Burkhard Vogel

Balanced
Phono-Amps
An Extension to the
‘The Sound of Silence’ Editions
Balanced Phono-Amps
Burkhard Vogel

Balanced Phono-Amps
An Extension to the ‘The Sound
of Silence’ Editions

123
Burkhard Vogel
Lab 6-11
Stuttgart
Germany

Additional material to this book can be downloaded from http://extras.springer.com.

Additional material to this book can be downloaded from http://extras.springer.com.

ISBN 978-3-319-18523-1 ISBN 978-3-319-18524-8 (eBook)

DOI 10.1007/978-3-319-18524-8

Library of Congress Control Number: 2015940724

Springer Cham Heidelberg New York Dordrecht London

© Springer International Publishing Switzerland 2016
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made.

Printed on acid-free paper

Springer International Publishing AG Switzerland is part of Springer Science+Business Media

(www.springer.com)
To Beate
Preface

In 2011, the second edition of my “The Sound of Silence” (TSOS-2) book appeared
on the markets. The integration of a broad range of valve solutions became the main
difference to the first edition (TSOS-1). In the July/August 2014 volume of their
JAES publication, the Audio Engineering Society published an article about “The
Vinyl Frontier”1 showing remarkable sales quantities of vinyl LPs in the UK: from
200,000 in 2009 to 780,000 in 2013. At the same time, I read in US, French and
German newspapers about an equally massive sales increase. Parallel to those
increase in LP sales, very interesting newly developed turntables and
phono-amplifiers of all kinds of technology entered (and still do) the markets. The
price range is huge too and a price of ≥15,000.00 EUR/≥18,000.00 $ for a
phono-amp or turntable is no longer impossible. Despite the still rather small
overall quantities, vinyl is back again, and it produces reasonable revenues and
profits.
Having studied the above-mentioned editions of my books, the observant reader
might stumble over the fact that the design of a fully balanced RIAA phono-amp is
missing. Finally, in these books, all mathematical- and design-oriented efforts led to
the semi-balanced “RIAA Phono-Amp Engine I” that includes different modules of
solid-state and triode-driven phono-amps; the triode module in the second edition
first. Semi-balanced, because Engine I offers balanced and/or un-balanced inputs,
followed by an un-balanced treatment of the RIAA transfer function creation—via
feedback path in the solid-state environment, via one passive network between two
triode gain stages. The outputs are balanced and un-balanced too.
The content of the herewith-presented TSOS-Extension2 shall fill the obvious
gap. No matter whether actively or passively configured, in this book on hand, fully
balanced means that each phono-amp stage ends up in a balanced—or in other
words symmetrical—solution, differentially amplified. Un-balanced/single-ended
intermediate solutions are not in the scope.

“The Vinyl Frontier”, Francis Rumsay, JAES Vol. 62, No. 7/8.
1

2
= TSOS-E.

vii
viii Preface

There are only two exceptions with un-balanced inputs:

1. In cases of input amplifiers for MM cartridges and the MM cartridge has a con-
nection from one of its output leads to the case (eg many Shure cartridges), and
2. In cases of turntables that offer un-balanced connectors and the user does not
want to install balanced cables.
In these cases, it makes sense to integrate un-balanced-in/balanced-out gain
stages via an external input.
Consequently, I call the presented rather complex phono-amp solution “RIAA
Phono-Amp Engine II”. It is thus a kind of platform fulfilling a high number of
design goals, focused mainly on MC cartridge usage. Among these goals, Engine II
offers the following:
• Many testing possibilities of very different active and/or passive amplifier
technologies and cartridge/turntable combinations.
• The selection of a simple-mode Engine II for private use or of a complete
test-purpose laboratory instrument.
• A deep insight into all design matters concerning electronic noise and stage
circuitry through extensive example calculations with Mathcad worksheets.
These worksheets include signal-to-noise ratio (SN) calculation approaches, and
all necessary calculation aspects concerning gain, input and output resistances, and
frequency and phase response-settings.
The inclusion of the TSOS-1/-2 indexes should ease follow-ups across the dif-
ferent books. Like in TSOS-1/-2, the lowest noise results and an excellent sound
production are still on top of my efforts.

Stuttgart, Germany Burkhard Vogel

March 2015
Contents

Part I The RIAA Phono-Amp Engine II

1 The Complete Engine II—Overview . . . . . . . . . . . . . . . . . . . . . . 3

1.1 Intro and Goals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 General Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Basic Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 Pictures of Cases and PCBs . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.4.1 Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.4.2 Printed Circuit Boards . . . . . . . . . . . . . . . . . . . . . . . 12
1.4.3 Front and Rear of the Engine . . . . . . . . . . . . . . . . . . 13
1.5 Power Supplies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.6 Mainboard . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2 The Triode Driven Central Amplifier Amp3 . . . . . . . . . . . . . . . . 21

2.1 General Design of Amp3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2 Gain and Noise Calculations . . . . . . . . . . . . . . . . . . . . . . . . 25
2.2.1 Gain of a DIF Followed by Two CFs . . . . . . . . . . . . 25
2.2.2 RIAA Transfer Function . . . . . . . . . . . . . . . . . . . . . 26
2.2.3 Noise and SN Calculations According to Fig. 1.2 . . . . 27
2.2.4 A Look into the Content of MCD-WS 3.1 . . . . . . . . . 29

3 Mathcad Worksheets Amp3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

3.1 MCD-WS: The Triode Gain Stage Amp3
with RIAA Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

4 The Solid-State (Op-Amp) Driven Central Amplifier Amp4 . . . . . 49

4.1 General Design of Amp4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.2 Gain and CMRR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.2.1 Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.2.2 RIAA Transfer Function . . . . . . . . . . . . . . . . . . . . . 54
4.2.3 CMRR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.3 Noise Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.3.1 Output Noise Voltage of the Amp4 Input Gain Stage
(2nd in Figs. 1.2 + 4.1) with OPs 3 & 4 . . . . . . . . . . 58

ix
x Contents

4.3.2 Output Noise Voltage of the Amp4 Output Gain

Stage (3rd in Figs. 1.2 + 4.1) with OPs 5 & 6 . . . . . . 59
4.3.3 Total Input and Output Noise Voltages of Amp4 . . . . 60
4.3.4 Noise and SN Calculations According to Fig. 1.2 . . . 60
4.3.5 A Look into the Content of MCD-WS 5.1 . . . . . . . . . 61
4.3.6 100 % Correlated Noise Voltages of Amp4 . . . . . . . . 62

5 Mathcad Worksheets Amp4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

5.1 MCD-WS: The Solid-State Gain Stage Amp4
with RIAA Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

6 The Op-Amp and Transformer Driven Output Stage Amp5 . . . . . 83

6.1 General Design and Gain of Amp5 . . . . . . . . . . . . . . . . . . . . 83
6.2 Power Supply. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
6.3 CMRR and Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
6.3.1 CMRR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
6.3.2 Noise and SNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
6.4 Additional Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

7 Mathcad Worksheets Amp5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

7.1 MCD-WS: The Op-Amp + Transformer Driven
Output Stage Amp5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

8 The Op-Amp and Transformer Driven Input Stage Amp1 . . . . . . 101

8.1 General Design and Gain of Amp1 . . . . . . . . . . . . . . . . . . . . 101
8.2 CMRR and Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
8.2.1 CMRR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
8.2.2 Noise and SNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
8.3 Measurement Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
8.4 Additional Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
8.4.1 DC Servo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
8.4.2 Wild Oscillation . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

9 Mathcad Worksheets Amp1 . . . . . ....................... 111

9.1 MCD-WS: The Transformer + Op-Amp Driven Amp1
(Real Data) . . . . . . . . . . . . . ....................... 112
9.2 MCD-WS: The Transformer + Op-Amp Driven Amp1
(Data Sheet Data) . . . . . . . . . ....................... 119

10 The BJT and Op-Amp Driven Input Amp2 . . . . . . . . . . . . . . . . . 125

10.1 General Design and Gain of Amp2 . . . . . . . . . . . . . . . . . . . . 125
10.2 CMRR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
Contents xi

10.3 Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 127

10.3.1 General Noise Aspects. . . . . . . . . . . . . . . . . ...... 127
10.3.2 The BJT Noise Model Reloaded . . . . . . . . . . ...... 128
10.3.3 In Search of the Slope Figures ‘x’ (2SC3329)
and ‘y’ (2SA1316) . . . . . . . . . . . . . . . . . . . . . . . . . 132
10.3.4 The SN Calculation Process . . . . . . . . . . . . . . . . . . . 137
10.3.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
10.4 Additional Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
10.4.1 Input Resistors R2, R3, R4 . . . . . . . . . . . . . . . . . . . 141
10.4.2 Wild Oscillation . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

11 Mathcad Worksheets Amp2 . . . . . . . . . . . . . . . . . . . . . . . . .... 143

11.1 MCD-WS: Evaluation of the 1/f-Noise Corner Frequency
of a 2SA1316 BJT . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 144
11.2 MCD-WS: Evaluation of the 1/f-Noise Corner Frequency
of a 2SC3329 BJT . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 148
11.3 MCD-WS: Amp2 SN and Gain Calculations—1/f-Noise
Based Version . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 152

12 Engine II Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

12.1 Visible and Audible Effects . . . . . . . . . . . . . . . . . . . . . . . . . 165
12.1.1 Visible Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
12.1.2 Audible Effects. . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
12.1.3 Test Records . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
12.1.4 Test Noise and Source Equipment. . . . . . . . . . . . . . . 168
12.1.5 Loudspeaker Situation and Headphones . . . . . . . . . . . 169
12.1.6 Listening Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
12.2 Measurement Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
12.2.1 Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
12.2.2 THD and IMD Matters, Left Channel . . . . . . . . . . . . 171
12.2.3 THD and IMD Matters, Right Channel . . . . . . . . . . . 175
12.2.4 General THD and IMD Matters . . . . . . . . . . . . . . . . 176
12.2.5 Summary Tables and Graphs . . . . . . . . . . . . . . . . . . 176
12.3 Conclusions and Final Remarks . . . . . . . . . . . . . . . . . . . . . . 180

Part II Knowledge Transfer

13 Selection of Draft Designs of Other Input Stages . . . . . . . . . .... 187

13.1 Intro . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 187
13.2 BJT/Op-Amps Driven MC Input Stage with Un-balanced
Input and Balanced Output. . . . . . . . . . . . . . . . . . . . . . .... 188
13.3 BJT/Op-Amps Driven MC Input Stage with Balanced
Transformer Input and Balanced Output. . . . . . . . . . . . . .... 189
xii Contents

13.4 Fully Triode Driven MC/MM Pre-Amp with Transformer

MC-Input and Balanced Output . . . . . . . . . . . . . . . . . . . . . . 191
13.5 Other Development Examples. . . . . . . . . . . . . . . . . . . . . . . . 195
13.5.1 The Joachim Gerhard Approach . . . . . . . . . . . . . . . . 195
13.5.2 The Ovidiu Popa Approach . . . . . . . . . . . . . . . . . . . 197
13.5.3 The Bob Cordell Approach . . . . . . . . . . . . . . . . . . . 199
13.6 The Output Stage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
13.7 Summary of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201

14 Mathcad Worksheets Draft Designs . . . . . . . . . . . . . . . . . ...... 205

14.1 MCD-WS: BJT/Op-Amp Driven MC Input Stage
with Un-Balanced Input and Balanced Output . . . . . . ...... 206
14.2 MCD-WS: BJT/Op-Amp Driven MC Input Stage
with Balanced Transformer Input and Balanced Output ...... 215
14.3 MCD-WS: Fully Triode Driven MC/MM Pre-Amp
with Transformer MC-Input and Balanced Output . . . . ...... 222

15 Measurement Tools and Trimming . . . . . . . ..... . . . . . . . . . . . 233

15.1 Computer Test Equipment . . . . . . . . . . ..... . . . . . . . . . . . 233
15.1.1 Intro . . . . . . . . . . . . . . . . . . . ..... . . . . . . . . . . . 233
15.1.2 Signal-to-Noise Ratio . . . . . . . ..... . . . . . . . . . . . 234
15.1.3 Distortion (THD) . . . . . . . . . . ..... . . . . . . . . . . . 236
15.1.4 IMD . . . . . . . . . . . . . . . . . . . ..... . . . . . . . . . . . 237
15.2 The Un-Balanced to Balanced Converter UBC . . . . . . . . . . . . 239
15.2.1 Circuit. . . . . . . . . . . . . . . . . . ..... . . . . . . . . . . . 239
15.2.2 F & P and SN Performance . . . ..... . . . . . . . . . . . 239
15.2.3 THD Performance. . . . . . . . . . ..... . . . . . . . . . . . 241
15.2.4 Output Resistances . . . . . . . . . ..... . . . . . . . . . . . 243
15.3 RIAA Encoder and Trimming . . . . . . . ..... . . . . . . . . . . . 244
15.3.1 Encoder . . . . . . . . . . . . . . . . . ..... . . . . . . . . . . . 244
15.3.2 Trimming Actions. . . . . . . . . . ..... . . . . . . . . . . . 244

16 The Very Low-Noise Balanced Measurement Amp PMMA . ..... 247

16.1 Intro . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 247
16.2 The Input Noise Voltage Density Question . . . . . . . . . ..... 249
16.3 The Roles of the MA Input Resistance and Input Noise
Current Density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
16.3.1 Influence of the DUT Output Resistance . . . . . . . . . . 251
16.3.2 The Input Noise Current Density Question . . . . . . . . . 252
16.3.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
16.4 The Final PMMA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
16.4.1 Principal Circuit Approach . . . . . . . . . . . . . . . . . . . . 256
16.4.2 Gain Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . 258
16.4.3 Noise Calculations—Rule-of-Thumb
SN Calculation Approach . . . . . . . . . . . . . . . ..... 260
Contents xiii

16.4.4 Noise Calculations—Detailed

SN Calculation Approach . . . . . . . . . . . . . . . . . . . . 261
16.5 The Complete PMMA Circuit . . . . . . . . . . . . . . . . . . . . . . . 264
16.6 PMMA Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
16.7 Practical Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
16.7.1 Ground Loop Avoidance and CMRR
of a Following Amp . . . . . . . . . . . . . . . . . . . . . . . . 268
16.7.2 Enclosure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
16.7.3 Room for Improvements . . . . . . . . . . . . . . . . . . . . . 269
16.8 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
16.8.1 DUT Output Resistance ≤10 Ω. . . . . . . . . . . . . . . . . 272
16.8.2 DUT Output Resistance >10 Ω. . . . . . . . . . . . . . . . . 272
16.8.3 Summary of Recommendations. . . . . . . . . . . . . . . . . 273
16.9 Final Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273

17 The Galvanically Isolated Measurement Amp PFMA . . . . . . . . . . 275

17.1 Intro . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
17.2 Ground Loop Avoidance . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
17.3 Additional PFMA Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
17.4 Gain and SN Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . 278
17.4.1 Gains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
17.4.2 Evaluation of Noise Voltages and SNs . . . . . . . . . . . 279
17.5 Enclosure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281

18 Mathcad Worksheets of Measurement Tools . . . . . . . . . . . . . . . . 283

18.1 MCD-WS: The UBC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
18.2 MCD-WS: The PMMA . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
18.3 MCD-WS: The PFMA . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302

19 A Unique MM Phono-Amp Noise Reduction Method . . . . . . . . . . 307

19.1 Intro . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
19.2 The Un-Balanced Noise Reduction Approach . . . . . . . . . . . . . 307
19.2.1 Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
19.2.2 The M1 ELS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
19.2.3 The M2 ELS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
19.2.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
19.2.5 Consequences. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
19.3 The Balanced Noise Reduction Approach . . . . . . . . . . . . . . . 314
19.3.1 Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
19.3.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
19.3.3 Consequences. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
19.4 SN Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
xiv Contents

