 𝜔𝜔𝑛𝑛2 = 0

𝜔𝜔𝑛𝑛 = √10 = 3.16

1
2𝜁𝜁𝜔𝜔𝑛𝑛 = 1; 𝜁𝜁 = = 0.16
2𝜔𝜔𝑛𝑛
Since this value of 𝜁𝜁 is less than the desired value of damping, 𝜁𝜁; we need a
compensator.
(b) Desired values are 𝜁𝜁 = 0.5, 𝑇𝑇𝑠𝑠 = 2.67 𝑠𝑠𝑠𝑠𝑠𝑠
4 4
𝜔𝜔𝑛𝑛 = = = 3 𝑟𝑟𝑟𝑟𝑟𝑟/𝑠𝑠
𝜁𝜁𝜁𝜁𝑠𝑠 (0.5)(2.67)
Desired characteristic equation: 𝑠𝑠 2 + (2𝜁𝜁𝜔𝜔𝑛𝑛 )𝑠𝑠 + 𝜔𝜔𝑛𝑛2 = 0
𝑠𝑠 2 + 3𝑠𝑠 + 9 = 0
Desired Poles, 𝑠𝑠1,2 = −1.5 ± 𝑗𝑗2.6

𝜃𝜃 = 𝑐𝑐𝑐𝑐𝑐𝑐 −1 𝜁𝜁 = 60°; 𝜁𝜁𝜔𝜔𝑛𝑛 = 1.5

Cross-over frequency (𝝎𝝎𝒄𝒄 )

10
𝐺𝐺(𝑠𝑠) =
𝑠𝑠 2 + 𝑠𝑠
10
𝐺𝐺(𝑗𝑗𝑗𝑗) =
−𝜔𝜔 2 + 𝑗𝑗𝑗𝑗
10
|𝐺𝐺(𝑗𝑗𝑗𝑗)| = =1
√𝜔𝜔 4 + 𝜔𝜔 2
𝜔𝜔4 + 𝜔𝜔2 − 100 = 0

𝜔𝜔2 = 0.5�−1 ± √1 + 400�; 𝜔𝜔 = 3.1, −10.5

𝜔𝜔𝑐𝑐 = 3.1
Compensator Design: 𝜁𝜁 = 0.5, 𝜔𝜔𝑛𝑛 = 3, 𝜔𝜔𝑐𝑐 = 3.1
1
𝛼𝛼 �𝑠𝑠 + �
𝐺𝐺𝑐𝑐 = 𝛼𝛼𝛼𝛼
1
�𝑠𝑠 + �
𝜏𝜏

139
 𝜙𝜙 = 100𝜁𝜁 = 50°
1 + 𝑠𝑠𝑖𝑖𝑖𝑖𝜙𝜙
𝛼𝛼 = = 7.549
1 − 𝑠𝑠𝑖𝑖𝑖𝑖𝜙𝜙
𝑀𝑀 = 10𝑙𝑙𝑙𝑙𝑙𝑙𝛼𝛼 = 8.78 ≈ 9 𝑑𝑑𝑑𝑑
𝑀𝑀� �
𝑙𝑙 = (10)� 10 = 7.94

1 𝑙𝑙 − 1 1 6.94
𝜏𝜏 = � 2 = � = 0.12
𝑗𝑗𝑐𝑐 𝛼𝛼 − 𝑙𝑙 3.1 49.05

(𝑠𝑠+1.1)
𝐺𝐺𝑐𝑐 = 7.55 (𝑠𝑠+8.33) ; Pole: 𝑠𝑠 = −8.3 > −𝜁𝜁𝑗𝑗𝑠𝑠

Desired Region

Design of Pre-filter
𝐺𝐺𝑐𝑐 𝐺𝐺 𝑠𝑠+𝑧𝑧 1
𝑇𝑇(𝑠𝑠) = ; Let 𝐺𝐺𝑐𝑐 = and 𝐺𝐺 =
1+𝐺𝐺𝑐𝑐 𝐺𝐺 𝑠𝑠+𝑝𝑝 𝑠𝑠

(𝑠𝑠+𝑧𝑧) (𝑠𝑠+𝑧𝑧)
𝐺𝐺𝑐𝑐 𝐺𝐺 = ; 𝑇𝑇(𝑠𝑠) =
𝑠𝑠(𝑠𝑠+𝑝𝑝) {𝑠𝑠(𝑠𝑠+𝑝𝑝)+(𝑠𝑠+𝑧𝑧)}

The zero of the system transfer function, 𝑇𝑇(𝑠𝑠) is euqal to the zero of the compensator
transfer function. Also the zero of the compensator affects the system characteristic
equation and hence the system response. To eliminate this effect of Zero in compensator,
a prefilter with unit magnitude (gain) is requred.

140
Prefilter with a P-I compensator
𝐾𝐾𝐼𝐼 𝐾𝐾𝑝𝑝 𝑠𝑠+𝐾𝐾𝐼𝐼 1
𝐺𝐺𝑐𝑐 (𝑠𝑠) = 𝐾𝐾𝑝𝑝 + = ; 𝐺𝐺(𝑠𝑠) =
𝑠𝑠 𝑠𝑠 𝑠𝑠

𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 𝐺𝐺 𝐾𝐾𝑝𝑝 𝑠𝑠 + 𝐾𝐾𝐼𝐼

𝑇𝑇(𝑠𝑠) = ; 𝐺𝐺𝑐𝑐 𝐺𝐺 =
(1 + 𝐺𝐺𝑐𝑐 𝐺𝐺) 𝑠𝑠 2
𝐾𝐾
𝐾𝐾𝑝𝑝 �𝑠𝑠 + 𝐾𝐾𝐼𝐼 �
𝐾𝐾𝑝𝑝 𝑠𝑠 + 𝐾𝐾𝐼𝐼 𝑝𝑝
𝑇𝑇(𝑠𝑠) = 𝐺𝐺𝑝𝑝 � � = 𝐺𝐺𝑝𝑝 � 2 �
𝑠𝑠 2 + 𝐾𝐾𝑝𝑝 𝑠𝑠 + 𝐾𝐾𝐼𝐼 �𝑠𝑠 + 𝐾𝐾𝑝𝑝 𝑠𝑠 + 𝐾𝐾𝐼𝐼 �

𝐾𝐾
�𝐾𝐾 𝐼𝐼 �
𝐾𝐾𝐼𝐼 𝑝𝑝
To eliminate the zero, ; the transfer function of prefilter, 𝐺𝐺𝑝𝑝 = 𝐾𝐾
𝐾𝐾𝑝𝑝 �𝑠𝑠+ 𝐼𝐼 �
𝐾𝐾𝑝𝑝

Then, the system transfer function would become,

𝐾𝐾𝐼𝐼
𝑇𝑇(𝑠𝑠) =
�𝑠𝑠 2 + 𝐾𝐾𝑝𝑝 𝑠𝑠 + 𝐾𝐾𝐼𝐼 �

Example:
Design a prefilter for given 𝑃𝑃. 𝑂𝑂. = 4%, 𝑇𝑇𝑠𝑠 = 0.5 𝑠𝑠𝑒𝑒𝑐𝑐
𝛿𝛿 100
𝜁𝜁 = ; 𝛿𝛿 = 𝑙𝑙𝑖𝑖 � � = 𝑙𝑙𝑖𝑖(25) = 3.22; 𝜁𝜁 = 0.716
√𝛿𝛿 2 + 𝜋𝜋 2 𝑃𝑃. 𝑂𝑂.
4
𝑗𝑗𝑠𝑠 = = 11.17 𝑒𝑒𝑡𝑡𝑑𝑑/𝑠𝑠; 2𝜁𝜁𝑗𝑗𝑠𝑠 = 16
𝜁𝜁𝑇𝑇𝑠𝑠
Desired characteristic equation,
𝑠𝑠 2 + (2𝜁𝜁𝑗𝑗𝑠𝑠 )𝑠𝑠 + 𝑗𝑗𝑠𝑠2 = 𝑠𝑠 2 + 16𝑠𝑠 + 124.77
Comparing with 𝑠𝑠 2 + 𝐾𝐾𝑝𝑝 𝑠𝑠 + 𝐾𝐾𝐼𝐼
𝐾𝐾𝐼𝐼
We can find 𝐾𝐾𝑝𝑝 = 16, 𝐾𝐾𝐼𝐼 = 125; = 7.8 ≈ 8
𝐾𝐾𝑝𝑝

8
So, the prefilter 𝐺𝐺𝑝𝑝 =
(𝑠𝑠+8)

Without prefilter,

�𝐾𝐾𝑝𝑝 𝑠𝑠 + 𝐾𝐾𝐼𝐼 � (16𝑠𝑠 + 125)

𝑇𝑇(𝑠𝑠) = =
�𝑠𝑠 2 + 𝐾𝐾𝑝𝑝 𝑠𝑠 + 𝐾𝐾𝐼𝐼 � (𝑠𝑠 2 + 16𝑠𝑠 + 125)

With prefilter,

141
 125
𝑇𝑇(𝑠𝑠) =
𝑠𝑠 2 + 16𝑠𝑠 + 125
So, if 𝑇𝑇(𝑠𝑠) has a zero (𝑠𝑠 + 𝑧𝑧) due to adding a compensator, the required prefilter to
𝑧𝑧
eliminate this zero is, 𝐺𝐺𝑝𝑝 =
(𝑠𝑠+𝑧𝑧)

Example:
Design a PI-compensator and prefilter so that the control system satisfies the desired
parameters of 𝑃𝑃. 𝑂𝑂. = 10% and 𝑇𝑇𝑠𝑠 = 2 𝑠𝑠𝑒𝑒𝑐𝑐
1
𝐺𝐺 (𝑠𝑠) = (𝑠𝑠−20) and 𝐻𝐻(𝑠𝑠) = 10

Compensator and Prefilter Design

𝐾𝐾𝐼𝐼 �𝐾𝐾𝑝𝑝 𝑠𝑠 + 𝐾𝐾𝐼𝐼 �
𝐺𝐺𝑐𝑐 (𝑠𝑠) = 𝐾𝐾𝑝𝑝 + =
𝑠𝑠 𝑠𝑠
𝐺𝐺𝑝𝑝 𝐺𝐺𝑐𝑐 𝐺𝐺 �𝐾𝐾𝑝𝑝 𝑠𝑠 + 𝐾𝐾𝐼𝐼 �
𝑇𝑇(𝑠𝑠) = ; 𝐺𝐺𝑐𝑐 𝐺𝐺 =
(1 + 𝐺𝐺𝑐𝑐 𝐺𝐺𝐻𝐻) 𝑠𝑠(𝑠𝑠 − 20)
𝐾𝐾
𝐺𝐺𝑝𝑝 𝐾𝐾𝑝𝑝 (𝑠𝑠 + 𝐼𝐼�𝐾𝐾 )
𝑝𝑝
𝑇𝑇(𝑠𝑠) =
𝑠𝑠(𝑠𝑠 − 20) + 10(𝐾𝐾𝑝𝑝 𝑠𝑠 + 𝐾𝐾𝐼𝐼 )
𝐾𝐾
𝐾𝐾𝑝𝑝 � 𝐼𝐼� �
𝐾𝐾
To eliminate the zero, �𝑠𝑠 + 𝐼𝐼�𝐾𝐾 � the required prefilter should be, 𝐺𝐺𝑝𝑝 = 𝐾𝐾
𝑝𝑝 �𝑠𝑠+� 𝐼𝐼�𝐾𝐾 ��
𝑝𝑝

With prefilter,
𝐾𝐾𝐼𝐼
𝑇𝑇(𝑠𝑠) =
𝑠𝑠(𝑠𝑠 − 20) + 10(𝐾𝐾𝑝𝑝 𝑠𝑠 + 𝐾𝐾𝐼𝐼 )

Characteristic equation,

𝑠𝑠(𝑠𝑠 − 20) + 10�𝐾𝐾𝑝𝑝 𝑠𝑠 + 𝐾𝐾𝐼𝐼 � = 0

𝑠𝑠 2 + �10𝐾𝐾𝑝𝑝 − 20�𝑠𝑠 + 10𝐾𝐾𝐼𝐼 = 0 ⟶ (1)

Desired characteristic equation,

142
 𝑠𝑠 2 + (2𝜁𝜁𝜔𝜔𝑛𝑛 )𝑠𝑠 + 𝜔𝜔𝑛𝑛2 = 0 ⟶ (2)
Desired parameters are 𝑃𝑃. 𝑂𝑂. = 10%, 𝑇𝑇𝑠𝑠 = 2 𝑠𝑠𝑠𝑠𝑠𝑠
100 𝛿𝛿
𝛿𝛿 = 𝑙𝑙𝑙𝑙 � � = 𝑙𝑙𝑙𝑙10 = 2.3; 𝜁𝜁 = = 0.59
𝑃𝑃. 𝑂𝑂. √𝛿𝛿 2 + 𝜋𝜋 2
4
𝜔𝜔𝑛𝑛 = = 3.39; 2𝜁𝜁𝜔𝜔𝑛𝑛 = 4
𝜁𝜁𝑇𝑇𝑠𝑠
Comparing (1) and (2),
10𝐾𝐾𝑝𝑝 − 20 = 2𝜁𝜁𝜔𝜔𝑛𝑛 = 4; 𝐾𝐾𝑝𝑝 = 2.4

10𝐾𝐾𝐼𝐼 = 𝜔𝜔𝑛𝑛2 = 11.5; 𝐾𝐾𝐼𝐼 = 1.15

𝐾𝐾 0.5 1
� 𝐼𝐼�𝐾𝐾 � = 0.5; 𝐺𝐺𝑝𝑝 = ; 𝐺𝐺𝑐𝑐 = � � (2.4𝑠𝑠 + 1.15)
𝑝𝑝 𝑠𝑠 + 0.5 𝑠𝑠
1.15
𝑇𝑇(𝑠𝑠) =
(𝑠𝑠 2 + 4𝑠𝑠 + 11.5)

Design of Lag Compensator

𝑠𝑠 + 𝑧𝑧
𝐺𝐺𝑐𝑐 (𝑠𝑠) = 𝐾𝐾 � �
𝑠𝑠 + 𝑝𝑝
𝑧𝑧 1
Let 𝛼𝛼 = 𝑎𝑎𝑎𝑎𝑎𝑎 𝜏𝜏 =
𝑝𝑝 𝑧𝑧

𝑧𝑧 1
𝑝𝑝 = =
𝛼𝛼 𝛼𝛼𝛼𝛼
𝑠𝑠 + (1/𝜏𝜏) (1 + 𝜏𝜏𝜏𝜏) (1 + 𝜏𝜏𝜏𝜏)
𝐺𝐺𝑐𝑐 = 𝐾𝐾 � � = 𝐾𝐾𝐾𝐾 = 𝐾𝐾1 ; 𝐾𝐾 = 𝐾𝐾𝐾𝐾
𝑠𝑠 + (1/𝛼𝛼𝛼𝛼) (1 + 𝛼𝛼𝛼𝛼𝛼𝛼) (1 + 𝛼𝛼𝛼𝛼𝛼𝛼) 1

Design process
For given, 𝐺𝐺(𝑠𝑠),
𝐿𝐿(𝑠𝑠) = 𝐺𝐺𝑐𝑐 (𝑠𝑠)𝐺𝐺(𝑠𝑠)
At 𝜔𝜔 = 𝜔𝜔𝑐𝑐 , |𝐿𝐿| in 𝑑𝑑𝑑𝑑 = 20𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙
𝜔𝜔𝑐𝑐
𝑧𝑧 = , (The zero is one decade below the crossover frequency)
10

143
Example
𝐾𝐾1
Given 𝐿𝐿(𝑠𝑠) = 𝐺𝐺𝑐𝑐 (𝑠𝑠)𝐺𝐺 (𝑠𝑠) =
𝑠𝑠(𝑠𝑠+2)

Desired parameters: velocity error constant, 𝐾𝐾𝑣𝑣 = 20 and crossover frequency, 𝜔𝜔𝑐𝑐 = 1.5
(𝐾𝐾1 /2)
𝐿𝐿(𝑠𝑠) =
𝑠𝑠(0.5𝑠𝑠 + 1)
𝐾𝐾𝑣𝑣 = 𝑠𝑠[𝐿𝐿(𝑠𝑠)]𝑠𝑠=0 = 𝐾𝐾1 /2; 𝐾𝐾1 = 2𝐾𝐾𝑣𝑣 = 40
𝐾𝐾𝑣𝑣
𝐿𝐿(𝑠𝑠) =
(0.5𝑠𝑠 2 + 𝑠𝑠)
𝐾𝐾𝑣𝑣
𝐿𝐿(𝑗𝑗𝑗𝑗) =
(−0.5𝜔𝜔 2 + 𝑗𝑗𝑗𝑗)
𝐾𝐾𝑣𝑣
|𝐿𝐿| =
[0.25𝜔𝜔 4 + 𝜔𝜔 2 ](0.5)
At 𝜔𝜔 = 𝜔𝜔𝑐𝑐 ,
𝑑𝑑𝑑𝑑 = 20𝑙𝑙𝑙𝑙𝑙𝑙|𝐿𝐿| = 20𝑙𝑙𝑙𝑙𝑙𝑙(𝐾𝐾𝑣𝑣 ) − 10𝑙𝑙𝑙𝑙𝑙𝑙{0.25𝜔𝜔𝑐𝑐4 + 𝜔𝜔𝑐𝑐2 } = 20𝑙𝑙𝑙𝑙𝑙𝑙(20) − 10𝑙𝑙𝑙𝑙𝑙𝑙(3.52)
= 26 − 5.5 = 20.5 𝑑𝑑𝑑𝑑
20𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 = 20.5 𝑑𝑑𝑑𝑑
𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 ≈ 1; 𝛼𝛼 = 10
𝜔𝜔𝑐𝑐 1.5 1 𝑧𝑧
𝑧𝑧 = = = 0.15, 𝜏𝜏 = = 6.67, 𝑝𝑝 = = 0.015
10 10 𝑧𝑧 𝛼𝛼
𝐾𝐾1 (1+𝜏𝜏𝜏𝜏) 40(1+6.67𝑠𝑠) 4(𝑠𝑠+0.15)
Compensator, 𝐺𝐺𝑐𝑐 = (1+𝛼𝛼𝛼𝛼𝛼𝛼)
= (1+66.67𝑠𝑠)
= (𝑠𝑠+0.015)
𝑠𝑠+𝑧𝑧 𝐾𝐾1
Or, 𝐺𝐺𝑐𝑐 (𝑠𝑠) = 𝐾𝐾 � � ; 𝐾𝐾 = = 40/10 = 4
𝑧𝑧+𝑝𝑝 𝛼𝛼

4(𝑠𝑠 + 0.15)
𝐺𝐺𝑐𝑐 (𝑠𝑠) =
(𝑠𝑠 + 0.015)

144
Example
𝐾𝐾1
Given 𝐿𝐿(𝑠𝑠) =
𝑠𝑠(𝑠𝑠+10)2

Design a Lag compensator for the desired parameters of 𝐾𝐾𝑣𝑣 = 20, 𝜔𝜔𝑐𝑐 = 1.3
𝐾𝐾
� 1�100�
𝐿𝐿(𝑠𝑠) =
𝑠𝑠(0.1𝑠𝑠 + 1)2
𝐾𝐾𝑣𝑣 = 𝑠𝑠[𝐿𝐿(𝑠𝑠)]𝑠𝑠=0 = 𝐾𝐾1 /100; 𝐾𝐾1 = 100 𝐾𝐾𝑣𝑣 = 2000
𝐾𝐾𝑣𝑣
𝐿𝐿(𝑠𝑠) =
(0.01𝑠𝑠 3 + 0.2𝑠𝑠 2 + 𝑠𝑠)
𝐾𝐾𝑣𝑣
𝐿𝐿(𝑗𝑗𝑗𝑗) =
−0.01𝜔𝜔 2 (𝑗𝑗𝑗𝑗) − 0.2𝜔𝜔 2 + 𝑗𝑗𝑗𝑗
𝐾𝐾𝑣𝑣
=
[−0.2𝜔𝜔 2 + 𝑗𝑗𝑗𝑗(1 − 0.01𝜔𝜔 2 )]
𝐾𝐾𝑣𝑣
|𝐿𝐿| =
[0.04𝜔𝜔 4 + 𝜔𝜔 2 (1 − 0.01𝜔𝜔 2 )2 ]0.5
𝑑𝑑𝑑𝑑 = 20𝑙𝑙𝑙𝑙𝑙𝑙|𝐿𝐿| = 20𝑙𝑙𝑙𝑙𝑙𝑙𝐾𝐾𝑣𝑣 − 10𝑙𝑙𝑙𝑙𝑙𝑙{𝜔𝜔2 + 0.02𝜔𝜔4 + 0.0001𝜔𝜔6 }
at 𝜔𝜔 = 𝜔𝜔𝑐𝑐 , 𝑑𝑑𝑑𝑑 = 24 = 20𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙
𝜔𝜔𝑐𝑐 1
𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 = 1.2; 𝛼𝛼 = 16; 𝑧𝑧 = = 0.13; 𝜏𝜏 = = 7.69
10 𝑧𝑧
𝑧𝑧 𝐾𝐾1 2000
𝑝𝑝 = = 0.008125, 𝐾𝐾 = = = 125
𝛼𝛼 𝛼𝛼 16
Compensator, 𝐺𝐺𝑐𝑐 = [125(𝑠𝑠 + 0.13)⁄(𝑠𝑠 + 0.008125)]

145
 Assessment
1. the design parameter in the design of compensator are,
a) 𝛼𝛼
b) 𝜏𝜏
c) 𝛼𝛼 and 𝜏𝜏
d) None of the above

2. In a lead compensator, if 𝑝𝑝 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 and 𝑧𝑧 = 𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧,

a) 𝛼𝛼 = 𝑝𝑝/𝑧𝑧
b) 𝛼𝛼 = 𝑧𝑧/𝑝𝑝
c) 𝛼𝛼 = 𝑝𝑝𝑝𝑝
d) 𝛼𝛼 = 𝑝𝑝 + 𝑧𝑧

3. In a lag compensator
a) 𝛼𝛼 = 𝑝𝑝/𝑧𝑧
b) 𝛼𝛼 = 𝑧𝑧/𝑝𝑝
c) 𝛼𝛼 = 𝑝𝑝𝑝𝑝
d) 𝛼𝛼 = 𝑝𝑝 + 𝑧𝑧

4. In a lead compensator design,

a) 𝜏𝜏 = 𝑧𝑧𝑧𝑧
b) 𝜏𝜏 = 1/𝑧𝑧
c) 𝜏𝜏 = 1/𝑝𝑝
d) 𝜏𝜏 = 𝑝𝑝𝑝𝑝

146
5. In a Lag compensator design,
a) 𝜏𝜏 = 1/𝑝𝑝𝑝𝑝
b) 𝜏𝜏 = 1/𝑧𝑧
c) 𝜏𝜏 = 1/𝑝𝑝
d) 𝜏𝜏 = 𝑝𝑝𝑝𝑝

6. In a lead compensator design, the requirement on pole is,

a) 𝑝𝑝 > 𝜁𝜁𝜔𝜔𝑛𝑛
b) 𝑝𝑝 < 𝜁𝜁𝜔𝜔𝑛𝑛
c) 𝑝𝑝 > −𝜁𝜁𝜔𝜔𝑛𝑛
d) 𝑝𝑝 < −𝜁𝜁𝜔𝜔𝑛𝑛

7. Pre-filter in general is added

a) to eliminate a pole
b) to eliminate a zero
c) to add a zero
d) to add a pole

8. for a PI compensator, the prefilter is,

𝐾𝐾
� 𝐼𝐼�𝐾𝐾 �
𝑝𝑝
a) 𝐺𝐺𝑝𝑝 = 𝐾𝐾
�𝑠𝑠+ 𝐼𝐼�𝐾𝐾 �
𝑝𝑝

𝐾𝐾
� 𝑝𝑝�𝐾𝐾 �
b) 𝐺𝐺𝑝𝑝 = 𝐾𝐾
𝐼𝐼

�𝑠𝑠+ 𝑝𝑝�𝐾𝐾 �
𝐼𝐼

𝐾𝐾
� 𝑝𝑝�𝐾𝐾 �
c) 𝐺𝐺𝑝𝑝 = 𝐾𝐾
𝐼𝐼

�𝑠𝑠+ 𝐼𝐼�𝐾𝐾 �
𝑝𝑝

𝐾𝐾
� 𝐼𝐼�𝐾𝐾 �
𝑝𝑝
d) 𝐺𝐺𝑝𝑝 = 𝐾𝐾
�𝑠𝑠+ 𝑝𝑝�𝐾𝐾 �
𝐼𝐼

147
 𝑠𝑠+12
9. Prefilter for a system with 𝑇𝑇(𝑠𝑠) = is,
𝑠𝑠 2 +7𝑠𝑠+12

a) 𝐺𝐺𝑝𝑝 = 12�(𝑠𝑠 + 12)

b) 𝐺𝐺𝑝𝑝 = 7�(𝑠𝑠 + 12)

c) 𝐺𝐺𝑝𝑝 = 1�(𝑠𝑠
+ 12)

d) 𝐺𝐺𝑝𝑝 = 4�(𝑠𝑠 + 12)

10. The velocity error constant for 𝐿𝐿(𝑠𝑠) is,

a) 𝐾𝐾𝑣𝑣 = [𝐿𝐿(𝑠𝑠)]𝑠𝑠=0
b) 𝐾𝐾𝑣𝑣 = 𝑠𝑠[𝐿𝐿(𝑠𝑠)]𝑠𝑠=0
c) 𝐾𝐾𝑣𝑣 = 𝑠𝑠 2 [𝐿𝐿(𝑠𝑠)]𝑠𝑠=0
1
d) 𝐾𝐾𝑣𝑣 = [𝐿𝐿(𝑠𝑠)]
𝑠𝑠

148
 10. State Variable Models and Design

Learning Objectives:
1) To model the 𝑛𝑛𝑡𝑡ℎ order system into ‘n’ first order systems using state variables.
2) To develop the “state” equation and “output” equation from a given nth order
differential equation or a transfer function.
3) To determine the Block diagram or signal flow graph from a given state equation
and output equation.
4) To check the system for controllability and observability.
5) To design an observer and a controller for given desired poles.