20 Mathcad Worksheets of the MM Noise Reduction . . . . . . . . . . . . 319

20.1 MCD-WS: The Un-Balanced Version . . . . . . . . . . . . . . . . . . 320
20.2 MCD-WS: The Balanced Version . . . . . . . . . . . . . . . . . . . . . 332

21 BJT Circuits in CE Configuration . . . . . . . . . . . . . . . . . . . . . ... 341

21.1 Intro . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 341
21.2 BJT—Bipolar Junction Transistor—Basics . . . . . . . . . . . . ... 341
21.2.1 Equations for Low Frequency Small
Signal Calculations . . . . . . . . . . . . . . . . . . . . . . . . . 341
21.2.2 Circuit Parameter Based Formulae . . . . . . . . . . . . . . 343
21.2.3 Noise of a BJT—Frequency Independent Version . . . . 343
21.2.4 Noise of a BJT—Frequency Dependent Version . . . . . 344
21.2.5 Noise of a Resistor R . . . . . . . . . . . . . . . . . . . . . . . 345
21.2.6 Noise Factor & Noise Figure . . . . . . . . . . . . . . . . . . 345
21.2.7 Signal-to-Noise Ratios SN . . . . . . . . . . . . . . . . . . . . 346
21.3 Basic (b) CEb Circuit. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346
21.3.1 Idle Gains Gb and Gb.rot. . . . . . . . . . . . . . . . . . . . . . 347
21.3.2 Gb(RL) = RL Dependent Gain Gb . . . . . . . . . . . . . . . 347
21.3.3 Input Resistance ri (O/P Open) . . . . . . . . . . . . . . . . . 347
21.3.4 Output Resistance ro.o (I/P = Open). . . . . . . . . . . . . . 347
21.3.5 Output Resistances ro.s and ro.s.rot (I/P = Shorted) . . . . 348
21.3.6 Operating Gains Gop(R0, RL) and Gops(f, R0, RL) . . . 348
21.3.7 Noise—Frequency Independent Version. . . . . . . . . . . 349
21.3.8 Noise—Frequency Dependent Version. . . . . . . . . . . . 351
21.4 CE Circuit CEcf with Current Feedback . . . . . . . . . . . . . . . . . 351
21.4.1 Gains Gcf and Gcf.rot . . . . . . . . . . . . . . . . . . . . . . . . 351
21.4.2 Input Resistances ri and ri.rot (O/P Open) . . . . . . . . . . 352
21.4.3 Output Resistances ro.o and ro.o.rot (I/P = Open) . . . . . 353
21.4.4 Output Resistances ro.s and ro.s.rot
(I/P = Shorted) . . . . . . . . . . . . . . . . . . . . . . . . . ... 353
21.4.5 Operating Gains Gop(R0, RL) and Gops(f, R0, RL) ... 353
21.4.6 Noise and SN. . . . . . . . . . . . . . . . . . . . . . . . . . ... 355
21.5 CE Type 2 Circuit CEvcf2 with Voltage Feedback
and Current Feedback . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
21.5.1 Gain Gvcf2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
21.5.2 Input Resistance ri (O/P Open) . . . . . . . . . . . . . . . . . 357
21.5.3 Output Resistance ro.s (I/P Shorted) . . . . . . . . . . . . . . 357
21.5.4 Output Resistance ro.o (I/P Open) . . . . . . . . . . . . . . . 358
21.5.5 Other Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 358
21.5.6 Noise and SN. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358
21.6 Correction of a TSOS-1 and TSOS-2 Gain Result . . . . . . . . . . 359
21.7 The CE in Series Configuration with an Op-Amp . . . . . . . . . . 360
21.7.1 Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
21.7.2 Noise and SN. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
Contents xv

22 Differential (DIF) Amps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365

22.1 Intro . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
22.2 The DIFA-OPA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366
22.2.1 Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366
22.2.2 Noise Calculations Version 1 . . . . . . . . . . . . . . . . . . 367
22.2.3 Noise Calculations Version 2 . . . . . . . . . . . . . . . . . . 368
22.2.4 Noise Calculations Version with One Input
Lead Grounded. . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
22.3 The DIFA-IC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
22.3.1 Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
22.3.2 Noise Calculations Version 1 . . . . . . . . . . . . . . . . . . 371
22.3.3 Noise Calculations Version 2 . . . . . . . . . . . . . . . . . . 372

23 Mathcad Worksheets of DIF Amps . . . . . . . . . . . . . . . . . . . . . . . 375

23.1 MCD-WS: DIFA-OPA . . . . . . . . . . . . . . . . . . . . . . . . . . . . 376
23.2 MCD-WS: DIFA-IC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 380

24 Old Stuff? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385

24.1 Intro . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385
24.2 The BRAUN Tracking Force Measurement Instrument . . . . . . 387
24.3 Professional Test and Calibration Records . . . . . . . . . . . . . . . 387
24.4 Final Note . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 390

Appendix 1: List of Mathcad Worksheets . . . . . . . . . . . . . . . . . . . . . . 391

Appendix 2: Useful Literature and Web Sites . . . . . . . . . . . . . . . . . . . 393

Index TSOS-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397

Index TSOS-2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403

Index TSOS-E . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415

Abbreviations and Symbols

(Expanded by the ones of TSOS-1 and TSOS-2)

A Appendix
A Amplitude (eg in an FFT diagram)
 Peak voltage of a voltage A
(A) A-weighting (eg in dBV(A))
a A-weighted (subscript)
a Valve anode or plate
AC Alternating current
ADC Analog–digital converter
AES Audio Engineering Society
Amp Amplifier
amp Amplifier (subscript)
ampx Amplifier x
arm Average metre reading
ass Assuming (subscript)
b Bypassed (subscript)
b Balanced (subscript)
b BJT base (subscript)
b′ BJT internal base (subscript)
ß Auxiliary symbol for the ratio of resistances
B Bandwidth (in general)
B BJT base
BE Subscript of a resistance R formed by BE = rbb′||RE
BJT Bipolar junction transistor
BNC Standardised high-frequency connection system
boost Booster
bp Band-pass filter
BTFMI BRAUN tracking force measurement instrument
Butt Butterworth
bv Big volume
C BJT collector
c BJT internal collector (subscript)
C Capacitance or capacitor

xvii
xviii Abbreviations and Symbols

c Corner (subscript)
c Valve cathode (also as subscript)
ca Contribution allowed
calc Calculated (subscript)
Cc Cathode capacitance
CCA CCS with CCG as anode load (also as subscript)
CCG Constant current generator (also as subscript)
CCIR Comité Consultatif International des Radiocommunications (became
later-on ITU-R)
CCS Common cathode gain stage (also as subscript)
CE Common emitter configuration
CF Cathode follower (also as subscript)
cf Corner frequency
CGS Common gate gain-stage (also as subscript)
Cheb Chebyshev
Cinch Audio connection system (equivalent to the RCA system)
CMRR Common mode rejection ratio
CMRRe CMRR in [dB]
CMS Computer (based) measurement system
c1 Proportional factor for RIAA network-type (E) calculations
D FET drain
d Decade (calculation of resistor excess noise)
dB Decibel
DA Digital-to-Analog
DC Direct current
DCS DC servo
dcs DC servo (subscript)
DDR Deutsche Demokratische Republik (see GDR, disappeared 1989)
dif Differential (subscript)
diff Difference, different
DIFA Differential gain-stage
DIFA Fully differential amplifier
DIN Deutsche Industrie Norm (German Industrial Standard)
D/S Douglas Self (author)
d.u.t. Device under test
DUT Device under test
dx Distortion harmonic number x (subscript)
e BJT internal emitter (subscript)
e Noise voltage
e 20*log(xyz) (subscript)
E BJT emitter
ein Equivalent input noise voltage density
EIN Equivalent rms input voltage
ENB Equivalent noise bandwidth
ENCD Equivalent noise current density
Abbreviations and Symbols xix

ENVD Equivalent noise voltage density

eon Output noise voltage density
EON Output rms noise voltage
EU Europe, European
EW Electronics World (magazine)
EW&WW Electronics World and Wireless World (older version of EW)
ex Excess noise voltage (subscript)
ex Excluding rumble (subscript)
f Frequency
fc Corner frequency (eg fc.e = noise voltage based, fc.i = noise current
based)
F Frequency response
Fc Worsening factor based on the low-frequency noise corner frequency
FET Field-effect transistor
ffm Frankfurt/Main (German financial hub)
FFT Fast Fourier Transformation
Fig Figure
FM Flank modulation
FS FFT size
fS Sample rate
G FET gate
G Gain
g Valve grid
Gb Gain of the bypassed gain-stage
GDR German democratic republic (disappeared 1989)
gm Mutual conductance
G0 Idle gain
Gu Gain of the un-bypassed gain-stage
Gx Gain of stage x
HF Hum figure
hFE BJT DC current gain
H&N Hum and noise
HP Unit width (of a 19″ case)
hp High-pass filter
HTGG-2 How to Gain Gain, 2nd ed
HU Unit height (of a 19″ case)
Hz hertz = cycles per second (cps)
I DC current
Ia Anode DC current
i AC current
i in (subscript)
IC Integrated circuit, also specifying a DIFA
Ic Cathode DC current
id Ideal
iL Input load (subscript)
xx Abbreviations and Symbols

IMD Intermodulation distortion

in eg with Rin = input resistance
i/p Input
IN Input load
ISCE Institute of sound and communications engineers, UK
ITU-R International Telecommunication Union—ITU Radiocommunication
Sector (formerly CCIT and CCIR)
J Jack
JAES Journal of the Audio Engineering Society
JFET Junction field-effect transistor
Jmp Jumper (incl. header)
JT Jensen transformers
K Kelvin [K]
k Bolzmann’s constant
k Kilo
L Inductance
L Left channel
L Load (subscript)
lat.mod Lateral modulated (subscript)
LB TR of the ex-GDR
lin Linear
LL Lundahl
LM Lateral modulation
log Logarithmic
LP Long play vinyl record
lp Low-pass filter
LTE Letter to the editor
LTP Long-tailed pair
lv Low volume
m Measured (subscript)
m Subscript of mutual conductance
MA Measurement amplifier
MC Moving coil (cartridge)
M/C Motchenbacher/Connelly (authors)
MCD Mathcad
mcd Mathcad
meas Measured (subscript)
MM Moving magnet (cartridge)
mr Metre reading (subscript)
MS Measurement system
MSR Maxi single vinyl record
N Noise (as subscript mostly used to express “in B20k”)
N Noise gain (subscript)
N Noise of a valve (subscript of noise resistance)
n Noise (as subscript mostly used to express “in B1”)
Abbreviations and Symbols xxi

n NPN (subscript)
n Secondary trafo turns divided by primary turns, thus, tr becomes 1:n
Nfc Nyquist frequency
NAB National Association of Radio and Television Broadcasters
(ex NARTB = a US organisation)
NCD Noise current density
ne Non-equalised (subscript)
NF Noise factor
NFe Noise figure (20log(NF))
NI Current noise index of resistors
NL Noise level
nom Nominal (subscript)
NVD Noise voltage density
o Out (also as subscript)
ocm Common mode pin of DIFA-ICs
OP Op-amp
op Op-amp (subscript)
o/p Output
OPA Op-amp (also specifying a DIFA)
ops Operating (subscript)
P Phase response
P Potentiometer
p Peak (subscript)
p Pentode (subscript)
p PNP (subscript)
p ponderé = weighted (subscript)
PA Power amp
pa Power amp (subscript)
par Parallel
PFMA Galvanically isolated measurement amp
pham Valve phono-amp
PL Plug
PMMA Fully balanced measurement amp
ppa Pre-preamp
Prof. German title for a university professor
PSU Power supply unit
q Quasi peak (subscript)
R Resistance or resistor (equivalent unit symbol for ohm [Ω])
R Right channel
Ra Anode resistor
ra Internal valve anode resistance
rbb′ BJT base spreading resistance (rbb on MCD worksheets)
rc Internal valve cathode resistance
rc Real case (subscript)
Rc Cathode resistor
xxii Abbreviations and Symbols

RCA see Cinch

re Real
real Real (subscript)
ref Reference, referenced to
Rel Relay
res Resolution (subscript)
Rf Feedback resistance
RG Gain setting resistance in DIFAs
Rg Bias setting grid resistor
Rgg Oscillation preventing grid resistor
RIAA Radio Industry Association of America, a standard setting US
organisation
riaa dto. as subscript
rms Root mean square (subscript)
rN Valve (tube) equivalent noise resistance (white noise only) (also as
subscript)
rNc Average equivalent noise resistance in a certain bandwidth that
includes the valve’s low-frequency noise (also as subscript)
Ro Output resistance
rot Rule of thumb (subscript)
RP Paralleled resistors
rpm Rotations per minute
rt Root
r1 Proportional factor for RIAA network-type (E) calculations
S Sample (subscript)
S S-filter (special noise measurement hp-filter)
S Siemens
S Source (subscript)
S Source of a FET
s Second
sec Secondary
seq Sequence or sequential
ser Serial
SN Signal-to-noise ratio
SNa Improvement of white noise after A-weighting (−2.046 dB in B20k)
SNa SN of a noise voltage after A-weighting
SNar Improvement of white noise after A-weighting and RIAA equalising
(−7.935 dB in B20k)
SNariaa SN of a noise voltage after RIAA equalisation and A-weighting
SNne SN non-equalised and non-weighted
SNr Improvement of white noise after RIAA equalising (−3.646 dB in
B20k)
SNra Improvement of an RIAA equalised noise voltage after A-weighting
(−4.289 dB in B20k = SNar − SNr)
SNriaa SN of a noise voltage after RIAA equalisation
Abbreviations and Symbols xxiii

SNsriaa SN of a noise voltage after RIAA equalisation and S-weighting

sol Solution
SR Single vinyl record
SRPP Shunt regulated push-pull
sst Schallplatten-Schnitt (vinyl cut)
succ-app(s) Successive approximation(s)
Sx Switch number x
T Transfer function
t Triode (subscript)
T-BNC BNC-based connection system for shielded twisted or paralleled
wires
TC Test and calibration
TCR Test and calibration record
THD Total harmonic distortion
tot Total (subscript)
TR Test record
tr Transformer turns ratio (eg: 3:11)
trafo Transformer
T/S Tietze/Schenk (authors)
TSOS-1 The sound of silence, 1st ed
TSOS-2 The sound of silence, 2nd ed
TSOS-E Balanced phono-amps, 1st ed. (E stands for extension to ….)
TSR TC records 1005 and 1007
TT Test terminal
Tx Time constant x
u Un-bypassed (subscript)
UBC Un-balanced to balanced converter
ub Un-balanced (subscript)
u0 Source signal voltage
v Signal voltage
Va Anode DC voltage between anode and cathode
VA Early voltage
Vc Cathode DC voltage between cathode and ground or cold end of the
grid resistor
Vcc DC supply voltage positive
VDC DC voltage
Vee DC supply voltage negative
VC-A Voltage-to-current-amp
ver.mod Vertical modulated (subscript)
VM Vertical modulation
VR Vinyl record
VV-A Voltage-to-voltage-amp
Vx Amplifying stage or device x
v0 Source signal voltage
WFe SN worsening figure based on the 1/f-noise corner frequency
xxiv Abbreviations and Symbols

wn White noise region (subscript)