10.1 State variable model for a given differential equation

The 𝑛𝑛𝑡𝑡ℎ order system is in general described by a single 𝑛𝑛𝑡𝑡ℎ order differential equation.
The state variables help to model the system into first order systems described by n-
number of first order differential equations.
For example, a third order system is described by
𝑦𝑦⃛ + 𝑎𝑎2 𝑦𝑦̈ + 𝑎𝑎1 𝑦𝑦̇ + 𝑎𝑎0 𝑦𝑦 = 𝑢𝑢(𝑡𝑡); 𝑤𝑤ℎ𝑒𝑒𝑒𝑒𝑒𝑒, 𝑢𝑢(𝑡𝑡) = Step input forcing function
𝑦𝑦⃛ = −𝑎𝑎0 𝑦𝑦 − 𝑎𝑎1 𝑦𝑦̇ − 𝑎𝑎2 𝑦𝑦̈ + 𝑢𝑢 = −𝑎𝑎0 𝑥𝑥1 − 𝑎𝑎1 𝑥𝑥2 − 𝑎𝑎2 𝑥𝑥3 + 𝑢𝑢
where 𝑥𝑥1 , 𝑥𝑥2 , 𝑥𝑥3 are defined as state variables.
𝑥𝑥1 = 𝑦𝑦; 𝑥𝑥2 = 𝑦𝑦̇ = 𝑥𝑥1̇ ; 𝑥𝑥3 = 𝑦𝑦̈ = 𝑥𝑥2̇ ; 𝑥𝑥3̇ = 𝑦𝑦⃛
We have three state variable since it is a third order system. An 𝑛𝑛𝑡𝑡ℎ order system would
require n-number of state variables.
The state equation for this system are,
𝑥𝑥1̇ = 𝑥𝑥2 ; 𝑥𝑥2̇ = 𝑥𝑥3 ; 𝑥𝑥3̇ = −𝑎𝑎0 𝑥𝑥1 − 𝑎𝑎1 𝑥𝑥2 − 𝑎𝑎2 𝑥𝑥3 + 𝑢𝑢
In matrix form,
𝑥𝑥1̇ 0 1 0 𝑥𝑥1 0
�𝑥𝑥2̇ � = � 0 0 𝑥𝑥
1 � � 2 � + �0� 𝑢𝑢
𝑥𝑥3̇ −𝑎𝑎0 −𝑎𝑎1 −𝑎𝑎2 𝑥𝑥3 1
{𝑥𝑥̇ } = [𝐴𝐴]{𝑥𝑥} + {𝐵𝐵}𝑢𝑢 is the state equation

149
In this system, 𝑥𝑥1 = 𝑦𝑦 relates the output ‘𝑦𝑦’ to the state variable 𝑥𝑥1 .
In matrix form,
𝑥𝑥1
𝑦𝑦 = [1 0 𝑥𝑥
0] � 2 �
𝑥𝑥3
𝑦𝑦 = [𝐶𝐶]𝑥𝑥 is the output equation. The state equation and the output equation represent a
system in stable variable model.

Solution to the state equation

{𝑥𝑥̇ } = [𝐴𝐴]{𝑥𝑥} + {𝐵𝐵}𝑢𝑢
Assuming {𝑥𝑥(0)} is the given initial condition, the Laplace transform of the state
equation gives,
[𝑆𝑆{𝑋𝑋(𝑠𝑠)} − {𝑋𝑋(0)}] = [𝐴𝐴]{𝑋𝑋(𝑠𝑠)} + {𝐵𝐵}𝑈𝑈(𝑠𝑠)
(𝑆𝑆[𝐼𝐼] − [𝐴𝐴]){𝑋𝑋(𝑠𝑠)} = {𝑋𝑋(0)} + {𝐵𝐵}𝑈𝑈(𝑠𝑠)
[𝑍𝑍(𝑠𝑠)]{𝑋𝑋(𝑠𝑠)} = {𝑋𝑋(0)} + {𝐵𝐵}𝑈𝑈(𝑠𝑠)
{𝑋𝑋(𝑠𝑠)} = [𝑍𝑍(𝑠𝑠)]−1 [{𝑋𝑋(0)} + {𝐵𝐵}𝑈𝑈(𝑠𝑠)]
= [𝜙𝜙(𝑠𝑠)][{𝑋𝑋(0)} + {𝐵𝐵}𝑈𝑈(𝑠𝑠)] = [𝜙𝜙(𝑠𝑠)]{𝑋𝑋(0)} + {𝜓𝜓(𝑠𝑠)}
where, [𝜙𝜙(𝑠𝑠)] = [𝑍𝑍(𝑠𝑠)]−1 = (𝑆𝑆[𝐼𝐼] − [𝐴𝐴])−1
𝜓𝜓(𝑠𝑠) = [𝜙𝜙(𝑠𝑠)]{𝐵𝐵}𝑈𝑈(𝑠𝑠)
To find the time response, 𝑋𝑋(𝑡𝑡); the inverse Laplace transform gives,
{𝑋𝑋(𝑡𝑡)} = [𝜙𝜙(𝑡𝑡)]{𝑋𝑋(0)} + {𝜓𝜓(𝑡𝑡)}
where [𝜙𝜙(𝑡𝑡)] is the State Transition Matrix or Fundamental matrix.

Special case
If 𝑢𝑢(𝑡𝑡) = 0, then 𝜓𝜓(𝑡𝑡) = 0
{𝑋𝑋(𝑡𝑡)} = [𝜙𝜙(𝑡𝑡)]{𝑋𝑋(0)}

150
 10.2 Time response for a given state equation

Let the state equation is,

𝑥𝑥 ̇ 0 −2 𝑥𝑥1 0
� 1� = � � � � = � � 𝑢𝑢(𝑡𝑡)
𝑥𝑥2̇ 1 −3 𝑥𝑥2 1
𝑥𝑥 (0) 1
Initial conditions, � 1 � = � �
𝑥𝑥2 (0) 1
{𝑥𝑥̇ } = [𝐴𝐴]{𝑥𝑥} + {𝐵𝐵}𝑢𝑢
𝑠𝑠 0 0 −2 𝑠𝑠 2
[𝑍𝑍(𝑠𝑠)] = 𝑆𝑆[𝐼𝐼] − [𝐴𝐴] = � �−� �=� �
0 𝑠𝑠 1 −3 −1 𝑠𝑠 + 3
1 𝑠𝑠 + 3 −2
[𝜙𝜙(𝑠𝑠)] = [𝑍𝑍(𝑠𝑠)]−1 = � �
𝐷𝐷𝐷𝐷𝐷𝐷 1 𝑠𝑠
𝐷𝐷𝐷𝐷𝐷𝐷 = 𝑠𝑠(𝑠𝑠 + 3) + 2 = 𝑠𝑠 2 + 3𝑠𝑠 + 2 = (𝑠𝑠 + 1)(𝑠𝑠 + 2)
𝑠𝑠 + 3 −2
⎡ ⎤
𝜙𝜙11 𝜙𝜙12 ⎢ (𝑠𝑠 + 1)(𝑠𝑠 + 2) (𝑠𝑠 + 1)(𝑠𝑠 + 2)⎥
[𝜙𝜙(𝑠𝑠)] = � �=
𝜙𝜙21 𝜙𝜙22 ⎢ 1 𝑠𝑠 ⎥
⎣(𝑠𝑠 + 1)(𝑠𝑠 + 2) (𝑠𝑠 + 1)(𝑠𝑠 + 2)⎦
Taking inverse L.T. for each element of the above matrix,
𝑠𝑠 + 3 𝑘𝑘1 𝑘𝑘2
𝜙𝜙11 (𝑠𝑠) = = +
(𝑠𝑠 + 1)(𝑠𝑠 + 2) 𝑠𝑠 + 1 𝑠𝑠 + 2
𝑘𝑘1 = (𝑠𝑠 + 1)𝜙𝜙11 (𝑠𝑠)|𝑠𝑠=−1 = 2; 𝑘𝑘2 = (𝑠𝑠 + 2)𝜙𝜙11 (𝑠𝑠)|𝑠𝑠=−2 = −1
2 1
𝜙𝜙11 (𝑠𝑠) = −
𝑠𝑠 + 1 𝑠𝑠 + 2
𝜙𝜙11 (𝑡𝑡) = 2𝑒𝑒 −𝑡𝑡 − 𝑒𝑒 −2𝑡𝑡
−2 𝑘𝑘1 𝑘𝑘2
𝜙𝜙12 (𝑠𝑠) = = +
(𝑠𝑠 + 1)(𝑠𝑠 + 2) 𝑠𝑠 + 1 𝑠𝑠 + 2
𝑘𝑘1 = (𝑠𝑠 + 1)𝜙𝜙12 (𝑠𝑠)|𝑠𝑠=−1 = −2; 𝑘𝑘2 = (𝑠𝑠 + 2)𝜙𝜙12 (𝑠𝑠)|𝑠𝑠=−2 = 2
2 2
𝜙𝜙12 (𝑠𝑠) = − +
𝑠𝑠 + 1 𝑠𝑠 + 2
𝜙𝜙12 (𝑡𝑡) = −2𝑒𝑒 −𝑡𝑡 + 2𝑒𝑒 −2𝑡𝑡

151
 1 1 1
𝜙𝜙21 (𝑠𝑠) = = −
(𝑠𝑠 + 1)(𝑠𝑠 + 2) 𝑠𝑠 + 1 𝑠𝑠 + 2
𝜙𝜙21 (𝑡𝑡) = 𝑒𝑒 −𝑡𝑡 − 𝑒𝑒 −2𝑡𝑡
𝑠𝑠 1 2
𝜙𝜙22 (𝑠𝑠) = =− +
(𝑠𝑠 + 1)(𝑠𝑠 + 2) 𝑠𝑠 + 1 𝑠𝑠 + 2
𝜙𝜙22 (𝑡𝑡) = −𝑒𝑒 −𝑡𝑡 + 2𝑒𝑒 −2𝑡𝑡
0 0
𝜓𝜓(𝑠𝑠) = [𝜙𝜙(𝑠𝑠)]{𝐵𝐵}𝑈𝑈(𝑠𝑠); {𝐵𝐵}𝑈𝑈(𝑠𝑠) = � � 𝑈𝑈(𝑠𝑠) = �1 �
1 �𝑠𝑠

For unit step, 𝑢𝑢(𝑡𝑡) = 1 and 𝑈𝑈(𝑠𝑠) = 1�𝑠𝑠

𝜓𝜓1 (𝑠𝑠) 𝜙𝜙 𝜙𝜙12 0
� � = � 11 �� �
𝜓𝜓2 (𝑠𝑠) 𝜙𝜙21 𝜙𝜙22 1�𝑠𝑠
1 −2 1 2 1
𝜓𝜓1 (𝑠𝑠) = (𝜙𝜙12 ) = =− + −
𝑠𝑠 𝑠𝑠(𝑠𝑠 + 1)(𝑠𝑠 + 2) 𝑠𝑠 𝑠𝑠 + 1 𝑠𝑠 + 2
𝜓𝜓1 (𝑡𝑡) = −1 + 2𝑒𝑒 −𝑡𝑡 − 𝑒𝑒 −2𝑡𝑡
1 1 𝑠𝑠 1 1 1
𝜓𝜓2 (𝑠𝑠) = (𝜙𝜙22 ) = � �= = −
𝑠𝑠 𝑠𝑠 (𝑠𝑠 + 1)(𝑠𝑠 + 2) (𝑠𝑠 + 1)(𝑠𝑠 + 2) 𝑠𝑠 + 1 𝑠𝑠 + 2
𝜓𝜓2 (𝑡𝑡) = 𝑒𝑒 −𝑡𝑡 − 𝑒𝑒 −2𝑡𝑡
𝑥𝑥 (𝑡𝑡) 𝑥𝑥 (0) 𝜙𝜙 𝜙𝜙12 1 𝜓𝜓 (𝑡𝑡)
� 1 � = [𝜙𝜙(𝑡𝑡)] � 1 � + {𝜓𝜓(𝑡𝑡)} = � 11 �� � + � 1 �
𝑥𝑥2 (𝑡𝑡) 𝑥𝑥2 (0) 𝜙𝜙21 𝜙𝜙22 1 𝜓𝜓2 (𝑡𝑡)
𝑥𝑥1 (𝑡𝑡) = 𝜙𝜙11 (𝑡𝑡) + 𝜙𝜙12 (𝑡𝑡) + 𝜓𝜓1 (𝑡𝑡) = 2𝑒𝑒 −𝑡𝑡 − 1
𝑥𝑥2 (𝑡𝑡) = 𝜙𝜙21 (𝑡𝑡) + 𝜙𝜙22 (𝑡𝑡) + 𝜓𝜓2 (𝑡𝑡) = 𝑒𝑒 −𝑡𝑡

152
 10.3 State variable for a given Transfer function

The state variable model can be developed from a given Transfer function of a system.
𝑌𝑌(𝑠𝑠)
In general, 𝑇𝑇(𝑠𝑠) = ; Let 𝑅𝑅(𝑠𝑠) = 𝑈𝑈(𝑠𝑠) (unit step input)
𝑅𝑅(𝑠𝑠)

Both 𝑌𝑌(𝑠𝑠) and 𝑅𝑅(𝑠𝑠) are polynomials in “s”

𝑌𝑌(𝑠𝑠) 𝑏𝑏𝑚𝑚 𝑆𝑆 𝑚𝑚 + 𝑏𝑏𝑚𝑚−1 𝑆𝑆 𝑚𝑚−1 + ⋯ + 𝑏𝑏1 𝑆𝑆 + 𝑏𝑏0 𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂
𝑇𝑇(𝑠𝑠) = = =
𝑈𝑈(𝑠𝑠) 𝑆𝑆 𝑛𝑛 + 𝑎𝑎𝑛𝑛−1 𝑆𝑆 𝑛𝑛−1 + ⋯ + 𝑎𝑎1 𝑆𝑆 + 𝑎𝑎0 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼
𝑏𝑏𝑚𝑚 𝑆𝑆 𝑚𝑚 + 𝑏𝑏𝑚𝑚−1 𝑆𝑆 𝑚𝑚−1 + ⋯ + 𝑏𝑏1 𝑆𝑆 + 𝑏𝑏0
= 𝑛𝑛
𝑆𝑆 [1 + 𝑎𝑎𝑛𝑛−1 𝑆𝑆 −1 + ⋯ + 𝑎𝑎1 𝑆𝑆 −(𝑛𝑛−1) + 𝑎𝑎0 𝑆𝑆 −𝑛𝑛 ]
𝑏𝑏𝑚𝑚 𝑆𝑆 −(𝑛𝑛−𝑚𝑚) + 𝑏𝑏𝑚𝑚−1 𝑆𝑆 −(𝑛𝑛−𝑚𝑚+1) + ⋯ + 𝑏𝑏1 𝑆𝑆 −(𝑛𝑛−1) + 𝑏𝑏0 𝑆𝑆 −𝑛𝑛
=
1 + 𝑎𝑎𝑛𝑛−1 𝑆𝑆 −1 + ⋯ + 𝑎𝑎1 𝑆𝑆 −(𝑛𝑛−1) + 𝑎𝑎0 𝑆𝑆 −𝑛𝑛

Example
Given
(2𝑠𝑠 2 + 8𝑠𝑠 + 6) 𝑌𝑌(𝑠𝑠)
𝑇𝑇(𝑠𝑠) = 3 2
=
(𝑠𝑠 + 8𝑠𝑠 + 16𝑠𝑠 + 6) 𝑈𝑈(𝑠𝑠)
a) Determine the state equation and output equation.
b) Draw the Block diagram model
c) Draw the Signal flow graph.

Solution:
The denominator is an 3𝑟𝑟𝑟𝑟 order polynomial and hence it is a third order system. So, it
will have three state variables, 𝑥𝑥1 , 𝑥𝑥2 and 𝑥𝑥3
(2𝑠𝑠 2 + 8𝑠𝑠 + 6)
𝑇𝑇(𝑠𝑠) =
𝑠𝑠 3 [1 + 8𝑠𝑠 −1 + 16𝑠𝑠 −2 + 6𝑠𝑠 −3 ]
2𝑠𝑠 −1 + 8𝑠𝑠 −2 + 6𝑠𝑠 −3 𝑏𝑏2 𝑠𝑠 −1 + 𝑏𝑏1 𝑠𝑠 −2 + 𝑏𝑏0 𝑠𝑠 −3
= =
1 + 8𝑠𝑠 −1 + 16𝑠𝑠 −2 + 6𝑠𝑠 −3 1 + 𝑎𝑎2 𝑠𝑠 −1 + 𝑎𝑎1 𝑠𝑠 −2 + 𝑎𝑎0 𝑠𝑠 −3
State Equations:
State variables are; 𝑥𝑥1 = 𝑦𝑦; 𝑥𝑥2 = 𝑦𝑦̇ = 𝑥𝑥1̇ ; 𝑥𝑥3 = 𝑦𝑦̈ = 𝑥𝑥2̇
State equations are;
𝑥𝑥1̇ = 𝑥𝑥2 ; 𝑥𝑥2̇ = 𝑥𝑥3 ; 𝑥𝑥3̇ = 𝑦𝑦⃛ = −𝑎𝑎0 𝑥𝑥1 − 𝑎𝑎1 𝑥𝑥2 − 𝑎𝑎2 𝑥𝑥3 + 𝑢𝑢 = −6𝑥𝑥1 − 16𝑥𝑥2 − 8𝑥𝑥3 + 𝑢𝑢

153
In matrix form,
𝑥𝑥1̇ 0 1 0 𝑥𝑥1 0
�𝑥𝑥2̇ � = � 0 0 1 � �𝑥𝑥2 � + �0� 𝑢𝑢
𝑥𝑥3̇ −6 −16 −8 𝑥𝑥3 1
Output Equation:
𝑥𝑥1 𝑥𝑥1
𝑦𝑦(𝑡𝑡) = [𝑏𝑏0 𝑏𝑏1 𝑏𝑏2 ] �𝑥𝑥2 � = [6 8 2] �𝑥𝑥2 �
𝑥𝑥3 𝑥𝑥3
The numerator polynomial gives the output equation while the denominator polynomial
gives the state equation.

Block diagram model:

Laplace transform of state variables:

𝑥𝑥2�
𝑥𝑥1̇ = 𝑥𝑥2 ; 𝑠𝑠𝑥𝑥1 = 𝑥𝑥2 ; 𝑥𝑥1 = 𝑠𝑠

𝑥𝑥3�
𝑥𝑥2̇ = 𝑥𝑥3 ; 𝑠𝑠𝑥𝑥2 = 𝑥𝑥3 ; 𝑥𝑥2 = 𝑠𝑠

𝑥𝑥3̇ = −6𝑥𝑥1 − 16𝑥𝑥2 − 8𝑥𝑥3 + 𝑢𝑢; 𝑠𝑠𝑥𝑥3 = −6𝑥𝑥1 − 16𝑥𝑥2 − 8𝑥𝑥3 + 𝑈𝑈

1
𝑥𝑥3 = (−6𝑥𝑥1 − 16𝑥𝑥2 − 8𝑥𝑥3 + 𝑈𝑈)
𝑠𝑠

Output equation 𝑦𝑦(𝑡𝑡) = 6𝑥𝑥1 + 8𝑥𝑥2 + 2𝑥𝑥3 ,

154
Block Diagram

Signal flow Graph

Example
4𝑠𝑠 + 12 4𝑠𝑠 −1 + 12𝑠𝑠 −2
𝑇𝑇(𝑠𝑠) = =
𝑠𝑠 2 + 8𝑠𝑠 + 12 1 + 𝑠𝑠 −1 + 12𝑠𝑠 −2
Standard format,
𝑏𝑏1 𝑠𝑠 −1 + 𝑏𝑏0 𝑠𝑠 −2
𝑇𝑇(𝑠𝑠) =
1 + 𝑎𝑎1 𝑠𝑠 −1 + 𝑎𝑎0 𝑠𝑠 −2
Denominator is a second order polynomial and hence it requires two state variable, 𝑥𝑥1
and 𝑥𝑥2
𝑥𝑥 ̇ 0 1 𝑥𝑥1 0 0 1 𝑥𝑥1 0
� 1� = � � �𝑥𝑥 � + � � 𝑢𝑢 = � � � � + � � 𝑢𝑢
𝑥𝑥2̇ −𝑎𝑎0 −𝑎𝑎1 2 1 −12 −8 𝑥𝑥2 1
Output equation:
𝑥𝑥 𝑥𝑥1
𝑦𝑦(𝑡𝑡) = [𝑏𝑏0 𝑏𝑏1 ] �𝑥𝑥1 � = [12 4] �𝑥𝑥 � = 12𝑥𝑥1 + 4𝑥𝑥2
2 2

155
State equation:
𝑥𝑥1̇ = 𝑥𝑥2 ; 𝑠𝑠𝑥𝑥1 = 𝑥𝑥2
𝑥𝑥2̇ = −12𝑥𝑥1 − 8𝑥𝑥2 + 𝑢𝑢; 𝑠𝑠𝑥𝑥2 = (𝑈𝑈 − 12𝑥𝑥1 − 8𝑥𝑥2 )

Block diagram

Signal flow Graph

1
𝑥𝑥2 = (𝑈𝑈 − 8𝑥𝑥2 − 12𝑥𝑥1 )
𝑠𝑠
1
𝑥𝑥1 = (𝑥𝑥2 )
𝑠𝑠
𝑌𝑌 = 12𝑥𝑥1 + 4𝑥𝑥2

Example:
𝑥𝑥 ̇ 0 1 𝑥𝑥1 0
� 1� = � � � � + � � 𝑢𝑢(𝑡𝑡)
𝑥𝑥2̇ (2 + 𝑘𝑘1 ) (3 − 𝑘𝑘2 ) 𝑥𝑥2 1
Determine 𝑘𝑘1 and 𝑘𝑘2 such that the roots of characteristic equation are −2 and −3.
Characteristic equation, |𝑧𝑧(𝑠𝑠)| = 0
0 1 𝑠𝑠 −1
[𝑧𝑧(𝑠𝑠)] = 𝑆𝑆[𝐼𝐼] − [𝐴𝐴] = � 𝑠𝑠 0� − � � = �
0 𝑠𝑠 (2 + 𝑘𝑘1 ) (3 − 𝑘𝑘2 ) −(2 + 𝑘𝑘1 ) (𝑠𝑠 − 3 + 𝑘𝑘2 )�
|𝑧𝑧(𝑠𝑠)| = 𝑠𝑠(𝑠𝑠 − 3 + 𝑘𝑘2 ) − (2 + 𝑘𝑘1 ) = 0
𝑠𝑠 2 − 3𝑠𝑠 + 𝑘𝑘2 𝑠𝑠 − 2 − 𝑘𝑘1 = 0
𝑠𝑠 2 + (𝑘𝑘2 − 3)𝑠𝑠 − (2 + 𝑘𝑘1 ) = 0 → (𝑖𝑖)

156
For given roots, the characteristic equation is
(𝑠𝑠 − 𝑝𝑝1 )(𝑠𝑠 − 𝑝𝑝2 ) = 0; 𝑝𝑝1 = −2, 𝑝𝑝2 = −3
(𝑠𝑠 + 2)(𝑠𝑠 + 3) = 𝑠𝑠 2 + 5𝑠𝑠 + 6 = 0 → (𝑖𝑖𝑖𝑖)
Comparing (𝑖𝑖) and (𝑖𝑖𝑖𝑖),
𝑘𝑘2 − 3 = 5 ⟹ 𝑘𝑘2 = 8
−2 − 𝑘𝑘1 = 6 ⟹ 𝑘𝑘1 = −8

Stability of state variable model

State equation: 𝑥𝑥̇ = [𝐴𝐴]𝑥𝑥 + 𝐵𝐵𝐵𝐵
[𝑍𝑍(𝑠𝑠)] = 𝑆𝑆[𝐼𝐼] − [𝐴𝐴]
Characteristic equation, 𝑞𝑞(𝑠𝑠) = |𝑍𝑍(𝑠𝑠)| = 0
Now, Routh – Horwitz criteria can be applied to the characteristic equation for stability
analysis.

Example:
𝑥𝑥1̇ 0 1 0 𝑥𝑥1 0
�𝑥𝑥2̇ � = � 0 0 1 � �𝑥𝑥2 � + �0� 𝑢𝑢
𝑥𝑥3̇ −6 −16 −8 𝑥𝑥3 1
{𝑥𝑥̇ } = [𝐴𝐴]{𝑥𝑥} + {𝐵𝐵}𝑢𝑢
𝑠𝑠 −1 0
[𝑍𝑍(𝑠𝑠)] = 𝑆𝑆[𝐼𝐼] − [𝐴𝐴] = �0 𝑠𝑠 −1 �
6 16 𝑠𝑠 + 8
Characteristic equation = |𝑍𝑍(𝑠𝑠)| = 0
The determinant of [𝑍𝑍(𝑠𝑠)] = 𝑠𝑠 3 + 8𝑠𝑠 2 + 16𝑠𝑠 + 6
Routh Array
𝑠𝑠 3 1 16
𝑠𝑠 2 8 6
𝑠𝑠1 15.25 0
𝑠𝑠 0 6
The system is stable since the first column of array is positive.

157
Example:
Determine ′𝑘𝑘′ for the system to be stable.
𝑥𝑥 ̇ −3 1 𝑥𝑥1 0
� 1� = � � �𝑥𝑥 � + � � 𝑢𝑢
𝑥𝑥2̇ −𝑘𝑘 1 2 1

[𝑍𝑍(𝑠𝑠)] = 𝑆𝑆[𝐼𝐼] − [𝐴𝐴] = �𝑠𝑠 + 3 −1

�
𝑘𝑘 𝑠𝑠 − 1
Characteristic equation, |𝑍𝑍(𝑠𝑠)| = (𝑠𝑠 + 3)(𝑠𝑠 − 1) + 𝑘𝑘 = 𝑠𝑠 2 + 2𝑠𝑠 + (𝑘𝑘 − 3)

Routh Array

𝑠𝑠 2 1 𝑘𝑘 − 3
𝑠𝑠1 2 0
𝑠𝑠 0 𝑘𝑘 − 3

For the first column to be positive, 𝑘𝑘 > 3. So, the value of “k” should be greater than
three for the system to be stable.

Design of State Variable System

Controllability and Observability:

A system is completely controllable if the determinant of the controllability matrix is not
zero. Similarity, a system is completely observable if the determinant of the observability
matrix is not zero.
Controllability:
The state equation of a system is given as,
{𝑥𝑥̇ } = [𝐴𝐴]{𝑥𝑥} + {𝐵𝐵}𝑢𝑢
The controllability matrix, [𝑝𝑝𝑐𝑐 ] = [{𝑝𝑝1 }{𝑝𝑝2 } ⋯ {𝑝𝑝𝑁𝑁 }]
Where {𝑝𝑝1 } = {𝐵𝐵}; {𝑝𝑝2 } = [𝐴𝐴]{𝑝𝑝1 }; {𝑝𝑝3 } = [𝐴𝐴]{𝑝𝑝2 }
{𝑝𝑝𝑁𝑁 } = [𝐴𝐴]{𝑝𝑝𝑁𝑁−1 }
If |𝑝𝑝𝑐𝑐 | = 0, the system is not controllable and if |𝑝𝑝𝑐𝑐 | ≠ 0, the system is controllable.