WW Wireless World (oldest version of EW magazine)
wyciwym What you calculate is what you measure
Wz SN worsening factor based on operating point setting components
Wz.e SN worsening figure
x BJT (NPN) noise current 1/f-noise slope power figure
XLR Standardised balanced connection system
y BJT (PNP) noise current 1/f-noise slope power figure
Z Impedance formed of different components (R and/or C and/or L)
z Indicates impedances, eg in Wz
Zf Feedback impedance
Zi(n) Input impedance
ZiL Input load impedance
µ Gain of a triode
µF µ-follower (also as subscript)
0 Symbolises source, eg R0 = source resistance
0 Source (subscript)
0s Reference of a sound-programme level (here: 0 dBV)
1 Double-triode system one (subscript)
2 Double-triode system two (subscript), etc.
5 i/p load of 5 Ω (subscript)
10 10 Hz (subscript)
20 i/p load of 20 Ω (subscript)
33 33 1/3 rpm (vinyl record)
40 i/p load of 40 Ω (subscript)
43 i/p load of 43 Ω (subscript)
45 45 rpm (vinyl record)
1k 1 kHz (subscript)
1k i/p load of 1 kΩ (subscript)
20k Like 1 k
45° Flank modulation
45°.mod Flank modulated (subscript)
|| Parallel
Physical Constants and Mathematical Sizes

B1 Bandwidth of 1 Hz
B10k Bandwidth of 10 kHz–20 kHz
B20k Bandwidth of 20 Hz–20 kHz (19,980 Hz)
B22k Bandwidth of 20 Hz–22 kHz
B26k Bandwidth of 20 Hz–26 kHz
k 1.38065 × 10−23 V•A•s K−1 = Boltzmann’s constant
q 1.6022 × 10−19 A•s = electron charge
SNne Non-equalised (ne) SN of a white noise source
SNa −2.046 dB = SN improvement by A-weighting of SNne in B20k
SNar −7.935 dB = SN improvement by RIAA equalising and A-weighting
of SNne in B20k
SNr −3.646 dB: SN improvement by RIAA equalising of SNne in B20k
SNra −4.289 dB: SN improvement by A-weighting of SNr in B20k
(SNr + SNra = SNar) Note: These SN improvements work only
in hum-free environments and with a white noise-based SNne
T 300 K for solid-state circuits
T 315 K inside valve amp cases
1k FFT size 210
2k FFT size 211
4k FFT size 212
8k FFT size 213
16k FFT size 214
32k FFT size 215
64k FFT size 216

xxv
List of Figures

Figures on text pages and on MCD Worksheets

Figure 1.1 Block diagram of the Engine II . . . . . . . . . . . . . . . . . . . .. 4

Figure 1.2 RIAA transfer function creation and its insertion
into an active and switchable amp chain . . . . . . . . . . . . . .. 7
Figure 1.3 Input alternative of Fig. 1.2. . . . . . . . . . . . . . . . . . . . . . .. 8
Figure 1.4 75 µs RIAA network at the input of Amps 3 & 4 . . . . . . .. 8
Figure 1.5 Situation of the 318 µs/3180 µs RIAA network
at the output of Amp3 . . . . . . . . . . . . . . . . . . . . . . . . . .. 10
Figure 1.6 Situation of the 318 µs/3180 µs RIAA network
at the output of Amp4 . . . . . . . . . . . . . . . . . . . . . . . . . .. 11
Figure 1.7 Front of Engine II, top two fully equal engine channels
L & R; bottom left and right two ±21 V/0.5 A PSU
insertion units, middle one triode PSU with +200 V/75
mA, 2 × 6.3 V/1.5 A insertion unit . . . . . . . . . . . . . . . . . . 11
Figure 1.8 Rear of Engine II. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Figure 1.9 Look on the fully equipped Mainboard of the right channel . . . 12
Figure 1.10 Rear and look into the case of one channel . . . . . . . . . . . . . 13
Figure 1.11 The first set of plug-in PCBs for Amp5 (fully left),
Amp 2 (middle), and Amp1 (fully right) . . . . . . . . . . . . . . . 14
Figure 1.12 Front of one channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Figure 1.13 Rear of one channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Figure 1.14 Main ±21 V power supply for all solid-state driven amps . . . 16
Figure 1.15 +200 V power supply for the triode driven section. . . . . . . . 17
Figure 1.16 Two +6.3 V regulated triode heater power supplies . . . . . . . 18
Figure 1.17 Wiring on the Mainboard and to the outside world . . . . . . . 19
Figure 2.1 Amp3 without RIAA networks . . . . . . . . . . . . . . . . . . . . . 22
Figure 2.2 Output noise voltage density of the amp sequence
Amp3 + Amp5 + Trafo with input shorted . . . . . . . . . . . . . 24
Figure 2.3 Calculated deviation from the exact RIAA transfer. . . . . . . . 25
Figure 3.1 Triode driven Amp3 incl. RIAA networks . . . . . . . . . . . . . . 32
Figure 3.2 Gain of Amp3 vs frequency . . . ....... . . . . . . . . . . . . . . 34
Figure 3.3 T2 defining network . . . . . . . . ....... . . . . . . . . . . . . . . 35
Figure 3.4 Bode plot of GT2(f). . . . . . . . . ....... . . . . . . . . . . . . . . 35

xxvii
xxviii List of Figures

Figure 3.5 T1 & T3 defining network. . . . . . . . . . . . . . . . . . . . . . . . . 36

Figure 3.6 Bode plot of GT1.3(f). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Figure 3.7 Normalized (1kHz at 0dB) RIAA transfer function . . . . . . . . 37
Figure 3.8 Bode plot of the gain of the combined RIAA network . . . . . . 37
Figure 3.9 Deviation from the exact RIAA transfer . . . . . . . . . . . . . . . 37
Figure 3.10 Bode plot of Amp3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Figure 3.11 = Fig. 2.3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Figure 3.12 Fig. 3.10’s phase response . . . . . . . . . . . . . . . . . . . . . . . . 38
Figure 3.13 Frequency response of the noise voltage density
of the T2(f) network . . . . . . . . . . . . . . . . . . . . . . . . ..... 40
Figure 3.14 Frequency response of the noise voltage density
of the T1(f) + T3(f) network . . . . . . . . . . . . . . . . . . ..... 41
Figure 3.15 a Amp3+Amp5 output noise voltage density
with i/p shorted. b = Fig. 2.2 . . . . . . . . . . . . . . . . . ..... 43
Figure 3.16 Comparison of the various output noise voltage
densities vs. frequency (trace 3 is nearly hidden
by trace 2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 46
Figure 3.17 Deviation from the exact RIAA transfer
of Amp1+Amp3+Amp5 . . . . . . . . . . . . . . . . . . . . . . ..... 46
Figure 4.1 Amp4 without RIAA networks . . . . . . . . . . . . . . . . ..... 50
Figure 4.2 Output noise voltage density of the amp sequence
Amp4 + Amp5 + Trafo with input shorted . . . . . . . . ..... 51
Figure 4.3 Calculated deviation from the exact RIAA transfer. . . ..... 51
Figure 4.4 Two-op-amp fully differential amplifier . . . . . . . . . . ..... 52
Figure 4.5 Basic Amp4 circuit with input stage around OPs 3 & 4,
output stage around OPs 5 & 6, and all relevant
signal voltages. . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 53
Figure 4.6 Symbol for a fully differential amplifier with balanced
electrometer input . . . . . . . . . . . . . . . . . . . . . . . . . ..... 54
Figure 4.7 General noise voltage and current situation of the
Amp4 input (2nd) gain stage . . . . . . . . . . . . . . . . . . ..... 57
Figure 4.8 General noise voltage and current situation of one
Amp4 subtractor of the 3rd gain stage . . . . . . . . . . . ..... 57
Figure 4.9 Detailed output noise voltage situation of the 2nd gain
stage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 63
Figure 5.1 Op-amp driven Amp4 incl. RIAA networks . . . . . . . . . . . . . 66
Figure 5.2 T2 defining network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Figure 5.3 Bode plot of GT2(f). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Figure 5.4 T1 & T3 defining network. . . . . . . . . . . . . . . . . . . . . . . . . 68
Figure 5.5 Bode plot of GT1(f) & GT3(f) . . . . . . . . . . . . . . . . . . . . . . . 69
Figure 5.6 Normalized (1 kHz at 0 dB) RIAA transfer function . . . . . . . 69
Figure 5.7 Bode plot of the combined RIAA network . . . . . . . . . . . . . . 70
Figure 5.8 Deviation from the exact RIAA transfer . . . . . . . . . . . . . . . 70
List of Figures xxix

Figure 5.9 Bode plot of Amp4 . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 70

Figure 5.10 = Fig. 4.3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 71
Figure 5.11 Fig. 5.9’s phase response . . . . . . . . . . . . . . . . . . . . . . ... 71
Figure 5.12 Noise voltage and current situation of the 2nd gain
stage (OP3 & OP4) . . . . . . . . . . . . . . . . . . . . . . . . . . ... 72
Figure 5.13 Noise voltage and current situation of the 3rd gain
stage (OP5 & OP6) . . . . . . . . . . . . . . . . . . . . . . . . . . ... 72
Figure 5.14 Frequency response of the noise voltage density
of the T2(f) network . . . . . . . . . . . . . . . . . . . . . . . . . . ... 73
Figure 5.15 Bode plot of the balanced o/p noise voltage density
of the 2nd gain stage . . . . . . . . . . . . . . . . . . . . . . . . . ... 74
Figure 5.16 Frequency response of the noise voltage density
of the T1(f)+T2(f) network . . . . . . . . . . . . . . . . . . . . . ... 75
Figure 5.17 a Amp4+Amp5 output noise voltage density with i/p
shorted. b = Fig. 4.2 . . . . . . . . . . . . . . . . . . . . . . . . . ... 76
Figure 5.18 Comparison of the various output noise voltage
densities vs. frequency (trace 3 is nearly hidden
by trace 2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 79
Figure 5.19 Deviation from the exact RIAA transfer
of Amp1+Amp4+Amp5 . . . . . . . . . . . . . . . . . . . . . . . . ... 79
Figure 5.20 Noise voltage densities at the output of the 2nd gain stage ... 81
Figure 6.1 Circuit of the engine’s output gain stage Amp5,
also showing additional offset trim variants . . . . . . . ...... 84
Figure 6.2 Plug-in connections between PCBs of Amps1, 2, 5
and the mainboard of Fig. 1.17 . . . . . . . . . . . . . . . ...... 85
Figure 6.3 Solid-state gain stage ±15 V PSU . . . . . . . . . . . . . ...... 85
Figure 6.4 Noise sources of Amp5’s input section (OPs 1 & 2) ...... 87
Figure 6.5 Noise sources of Amp5’s balanced output section
(OPs 3 & 4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 87
Figure 6.6 Noise sources of Amp5’s un-balanced output
stage (OP5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 88
Figure 7.1 Op-amp driven Amp5 incl. balanced & un-balanced
output. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 92
Figure 7.2 = Fig. 6.4 Noise voltage and current situation
of the 1st gain stage (OP1 & OP2) . . . . . . . . . . . . ...... 94
Figure 7.3 = Fig. 6.5 Noise voltage and current situation
of the 2nd gain stage (OP3 stage & the equivalent
OP4 stage) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 94
Figure 7.4 Bode plot of the balanced o/p noise voltage density
of the 1st gain stage. . . . . . . . . . . . . . . . . . . . . . . ...... 95
Figure 7.5 = Fig. 6.6 Noise voltage and current situation
of the un-balanced output gain stage (OP5) . . . . . . ...... 96
Figure 7.6 Amp5 output noise voltage density with i/p shorted . ...... 96
xxx List of Figures

Figure 8.1 Input gain stage alternative Amp1 . . . . . . . . . . . . . . ..... 102

Figure 8.2 Noise sources of Amp1 . . . . . . . . . . . . . . . . . . . . . ..... 104
Figure 8.3 Components of the Fig. 8.2 input load impedance Zi . ..... 105
Figure 8.4 Frequency responses of Amp1, based on five different
Amp1 input resistances Ri(=S1–S4 settings) . . . . . . . ..... 106
Figure 8.5 Phase responses of Amp1, based on five Amp1
input resistances Ri(=S1–S4 settings) . . . . . . . . . . . . ..... 106
Figure 8.6 Output voltage frequency responses for five different
Amp1 input voltages in 10 dB steps from 50 µVrms
to 5.0 mVrms in a 10 Hz–20 kHz band . . . . . . . . . . . ..... 108
Figure 9.1 = Fig. 8.1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Figure 9.2 = Fig. 8.2 Noise sources of Amp1 . . . . . . . . . . . . . . . . . . . 113
Figure 9.3 = Fig. 8.3 Components of the Fig. 9.2 input load Zi . . . . . . 113
Figure 9.4 Amp1 input noise voltage density with R0 = 20 Ω . . . . . . . . 114
Figure 9.5 Amp1 input noise voltage density with R0 = 0 Ω . . . . . . . . . 115
Figure 9.6 Input referred and A-weighted SN vs. input load . . . . . . . . . 117
Figure 9.7 Zoomed Fig. 9.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Figure 9.8 NF of Amp1 vs. R0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Figure 9.10 = Fig. 8.1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Figure 9.11 = Fig. 8.2 Noise sources of Amp1 . . . . . . . . . . . . . . . . . . . 120
Figure 9.12 = Fig. 8.3 Components of the Fig. 9.11 input load Zi. . . . . . 120
Figure 9.13 Amp1 input noise voltage density with R0 = 20 Ω . . . . . . . . 121
Figure 9.14 Amp1 input noise voltage density with R0 = 0 Ω . . . . . . . . . 122
Figure 9.15 NF of Amp1 vs. R0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
Figure 10.1 Input gain stage alternative Amp2 . . . . . . . . . . . . . . . .... 126
Figure 10.2 Booster for insertion into Fig. 10.1’s points A + A’ and
B + B’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 127
Figure 10.3 SSM2210 noise voltage density versus frequency and
collector current. . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 129
Figure 10.4 SSM2210 noise current density versus frequency and
collector current. . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 129
Figure 10.5 General simplified BJT noise model for the audio band .... 129
Figure 10.6 Transfer of the Fig. 10.5 model into a strictly input
referred one . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 130
Figure 10.7 Transfer of the Fig.10.6 noise model into the final BJT
noise model with only two equivalent noise sources
in place . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Figure 10.8 NF of 2SC3329 at 10 Hz . . . . . . . . . . . . . . . . . . . . . . . . . 132
Figure 10.9 NF of 2SC3329 at 1 kHz . . . . . . . . . . . . . . . . . . . . . . . . . 133
Figure 10.10 NF of 2SA1316 at 10 Hz . . . . . . . . . . . . . . . . . . . . . . . . . 134
Figure 10.11 NF of 2SA1316 at 1 kHz . . . . . . . . . . . . . . . . . . . . . . . . . 135
List of Figures xxxi