158
Observability:
The output equation of a system is, 𝑦𝑦 = [𝑐𝑐]{𝑥𝑥}
[𝑝𝑝1 ]
The observability matrix, [𝑝𝑝𝑜𝑜 ] = � ⋮ �
[𝑝𝑝𝑁𝑁 ]
Where, [𝑝𝑝1 ] = [𝐶𝐶]; [𝑝𝑝2 ] = [𝑝𝑝1 ][𝐴𝐴]; [𝑝𝑝3 ] = [𝑝𝑝2 ][𝐴𝐴]
[𝑝𝑝𝑛𝑛 ] = [𝑝𝑝𝑛𝑛−1 ][𝐴𝐴]
If |𝑝𝑝𝑜𝑜 | = 0, the system is not observable and if |𝑝𝑝𝑜𝑜 | ≠ 0, the system is observable.

Example:
𝑦𝑦⃛ + 4𝑦𝑦̈ + 3𝑦𝑦̇ + 2𝑦𝑦 = 𝑢𝑢
Check for controllability of the system
𝑦𝑦⃛ = 𝑢𝑢 − 2𝑦𝑦 − 3𝑦𝑦̇ − 4𝑦𝑦̈ ; let 𝑥𝑥1 = 𝑦𝑦, 𝑥𝑥2 = 𝑦𝑦̇ , 𝑥𝑥3 = 𝑦𝑦̈
𝑦𝑦⃛ = 𝑢𝑢 − 2𝑥𝑥1 − 3𝑥𝑥2 − 4𝑥𝑥3
State equation: 𝑥𝑥1̇ = 𝑥𝑥2 ; 𝑥𝑥2̇ = 𝑥𝑥3 ; 𝑥𝑥3̇ = 𝑦𝑦⃛ = 𝑢𝑢 − 2𝑥𝑥1 − 3𝑥𝑥2 − 4𝑥𝑥3
𝑥𝑥1̇ 0 1 0 𝑥𝑥1 0
�𝑥𝑥2̇ � = � 0 0 𝑥𝑥
1 � � 2 � + �0� 𝑢𝑢
𝑥𝑥3̇ −2 −3 −4 𝑥𝑥3 1
[𝑝𝑝𝑐𝑐 ] = [{𝑝𝑝1 }{𝑝𝑝2 }{𝑝𝑝3 }]
0
{𝑝𝑝1 } = {𝐵𝐵} = �0�
1
0 1 0 0 0
{𝑝𝑝2 } = [𝐴𝐴]{𝑝𝑝1 } = � 0 0 1 � � 0� = � 1 �
−2 −3 −4 1 −4
0 1 0 0 1
{𝑝𝑝3 } = [𝐴𝐴]{𝑝𝑝2 } = � 0 0 1 � � 1 � = �−4�
−2 −3 −4 −4 13
0 0 1
[𝑝𝑝𝑐𝑐 ] = �0 1 −4�
1 −4 13
|𝑝𝑝𝑐𝑐 | = 1(0 − 1) = −1
Since |𝑝𝑝𝑐𝑐 | ≠ 0, the system is controllable.

159
Example:
𝑥𝑥 ̇ −2 0 𝑥𝑥1 1
� 1� = � � �𝑥𝑥 � + � � 𝑢𝑢
𝑥𝑥2̇ 𝑑𝑑 −3 2 0
Find ‘𝑑𝑑’ such that the system is controllable.
[𝑝𝑝𝑐𝑐 ] = [{𝑝𝑝1 }{𝑝𝑝2 }]

{𝑝𝑝1 } = {𝐵𝐵} = �1�

0

{𝑝𝑝2 } = [𝐴𝐴]{𝑝𝑝1 } = �−2 0 1 −2

�� � = � �
𝑑𝑑 −3 0 𝑑𝑑
1 −2
[𝑝𝑝𝑐𝑐 ] = � �
0 𝑑𝑑
|𝑝𝑝𝑐𝑐 | = 𝑑𝑑, for the system to be controllable
|𝑝𝑝𝑐𝑐 | ≠ 0, so 𝑑𝑑 ≠ 0 or the value of “d” should be positive.

Example:
For a given output equation, 𝑦𝑦 = [𝐶𝐶]{𝑥𝑥}, where [𝐶𝐶] = [1 0 0].
and given state equation as
{𝑥𝑥̇ } = [𝐴𝐴]{𝑥𝑥} + {𝐵𝐵}𝑢𝑢
0 1 0 0
where [𝐴𝐴] = � 0 0 1 � and {𝐵𝐵} = �0�
−2 −3 −4 1
Determine if the system is completely observable.
[𝑝𝑝1 ]
[𝑝𝑝𝑜𝑜 ] = �[𝑝𝑝2 ]�
[𝑝𝑝3 ]

[𝑝𝑝1 ] = [𝐶𝐶] = [1 0 0]
0 1 0
[𝑝𝑝2 ] = [𝑝𝑝1 ][𝐴𝐴] = [1 0 0] � 0 0 1 � = [0 1 0]
−2 −3 −4
0 1 0
[𝑝𝑝3 ] = [𝑝𝑝2 ][𝐴𝐴] = [0 1 0] � 0 0 1 � = [0 0 1]
−2 −3 −4

160
 1 0 0
[𝑝𝑝𝑜𝑜 ] = �0 1 0� ; |𝑝𝑝𝑜𝑜 | = 1
0 0 1
Since 𝑝𝑝𝑜𝑜 ≠ 0, the system is completely observable.

Example:

Given the state equation and output equation as

𝑥𝑥 ̇ 2 0 𝑥𝑥1 1 𝑥𝑥1
� 1� = � � � � + � � 𝑢𝑢; 𝑦𝑦 = [1 0] �𝑥𝑥 � ; determine the controllability and
𝑥𝑥2̇ −1 1 𝑥𝑥2 −1 2
observability of the system.
Controllability Check:
[𝑝𝑝𝑐𝑐 ] = [{𝑝𝑝1 }{𝑝𝑝2 }]

{𝑝𝑝1 } = {𝐵𝐵} = � 1 �
−1

{𝑝𝑝2 } = [𝐴𝐴]{𝑝𝑝1 } = �−2 0 1 2

�� � = � �
−1 1 −1 −2

[𝑝𝑝𝑐𝑐 ] = � 1 2
� ; |𝑝𝑝𝑐𝑐 | = 0;
−1 −2
So, the system is not controllable.
Observability Check:
[𝑝𝑝1 ]
[𝑝𝑝𝑜𝑜 ] = � �
[𝑝𝑝2 ]
[𝑝𝑝1 ] = [𝐶𝐶] = [1 0]

[𝑝𝑝2 ] = [𝑝𝑝1 ][𝐴𝐴] = [1 2 0

0] � � = [2 0 ]
−1 1

[𝑝𝑝𝑜𝑜 ] = �1 0 |𝑝𝑝 |
� ; 𝑜𝑜 = 0
2 0
So, the system is not observable.

161
 Design of state variable control system
In state variable model control system requires the design of a controller and design of an
observer. The design is based on pole placement technique for an acceptable closed loop
feedback system performance. The system should be completely controllable and
observable first for a viable design process.

Controller Design
The design objective is to find the controller gain matrix, [𝐾𝐾] for given pole location.
Then, the system input, 𝑢𝑢(𝑡𝑡) = −[𝐾𝐾]{𝑥𝑥}, where [𝐾𝐾] is a row matrix.
{𝑥𝑥̇ } = [𝐴𝐴]{𝑥𝑥} + {𝐵𝐵}𝑢𝑢; 𝑢𝑢 = −[𝐾𝐾]{𝑥𝑥}

{𝑥𝑥̇ } = �[𝐴𝐴] − {𝐵𝐵}[𝐾𝐾]�{𝑥𝑥}

{𝑥𝑥̇ } = [𝐴𝐴1 ]{𝑥𝑥}; [𝐴𝐴1 ] = [𝐴𝐴] − {𝐵𝐵}[𝐾𝐾]
Let the solution is, {𝑥𝑥} = 𝑒𝑒 𝑠𝑠𝑠𝑠
𝑥𝑥̇ = 𝑠𝑠𝑒𝑒 𝑠𝑠𝑠𝑠 = 𝑠𝑠𝑠𝑠
𝑠𝑠𝑠𝑠 = 𝐴𝐴1 𝑥𝑥
〈𝑠𝑠[𝐼𝐼] − [𝐴𝐴1 ]〉{𝑥𝑥} = 0
So, the determinant |𝑠𝑠𝑠𝑠 − 𝐴𝐴1 | = 0 → (𝑖𝑖)
The determinant gives the characteristic equation.
For a second order system, if the given poles are 𝑠𝑠1 and 𝑠𝑠2 , the desired characteristic
equation is (𝑠𝑠 − 𝑠𝑠1 )(𝑠𝑠 − 𝑠𝑠2 ) = 0 → (𝑖𝑖𝑖𝑖)
The controller gain can be determined by comparing the equation (𝑖𝑖) and equation (𝑖𝑖𝑖𝑖).

Example:
𝑥𝑥 ̇ 0 1 𝑥𝑥1 0 𝑥𝑥1
� 1� = � � �𝑥𝑥 � + � � 𝑢𝑢; 𝑦𝑦 = [1 0] �𝑥𝑥 �
𝑥𝑥2̇ 0 0 2 1 2

Design a controller such that the desired poles are 𝑠𝑠1 = −1 + 𝑗𝑗 and 𝑠𝑠2 = −1 − 𝑗𝑗.
[𝐴𝐴1 ] = [𝐴𝐴] − {𝐵𝐵}[𝐾𝐾1 𝐾𝐾2 ]
0 0 0 1
[𝐴𝐴1 ] = �0 1 0 𝐾𝐾2 ] = �0 1� − �
� − � � [𝐾𝐾1 �=� �
0 0 1 0 0 𝐾𝐾1 𝐾𝐾2 −𝐾𝐾1 −𝐾𝐾2

162
 𝑠𝑠 0 0 1 𝑠𝑠 −1
𝑠𝑠[𝐼𝐼] − [𝐴𝐴1 ] = � �−� �=� �
0 𝑠𝑠 −𝐾𝐾1 −𝐾𝐾2 𝐾𝐾1 𝑠𝑠 + 𝐾𝐾2
|𝑠𝑠𝑠𝑠 − 𝐴𝐴1 | = 0; 𝑠𝑠 2 + 𝑠𝑠𝐾𝐾2 + 𝐾𝐾1 = 0 → (𝑖𝑖)
For given poles, (𝑠𝑠 − 𝑠𝑠1 )(𝑠𝑠 − 𝑠𝑠2 ) = 𝑠𝑠 2 − (𝑠𝑠1 + 𝑠𝑠2 )𝑠𝑠 + 𝑠𝑠1 𝑠𝑠2 = 0 → (𝑖𝑖𝑖𝑖)
Comparing the equations (𝑖𝑖) and (𝑖𝑖𝑖𝑖),
𝐾𝐾1 = 𝑠𝑠1 𝑠𝑠2 = (−1 + 𝑗𝑗)(−1 − 𝑗𝑗) = 1 − (−1) = 2
𝐾𝐾2 = −(𝑠𝑠1 + 𝑠𝑠2 ) = −(−1 + 𝑗𝑗 − 1 − 𝑗𝑗) = 2
The controller gain matrix, [𝐾𝐾] = [2 2]
Comparing the equation (𝑖𝑖𝑖𝑖) with the standard characteristic equation of
𝑠𝑠 2 + 2𝜁𝜁𝜔𝜔𝑛𝑛 𝑠𝑠 + 𝜔𝜔𝑛𝑛 2 = 0

𝜔𝜔𝑛𝑛 = �𝑠𝑠1 𝑠𝑠2 = �𝐾𝐾1 = √2

(𝑠𝑠1 + 𝑠𝑠2 ) 𝐾𝐾2 1
𝜁𝜁 = − = =
2𝜔𝜔𝑛𝑛 2�𝐾𝐾1 √2

Example:
Design a controller for a given third order system of 𝑦𝑦⃛ + 5𝑦𝑦̈ + 3𝑦𝑦̇ + 2𝑦𝑦 = 𝑢𝑢 such that
the third poles are, 𝑠𝑠1 = −4.8, 𝑠𝑠2,3 = −4.8 ± 𝑗𝑗3.6

𝑦𝑦⃛ = 𝑢𝑢 − 2𝑦𝑦 − 3𝑦𝑦̇ − 5𝑦𝑦̈

State equation:
𝑥𝑥1̇ 0 1 0 𝑥𝑥1 0
�𝑥𝑥2̇ � = � 0 0 𝑥𝑥
1 � � 2 � + �0� 𝑢𝑢
𝑥𝑥3̇ −2 −3 −5 𝑥𝑥3 1
Output equation:
𝑥𝑥1
𝑦𝑦 = [1 0 𝑥𝑥
0] � 2 �
𝑥𝑥3
Let the controller gain matrix, [𝐾𝐾] = [𝐾𝐾1 𝐾𝐾2 𝐾𝐾3 ]
𝑥𝑥1
Then the input equation, 𝑢𝑢 = −[𝐾𝐾1 𝐾𝐾2 𝐾𝐾3 ] �𝑥𝑥2 �
𝑥𝑥3
{𝑥𝑥̇ } = [𝐴𝐴1 ]{𝑥𝑥}

163
 0 1 0 0 0 0
[𝐴𝐴1 ] = [𝐴𝐴] − {𝐵𝐵}[𝐾𝐾] = � 0 0 1 �−� 0 0 0�
−2 −3 −5 𝐾𝐾1 𝐾𝐾2 𝐾𝐾3
0 1 0
=� 0 0 1 �
−2 − 𝐾𝐾1 −3 − 𝐾𝐾2 −5 − 𝐾𝐾3
𝑠𝑠 −1 0
|𝑠𝑠𝑠𝑠 − 𝐴𝐴1 | = � 0 𝑠𝑠 −1 �=0
2 + 𝐾𝐾1 3 + 𝐾𝐾2 𝑠𝑠 + 5 + 𝐾𝐾3
𝑠𝑠 3 + (5 + 𝐾𝐾3 )𝑠𝑠 2 + (3 + 𝐾𝐾2 )𝑠𝑠 + (2 + 𝐾𝐾1 ) = 0 → (𝑖𝑖)
The characteristic equation for given poles,
(𝑠𝑠 − 𝑠𝑠1 )(𝑠𝑠 − 𝑠𝑠2 )(𝑠𝑠 − 𝑠𝑠3 ) = 0
𝑠𝑠 3 + 14.4𝑠𝑠 2 + 82.08𝑠𝑠 + 172.8 = 0 → (𝑖𝑖𝑖𝑖)
Comparing the equations (𝑖𝑖) and (𝑖𝑖𝑖𝑖),
𝐾𝐾1 + 2 = 172.8; 𝐾𝐾1 = 170.8
𝐾𝐾2 + 3 = 82.08; 𝐾𝐾2 = 79.08
𝐾𝐾3 + 5 = 14.4; 𝐾𝐾3 = 9.4
Controller gain matrix, [𝐾𝐾] = [170.8 79.1 9.4]
The equation (𝑖𝑖𝑖𝑖) can be written as,
(𝑠𝑠 + 𝜁𝜁𝜔𝜔𝑛𝑛 )(𝑠𝑠 2 + 2𝜁𝜁𝜔𝜔𝑛𝑛 𝑠𝑠 + 𝜔𝜔𝑛𝑛 2 ) = 0
𝑠𝑠 3 + (3𝜁𝜁𝜔𝜔𝑛𝑛 )𝑠𝑠 2 + 𝜔𝜔𝑛𝑛 2 (1 + 2𝜁𝜁 2 )𝑠𝑠 + (𝜁𝜁𝜔𝜔𝑛𝑛 3 ) = 0
Comparing, 𝜁𝜁𝜔𝜔𝑛𝑛 3 = 172.8
�1�2�
3(172.8)
3𝜁𝜁𝜔𝜔𝑛𝑛 = 14.4; 𝜔𝜔𝑛𝑛 = � � =6
14.4
14.4
𝜁𝜁 = = 0.8
3(6)
𝜔𝜔𝑛𝑛 2 (1 + 2𝜁𝜁 2 ) = 82.08
The required Percent Overshoot and settling time can be determined from the values of
𝜔𝜔𝑛𝑛 and 𝜁𝜁.
−𝜁𝜁𝜁𝜁
4 �1−𝜁𝜁2
𝑇𝑇𝑠𝑠 = = 0.83 𝑠𝑠𝑠𝑠𝑠𝑠; 𝑃𝑃. 𝑂𝑂. = 100𝑒𝑒 = 1.52%
𝜁𝜁𝜔𝜔𝑛𝑛

164
 Ackermann’s Formula Method
This method is viable for higher order systems and it can be easily applied with computer
for the determination of controller gain matrix, [𝐾𝐾]
For the 𝑛𝑛𝑡𝑡ℎ order system, [𝐾𝐾] = [𝐾𝐾1 𝐾𝐾2 ⋯ 𝐾𝐾𝑛𝑛 ]

If the desired poles are, 𝑠𝑠1 , 𝑠𝑠2 ⋯ 𝑠𝑠𝑛𝑛 ;

The desired characteristic equation is,
𝑞𝑞(𝑠𝑠) = (𝑠𝑠 − 𝑠𝑠1 )(𝑠𝑠 − 𝑠𝑠2 ) ⋯ (𝑠𝑠 − 𝑠𝑠𝑛𝑛 ) = 𝑠𝑠 𝑛𝑛 + 𝛼𝛼𝑛𝑛−1 𝑠𝑠 𝑛𝑛−1 + ⋯ + 𝛼𝛼1 𝑠𝑠 + 𝛼𝛼0
Then, Ackermann’s formula for the controller gain matrix is,
[𝐾𝐾] = [0 0 ⋯ 1][𝑃𝑃𝑐𝑐 ]−1 [𝑞𝑞(𝐴𝐴)]
where [𝑃𝑃𝑐𝑐 ] is the controllability matrix and
[𝑞𝑞(𝐴𝐴)] = [𝐴𝐴]𝑛𝑛 + 𝛼𝛼𝑛𝑛−1 [𝐴𝐴]𝑛𝑛−1 + ⋯ 𝛼𝛼1 [𝐴𝐴] + 𝛼𝛼0 [𝐼𝐼]
This formula requires the inverse of the controllability matrix and hence it should have
non-zero determinant. So, the system should be controllable, and it should be verified
first before applying this method.

Example:
𝑥𝑥 ̇ 0 1 𝑥𝑥1 0
� 1� = � � �𝑥𝑥 � + � � 𝑢𝑢
𝑥𝑥2̇ 0 0 2 1
Design a controller by Ackermann’s method for desired poles of 𝑠𝑠1,2 = −1 ± 𝑗𝑗
[𝐾𝐾1 𝐾𝐾2 ] = [0 1][𝑃𝑃𝑐𝑐 ]−1 [𝑞𝑞(𝐴𝐴)]
[𝑃𝑃𝑐𝑐 ] = [{𝑃𝑃1 } {𝑃𝑃2 }]
0
{𝑃𝑃1 } = {𝐵𝐵} = � �
1
1
{𝑃𝑃2 } = [𝐴𝐴]{𝑃𝑃1 } = � �
0

[𝑃𝑃𝑐𝑐 ] = �0 1 |𝑃𝑃 |
� ; 𝑐𝑐 = −1
1 0

[𝑃𝑃𝑐𝑐 ]−1 = (−1) � 0 −1� = �0 1

�
−1 0 1 0
𝑞𝑞(𝑠𝑠) = (𝑠𝑠 − 𝑠𝑠1 )(𝑠𝑠 − 𝑠𝑠2 ) = 𝑠𝑠 + 2𝑠𝑠 + 2
For second order system, 𝑞𝑞(𝑠𝑠) = 𝑠𝑠 2 + 𝛼𝛼1 𝑠𝑠 + 𝛼𝛼0
Comparing these two equations, 𝛼𝛼1 = 2, 𝛼𝛼0 = 2

165
 [𝑞𝑞(𝐴𝐴)] = [𝐴𝐴]2 + 𝛼𝛼1 [𝐴𝐴] + 𝛼𝛼0 [𝐼𝐼] = �0 1� �0 1� + 2 �0 1� + 2 �1 0 2 2
�=� �
0 0 0 0 0 0 0 1 0 2

𝐾𝐾2 ] = [0 0 1 2 2
[𝐾𝐾1 1] � �� � = [2 2]
1 0 0 2

Observer Design
The design objective is to determine the observer gain matrix, [𝐿𝐿] for given desired pole
location. If a system is completely observable with a given set of outputs, it is possible to
determine (or to estimate) the states that are not directly measured (or observed).
For given state equation, {𝑥𝑥̇ } = [𝐴𝐴]{𝑥𝑥} + {𝐵𝐵}𝑢𝑢 and output equation, 𝑦𝑦 = [𝐶𝐶]{𝑥𝑥}
“Luenberger” provides the full state observer as,

{𝑥𝑥�}̇ = 𝐴𝐴𝑥𝑥� + 𝐵𝐵𝐵𝐵 + 𝐿𝐿(𝑦𝑦 − 𝐶𝐶𝑥𝑥�)

where {𝑥𝑥�}̇ is the ‘estimated’ value of states {𝑥𝑥} and {𝐿𝐿} is the observer gain matrix. The
goal of observer is to provide and estimate 𝑥𝑥� such that 𝑥𝑥� → 𝑥𝑥 as 𝑡𝑡 → ∞. Since we don’t
know 𝑥𝑥(𝑡𝑡0 ) precisely, we must provide an initial estimate, 𝑥𝑥�(𝑡𝑡0 ). Now, the error in
estimation can be defined as, 𝑒𝑒(𝑡𝑡) = 𝑥𝑥(𝑡𝑡) − 𝑥𝑥�(𝑡𝑡) and 𝑒𝑒(𝑡𝑡) → 0 as 𝑡𝑡 → ∞.

𝑒𝑒̇ = 𝑥𝑥̇ − 𝑥𝑥�̇

𝑥𝑥̇ = Ax + Bu; 𝑥𝑥�̇ = 𝐴𝐴𝑥𝑥� + 𝐵𝐵𝐵𝐵 + 𝐿𝐿[𝐶𝐶𝐶𝐶 − 𝐶𝐶𝑥𝑥�] = 𝐴𝐴𝑥𝑥� + 𝐵𝐵𝐵𝐵 + 𝐿𝐿𝐿𝐿𝐿𝐿
𝑒𝑒̇ = 𝐴𝐴(𝑥𝑥 − 𝑥𝑥�) − 𝐿𝐿𝐿𝐿𝐿𝐿; 𝑥𝑥 − 𝑥𝑥� = 𝑒𝑒
𝑒𝑒̇ = (𝐴𝐴 − 𝐿𝐿𝐿𝐿)𝑒𝑒
Let 𝑒𝑒(𝑡𝑡) = 𝑒𝑒 𝑠𝑠𝑠𝑠 ; 𝑒𝑒̇ = 𝑠𝑠𝑠𝑠 𝑠𝑠𝑠𝑠
𝑠𝑠𝑠𝑠 𝑠𝑠𝑠𝑠 = (𝐴𝐴 − 𝐿𝐿𝐿𝐿)𝑒𝑒 𝑠𝑠𝑠𝑠
𝑒𝑒 𝑠𝑠𝑠𝑠 [𝑠𝑠𝑠𝑠 − (𝐴𝐴 − 𝐿𝐿𝐿𝐿)] = 0
So, the determinant |𝑠𝑠𝑠𝑠 − (𝐴𝐴 − 𝐿𝐿𝐿𝐿)| = 0

166
Example:
Given,
𝑥𝑥 ̇ 2 3 𝑥𝑥1 0 𝑥𝑥1
� 1� = � � �𝑥𝑥 � + � � 𝑢𝑢; and 𝑦𝑦 = [1 0] �𝑥𝑥 �
𝑥𝑥2̇ −1 4 2 1 2

Design an observer such that the poles are 𝑠𝑠1,2 = −8 ± 6𝑗𝑗

𝐿𝐿 𝐿𝐿 0
𝐿𝐿𝐿𝐿 = � 1 � [1 0] = �𝐿𝐿1 �
𝐿𝐿2 2 0
(2 − 𝐿𝐿1 ) 3
[𝐴𝐴 − 𝐿𝐿𝐿𝐿] = � �
(−1 − 𝐿𝐿2 ) 4
𝑠𝑠 − (2 − 𝐿𝐿1 ) −3
𝑠𝑠[𝐼𝐼] − [𝐴𝐴 − 𝐿𝐿𝐿𝐿] = � �
(1 + 𝐿𝐿2 ) 𝑠𝑠 − 4
Determinant, |𝑠𝑠𝑠𝑠 − (𝐴𝐴 − 𝐿𝐿𝐿𝐿)| is,
𝑠𝑠 2 + 𝑠𝑠(𝐿𝐿1 − 6) + (3𝐿𝐿2 − 4𝐿𝐿1 + 11) = 0 → (𝑖𝑖)
For given poles, the characteristic equation is
(𝑠𝑠 − 𝑠𝑠1 )(𝑠𝑠 − 𝑠𝑠2 ) = 0
𝑠𝑠 2 − (𝑠𝑠1 + 𝑠𝑠2 )𝑠𝑠 + 𝑠𝑠1 𝑠𝑠2 = 0
𝑠𝑠 2 + 16𝑠𝑠 + 100 = 0 → (𝑖𝑖𝑖𝑖)
Comparing the equations (𝑖𝑖) and (𝑖𝑖𝑖𝑖),
𝐿𝐿1 − 6 = 16; 𝐿𝐿1 = 22
3𝐿𝐿2 − 4𝐿𝐿1 + 11 = 100; 𝐿𝐿2 = 177/3 = 59
The observer gain matrix,
𝐿𝐿 22
� 1� = � �
𝐿𝐿2 59
Comparing the equation (𝑖𝑖𝑖𝑖) with standard form,
𝑠𝑠 2 + 2𝜁𝜁𝜔𝜔𝑛𝑛 𝑠𝑠 + 𝜔𝜔𝑛𝑛 2 = 0
𝜔𝜔𝑛𝑛 = 10, 𝜁𝜁 = 0.8
From these values, appropriate 𝑃𝑃. 𝑂𝑂. and 𝑇𝑇𝑠𝑠 can be determined.