Figure 10.12 Amp2 with all noise calculation relevant active

and passive components . . . . . . . . . . . . . . . . . . . . . ..... 136
Figure 10.13 Input load situation of a BJT. . . . . . . . . . . . . . . . . . ..... 136
Figure 10.14 Upper (N) half of Amp2 (excl. input network Z1(f,R0)),
showing all relevant noise sources
(frequency independent) . . . . . . . . . . . . . . . . . . . . . ..... 138
Figure 10.15 Input resistor alternatives (input Cs not shown) . . . . . ..... 141
Figure 11.1 2SA1316 current noise density curve and its tangents,
after succ-apps and the chosen input load . . . . . . . . . .... 145
Figure 11.2 2SA1316 voltage noise density curve and its tangents,
after succ-apps and the chosen input load . . . . . . . . . .... 145
Figure 11.3 2SC3329 current noise density curve and its tangents,
after succ-apps and the chosen input load . . . . . . . . . .... 149
Figure 11.4 2SC3329 voltage noise density curve and its tangents,
after succ-apps and the chosen input load . . . . . . . . . . . . . 149
Figure 11.5 = Fig. 10.12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
Figure 11.6 Impedance of the input network ZiL(f). . . . . . . . . . . . . . . . . 154
Figure 11.7 Input impedances Zi(f) & Zin(f) . . . . . . . . . . . . . . . . . . . . . 155
Figure 11.8 Noise situation of Amp2, transferred into the upper
half of the amp and into the lower half, and vice versa .... 156
Figure 11.9 = Fig. 10.14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 159
Figure 11.10 Input referred noise voltage densities vs. frequency
based on two different input loads . . . . . . . . . . . . . . . .... 160
Figure 11.11 Amp2 equivalent input noise voltage densities vs. R0
based on two different frequencies . . . . . . . . . . . . . . . .... 160
Figure 11.12 Amp2 equivalent input noise voltage densities vs.
frequency showing 1/f-noise based on two different
input loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 160
Figure 11.13 Noise Figure vs. R0 . . . . . . . . . . . . . . . . . . . . . . . . . .... 161
Figure 11.14 Amp2 output noise voltage densities vs. frequency
RIAA-equalized + A-weighted + un-weighted . . . . . . . .... 163
Figure 11.15 Input referred un-weighted + A-weighted &
RIAA-equalized SN vs. R0 . . . . . . . . . . . . . . . . . . . . . .... 163
Figure 12.1 a Frequency and phase response of the left channel’s
Amps 1 & 2 via Amp3 and Amp5 + Trafo, fed by a
generator output resistance of 10.8 Ω. b Frequency and
phase response of the left channel’s Amp1 via Amp3
and Amp4 and Amp5 + Trafo, fed by a generator
output resistance of appr. 0 Ω . . . . . . . . . . . . . . . . . . . . . . 166
Figure 12.2 Engine II output noise voltage density curve of the left
channel, input loaded with 20 Ω and Amp1 + Amp4 +
Amp5 + Trafo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
xxxii List of Figures

Figure 12.3 Engine II output noise voltage density curve of the left
channel, input loaded with 20 Ω and Amp2 + Amp3 +
Amp5 + Trafo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 171
Figure 12.4 Engine II output noise voltage density curve of the left
channel, external input shorted and Amp3 + Amp5 + Trafo. .. 172
Figure 12.5 Left channel’s Amp1 + Amp3 + Amp5 + Trafo
distortion measurement result of a 1 kHz signal
via the central triode path . . . . . . . . . . . . . . . . . . . . . . . .. 173
Figure 12.6 Left channel’s Amp1 + Amp4 + Amp5 + Trafo
distortion measurement result of a 1 kHz signal
via the central op-amp path . . . . . . . . . . . . . . . . . . . . . . .. 174
Figure 12.7 Same as Fig. 12.5, however, Amp1 is replaced by Amp2 . .. 174
Figure 12.8 Same as Fig. 12.6, however, Amp3 is replaced by Amp4 . .. 175
Figure 12.9 Frequency and phase responses of Amp1 via Amps 3 & 4,
fed by a generator with an output resistance of 0 Ω . . . . . . . .. 181
Figure 12.10 Frequency and phase responses of Amp1 via Amps 3 & 4,
fed by a generator with an output resistance of 10.8 Ω . . . . . .. 182
Figure 12.11 Frequency and phase responses of Amp2 via Amps 3 & 4,
fed by a generator with an output resistance of 10.8 Ω . . . . . .. 183
Figure 13.1 BJT driven input stage alternative for MC purposes . . ..... 188
Figure 13.2 Balanced transformer input and BJT/Op-Amp driven
input stage alternative for low-output MC cartridges. . ..... 190
Figure 13.3 Fully triode driven MC/MM input stage alternative
with transformer (Lundahl LL9226) MC-input and
balanced output . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 192
Figure 13.4 Noise model of the Fig. 13.3 MC input stage . . . . . . ..... 194
Figure 13.5 Principal Joachim Gerhard design with high-Z
(Principally, Mr. Gerhard’s low-Z input approach
looks the same; details see footnote 11) voltage
driven input, turned into a linear input amp for
Engine II purposes . . . . . . . . . . . . . . . . . . . . . . . . . ..... 196
Figure 13.6 Principal Ovidiu Popa design with additional and new
output stage, turned into a linear input amp for
Engine II purposes . . . . . . . . . . . . . . . . . . . . . . . . . ..... 198
Figure 13.7 Principal Bob Cordell design with an additional and
new output stage . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 199
Figure 14.1
= Fig. 13.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
Figure 14.2
Impedance of the input network Z1(f) . . . . . . . . . . . . . . . . . 207
Figure a Frequency dependent gain. b Phase response of the gain .
14.3 . . 208
Figure Input impedance Zin(f) . . . . . . . . . . . . . . . . . . . . . . . . .
14.4 . . 209
Figure 14.5
Frequency and R0 dependent equivalent input noise
voltage density for two different input loads . . . . . . . . . . .. 212
Figure 14.6 Frequency and R0 dependent equivalent output noise
voltage density for two different input loads . . . . . . . . . . .. 212
List of Figures xxxiii

Figure 14.7 R0 dependent average input noise voltage density . . . ..... 213
Figure 14.8 R0 dependent, A-weighted, and RIAA equalized output
referred SNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 214
Figure 14.9 = Fig. 13.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 215
Figure 14.10 Input referred noise voltage density based on two
different input loads . . . . . . . . . . . . . . . . . . . . . . . . ..... 219
Figure 14.11 R0 dependent average input noise voltage density . . . ..... 220
Figure 14.12 R0 dependent, A-weighted, and RIAA equalized output
referred SNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 221
Figure 14.17 = Fig. 13.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 222
Figure 14.18 = Fig. 13.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 226
Figure 14.19 Output referred A-weighted and RIAA equalized SN vs.
R0 of the MC input stage . . . . . . . . . . . . . . . . . . . . ..... 230
Figure 14.20 Output noise voltage densities of the two phono-amp
input stage versions, inputs loaded. . . . . . . . . . . . . . ..... 231
Figure 15.1 Input referred noise voltage density curve of the Clio
8.5 measurement system, balanced input shorted . . . . . . . .. 234
Figure 15.2 Same as Fig. 15.1, however, smoothed by ½ Octave . . . . .. 235
Figure 15.3 Worsening figure We(B) as function of the difference
B of two SNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 236
Figure 15.4 Zoomed Fig. 15.3 with B ≤ 5 dB . . . . . . . . . . . . . . . . . .. 236
Figure 15.5 Clio’s 1 kHz 0 dBV signal and its distortion artefacts in B20k . .. 237
Figure 15.6 The Clio IMD measurement result with 250 Hz/8 kHz
and 4:1 signal levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
Figure 15.7 The un-balanced to balanced converter UBC . . . . . . . . . . . . 239
Figure 15.8 Booster alternatives for Fig. 15.7 . . . . . . . . . . . . . . . . . . . . 240
Figure 15.9 THD at UBC’s PL01 output, fed by a 0 dBV/1 kHz signal . . . 242
Figure 15.10 THD at UBC’s J03 output, fed by a 0 dBV/1 kHz signal . . . 242
Figure 15.11 THD at UBC’s J03 output, fed by a −20 dBV/1 kHz signal . . . 243
Figure 15.12 Test-board case with Amp1. . . . . . . . . . . . . . . . . . . . . . . . 245
Figure 15.13 Circuit of the test-board . . . . . . . . . . . . . . . . . . . . . . . . . . 245
Figure 15.14 RIAA encoder for RIAA transfer function trimming
actions via external input of the Engine II. . . . . . . . . . . . .. 246
Figure 16.1 Output voltage noise density curve of the NAD M51 . ..... 248
Figure 16.2 a Worsening Figure We(B) as function of the
difference B of two SNs. b Zoomed version of
Fig. 16.2a for correction purposes of two SNs
with values that are close together (B ≤ 5 dB). . . . . . ..... 250
Figure 16.3 Noise voltage density situation with a corner frequency
of 1 kHz (incl. tangents) . . . . . . . . . . . . . . . . . . . . . ..... 250
Figure 16.4 General output situation of the DUT (left) and balanced
input of measurement instrument (right) . . . . . . . . . . ..... 252
Figure 16.5 Effective output noise voltage of the DUT . . . . . . . . ..... 253
xxxiv List of Figures

Figure 16.6 The creation of the total input noise voltage of the
measurement amp . . . . . . . . . . . . . . . . . . . . . . . . . ..... 253
Figure 16.7 a DUT output resistance versus various MA total input
noise voltage variants. b Zoomed version of Fig. 16.7a
(for Rin.tot = 18.182 kΩ only) . . . . . . . . . . . . . . . . . ..... 255
Figure 16.8 Principal INA circuit . . . . . . . . . . . . . . . . . . . . . . . ..... 256
Figure 16.9 Principal final MA. (+) and (−) at the input and outputs
indicate the phase relationship . . . . . . . . . . . . . . . . . ..... 257
Figure 16.10 Relevant circuits for the gain calculation process . . . . ..... 257
Figure 16.11 Circuit for rule-of-thumb SN calculation . . . . . . . . . . ..... 260
Figure 16.12 Noise model of one-half input stage (Amp A with OP1
and OP2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 262
Figure 16.13 Noise model of one of the two summing stages (OP5
and OP6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 262
Figure 16.14 Noise model of the balanced to un-balanced
converter (OP7) . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 263
Figure 16.15 Input load dependency of the output referred SN . . . . ..... 263
Figure 16.16 Complete MA circuit . . . . . . . . . . . . . . . . . . . . . . . ..... 264
Figure 16.17 RG alternatives for gains of a ×10 and b ×100 (for only
one input stage) . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 265
Figure 16.18 Output noise voltage density curves, including input Cs
lower trace input shorted, upper trace input loaded
with 1 kΩ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 267
Figure 16.19 Same as Fig. 16.18 with linear frequency range . . . . . ..... 268
Figure 16.20 Frequency (top) and phase response (bottom) (bal
in/bal out) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 269
Figure 16.21 The PMMA in its enclosure . . . . . . . . . . . . . . . . . . ..... 270
Figure 16.22 Plug-in input stages of the PMMA. . . . . . . . . . . . . . ..... 270
Figure 17.1 PFMA circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
Figure 17.2 Output noise voltage density of the PFMA, input shorted . . . 276
Figure 17.3 F & P of the PFMA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Figure 17.4 Noise sources of the PFMA . . . . . . . . . . . . . . . . . . . . . . . 279
Figure 17.5 Evaluation of the OP1 input load . . . . . . . . . . . . . . . . . . . . 280
Figure 17.6 R0 dependency of the PFMA’s input referred SN. . . . . . . . . 280
Figure 17.7 PFMA and its enclosure . . . . . . . . . . . . . . . . . . . . . . . . . . 281
Figure 18.1 = Fig 15.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
Figure 18.2 UBC input impedance . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
Figure 18.3 Output noise voltage density at PL01 . . . . . . . . . . . . . . . . . 286
Figure 18.4 = Fig. 16.15 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
Figure 18.5 = Fig. 16.12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
Figure 18.6 Noise situation after paralleling of two i/p stages . . . . . . . . 293
Figure 18.7 Output noise voltage density of the whole 1st stage . . . . . . . 294
Figure 18.8 = Fig. 16.13 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294
List of Figures xxxv

Figure 18.9 = Fig. 16.14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 296

Figure 18.10 = Fig. 16.15 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 297
Figure 18.11 = Fig. 16.11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 298
Figure 18.12 Graph for interpolation purposes of the measurement
correction figure D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
Figure 18.13 = Fig. 16.10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
Figure 18.14 = Fig. 17.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
Figure 18.15 = Fig. 17.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
Figure 18.16 = Fig. 17.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
Figure 18.17 PFMA input noise voltage density vs. frequency . . . . . . . . . 305
Figure 18.18 Input referred R0 dependent SN. . . . . . . . . . . . . . . . . . . . . 305
Figure 19.1 Principal situation of an MM cartridge attached to an
un-balanced phono-amp input . . . . . . . . . . . . . . . . . ..... 308
Figure 19.2 Noise model of Fig. 19.1 . . . . . . . . . . . . . . . . . . . . ..... 308
Figure 19.3 Principal situation of an MM cartridge attached to a
balanced phono-amp input . . . . . . . . . . . . . . . . . . . ..... 308
Figure 19.4 Noise model of Fig. 19.3 . . . . . . . . . . . . . . . . . . . . ..... 309
Figure 19.5 Test MM phono-amp à la Fig. 19.1 . . . . . . . . . . . . . ..... 310
Figure 19.6 Standard Cartridge noise voltages at the input of the
test phono-amp à la Fig. 19.5 . . . . . . . . . . . . . . . . . ..... 310
Figure 19.7 R1 replaced by an M1 ELS. . . . . . . . . . . . . . . . . . . ..... 311
Figure 19.8 R10 split into two sections to produce a virtual ground
in-between them . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
Figure 19.9 R1 replaced by an M2 ELS. . . . . . . . . . . . . . . . . . . . . . . . 313
Figure 19.10 Input situation of a balanced phono-amp. . . . . . . . . . . . . . . 314
Figure 19.11 ELS of R1a + R1b. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
Figure 20.1 = Fig. 19.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320
Figure 20.2 = Fig. 19.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320
Figure 20.3 = Fig. 19.9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320
Figure 20.4 Input impedance of the Fig. 20.1 phono-amp input load . . . . 322
Figure 20.5 Phase of the Fig. 20.1 phono-amp input load . . . . . . . . . . . 322
Figure 20.6 Noise voltage density of the three different input
voltage dividers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 324
Figure 20.7 = Fig. 19.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 324
Figure 20.8 Total input noise voltage densities at the input of the
phono-amp (A) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326
Figure 20.9 Derivation sequence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
Figure 20.10 = Fig. 19.10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
Figure 20.11 = Fig. 19.11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
Figure 20.12 Input impedance of the Fig. 20.1 phono-amp input load . . . . 333
Figure 20.13 Phase of the Fig. 20.12 phono-amp input load . . . . . . . . . . 333
Figure 20.14 Noise voltage density of the three different input
voltage dividers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 335
xxxvi List of Figures

Figure 20.15 = Fig. 19.12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335