167
 Ackermann’s Formula method
For the 𝑛𝑛𝑡𝑡ℎ order system,
𝐿𝐿1
𝐿𝐿2
{𝐿𝐿} = � �
⋮
𝐿𝐿𝑛𝑛
If the desired poles are 𝑠𝑠1 , 𝑠𝑠2 , ⋯ 𝑠𝑠𝑛𝑛 ;
The desired characteristic equation is,
𝑝𝑝(𝑠𝑠) = 𝑠𝑠 𝑛𝑛 + 𝛽𝛽𝑛𝑛−1 𝑠𝑠 𝑛𝑛−1 + ⋯ + 𝛽𝛽1 𝑠𝑠 + 𝛽𝛽0
Then, Ackermann’s formula for the observer gain matrix is,
0
−1 0
{𝐿𝐿} = [𝑝𝑝(𝐴𝐴)][𝑃𝑃0 ] � �
⋮
1
where [𝑃𝑃0 ] is the observability matrix and
[𝑝𝑝(𝐴𝐴)] = [𝐴𝐴]𝑛𝑛 + 𝛽𝛽𝑛𝑛−1 [𝐴𝐴]𝑛𝑛−1 + ⋯ + 𝛽𝛽1 [𝐴𝐴] + 𝛽𝛽0 [𝐼𝐼]
This formula requires the inverse of the observability matrix and hence it should have
non-zero determinant. So, the system should be observable, and it should be verified first
before applying this method.

Example:
Given,
2 3 0
[𝐴𝐴] = � � ; {𝐵𝐵} = � � ; [𝐶𝐶] = [1 0]
−1 4 1
Design an observer such that the poles are, 𝑠𝑠1,2 = −8 ± 6𝑗𝑗
[𝑝𝑝1 ]
[𝑃𝑃0 ] = � �
[𝑝𝑝2 ]
[𝑝𝑝1 ] = [𝐶𝐶] = [1 0]
[𝑝𝑝2 ] = [𝑝𝑝1 ][𝐴𝐴] = [2 3]
1 0
[𝑃𝑃0 ] = � � ; |𝑃𝑃0 | = 3; system is observable.
2 3
1 1 0
[𝑃𝑃0 ]−1 = � � � 3 0
� = �− 2
3 −2 1 �3 �3�
1

168
For given poles (𝑠𝑠1 = −8 + 6𝑗𝑗, 𝑠𝑠1 = −8 − 6𝑗𝑗);
𝑝𝑝(𝑠𝑠) = 𝑠𝑠 2 + 𝛽𝛽1 𝑠𝑠 + 𝛽𝛽0 = 𝑠𝑠 2 + (𝑠𝑠1 + 𝑠𝑠2 )𝑠𝑠 + 𝑠𝑠1 𝑠𝑠2
𝛽𝛽0 = 𝑠𝑠1 𝑠𝑠2 = 100
𝛽𝛽1 = −(𝑠𝑠1 + 𝑠𝑠2 ) = 16
[𝑝𝑝(𝐴𝐴)] = [𝐴𝐴]2 + 𝛽𝛽1 [𝐴𝐴] + 𝛽𝛽0 [𝐼𝐼]
1 0
[𝑃𝑃(𝐴𝐴)] = � 2 3 2 3
�� � + 16 �
2 3
� + 100 � �=�
133 66
�
−1 4 −1 4 −1 4 0 1 −22 177
𝐿𝐿 0 133 66 1 0 0 22
� 1 � = [𝑝𝑝(𝐴𝐴)][𝑃𝑃0 ]−1 � � = � � �− 2 1� � �1� = �59�
𝐿𝐿2 1 −22 177 �3 3
𝐿𝐿 22
The observer gain matrix, � 1 � = � �
𝐿𝐿2 59

169
 Assessment

1. A fourth order system with state variables reduces to

(a) 2 second order system
(b) 1 third order and 1 first order system
(c) 4 first order system
(d) None of the above

2. The size of [𝐴𝐴] matrix in the state equation of a third order system is,
(a) 3 × 3
(b) 3 × 1
(c) 1 × 1
(d) 1 × 3

3. The size of [𝐶𝐶] matrix in the output equation of a third order system is,
(a) 3 × 3
(b) 3 × 1
(c) 1 × 1
(d) 1 × 3

4. If the 𝑇𝑇. 𝐹𝐹. of a system is given, the state equation can be obtained from
(a) Denominator polynomial
(b) Input function in time domain
(c) Numerator polynomial
(d) Output function in time domain

170
5. The output equation can be obtained from
(a) Denominator polynomial
(b) Input function in time domain
(c) Numerator polynomial
(d) Output function in time domain

6. The Routh array for stability requires

(a) characteristic equation
(b) |𝑍𝑍(𝑠𝑠)| = 0
(c) 𝑆𝑆[𝐼𝐼] − [𝐴𝐴] = 0
(d) All of the above

7. A system is controllable if
(a) [𝑃𝑃𝑂𝑂 ] = 0
(b) [𝑃𝑃𝐶𝐶 ] = 0
(c) [𝑃𝑃𝑂𝑂 ] ≠ 0
(d) [𝑃𝑃𝐶𝐶 ] ≠ 0

8. A system is observable if,

(a) [𝑃𝑃𝑂𝑂 ] = 0
(b) [𝑃𝑃𝐶𝐶 ] = 0
(c) [𝑃𝑃𝑂𝑂 ] ≠ 0
(d) [𝑃𝑃𝐶𝐶 ] ≠ 0

171
9. Ackermann’s formula used in design requires
(a) [𝑃𝑃𝑂𝑂 ] ≠ 0 and [𝑃𝑃𝐶𝐶 ] ≠ 0
(b) [𝑃𝑃𝑂𝑂 ] = 0 and [𝑃𝑃𝐶𝐶 ] ≠ 0
(c) [𝑃𝑃𝑂𝑂 ] = 0 and [𝑃𝑃𝐶𝐶 ] = 0
(d) [𝑃𝑃𝑂𝑂 ] ≠ 0 and [𝑃𝑃𝐶𝐶 ] = 0

10. Controller gain in Ackermann’s formula is a

(a) Column matrix
(b) Square matrix
(c) Row matrix
(d) Single value

11. Observer gain in Ackermann’s formula is a

(a) Column matrix
(b) Square matrix
(c) Row matrix
(d) Single value

172
 APPENDIX
MATLAB and SIMULINK

MATLAB
MATLAB is a tool for engineering computations and it is based on matrices. For
problems related to control system, MATLAB has a collection of special files (M-files) in
the “Control System Toolbox”
Writing a matrix in MATLAB:
1 2 3
[𝐴𝐴] = � 6 7 8�
11 12 13
In MATLAB write row by row >> 𝐴𝐴 = [1 2 3; 6 7 8; 11 12 13]
“>>” denotes the command prompt on the screen. Each row is separated by a semi-colon.
Writing a vector:
If a vector starts from 10 and increasing in steps of 5 up to 50, it can be written as
≫ 𝑥𝑥 = 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙(𝑛𝑛1, 𝑛𝑛2, 𝑛𝑛)
= 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙(10, 50, 5)
𝑥𝑥 = 10 15 20 25 30 35 40 45 50
Addition and Subtraction
2 3
>> 𝐴𝐴 = [2 3; 4 5; 6 7] ⇒ �4 5�
6 7
1 0
>> 𝐵𝐵 = [1 0; 3 4; 0 5] ⇒ �3 4�
0 5
3 3
>> 𝐶𝐶 = 𝐴𝐴 + 𝐵𝐵 ⇒ 𝐶𝐶 = 7 9
6 12
1 3
≫ 𝐶𝐶 = 𝐴𝐴 − 𝐵𝐵 ⇒ 1 1
6 2

173
Multiplication
1
≫ 𝑋𝑋 = [1; 3; 5] ⇒ 𝑋𝑋 = 3
5
2
≫ 𝑌𝑌 = [2; 4; 6] ⇒ 𝑌𝑌 = 4
6
2
′
≫ 𝑍𝑍 = 𝑋𝑋 ∗ 𝑌𝑌 ⇒ [1 3 ]
5 �4� = 44
6
1 2 4 6
≫ 𝑍𝑍 = 𝑋𝑋 ∗ 𝑌𝑌 ′ ⇒ �3� [2 4 6] = 6 12 18
5 10 20 30

A vector with complex number

≫ 𝑌𝑌 = [6 + 5 ∗ 𝑖𝑖 3 + 4 ∗ 𝑖𝑖 1 − 𝑖𝑖]; ⇒ [6 + 5𝑗𝑗 3 + 4𝑗𝑗 1 − 𝑗𝑗]

≫ 𝑌𝑌.^ 2 ⇒ [11 + 60𝑖𝑖 −7 + 24𝑖𝑖 0 − 2𝑖𝑖 ]

Roots of characteristic equation

𝑠𝑠 3 + 6𝑠𝑠 2 + 11𝑠𝑠 + 6 = 0
≫ 𝑝𝑝 = [1 6 11 6];
≫ 𝑟𝑟 = 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟(𝑝𝑝)
−3
𝑟𝑟 = −2 ⇒ 𝑠𝑠1 = −1, 𝑠𝑠2 = −2, 𝑠𝑠3 = −3
−1
𝑠𝑠 3 + 6𝑠𝑠 2 + 11𝑠𝑠 + 6 = (𝑠𝑠 − 𝑠𝑠1 )(𝑠𝑠 − 𝑠𝑠2 )(𝑠𝑠 − 𝑠𝑠3 ) = (𝑠𝑠 + 1)(𝑠𝑠 + 2)(𝑠𝑠 + 3)

Given the roots, find the characteristic equation

≫ 𝑟𝑟 = [−3 −2 −1];
≫ 𝑞𝑞 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝(𝑟𝑟)
𝑞𝑞 = 1 6 11 6
𝑞𝑞 = 𝑠𝑠 3 + 6𝑠𝑠 2 + 11𝑠𝑠 + 6

174
Eigenvalues
0 1 0
[𝐴𝐴] = �−1 0 2�
3 0 5
≫ A = [0 1 0; −1 0 2; 3 0 5]

5.213
≫ 𝑒𝑒𝑒𝑒𝑒𝑒[𝐴𝐴] ⇒ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸: −0.1065 + 1.4487𝑖𝑖
−0.1065 − 1.4487𝑖𝑖

Eigenvalues and Eigenvectors

0 1 0
[𝐴𝐴] = � 0 0 1�
−6 −11 −6
≫ 𝐴𝐴 = [0 1 0; 0 0 1; −6 −11 −6];
≫ [𝑋𝑋, 𝐷𝐷] = 𝑒𝑒𝑒𝑒𝑒𝑒(𝐴𝐴)
−1 0 0
Eigenvalues: 𝐷𝐷 = 0 −2 0
0 0 −3
−0.5774 0.2182 −0.1048
Eigen vector: 𝑋𝑋 = 0.5774 −0.4364 0.3145
−0.5774 0.8729 −0.9435

Inverse of a matrix
1 1 2
[𝐴𝐴] = �3 4 0�
1 2 5
≫ 𝐴𝐴 = [1 1 2; 3 4 0; 1 2 5];
≫ 𝑖𝑖𝑖𝑖𝑖𝑖(𝐴𝐴)
2.2222 −0.1111 −0.8889
[𝐴𝐴]−1 = � −1.6667 0.3333 0.6667 �
0.2222 −0.1111 0.1111
NOTE: All function names MUST be in LOWER CASE.
Example: 𝑖𝑖𝑖𝑖𝑖𝑖(𝐴𝐴), 𝑒𝑒𝑒𝑒𝑒𝑒(𝐴𝐴), 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝(𝐴𝐴) etc.

175
Partial Fraction
𝐵𝐵(𝑙𝑙) 𝑟𝑟1 𝑟𝑟2 𝑟𝑟𝑛𝑛
= + + ⋯+ + 𝑘𝑘
𝐴𝐴(𝑙𝑙) 𝑙𝑙 − 𝑙𝑙1 𝑙𝑙 − 𝑙𝑙2 𝑙𝑙 − 𝑙𝑙𝑛𝑛
Example:
𝐵𝐵(𝑙𝑙) 2𝑙𝑙 3 + 5𝑙𝑙 2 + 3𝑙𝑙 + 6
=
𝐴𝐴(𝑙𝑙) 𝑙𝑙 3 + 6𝑙𝑙 2 + 11𝑙𝑙 + 6

𝐵𝐵(𝑙𝑙) −6 −4 3
= + + +2
𝐴𝐴(𝑙𝑙) 𝑙𝑙 + 3 𝑙𝑙 + 2 𝑙𝑙 + 1

Given the above relation, the original function can be obtained from

176
Poles and Zeros of a Transfer Function
4𝑙𝑙 2 + 16𝑙𝑙 + 12
𝑇𝑇(𝑙𝑙) =
𝑙𝑙 4 + 12𝑙𝑙 3 + 44𝑙𝑙 2 + 48𝑙𝑙

MATLAB Command:

‘𝑧𝑧’, ‘𝑙𝑙’, ‘𝑘𝑘’ refer to Zeros, Poles, Gain, respectively. 𝑧𝑧2 ≈ −3, 𝑧𝑧1 ≈ −1
(𝑙𝑙 + 1)(𝑙𝑙 + 3)
𝑇𝑇(𝑙𝑙) = (4)
𝑙𝑙(𝑙𝑙 + 2)(𝑙𝑙 + 4)(𝑙𝑙 + 6)

The command to find T(s) from given z, p, k:

177
System Response
Let the Transfer Function for a second order system is,
2(𝑙𝑙 + 5) 𝑌𝑌(𝑙𝑙)
𝑇𝑇(𝑙𝑙) = =
𝑙𝑙 2 + 2𝑙𝑙 + 10 𝑅𝑅(𝑙𝑙)
For unit step input, 𝑅𝑅(𝑙𝑙) = 𝑈𝑈(𝑙𝑙)
𝑌𝑌(𝑙𝑙) 2𝑙𝑙 + 10
= 2
𝑈𝑈(𝑙𝑙) 𝑙𝑙 + 2𝑙𝑙 + 10
Let the time interval for response is 0 ≤ 𝑟𝑟 ≤ 8 in steps of every 0.01 second.