Figure 20.16 Total input noise voltage densities at the input of the
phono-amp (points A-B) . . . . . . . . . . . . . . . . . . . . . . . . . . 337
Figure 21.1 BJT model for low-frequency small signal calculation
purposes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342
Figure 21.2 BJT noise model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343
Figure 21.3 Basic common emitter circuit CE. . . . . . . . . . . . . . . . . . . . 346
Figure 21.4 CEb’s operational model . . . . . . . . . . . . . . . . . . . . . . . . . . 348
Figure 21.5 Fig. 21.4 with all relevant noise sources . . . . . . . . . . . . . . . 350
Figure 21.6 Common emitter circuit CEcf with cf via RE . . . . . . . . . . . . 351
Figure 21.7 CEcf ’s operational model . . . . . . . . . . . . . . . . . . . . . . . . . 354
Figure 21.8 Noise model of Fig. 21.7 with all relevant noise sources. . . . 355
Figure 21.9 Common emitter circuit CEvcf2 with vf via R2 and cf
via RE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 356
Figure 21.10 Derivation of the gain equation for Fig. 21.9. . . . . . . . . ... 357
Figure 21.11 Noise model of the Fig. 21.9 CEvcf2 gain stage . . . . . . . ... 358
Figure 21.12 Principal circuit of a closed loop arrangement with a
BJT followed by an op-amp . . . . . . . . . . . . . . . . . . . . ... 360
Figure 21.13 Frequency independent noise model of the modified
Fig. 21.12 amp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 362
Figure 22.1 Situation of the signal voltages in a DIFA . . . . . . . . . . . . . 366
Figure 22.2 General DIF-OPA circuit with input load . . . . . . . . . . . . . . 366
Figure 22.3 Relevant noise sources of Fig. 22.2 . . . . . . . . . . . . . . . . . . 368
Figure 22.4 Adapted Fig. 22.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
Figure 22.5 Typical amplifier circuit with a DIFA-IC . . . . . . . . . . . . . . 370
Figure 22.6 DIFA-IC with input load . . . . . . . . . . . . . . . . . . . . . . . . . 371
Figure 22.7 Noise sources of a DIFA-IC . . . . . . . . . . . . . . . . . . . . . . . 372
Figure 23.1
= Fig. 22.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 376
Figure 23.2
= 22.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 377
Figure 23.3
= Fig. 22.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 377
Figure 23.4
Input referred and input loaded noise voltage densities
of the fully differential input and the grounded input . ..... 378
Figure 23.5 = 22.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 380
Figure 23.6 = Fig. 22.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 380
Figure 23.7 Input referred differential noise voltage density,
input shorted or output resistance of preceding
gain stage *0R . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 381
Figure 24.1 BRAUN tracking force measurement instrument . . . . . . . . . 386
Figure 24.2 The BTFMI in action . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
List of Tables

Table 8.1 Detailed frequency and phase measurement results of Amp1,

based on five Amp1 input resistances Ri(= S1–S4 settings) . . . 107
Table 8.2 Figure 8.6 delta data at 10 Hz . . . . . . . . . . . . . . . . . . . . . . 107
Table 10.1 Amp2 non-weighted SN results for two different
input loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Table 10.2 Same as Table 10.1 after RIAA equalization
and A-weighting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Table 12.1 Relevant Engine II SN results . . . . . . . . . . . . . . . . . . . . . . 177
Table 12.2 Relevant Engine II THD figures . . . . . . . . . . . . . . . . . . . . . 178
Table 12.3 Relevant Engine II IMD figures . . . . . . . . . . . . . . . . . . . . . 179
Table 12.4 Frequency and phase response measurement results
of Amp1 and Amp2, fed by different generator
output resistances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
Table 13.1 Draft design SNs compared with Engine II results . . . . . . . . 202
Table 16.1 PMMA performance results . . . . . . . . . . . . . . . . . . . . . . . . 267
Table 16.2 Increasing input loads yield increasing differences
between measured and calculated output referred SNs . . . . . . 271
Table 24.1 TCRs of the Deutsche Grammophon Gesellschaft . . . . . . . . . 388
Table 24.2 DIN TCRs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
Table 24.3 TCRs of the VEB-Deutsche Schallplatten . . . . . . . . . . . . . . 389

xxxvii
 Part I
The RIAA Phono-Amp Engine II
The Complete Engine II—Overview
1

Sometimes, the development of a very flexible fully balanced hybrid MC phono-

amp solution may lead to rather unexpected and surprising results: Vive la
différence?

1.1 Intro and Goals

My first RIAA Phono-Amp Engine (I) became an essential part of my books “The
Sound of Silence” (TSOS-1/-2).1,2 It became essential because with it, I could
transfer all presented theory about electronic noise from math to a real thing. I could
calculate it and I could measure it, with results that only show rather small devi-
ations from the calculated values. However, I based the main amplifier chain design
on an un-balanced RIAA transfer creation that additionally includes balanced/un-
balanced inputs and outputs. The point I did not solve very elegantly was switching
between the various input-output possibilities. Thus, the handling of quick com-
parisons between different cartridges and turntables became a hard job. In addition,
I had to house the triode amplifier solution in a separate external case.
In contrast to my yesterday approach, the new Engine II design will deal with a
much more elegant approach, expressed by several standards and goals as follows:
A. Basically, the new engine should allow the user
1. to compare two different MC cartridges mounted on two turntables of the
same type or on one turntable with two equal tonearms
2. to compare two different turntables/tonearms (or one turntable with two
different tonearms) equipped with the same type of MC cartridge
3. to compare the sound of triode driven versus solid-state driven RIAA
transfer creation stages

1
“The Sound of Silence”, 1st ed.
2
“The Sound of Silence”, 2nd ed.

© Springer International Publishing Switzerland 2016 3

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_1
4 1 The Complete Engine II—Overview

4. to enhance the number of inputs by additional external amplifier stages for

broader comparison needs, including MM cartridges and gain stages, with
un-balanced inputs but always with balanced outputs
5. to test various technologies of balanced input and output gain stages, eg such
as transformer or op-amp or JFET or BJT or valve driven ones

Fig. 1.1 Block diagram of the Engine II

1.1 Intro and Goals 5

B. Additionally, the requirements concerning the chosen electronic solution should

hit or become even better than the following points/goals:
1. Fully balanced from input to output—plus one un-balanced output
2. A-weighted input referred SN: better than or equal to −79 dB(A) in
B20k (20 Hz–20 kHz), ref. 0.5 mVrms, input load = 20 Ω
Note: the resulting A-weighted and RIAA equalized SNariaa of an extremely
low-noise LP vinyl record—eg DMM cut LPs with −72.5 dB(A)—and the
phono-amp would then become a combined and weighted SN of −71.6 dB(A);
the resulting SNariaa of an average vinyl LP would become −70.5 dB(A)3
3. Nominal output level balanced and un-balanced: 0 dBV ≡ 1 Vrms, but also
changeable to 0 dBu to +6 dBu
4. Nominal gain Gnom = 2000 (+66 dB) with trimming possibility for each
channel: from min. 700 to max. 3000, hence, the input sensitivity reaches
from 0.33 to 1.45 mVrms referenced to DIN 0 dB (8 cm/s/1 kHz/flank mod.)
or from 0.2 to 0.9 mVrms referenced to a velocity of 5 cm/s/1 kHz/lateral
modulated
5. Phase response in B20k: ≤ ±30°
6. Frequency response in B20k: ≤ ±0.1 dB
7. RIAA transfer function creation: fully passive
8. With nominal gain overload margin ≤20 dB for all frequencies at all points
of the signal chain; any change of the nominal gain by ±x dB yields an
overload margin of 20 dB ∓ x dB
9. Slew rate >1.8 µV/s
10. Easy change of input and output gain stages
11. Power supply: ±15 V regulated for solid-state and +200 V/+6.3 V regulated
for triode stages
12. Cases: one per channel, one for each PSU, 19″ format
Fulfilment of the above given points has led to the block diagram given in
Fig. 1.1. The next chapters and sections will give the details.

1.2 General Concept

For MC purposes, the Fig. 1.1 input section (Amps 1 & 2) is composed of two low-
noise and high gain linear amps of different or equal configuration and very flat
frequency and phase response. In addition, their distortion level is rather low (more
on distortion etc. see further down in Chap. 12—Engine II Performance). An
external input allows the integration of additional linear amp stages. They may
serve for MC or MM purposes with balanced or un-balanced inputs, however,
always with balanced outputs.

3
TSOS-2, Chap. 11, Fig. 11.5/TSOS-1, Chap. 3, Fig. 3.95.
6 1 The Complete Engine II—Overview

The nominal signal level (100 mV) at the input section’s output and center
section’s (Amps 3 & 4) input comes from two sources. Firstly, it’s a level that
allows the design of internal output stages of Amps 1 & 2 which practically do not
hurt the noise level produced by the input stages of Amps 1 & 2, and secondly, the
overload requirement can also sufficiently be fulfilled in the center section of
Amps 3 (triodes) & 4 (op-amps).
Many comparisons of phono-amps suffer from the inequality of the components
used to create and electronically handle the RIAA transfer function. I think that,
besides a flat frequency & phase response and a certain distortion & intermodulation
level, most of the amplifier’s personal sound is created by these components and
their application in the amp chain. Therefore, to ensure absolute equal conditions
I have chosen only one 75 µs network and only one 318 µs & 3180 µs network.
These networks are switched by relays to the in- and outputs of Amps 3 & 4, the only
amps that cannot be changed for further experiments. They are rather fast broadband
linear valve or op-amp driven fully balanced amplifying devices, fixed on the main
board, producing only minimal additional noise and THD & IMD. Their common
mode rejection ratio is outstanding too.
I designed the output stage Amp5 as low-noise and as low-distortion as possible.
Therefore, it works with rather expensive OPA627 op-amps plus a switchable 1:1
high-quality output transformer at its output. Its balanced and un-balanced paths
have nearly no output resistance, very low THD & IMD, and very flat frequency
and phase response in B20k. Additionally, Amp5 does not add neither noise nor hum
to the noise level of the preceding gain stages Amps 1–4.

1.3 Basic Considerations

The way to produce the RIAA transfer function RIAA(f)4 form the main difference
between the two engine versions:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 þ ð2pf T3Þ2
RIAAðf Þ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 þ ð2pf T1Þ2 þ 1 þ ð2pf T2Þ2
ð1:1Þ
T1 ¼ 3180 ls
T2 ¼ 75 ls
T3 ¼ 318 ls

Engine I offers three solid-state solutions with creation of the transfer function
via overall feedback plus a valve solution with only one passive network
(2-pham5 concept = two gain stages and the whole RIAA network in the middle).

4
TSOS-2 Chap. 2, TSOS-1 Chap. 2.
5
TSOS-2 Sect. 17.7.
1.3 Basic Considerations 7

Fig. 1.2 RIAA transfer function creation and its insertion into an active and switchable amp chain

No matter which type of amplification we use (valve or solid-state) Engine II

always sets on a fully passive approach. We have the RIAA network split into two
sub-networks (3-pham6 concept = three gain stages with split RIAA networks
between them). Hence, we have an input amp followed by the 75 µs sub-net at the
central amp’s entrance, the 318 µs & 3180 µs sub-net at the central amp’s output,
loaded by an adequate output amp’s input resistance, here the one of Amp5.
The chosen sequence is highly favourable according to the overload question:
before the high frequencies (up to appr. +20 dB at 20 kHz) enter the central Amps 3
or 4 they got damped by the 75 µs lp to the max. 100 mVrms level. In addition, it
damps noise from the linear input Amps 1 & 2 too.
I have chosen the Fig. 1.2 design because it allows the RIAA transfer creation
with utmost precision and it fulfils the overload goal in both signal paths.
Figure 1.3 shows an input alternative with fully galvanically isolated input
connection via two 1:1/600 Ω:600 Ω transformers, however and after many checks,
without disturbance of the overall picture of Fig. 1.2. Concerning external i/p-loads,
it simply adds further flexibility.
Trimming pot P1, C5, and (R25 + R26) form the 75 µs network. In conjunction
with C12, R35, (R36 + R77) || Amp5’s (R1 + R2), and with trimming pots
P3 (+ Ro.cf3 + Ro.cf4 = cathode follower output resistances) for the triode driven
Amp3 and P4 + (R31 + R32) for the op-amp driven Amp4 we get the 318 µs &
3180 µs network. It only works perfect with the input resistance (R1 + R2) of the
following output stage Amp5 taken into account. At the output of Amp4, we find
R31 and R32. They stand for their output resistance counterparts Ro.cf3 and Ro.cf4 in
the triode path.
The detailed calculation of the time constants follows next; the detailed calcu-
lation of the triode output resistances follows in the Amp3 chapter.

6
TSOS-2 Sect. 17.7.
8 1 The Complete Engine II—Overview

Fig. 1.3 Input alternative of

Fig. 1.2

Because of the given input resistances of Amps 3 & 4 (R39a–d) and their given
equal balanced input capacitances Ci.dif, together with the nearly 0 Ω output
resistance of OPs 1 & 2 (see Ro.op1 & Ro.op2 in Fig. 1.4), we obtain T2 for Amp3
and Amp4:
Amp3 (Ro.op1 = Ro.op2 = 0 Ω):

 
ðP1 þ R25ÞR39a R26 R39b
T2 ¼ 75ls ¼ ðC9 þ Ci:dif Þ þ ð1:2Þ
P1 þ R25 þ R39a R26 þ R39b

Fig. 1.4 75 µs RIAA network at the input of Amps 3 & 4

1.3 Basic Considerations 9

Amp3 (Ro.op1 = Ro.op2 > 0 Ω):

2    3
P1 þ R25 þ R23jjRo:op1 R39a
6  
6 P1 þ R25 þ R23jjRo:op1 þ R39a 7
7
T2 ¼ 75 ls ¼ ðC9 þ Ci:dif Þ6   7 ð1:3Þ
4 R26 þ R24jjRo:op2 R39b 5
þ 
R26 þ R24jjRo:op2 þ R39b

Amp4 (Ro.op1 = Ro.op2 = 0 Ω):

 
ðP1 þ R25ÞR39c R26 R39d
T2 ¼ 75 ls ¼ ðC9 þ Ci:dif Þ þ ð1:4Þ
P1 þ R25 þ R39c R26 þ R39d

Amp4 (Ro.op1 = Ro.op2 > 0 Ω):

2    3
P1 þ R25 þ R23jjRo:op1 R39c
6  
6 P1 þ R25 þ R23jjRo:op1 + R39c 7
7
T2 ¼ 75 ls ¼ ðC9 þ Ci:dif Þ6   7 ð1:5Þ
4 R26 þ R24jjRo:op2 R39d 5
þ 
R26 + R24jjRo:op2 þ R39d

Note: The Amp4 input capacitance Ci.dif (*17.3 pF) must physically be added
to the Amp4 input. Here I propose the value of 15 pF ceramic, because the OPAs
add their sequence connected input capacitances too.
With the given input resistance of Amp5 (=Amp5’s R1 + R2 in Figs. 1.2 and
1.5) we obtain the following equations for the other two time constants T1 & T3:
Equal for both, Amps 3 & 4:

T3 ¼ 318 ls ¼ C12 R35 ð1:6Þ

Amp3 only (Fig. 1.5):

8 2 1 31 9
> >
>
< R ðf Þ =
>
6
T1:amp3
  7

T1 ¼ 3180 ls ¼ C12 R35 þ 4 1 5
> R1amp5 R36 R2amp5 R37 >
>
: þ þ >
;
R1 þ R36 R2
amp5 þ R37
amp5
ð1:7Þ
10 1 The Complete Engine II—Overview

Fig. 1.5 Situation of the 318 µs/3180 µs RIAA network at the output of Amp3

RT1:amp3 ðf Þ ¼ P3 þ Ro:cf:dif ðf Þ þ R33 þ R34 ð1:8Þ

The equation to calculate the frequency dependent output resistance Ro.cf.dif(f) of

the triode driven Amp3 will be given further down in the Amp3 Chap. 2. If the
capacitances in the signal path at the output of the triode cathode followers do not
hurt a flat frequency and phase response, we can take the frequency independent
value of the cathode follower’s output resistance Ro.cf.dif. Hence, in (1.7) we could
work without magnitude and RT1.Amp3.
Amp4 only (Fig. 1.6):

8 2 31 9
>
> R1 >
>
< T1:amp4 =
6   7
T1 ¼ 3180 ls ¼ C12 R35 þ 4 R1 R36 R2 R37 1 5
>
> þ
amp5
þ
amp5 >
>
: R1amp5 þ R36 R2amp5 þ R37 ;

ð1:9Þ

RT1:amp4 ¼ P4 þ R31 þ R32 þ R33 þ R34 ð1:10Þ

In contrast to the Amp3 output, we have no capacitances in the signal path.