Performance Parameters
(Rise time, Peak time, Percent Overshoot, Settling Time)
25
𝑇𝑇(𝑙𝑙) =
𝑙𝑙 2 + 6𝑙𝑙 + 25
Step response for 0 ≤ 𝑟𝑟 ≤ 5
Time interval, ∆𝑟𝑟 = 0.005 𝑙𝑙𝑙𝑙𝑙𝑙

178
Create a new script in MATLAB then type the codes. After running it, the results can be
found in workspace. Correct answer is showing below.

Impulse and Ramp responses as a function of step response

𝑌𝑌(𝑠𝑠) 1
Let = = 𝑇𝑇(𝑙𝑙)
𝑅𝑅(𝑠𝑠) 𝑠𝑠 2 +0.2𝑠𝑠+1

Impulse: 𝑅𝑅(𝑙𝑙) = 1
Step: 𝑅𝑅(𝑙𝑙) = 1/𝑙𝑙
1
Impulse response = [𝑙𝑙𝑇𝑇(𝑙𝑙)] = step response of [𝑙𝑙𝑇𝑇(𝑙𝑙)]
𝑠𝑠
𝑙𝑙
𝑇𝑇1 (𝑙𝑙) = 𝑙𝑙𝑇𝑇(𝑙𝑙) =
𝑙𝑙 2 + 0.2𝑙𝑙 + 1

179
 1 1 1 1
Ramp: 𝑅𝑅(𝑙𝑙) = , Ramp response = � 𝑇𝑇(𝑙𝑙)� = step response of � 𝑇𝑇(𝑙𝑙)�
𝑠𝑠 2 𝑠𝑠 𝑠𝑠 𝑠𝑠
1
Let 𝑇𝑇(𝑙𝑙) = ; 0 ≤ 𝑟𝑟 ≤ 7; ∆𝑟𝑟 = 0.1
𝑠𝑠 2 +𝑠𝑠+1

1 1
𝑇𝑇1 (𝑙𝑙) = [𝑇𝑇(𝑙𝑙)] = 3
𝑙𝑙 𝑙𝑙 + 𝑙𝑙 2 + 𝑙𝑙

180
Dynamic System with initial conditions
Initial conditions:
𝑥𝑥(0) = 𝑥𝑥0 , 𝑥𝑥̇ 0 (0) = 𝐸𝐸0
Given:
𝑚𝑚 = 1𝑘𝑘𝑒𝑒, 𝑏𝑏 = 3𝑘𝑘𝑒𝑒/𝑙𝑙, 𝑘𝑘 = 2𝑁𝑁/𝑚𝑚, 𝑥𝑥0 = 0.1𝑚𝑚, 𝐸𝐸0 = 0.05𝑚𝑚/𝑙𝑙
Equation of motion: 𝑚𝑚𝑥𝑥̈ + 𝑏𝑏𝑥𝑥̇ + 𝑘𝑘𝑥𝑥 = 0
Laplace Transform of the above equation gives,

(𝑚𝑚𝑙𝑙 + 𝑏𝑏)𝑥𝑥0 + 𝑚𝑚𝐸𝐸0 (𝑙𝑙 + 3)(0.1) + (1)(0.05) 0.1𝑙𝑙 + 0.35

𝑋𝑋(𝑙𝑙) = = = 2
𝑚𝑚𝑙𝑙 2 + 𝑏𝑏𝑙𝑙 + 𝑘𝑘 𝑙𝑙 2 + 3𝑙𝑙 + 2 𝑙𝑙 + 3𝑙𝑙 + 2

1
Response = [𝑙𝑙𝑋𝑋(𝑙𝑙)] = step response of [𝑙𝑙𝑋𝑋(𝑙𝑙)]
𝑠𝑠
0.1𝑠𝑠 2 +0.35𝑠𝑠
Let 𝑋𝑋1 (𝑙𝑙) = 𝑙𝑙𝑋𝑋(𝑙𝑙) =
𝑠𝑠2 +3𝑠𝑠+2

Step response of 𝑋𝑋1 (𝑙𝑙)

181
Root Locus
Commands that are useful:
“conv”: convolution (To multiply polynomials)
“deconv”: deconvolution (To divide polynomials)
“rlocus”: To plot root locus diagram

Example: Transfer function:

𝐾𝐾(𝑙𝑙 + 3)
, 𝐾𝐾 = 1
𝑙𝑙(𝑙𝑙 + 1)(𝑙𝑙 2 + 4𝑙𝑙 + 16)
The range for x and y axes: −6 ≤ 𝑥𝑥 ≤ 6, −6 ≤ 𝑝𝑝 ≤ 6
Zero: 𝑙𝑙 = −3 Poles: Multiply 𝑙𝑙(𝑙𝑙 + 1)(𝑙𝑙 2 + 4𝑙𝑙 + 16) and then find the roots.
Let 𝑙𝑙 = 𝑙𝑙 2 + 𝑙𝑙, 𝑏𝑏 = 𝑙𝑙 2 + 4𝑙𝑙 + 16

𝑙𝑙 = 𝑙𝑙 4 + 5𝑙𝑙 3 + 20𝑙𝑙 2 + 16𝑙𝑙

Roots of 𝑙𝑙 2 + 4𝑙𝑙 + 16
𝑃𝑃𝑟𝑟𝑙𝑙𝑙𝑙𝑙𝑙: 0, −1, −2 ± 3.4641𝑙𝑙
𝑍𝑍𝑙𝑙𝑟𝑟𝑟𝑟: − 3

182
Root locus diagram

Bode Plot
25
𝐺𝐺(𝑙𝑙) =
𝑙𝑙 2 + 4𝑙𝑙 + 25

183
Nyquist Plot
1
𝐺𝐺(𝑙𝑙) =
𝑙𝑙 2 + 0.8𝑙𝑙 + 1

Phase Margin and Gain Margin

These are a measure of closeness of the polar plot to (−1 + 𝑗𝑗0) point.
𝐺𝐺𝑚𝑚: Gain Margin
𝑃𝑃𝑚𝑚: Phase Margin
𝑤𝑤𝑙𝑙𝑙𝑙: Phase cross over frequency (𝜔𝜔𝑜𝑜 )
𝑤𝑤𝑙𝑙𝑒𝑒: Gain cross over frequency (𝜔𝜔𝑐𝑐 )
Command:

Example:
20(𝑙𝑙 + 1)
𝐺𝐺(𝑙𝑙) =
𝑙𝑙(𝑙𝑙 + 5)(𝑙𝑙 2 + 2𝑙𝑙 + 10)

184
State-space Control
Controllability and Observability:
𝐶𝐶𝐶𝐶𝑁𝑁𝑇𝑇 = 𝑙𝑙𝑟𝑟𝑟𝑟𝑏𝑏(𝐴𝐴, 𝐵𝐵)
𝐶𝐶𝐵𝐵𝑂𝑂𝐸𝐸𝑅𝑅 = 𝑟𝑟𝑏𝑏𝑙𝑙𝐸𝐸(𝐴𝐴, 𝐶𝐶)
{𝑥𝑥̇ } = [𝐴𝐴]{𝑥𝑥} + {𝐵𝐵}𝐸𝐸
𝑝𝑝 = [𝐶𝐶]{𝑥𝑥}

Example:

0 1 0 0
Given: [𝐴𝐴] = � 0 0 1 � {𝐵𝐵} = �0� [𝐶𝐶] = [5 6 1]
−6 −11 −6 1

System is controllable since the det (𝐶𝐶𝐶𝐶𝑁𝑁𝑇𝑇) = −1

System is NOT observable since the det (𝐶𝐶𝐵𝐵𝑂𝑂𝐸𝐸𝑅𝑅) = 0

185
 List of MATLAB Commands

A selective list of commands and their description that may be useful in the Control
System course is given below.

Command Description
𝑎𝑎𝑎𝑎𝑎𝑎 Absolute value, Magnitude of complex number
𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 Phase angle of a complex number
𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 Arctangent [𝑡𝑡𝑡𝑡𝑡𝑡−1 ( )]
𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 Manual axis scaling
𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 Bode Plot
𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 Clear workspace
𝑐𝑐𝑐𝑐𝑐𝑐 Clear current figure
𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 Convolution, Multiplication of polynomials
𝑐𝑐𝑐𝑐𝑐𝑐 Cosine
𝑐𝑐𝑐𝑐𝑐𝑐ℎ Hyperbolic Cosine
𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 Controllability Matrix computation
𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 Deconvolution, Division of polynomials
𝑑𝑑𝑑𝑑𝑑𝑑 Determinant
𝑒𝑒𝑒𝑒𝑒𝑒 Eigenvalues and eigenvectors
𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 Laplace Transform frequency response
𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 Toggles the major lines of current axes
ℎ𝑒𝑒𝑒𝑒𝑒𝑒 Lists all help topics
𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 Imaginary part
𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 Impulse response
𝑖𝑖𝑖𝑖𝑖𝑖 Inverse of a matrix
𝑙𝑙𝑙𝑙𝑙𝑙 Natural logarithm
𝐿𝐿𝐿𝐿𝐿𝐿10 Log base 10
𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 Logarithmically spaced vector
𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 Simulate time response
𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛 Nyquist Plot
𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 Compute Observability matrix
𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 Linear x-y Plot
𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 Compute polynomial from roots
𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 Partial fraction
𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 Root Locus Plot
𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 Compute roots from polynomial
𝑠𝑠𝑠𝑠𝑠𝑠 Sine function
𝑠𝑠𝑠𝑠𝑠𝑠ℎ Hyperbolic sine
𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 Square root
𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 Unit-step response

186
 𝑠𝑠𝑠𝑠𝑠𝑠 Addition
𝑡𝑡𝑡𝑡𝑡𝑡 Tangent
𝑡𝑡𝑡𝑡𝑡𝑡ℎ Hyperbolic Tangent
𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 To position Text arbitrarily
𝑡𝑡𝑡𝑡 To find transfer function
𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 Plot title
𝑥𝑥𝑥𝑥𝑥𝑥𝑥𝑥𝑥𝑥𝑥𝑥 X-axis label
𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦 Y-axis label

187
SIMULINK
SIMULINK is a simulation tool to study the control system. The graphical output helps
to visually see the behavior of the system. The effect of changing the control gain or
transfer function parameters can be easily observed through rapidly changing output in
this simulation model. It is very useful in tuning the control system to satisfy the desired
performance parameters.
Model Building
The block diagram of a control system can be simulated with blocks available in
SIMULINK library. Some of the useful blocks from the library are given below.
Sources:

Sink:

A block has connecting point on the right side or left side or on both sides as shown
below.

The input port is the connecting point on the left and it brings input to the block. The
output port is the connecting point on the right and it takas output to another block or
scope.

188
Commonly used blocks:

Continuous Blocks

189
Working with Blocks

If poles and zeros are known, ZERO-POLE block can be used to simulate the transfer
function.
Let 𝑃𝑃𝑟𝑟𝑙𝑙𝑙𝑙: 𝑙𝑙 = −2, 𝑙𝑙 = −3 and 𝑍𝑍𝑙𝑙𝑟𝑟𝑟𝑟: 𝑙𝑙 = −6

190
Example:

In the MATLAB window type SIMULINK or choose SIMULINK from the top MENU
BAR, Choose feedback controller. First create a working area by selecting “NEW” under
File menu. Under “NEW” choose Blank model.
A working window named “untitled1” will appear, under “Tools” menu, choose Library
Browser.
Now you can start building a model by dragging the blocks from library and dropping
into the working window and connecting them. To connect two blocks, click on the
output port of one block and drag the mouse to the input port of another block. In the
example there are six blocks in forward loop and one block in feedback loop. Drag STEP,
SUM, GAIN, ZERO-POLE, TRANSFER FCN, and SCOPE blocks one at a time into the
working window. Connect all the blocks as shown. Drag another gain block and double-
click on it. From drop down options, choose “FLIP IT” to flip the Block and place it in
the feedback loop.
Open the step function block by double-clicking on it. Enter “0” for initial starting time.
Enter the initial value and final value as “0” and “1” respectively.

191
In the summing block double-click to open it and change the feedback sign from
‘plus(+)’ to ‘minus(−)’.
In the gain block, open the block and enter the value 10.
Open the zero-pole block and enter the Zero-Pole vectors as

Open the Transfer function block and enter numerator and denominator as a vector of
polynomial coefficients.

Open the feedback gain loop and enter “1”.

Running Simulation
From the simulation menu select “START” or select “RUN” on the Menu bar. To view
the result, open the SCOPE block by double-clicking on it. The output will be displayed
in a graph.

Save the model

It is important to save the model by selecting “save as” and giving a name. Then you can
come back to the model in future by choosing “open” in FILES Menu. It will show saved
Files. You can open a file. It is easy to change the gain or transfer function by opening
the respective block and entering the new values. It helps to study the output when the
parameters change without building another model or writing a program code.

192