Therefore, the auxiliary resistance RT1.Amp4 is fully frequency independent. In cases
of OP 5 & 6 output resistances >0 Ω we must add these values to R31 and R32 in
the above given equations.
1.4 Pictures of Cases and PCBs 11

Fig. 1.6 Situation of the 318 µs/3180 µs RIAA network at the output of Amp4

1.4 Pictures of Cases and PCBs

1.4.1 Cases

The Engine II is a rather complex and expensive approach: to get it done in 2013 I
had to spend roughly 2500 EUR net. Therefore, some pics may give an idea about
the many things that must work together and produce the shown results.
In the top 19″ 3 UH-84 HP case of Fig. 1.7 we find the two engine amplifier
channels for stereo use: green LEDs for the left channel, red LEDs for the right
channel. Both channels are totally equal. They are housed in two 3 UH-42 HP
fischer insertion cases (Fig. 1.8 shows the rear of Fig. 1.7).

Fig. 1.7 Front of Engine II, top two fully equal engine channels L & R; bottom left and right two
±21 V/0.5 A PSU insertion units, middle one triode PSU with +200 V/75 mA, 2 × 6.3 V/1.5 A
insertion unit
12 1 The Complete Engine II—Overview

Fig. 1.8 Rear of Engine II

In the bottom 19″ 3 UH-84 HP case, together with the two ±21 V PSU devices
(cases: 3 UH-21 HP at the fully left and right sides) an Amp3 200 V plus 2 × 6.3 V
PSU (3 UH-42 HP case in the middle) is housed in a separate 19″ 3 UH-84 HP case.
The whole PSU case is located roughly 1 m away from the upper 19″ 3 UH-84 HP
case.

1.4.2 Printed Circuit Boards

Figure 1.9 gives a look on the completely equipped right channel main board before
insertion into the case. From right to left side we find the plugged-in PCBs of

Fig. 1.9 Look on the fully equipped Mainboard of the right channel
1.4 Pictures of Cases and PCBs 13

Fig. 1.10 Rear and look into

the case of one channel

Amp1, Amp2, and Amp5. In the top middle, there is Amp4 and the two double-
triodes of the triode path Amp3 fill the lower part of the middle region.
Figure 1.10 shows the rear of one channel. The observer also gets an impression
about the crush inside the case. Nevertheless, many holes in the top and bottom
metal plates ensure enough cooling.
Figure 1.11 gives an impression of the PCBs of the plug-in amps, the ones I used
to create the measurement and listening results of this book. However, the shown
Amps 1 & 2 versions need some improvements to overcome disadvantages I will
describe in the following chapters.

1.4.3 Front and Rear of the Engine

The many switching possibilities of one chain need careful arrangement of the
switches on the amplifier’s front. I’ve chosen a set-up that follows—from left to
right—the logic given in Fig. 1.1. Figure 1.12 presents the solution for the left
channel. The right one looks the same with the exception of red LEDs.
The decision to design and further on to work with two fully separate channels
comes from the fact that mono signals can easily be compared by application of
14 1 The Complete Engine II—Overview

Fig. 1.11 The first set of plug-in PCBs for Amp5 (fully left), Amp 2 (middle), and Amp1 (fully
right)

Fig. 1.12 Front of one channel

(a) two different triode operating point settings, or

(b) different double-triodes, and/or
(c) different op-amp types.
J04 at the rear offers ±21 V for external solid-state input amp purposes. There is
enough space to add female sockets for external triode anode and heater supply, if
1.4 Pictures of Cases and PCBs 15

Fig. 1.13 Rear of one channel

the triode PSU offers enough energy. Another solution for that could be extra PSUs
for the external amps.
The front and rear of the PSU units are not shown in detail here. The placements
of the sockets on each rear follow the logic of the needs of the two channels, as of
Fig. 1.8. However, there is only one central mains socket for the triode PSU in the
middle of Fig. 1.8. The mains connection for the two ±21 V units comes from two
additional and un-switched output sockets on the rear’s right side of the triode PSU.
In such a complex amp environment the right grounding is not easy to find.
Therefore, at the rear of each channel and in the middle of the triode PSU rear we
find ground lifts (see Figs. 1.8 and 1.13). The same applies to the motherboard with
its numerous ground lift jumpers (see Fig. 1.17). Additional cable sockets on the
rear of the channels allow further grounding actions between turntable(s), housing
shields, and amps.

1.5 Power Supplies

I do not describe the different PSU units in detail here because the main emphasis
lies on the amplifier action. However, here come the facts & figures on how the
whole engine gets powered. In contrast to the findings of John Walton in L|A
Vol. 4,7 I always prefer 317 / 337 types. Each of the many solid-state amps, plug-in
ones à la Fig. 1.11 as well as the ones on the main PCB, has its own stabilized

7
“A comparative overview of power supply regulator designs with listening tests”, John Walton,
Linear Audio Vol. 4, ISBN 978-949092905-3.
 16

Fig. 1.14 Main ±21 V power supply for all solid-state driven amps
1 The Complete Engine II—Overview
 1.5 Power Supplies
17

Fig. 1.15 +200 V power supply for the triode driven section
 18

Fig. 1.16 Two +6.3 V regulated triode heater power supplies

1 The Complete Engine II—Overview
 1.5 Power Supplies
19

Fig. 1.17 Wiring on the Mainboard and to the outside world

20 1 The Complete Engine II—Overview

±15 V sub-PSU, each channel is fed by a ±21 V main PSU, shown in Figs. 1.7 and
1.8 on the fully left and fully right side of the lower half. All op-amps have their
blocking Cs located rather close to the IC. The ±21 V circuit is given in Fig. 1.14.
A view words about the +200 V PSU. There is only one of them. I took a
transformer with two 80 V/75 mA output windings. Each of them feeds a rectifier
bridge followed by an integrated stabilizing circuit LT783KC with a trimmed
output voltage of +100 V. Both output voltages stacked lead to a very stable and
rather low-noise and low-hum output voltage of +200 V for both channels. There is
no need of the same configuration for the other channel. One unit is enough here
because there is a rather high filtering effect of the Amp3 anode voltage C-R-C
chain. Between the two channels any interaction is cut down to an immeasurable
level. Figure 1.15 shows the circuit.
In that unit I’ve also installed two 6.3 V heater supplies, one for the first double-
triode’s low heater potential around 0 V DC, the other one for the output triode with
its heater potential around +90 V DC. Both heater voltages are regulated too.
Figure 1.16 gives the details.

1.6 Mainboard

Figure 1.17 shows the circuit of the Mainboard. Its fully equipped version is shown
in Fig. 1.9. Switches S1, S2, S7, S8 (and S2 on Figs. 1.15 and 1.16) are the before
mentioned ground lifts that can be switched at the outside of the case rears.
Depending on the hum amount represented in an output FFT diagram (with input
shorted) jumpers Jmp1, Jmp4, Jmp5, and Jmp6 can be set to suppress most of the
mains interferences on the ground lines. However, because of the fully balanced
layout there should not be a need for that. I guess the rest of the circuit is self-
explanatory.
The Triode Driven Central Amplifier
Amp3 2

2.1 General Design of Amp3

The two central Amps 3 & 4 are the heart of the whole amplifier arrangement.
Therefore, I will start my explanations with these, followed by the output stage
Amp5 and the two input stages Amps 1 & 2.
Before I entered into the here presented design of Fig. 2.1, I had some tests on
the most useful balanced triode driven solution, however, always in conjunction
with the placement of the RIAA network. In addition, the solution should be as low-
noise as possible, thus, increasing the noise level of the preceding gain stage by not
more than point B.2. of Chap. 1 would allow. On the other hand, it should have
outstanding CMRR and easy balance trimming.
Based on the findings in the Differential Gain Stage (DIF) chapter of the 2nd
edition of my How to Gain Gain book (HTGG-2)1 I opted for a DIF input stage
followed by a CF (cathode follower) output stage. The easy handling of a CCSCF
gain stage (Common Cathode gain Stage CCS followed by a CF) led to the shown
Fig. 2.1 configuration without RIAA networks. I used such a CCSCF as output
stage in the triode driven Module 4 phono-amp of Engine I in TSOS-2.2 To get a
rather high CMRR the DIF stage’s DC current comes from a solid-state current
generator (a sink here), formed by two BJTs. It creates a very high dynamic
resistance between t1 & t2 cathode and ground.
The input section is the DIF formed by a gm-selected low-noise double-triode
E88CC/6922 (E188CC/7308 work well too). In each triode system the anode current
is equal and trimmed to 2 mA by P2 of the current sink T1 & T2 (480 mV between
test points TP3 & TP4). Trimming of P5 optimizes CMRR further (calculated appr.
100 dB). It ensures equal signal levels at the cathodes of t4 & t3. Each of the
following CF stages is powered by appr. 90 V/2 mA too. Here, instead of the shown

“How to Gain Gain”, 2nd ed., B. Vogel, (HTGG-2), Chap. 30.

1

2
TSOS-2, Chap. 17.

© Springer International Publishing Switzerland 2016 21

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_2
22 2 The Triode Driven Central Amplifier Amp3

Fig. 2.1 Amp3 without RIAA networks

2.1 General Design of Amp3 23

E88CC I also tried non-selected NOS 7308 s. Their noise level is rather low and their
triode systems do not differ very much. Finally, I took the 7308 s.3
The gain Gamp3 becomes measured appr. 16 and the whole design looks rather
simple. In addition, with a perfect trim we can get a 1 kHz THD ≤ 0.010 %
(stronger d2 than d3) and IMD always <0.010 % (I’ve measured 4 different
80 %/20 % frequency pairs). My Clio sinus generator offers a min. 1 kHz THD
level of 0.002 % rounded4 through my un-balanced to balanced converter (see
Footnote 4), strictly THD only and not THD + N! I could calculate the real 1 kHz
THD with distortion spike level figures taken from the FFT diagram: 0.00159 %
(more about distortions etc. see Chap. 12—Engine II Performance).
The gain stage fulfils the overload goal. I measured 46 Vpp before soft clipping.
With an input signal level of 100 mVrms + 20 dB overload margin = 1 Vrms and a
gain of 16 we need a max. voltage swing of 16 V * 2 * √2 = 45.255 Vpp.
Based on the following considerations we can roughly check the extra-generated
noise level of the sequence of Amp3 & Amp5: with input loaded by 20 Ω, Amp2 (its
SN looks worse than the one of Amp1) alone generates a measured (m) output
referred non-equalized (ne) SNne.o.m = −73.1 dB ref. 100 mV/B20k, almost white
noise. It includes a tiny amount of 1/f-noise.5 Multiplication by 10 (theoretically
through a no-noise amp-stage) leads to a total output referred SNne.o.tot = −73.1 dBV
at the engine’s output. Now, after application of the B20k RIAA function and
A-weighting SN-improvement figure SNar ≈ −8 dB6 for purely white noise gener-
ating devices we obtain the guessed output referred A-weighted and equalized (ariaa)
SNariaa.o = −81.1 dBV(A). With the sequence of 20 Ω + Amp2 + Amp3 +
Amp5 + Trafo at the output of the Engine I’ve measured SNariaa.o.m = −79.9 dBV(A)
for the left channel and −80.2 dBV(A) for the right one. With that, the goal of an input
referred SNariaa.i = −79.0 dBV(A) won’t get into trouble.
Nevertheless, via shorted external input, the output referred SN of the amp
sequence Amp3 + Amp5 + Trafo becomes measured (calculated) −99.0 dBV(A)
(−100.2 dBV(A)). Figure 2.27 shows the curve of the noise voltage density at the
output of the before given sequence, based on data-sheet data. It also shows a kind
of 1/f-noise characteristic. It is generated by two sources:
(a) by an assumed 1/f-noise corner frequency of fc.e1.2 = 1 kHz of the DIF input
triodes (high influence on the overall noise voltage) and an fc.e3.4 = 10 kHz of
the two output CFs (rather low influence on the overall noise voltage), and

3
I deeply have to thank my friend Klaus Burosch (www.burosch.de) for his courteous support
concerning his huge collection of NOS and brand new valves. All used (and many more) valves
had to pass the test arrangement I’ve presented in Jan Didden’s Linear Audio Vol. 4 “The Glowing
NoiseMaker—on the demystification of triode noise” or in HTGG-2, Sect. 2.3.
4
Details see Chap. 15.
5
Additionally see my remarks on Amp2’s SN in Chapter 10.
6
TSOS-2, Chapter 15, TSOS-1 Chapter 6.
7
Details see next Chapter and MCD-WS 3.1.
24 2 The Triode Driven Central Amplifier Amp3

0
20
40
60
⎛ en.o.amp3.5( f) ⎞
[dBV]

80
20⋅log ⎜ ⎟ 100
⎝ 1V ⎠ 120
140
160
180
200
1 .10 1 .10 1 .10
3 4 5
10 100
f
[Hz]

Fig. 2.2 Output noise voltage density of the amp sequence Amp3 + Amp5 + Trafo with input
shorted

(b) by the RIAA network effect of the 318 µs/3180 µs network at the output of
Amp3. The 75 µs input network has practically no effect on the Amp3 noise
generation. It only filters the incoming noise voltage from preceding gain
stages.
To calculate the component values for the T1/T3 RIAA network we need the
differential o/p resistance Ro.cf.dif(f) of t3 & t4 (see (1.6)–(1.8) in the previous
chapter). Because of C44a–C45b it is frequency dependent and in consideration of
R48 & R49 it is the sum of the equal o/p resistances Ro.cf3(f) + Ro.cf4(f). The
relevant equations look as follows:

Ro:cf3 ðf Þ ¼ Ro:cf4 ðf Þ
2 31
6 1 1 7 ð2:1Þ
¼4 þ
1 5
R48
Ro:cf3 þ 2jpf C44a
2

Ro:cf:dif ðf Þ ¼ Ro:cf3 ðf Þ þ Ro:cf4 ðf Þ ð2:2Þ

ra3 R47
Ro:cf3 ¼
ra3 þ ð1 þ l3 ÞR47 ð2:3Þ
¼ Ro:cf4

According to the goals C44a–C45b must be chosen of a size that should not hurt
a flat frequency and phase response in B20k. Then, with ra3 = ra4 = 8.836 kΩ,
gm3 = gm4 = 3.5 mS, and µ3 = µ4 = 29 we’ll get Ro.cf.dif = 549.2 Ω. I’ve chosen
Panasonic FC 63 V types. With them, the deviation from the flatness becomes a
calculated −0.025 dB/+0.2° at 20 Hz only. Figure 2.3 shows the calculated devi-
ation from the exact RIAA transfer if we would consider the RIAA networks. The
measured frequency and phase response will be given in Chap. 12.
2.1 General Design of Amp3 25

0.1

0.05

[dB]
Dev2 e (f) 0

0.05

0.1
1 .10 1 .10 1 .10
3 4 5
10 100
f
[Hz]

Fig. 2.3 Calculated deviation from the exact RIAA transfer

Figure 2.1 also shows the constant current sink around BJTs T1 & T2. The
actual noise voltage of this CCsi is of minor importance. What hits the DIF most
is the noise current mainly produced by T1’s collector current. Multiplied by the
cathode input resistance of the DIF we have an enormous noise voltage that is
amplified by the here effective grounded grid gain stages(CGS) formed by t1 & t2,
in this case leading to 100 % correlated noise voltages of equal amplitude at the
anodes of t1 & t2. Hence, at the differential output of Amp3 we find the CCsi
generated noise voltage with a doubled level! To suppress it we need a following
Amp5 with rather high CMRR. Chapter 6 gives the details of Amp5.

2.2 Gain and Noise Calculations

The Mathcad worksheet (MCD WS-3.1) of the next chapter gives all the details of a
rather extensive calculation course. All results are based on data-sheet data. I’ve
also gone through the calculation with actual data. Selected low-noise triodes
should have very low 1/f-noise corner frequencies and far better (higher) gm-values
than the ones of the data-sheets. Fortunately, they do not differ very much from the
ones gained by application of data-sheet data together with the assumed data for the
1/f-noise corner frequency. I guess it is clear that higher 1/f-noise corner fre-
quencies will automatically lead to worsened SNs.
The complete calculation of the gain and noise production of a DIF can easily be
studied in HTGG-2, Chap. 30. However, for a better understanding I will repeat the
equivalent circuit and the main equations here.

2.2.1 Gain of a DIF Followed by Two CFs

The DIF’s idle gain G0.dif:

R40 þ 0:5P7
G0:dif ¼ G0:t1:2 ¼ l1 ð2:4Þ
ra1 þ R40 þ 0:5P7
26 2 The Triode Driven Central Amplifier Amp3

The CF’s frequency dependent gain with output load RL(f) (because of its tiny
influence here the impedance of Fig. 1.2’s C12 is set to 0 Ω):

R47
Gcf4 ðf Þ ¼ l4 ð2:5Þ
ra4 R47
ra4 þ ð1 þ l4 ÞR47 þ R
L:t4 ðf Þ

2 31
1 6 1 1 7
RL:t4 ðf Þ ¼ þ4 þ
1 5
2jpf0:5C44a R48 1 1 ð2:6Þ
R34 þ ½0:5R35 þ R36
¼ RL:t3 ðf Þ

G0.dif is the DIF’s idle gain because its anode has an infinite load by the fol-
lowing t3/t4 grids. RL.t4(f) is the frequency dependent load at the cathode of t4. The
same applies to RL.t3(f).

Gcf3 ðf Þ ¼ Gcf4 ðf Þ ð2:7Þ

Gcf ðf Þ ¼ Gcf3 ðf ÞGcf4 ðf Þ ð2:8Þ

⇒ The frequency dependent DIFCF gain Gdifcf(f) thus becomes:

Gdifcf ðf Þ ¼ G0:dif Gcf ðf Þ ð2:9Þ

⇒ The balanced gain Gop1.2 of the two op-amps OPs 1 & 2 is 1. Hence, the
Amp3 gain Gamp3(f) without RIAA transfer becomes:

Gamp3 ðf Þ ¼ Gdifcf ðf ÞGop1:2 ð2:10Þ

2.2.2 RIAA Transfer Function

From Figs. 1.4 and 1.5 and (2.1)–(2.3) we can derive the frequency dependent and
RIAA transfer loaded gains GT2(f) and GT1.3(f) of the Amp3 input and output
networks as follows:

M
GT2 ðf Þ ¼
1
M þ R25 þ R26 þ P1 þ 2 R1 o:op1 þ R23 1
ð2:11Þ
h i1
M ¼ 2jpf ðC9 þ Ci:dif Þ þ ðR39a þ RR39bÞ1
2.2 Gain and Noise Calculations 27


1 1
1 1
½2jpfC12 þ R35 þ RL:dif
GT1:3 ðf Þ ¼ 
1 1 ð2:12Þ
1 1
½2jpfC12 þ R35 þ RL:dif þ RT1:eff ðf Þ

 1 1
RL:dif ¼ R1amp5 þ R2amp5 þ ½R36 + R371 ð2:13Þ

RT1:eff ðf Þ ¼ Ro:cf:dif ðf Þ þ P3 þ R33 þ R34 ð2:14Þ

⇒ With C9 and C12 carefully selected according to Figs. 1.4 and 1.5 the transfer
function Tamp3(f) of the whole Amp3, including RIAA transfer function, thus
becomes:

Tamp3 ðf Þ ¼ Gamp3 ðf ÞGT2 ðf ÞGT1:3 ðf Þ ð2:15Þ

2.2.3 Noise and SN Calculations According to Fig. 1.2

The calculation of the noise voltage of the DIFCF alone makes no sense, as long as
there are influential factors at its input (OPs 1 & 2 + T2(f)) and at its output
(T1(f) + T3(f) + Amp5). All together, they generate a noise voltage that can be
measured at the output of Amp5, and thus be compared with the calculated results.
The calculation course follows the mathematical course given in MCD-WS 3.1,
“6. Noise and SN calculations”. However, here comes the short version. It tackles
the major factors.
To calculate the output noise voltage density en.o.amp3.5(f) at the o/p of Amp5 and
with Amp3 input shorted the rather complex looking equation looks as follows:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u220 1 3 3
u e ðf Þ2 G ðf Þ2
u 6 6 @ n:o:op1:2 2 T2 AjG0:dif j2 7 7
u 6 6 þ en:Z:T2 ðf Þ 7 7
u66 7 7
u 6 6 þ 2en:rN1 ðf Þ 2
7Gcf ðf Þ 7
2
u66 7 7GT1:3 ðf Þ
2
u 6 4 þ 2e2 2
5
u6 n:Rgg3 þ 2en:Ra:eff ðf Þ 7
7
u
en:o:amp3:5 ðf Þ ¼ Gamp5 u 4 þ2en:rN3 ðf Þ2 5
u
u 2en:Rc:eff ðf Þ 2
u
2
u
u þ 2en:ccsi Gcgs:1 Gcf ðf ÞGT1:3 ðf Þ
u
t CMRRamp5
þ en:Z:T1:3 ðf Þ þ i2n:i:amp5 ZT1:3 ðf Þ2 þ e2n:i:amp5
ð2:16Þ
28 2 The Triode Driven Central Amplifier Amp3

R40 þ 0:5P7
Gcgs:1 ¼ ð1 þ l1 Þ
ra1 þ R40 þ 0:5P7 ð2:17Þ
¼ Gcgs:2

According to (2.16), Figs. 1.2 and 2.1 it includes the following 100 % un-
correlated noise sources:

• OP1 & OP2 (frequency dependent = fd): en.o.op1.2(f)

• T2(f) network (fd): en.Z.T2(f)
• t1 & t2 noise (fd)a: en.rN1(f) = en.rN2(f)
• Rgg3 = R44 + R45: en.Rgg3
• Ra = R40 + R41 (incl. fd excess noise): en.Ra.eff(f)
• t3 & t4 noise (fd) :
b
en.rN3(f) = en.rN4(f)
• Rc = R46 + R47 (incl. fd excess noise): en.Rc.eff(f)
• T1(f) & T3(f) network (fd): en.Z.T1.3(f) & ZT1.3(f)
• Amp5 i/p noise current in.i.amp5 (no 1/f-noise!)
• Amp5 input referred noise voltage: en.i.amp5 (=average value in B20k)
a
I’ve chosen a low-noise double-triode here that has, in both systems, a 1/f-noise corner frequency
of 1 kHz; a change to 10 kHz would worsen the calculated output referred SN of −100.2 dBV(A)
by appr. 3.5 dB, a change to 100 Hz would improve SNs by appr. 0.6 dB
b
For calculation purposes, I’ve chosen a noisy double-triode here that has in both systems a
1/f-noise corner frequency of 10 kHz; a change to 1 kHz would improve the result of (2.16) by
0.014 dB only
pffiffiffiffiffiffiffiffiffi
Note: By integration over B20k and division by B20k the frequency dependency
of the input referred noise voltage density of Amp5 (see respective MCD-WS in
Chap. 7) can be turned into one single average density value. Hence, and in other
words, we gain the rms value of the noise voltage in B20k and after division by
pffiffiffiffiffiffiffiffiffi
B20k we’ll get the average density value, however, guilty in B20k only!
• Generated by the noise current of the constant current sink CCsi and its BJTs T1
& T2 and multiplied by the equal gains Gcgs.1 of the Common Grid Stages
(CGS) t1 & t2 the 100 % correlated noise voltage en.ccsi is damped by the
CMRR of Amp5 (see Chaps. 6 and 7). Because there is 100 % un-correlation
between this term and all the other ones, it is integrated into (2.16) too.
I must point out that some terms in (2.16) do not add significant values to the
total sum underneath the root. Nevertheless, I keep them for universal usage with
other than the chosen components.
The resulting noise voltage density multiplied by the A-weighting function,
referenced to 1 Vrms nominal signal output voltage, and integrated over the band-
width of B20k, will lead to the A-weighted output referred SNa.o.amp3.5 in B20k,
expressed in dBV(A).
2.2 Gain and Noise Calculations 29

With the exception of the DIF, the detailed calculation approaches of the dif-
ferent terms in (2.16) can completely be studied in TSOS-2. TSOS-1 is not a help at
all because it doesn’t cover the triode math approaches.

2.2.4 A Look into the Content of MCD-WS 3.1

MCD-WS 3.1 shows some additional interesting results:

• Very important for external amplifiers Point 6.5.2 shows the calculation of the
Amp3 CMRR.
• Point 7. covers the math of an extremely low-noise input load of the Fig. 2.1
arrangement with Amp3. Here, I’ve chosen a pre-amp with a gain of 200, an i/p
referred noise voltage density of 0.2 nV/rtHz and noise current density of only
2.4 pA/rtHz. The i/p load is 20 Ω. Now we can compare the A-weighted and
RIAA equalized SN result (=−82.523 dBV(A)) with the one of Point 8.
• Point 8. covers the math of the Point 7. low-noise Amp1, followed by a no-noise
arrangement à la Fig. 2.1. The SN result becomes −82.582 dBV(A).
• Hence, the difference is appr. 0.06 dB only. It is nothing else but the Noise
Figure. In other words: a further chase for extremely low-noise solutions makes
no sense for input loads ≥20 Ω. We will see later on in Chap. 10 what it will
mean for input loads <20 Ω.
• Point 9. and 10. show calculations of the Noise Figure NF of the amp chain
Amp3 + Amp5, fed by a lowest-noise input amp: 9. for MC and 10. for MM
cartridge purposes. These NFs are all <0.1 dB. Hence, together with its input
and output loads the noise impact of the here presented Amp3 is completely
ignorable.
• Point 11. gives up the shorted input and replaces the shortage by an output
resistance of a preceding gain stage, here 1 kΩ. Because of the 75 µs lp at the
input, the noise impact becomes marginal too.
Mathcad Worksheets Amp3
3

Contents

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks

pffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffi
Note 1: MCD 11 has no built-in unit “rtHz” or “ Hz”. To get 1 Hz based
voltage noise and current noise densities the rms noise voltage and
current in a specific frequency range B > 1 Hz must be multiplied by
pffiffiffiffiffiffiffiffiffi
1 Hz and divided by the root of that specific frequency range √B!
Note 2: MCD 11 offers no “dB” unit. This is available from MCD 13 on!

© Springer International Publishing Switzerland 2016 31

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_3
32 3 Mathcad Worksheets Amp3

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks Page 1

Calculations of the trode driven Amp3 with T1/T3 RIAA network at the output, T2 RIAA
network at the input, and based on triode data-sheet figures :

Fig. 3.1 Triode driven Amp3 incl. RIAA networks

1. Genaral data :

D IF (t 1+t2) with E88CC - 2x2mA: 2xCF (t 3+t4) wit h E88CC - 2x2mA

At the triode stage input: OP1 OPA627 OP2 OP1

23 1
k 1.38065 10 VAsK T 315K B20k 19980 Hz B1 1Hz

f 20Hz 25Hz 20000 Hz h 1000Hz

Assumed: fc1 1000Hz fc2 fc1 fc3 10000 Hz fc4 fc3

fc1 ln( 1000 ) B20k

Fc1 Fc1 1.346 Fc2 Fc1
B20k

fc3 ln( 1000 ) B20k

Fc3 Fc3 4.457 Fc4 Fc3
B20k

µ1 29 µ2 µ1 µ3 29 µ4 µ3

3 3
gm1 3.5 10 S gm2 gm1 gm3 3.5 10 S gm4 gm3

µ1 3 µ3
ra1 ra1 8.286 10 ra2 ra1 ra3 ra4 ra3
gm1 gm3

Vg1 2.8V Vg2 Vg1 Vg3 2.8V Vg4 Vg3

3 Mathcad Worksheets Amp3 33

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks Page 2

Va1 90V Va2 Va1 Va3 90V Va4 Va3

3 3
Ia1 2 10 A Ia2 Ia1 Ia3 2 10 A Ia4 Ia3

2. Gain stage (GS) component data :

3 3 3
R20 20 10 R21 10 10 R22 R21 R23 4.99 10 R24 R23
3 3 3
R25 750 R26 787 R33 5.62 10 R34 6.81 10 R35 1.4 10
3 3 3
R36 100 10 R37 R36 R39a 100 10 R39b R39a R40 18.4 10
3
R41 R40 R42 145 R43 130 10 R44 301 R45 R44
3
143 10 3 3
R46 R46 47.667 10 R47 R46 R48 11 10 R49 R48
3
3
3 6.65 10 3
R50 22.1 10 R51 R50 R52 R52 1.663 10
4

3
R1amp5 100 10 R2amp5 R1amp5
6
C5 100 10 F C6 C5 C7 C5 C8 C5
9 9
C9 47 10 F C12 227.143 10 F
6
C44a 220 10 F C44b C44a C45a C44a C45b C44a
3 3
P5 20 10 P6 P5 P7 2 10 P8 50

Note: For further calculation purposes only: R25 = R25+P1, R33 = R33+P3, R42 = R42+P8, R40&41 = R40&41+0.5*P7
Exact values for P1, P3, C9, C12 see further down

3. DIFCF = CCSCF in balanced mode :

3.1 DIF :
R40
G0.dif µ1 G0.dif 19.996
ra1 R40

ra1 R40 3
Ro.dif 2 Ro.dif 11.426 10
ra1 R40

ra1 R40 3
Ro.a1 Ro.a1 5.713 10
ra1 R40

12 12 12
Cga1 1.4 10 F Cgc1 3.1 10 F Cstray1 2 10 F

12
Ci1 G0.dif 1 Cga1 Cgc1 Cstray1 Ci1 34.494 10 F

12
Ci.dif 0.5 Ci1 Ci.dif 17.247 10 F
34 3 Mathcad Worksheets Amp3

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks Page 3

3.2 CF :
1
1 1 1 3
RL.t4( f) RL.t4( h) 4.461 10
2j f 0.5 C44a R48 1
1 1
R34
0.5 R35 R36

RL.t3( f) RL.t4( f)

R47
Gcf.4( f) µ4
ra4 R47 Gcf.4( h) 0.905
ra4 1 µ4 R47
RL.t4( f)
Gcf.4( 20Hz) 0.905
Gcf.3( f) Gcf.4( f)

Gcf ( f) Gcf.3( f) Gcf.4( f) Gcf ( h) 0.82

ra4 R47
Ro.cf.t4 Ro.cf.t4 274.599
ra4 1 µ4 R47

1
1 1
Ro.cf.t4.eff( f) Ro.cf.t4.eff( h) 267.915
1 R48
Ro.cf.t4 2j f 0.5 C44a

Ro.cf.t3.eff( f) Ro.cf.t4.eff( f)

Ro.cf.dif( f) Ro.cf.t4.eff( f) Ro.cf.t3.eff( f) Ro.cf.dif( h) 535.83

3.3 Gain of DIFCF :

Gdifcf( f) G0.dif Gcf ( f) Gdifcf( h) 16.390411

3.4 Gain and Ro of i/p buffers OP1and OP2 : Gop1.2 1 Ro.op1 0.1 Ro.op2 Ro.op1

3.5 Overall gain of Amp3 without RIAA network :

Gamp3( f) Gdifcf( f) Gop1.2 Gamp3( h) 16.390411

25

20
[times]

Gamp3( f) 15

10
5 Fig. 3.2
3 4 5 Gain of Amp3 vs frequency
10 100 1 10 1 10 1 10
f
[Hz]
3 Mathcad Worksheets Amp3 35

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks Page 4

4. RIAA time constants T and transfer gains G T:

4.1 T2 = 75µs : P1 70.8

Fig. 3.3
T2 defining network

succ-apps of P1 should bring in line the value of T2 with the RIAA network component values (because of the
marginal influence the voltage divider respective effects of R o.op 1 /Ro.op 2 and R23/R26 are ignored here!):

1
1
1
1 1 1
T2 C9 Ci.dif Ro.op1 R23 R25 P1 R39a
1
1
1
1 1 1
Ro.op2 R24 R26 R39b

6
T2 75 10 s

1
1
2j f C9 Ci.dif ( R39a R39b)
GT2( f)
1 1
1 1 1
2j f C9 Ci.dif ( R39a R39b) R25 R26 P1 2 Ro.op1 R23

1
GT2 0.701
2 T2

1
20 log GT2 3.08 [dB]
GT2.e( f) 20 log GT2( f) 2 T2

0
2
4
6
8
[dB]

GT2.e( f) 10
12 Fig. 3.4
14
16 Bode plot of GT2(f)
18
20
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]
36 3 Mathcad Worksheets Amp3

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks Page 5

3 9
4.2 T1 = 3180µs, T3 = 318µs : P3 1.450653 10 C12 227.1428575 10 F

Fig. 3.5
T1 & T3 defining network

succ-apps of 1. C12 & 2. P3 should bring in line the values of T1 & T3 with the required RIAA values
:
6
T3 C12 R35 T3 318 10 s
3
RT1.eff( f) Ro.cf.dif( f) P3 R33 R34 RT1.eff( h) 14.416 10

1
1
1 1
1 1 1 1 1
T1 C12 R35 RT1.eff( h) R1amp5 R36 R2amp5 R37

3
T1 3.18 10 s
1
1 1 3
RL.dif RL.dif 100 10
R1amp5 R2amp5 R36 R37

1
1
1 1
R35
2j f C12 RL.dif
GT1.3( f) GT1.3( h) 0.098
1
1
1 1
R35 RT1.eff( f)
2j f C12 RL.dif 20 log GT1.3( 20Hz) 1.829 [dB]

1 1
20 log GT1.3 4.15 [dB] 20 log GT1.3 18.203 [dB]
2 T1 2 T3

GT1.3.e ( f) 20 log GT1.3( f) 20 log GT1.3( 20kHz) 21.167 [dB]

0
2
4
6
8
[dB]

GT1.3.e( f) 10 Fig. 3.6

12 Bode plot of GT1.3(f)
14
16
18
20
22
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]
3 Mathcad Worksheets Amp3 37

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks Page 6

4.3 RIAA transfer :

1
Griaa( f) GT2( f) GT1.3( f) Griaa.e( f) 20 log Griaa( f) Griaa( h)

20
16
12
8
4
[dB]

Griaa.e ( f) 0
4
8 Fig. 3.7
12 Normalized (1kHz at 0dB)
16 RIAA transfer function
20
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

1
[times]

Griaa ( f) 0.1 Fig. 3.8

Bode plot of the gain of the
combined RIAA network

0.01
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

2
6
1 2 h 318 10 s
R1000
2 2
6 6
1 2 h 3180 10 s 1 2 h 75 10 s

2
6
1 2 f 318 10 s 1
R0 ( f) R1000
2 2 Dev1e ( f) 20 log R0 ( f) Griaa.e( f)
6 6
1 2 f 3180 10 s 1 2 f 75 10 s

0.1
0.075
0.05
0.025
[dB]

Dev1e ( f) 0
0.025 Fig. 3.9
0.05 Deviation from the exact
0.075 RIAA transfer
0.1
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]
38 3 Mathcad Worksheets Amp3

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks Page 7

5. Transfer function T(f) of Amp3

General assumption: the corner frequency of the i/p hp R21, R22, C5 ... C8 is chosen that there is no influence on a flat
phase and frequency response in B20k !

Tamp3( f) Gamp3( h) GT2( f) GT1.3( f) Tamp3( h) 1.435716

Tamp3.e( f) 20 log Tamp3( f) Tamp3.e( h) 3.141371 [dB]

30
20
10
[dB]

Tamp3.e( f) Fig. 3.10

0
Bode plot of Amp3
10
20
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]
1
Dev2e ( f) 20 log R0 ( f) 20 log Tamp3( f) Tamp3( h)

0.1
0.05
[dB]

Dev2e ( f) 0
0.05

0.1 Fig. 3.11 = Fig. 2.3

3 4 5
10 100 1 10 1 10 1 10 Fig. 3.10's deviation from the
f .
[Hz] exact RIAA transfer

6
1 2j f 318 10 s 3
RIAA( f) RIAA( h) 101.03 10
6 6
1 2j f 3180 10 s 1 2j f 75 10 s

1
Im Tamp3( f) RIAA( f)
amp3( f) atan
Re Tamp3( f) RIAA( f)
1 amp3( h) 0

amp3( f)
[°]

0
deg Fig. 3.12
1 Fig. 3.10's phase response

2
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]
3 Mathcad Worksheets Amp3 39

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks Page 8

6. Noise and SN calculations :

3
6.1 General data : vi.ref 100 10 V

3 3
vo.ref vi.ref Tamp3( h) vo.ref 143.572 10 V vo.ref.m 152.8 10 V

Op-amps:
9 15
en.i.op1 4.8 10 V en.i.op2 en.i.op1 in.i.op1 2.5 10 A in.i.op2 in.i.op1

fc.e.1 40Hz fc.i.1 0.1Hz

fc.e.1
en.i.op1( f) en.i.op1 1 en.i.op2( f) en.i.op1( f)
f

Triodes:

3.06 9
rN1 rN1 874.286 en.rN1 4 k T rN1 B1 en.rN1 3.9 10 V
gm1

rN2 rN1 en.rN2 en.rN1

3.06 9
rN3 rN3 874.286 en.rN3 4 k T rN3 B1 en.rN3 3.9 10 V
gm1

rN4 rN3
en.rN4 en.rN3

fc1
en.rN1( f) en.rN1 1 9
f en.rN1( h) 5.515 10 V

en.rN2( f) en.rN1( f)

fc3
en.rN3( f) en.rN3 1 9
f en.rN3( h) 12.935 10 V

en.rN4( f) en.rN3 ( f)

Resistances:

Ra Ra1 Ra2 Rc Rc3 Rc4 NIe.a 24 [dB] NIe.c 24 [dB]

Ra R40 VRa Ia1 Ra VRa 36.8 V

9
en.R40 4 k T B1 R40 en.R40 17.891 10 V

NIe.a
10 12 V 2
10 10 Ra 9
en.R40ex( f) B1 en.R40ex( h) 48.388 10 V
ln( 10) f
40 3 Mathcad Worksheets Amp3

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks Page 9

2 2 9
en.Ra( f) en.R40 en.R40ex( f) en.Ra( h) 51.59 10 V

ra1 9
en.Ra.eff( f) en.Ra( f) en.Ra.eff( h) 16.018 10 V
ra1 Ra

9
Rgg3 R44 en.Rgg3 4 k T B1 R44 en.Rgg3 2.288 10 V

Rgg4 Rgg3 en.Rgg4 en.Rgg3

Rc R46 VRc Ia1 Rc VRc 95.333 V

ra3
rc rc3 rc4 rc rc 276.19
1 µ1

9
en.R46 4 k T B1 R46 en.R46 28.796 10 V

NIe.c
10 12 V 2
10 10 Rc 9
en.R46ex( f) B1 en.R46ex( h) 125.354 10 V
ln( 10) f

2 2 9
en.Rc( f) en.R46 en.R46ex( f) en.Rc( h) 128.619 10 V

rc 12
en.Rc.eff( f) en.Rc( f) en.Rc.eff( h) 740.95 10 V
Rc rc

6.2 Noise Voltage of the T2 network :

1
1 1
ZT2 ( R25 R26 P1) ( R39a R39b) 3
ZT2 1.595 10
9
en.Z.T2( f) GT2( f) 4 k T B1 ZT2 en.Z.T2( h) 4.727 10 V

8
1 10

Fig. 3.13
[V/rtHz]

Frequency response of
en.Z.T2( f) 1 10 9
the noise voltage density
of the T2(f) network

10
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

ZT2( f) ZT2 GT2( f)

3 Mathcad Worksheets Amp3 41

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks Page 10

6.3 Noise Voltage of the T1/T3 network at the o/p :

1
1 1 1
ZT1.3( f) 3
Ro.cf.dif( f) P3 R33 R34 1 RL.dif ZT1.3( h) 1.407 10
R35
2j f C12
12
en.Z.T1.3( f) GT1.3( f) 4 k T B1 ZT1.3( f) en.Z.T1.3( h) 482.963 10 V

7
1 10

8
1 10
[V/rtHz]

en.Z.T1.3( f) Fig. 3.14

1 10
9 Frequency response of
the noise voltage density
of the T1(f)+T3(f) network
10
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

6.4 Relevant Amp5 data :

2.5 15 9
in.i.amp5 10 A average noise voltage in B20k : en.i.amp5 8.103 10 V
2

1V
Gamp5 Gamp5 6.965
vo.ref

0.1 3
Gcm.amp5 12 Gcm.amp5 1 10
200
Gamp5 3
CMRRamp5 CMRRamp5 6.965 10
Gcm.amp5

CMRRamp5.e 20 log CMRRamp5 CMRRamp5.e 76.859 [dB]

6.5 Impact of the current sink CCsi of BJTs T1 and T2 and CMRR of the gain stage:
6.5.1 CCsi : 19
q 1.6022 10 As
3
IC1 4 10 A VCE1 2.2V VA1 100V hfe1 500 RE R42 P8 RE 195

q IC1 3 VA1 3
gmt1 gmt1 147.361 10 S rbb 13.74 rce1 rce1 25 10
kT IC1

hfe1 3 ra1 R40

rbe1 rbe1 3.393 10 rc.1 0.5 rc.1 444.762
gmt1 1 µ1
42 3 Mathcad Worksheets Amp3

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks Page 11

3
IC2 1 10 A VCE2 1.2V VA2 100V hfe2 500

q IC2 3 VA2 3
gmt2 gmt2 36.84 10 S rce2 rce2 100 10
kT IC2
hfe2 3
rbe2 rbe2 13.572 10
gmt2

rbe1
hfe1
rce1 3
rccsi rce1 1 rccsi 695.418 10
rbe1
1
RE rbe2
RE rbe2

in.C1 2 q IC1 B1 12
in.C1 35.802 10 A

Gbjt gmt1 rc.1 Gbjt 65.541

2 2
2k T 9
en.bjt B 4 k T B1 rbb RE en.bjt 1.921 10 V
q IC1 1

2
2 2 rc.1 12 9
en.bjt Gbjt 80.472 10 V in.C1 rc.1 15.923 10 V
rc.1 rccsi

=>
2
2 2 2 rc.1 9
en.ccsi in.C1 rc.1 en.bjt Gbjt en.ccsi 15.923 10 V
rc.1 rccsi

R40
Gcgs.1 1 µ1 Gcgs.1 20.685
ra1 R40

6.5.2 CMRR of Amp3 :

3
CMRRamp3.rot rccsi gm1 CMRRamp3.rot 2.434 10

CMRRamp3.rot.e 20 log CMRRamp3.rot CMRRamp3.rot.e 67.726 [dB]

2 rccsi 2 rccsi 3
CMRRamp3 1 µ1 CMRRamp3 1.565 10
ra1 Ra ra1 Ra

CMRRamp3.e 20 log CMRRamp3 CMRRamp3.e 63.888 [dB]

3 Mathcad Worksheets Amp3 43

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks Page 12

6.6 Output noise voltage of Amp3 + Amp5 with i/p shorted (amp3.5) :

2 2 9
en.o.op1.2 ( f) Gop1.2 en.i.op1( f) en.i.op2( f) en.o.op1.2 ( h) 6.923 10 V

2 2 2 2 2 2
en.o.amp3.5( f) Gamp5 en.o.op1.2 ( f) GT2( f) en.Z.T2( f) G0.dif Gcf ( f) GT1.3( f)
2
2 en.rN1( f)
2 2 2
2 en.Rgg3 2 en.Ra.eff( f) 2 en.rN3( f)
2
2 en.Rc.eff( f)
2
en.ccsi 2 Gcgs.1 Gcf ( f) GT1.3( f)
CMRRamp5
2 2 2 2
en.Z.T1.3( f) in.i.amp5 ZT1.3( f) en.i.amp5

9
en.o.amp3.5( h) 123.309 10 V

5
1 10

6
1 10
[V/rtHz]

en.o.amp3.5( f)
7
1 10

8
1 10
3 4 5
10 100 1 10 1 10 1 10
f Fig. 3.15a &
[Hz] Fig. 3.15b = Fig. 2.3
Amp3+Amp5 output
noise voltage density
0
20 with i/p shorted
40
60
80
[dBV]

en.o.amp3.5( f)
20 log 100
1V 120
140
160
180
200
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

20000Hz
1 2 6
eN.o.amp3.5 en.o.amp3.5( f) df eN.o.amp3.5 22.905 10 V
B1
20Hz

Tamp3( h) Gamp5 10

en.o.amp3.5( f) 9
en.i.amp3.5( f) en.i.amp3.5( h) 12.331 10 V
Tamp3( h) Gamp5
44 3 Mathcad Worksheets Amp3

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks Page 13

6.7 SNs of the sequence of Amp3 + Amp5 :

20000Hz
1 2
en.o.amp3.5( f) df
B1
20Hz
SNo.amp3.5 20 log SNo.amp3.5 92.801 [dBV]
1V

measured with high CMRR instrument: -90.43 [dBV]

B(f) = A(f) = A-weighting transfer function

1.259
B( f)
2 2 2 2
20.6Hz 107.7Hz 737.9Hz f
1 1 1 1
f f f 12200 Hz

20000Hz
1 2 2
en.o.amp3.5( f) B( f) df
B1
20Hz
SNa.o.amp3.5 20 log SNa.o.amp3.5 100.165 [dBV(A)]
1V

SNa.o.amp3.5.m 99.00 [dBV(A)]

20000Hz
1 2 2
en.i.amp3.5( f) B( f) df
B1
20Hz
SNa.i.amp3.5 20 log SNa.i.amp3.5 100.165 [dB(A)]
vi.ref

7. Amp3 + Amp5 with i/p loaded by the noise voltage of a preceding MC amp1 with extremely
low input noise voltage & current density :
12
9 2.4 10 A
Gamp1 200 R0 20 en.i.0 0.2 10 V 2 in.i.0
2
3 3
T 315K vo.ref.mc 100 10 V vi.ref.mc 0.5 10 V

12
en.R0 4 k T B1 R0 en.R0 589.851 10 V

2 2 2 2 12
en.i.0.eff en.i.0 en.R0 in.i.0 R0 en.i.0.eff 655.039 10 V

9
en.o.amp1 Gamp1 en.i.0.eff en.o.amp1 131.008 10 V

B20k 6
eN.o.amp1 en.o.amp1 eN.o.amp1 18.518 10 V
B1

2 2 2 2 6
en.o.mc( f) en.o.amp1 Tamp3( f) Gamp5 en.o.amp3.5( f) en.o.mc( h) 1.313 10 V
3 Mathcad Worksheets Amp3 45

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks Page 14

en.o.mc( f)
en.i.mc( f) 12
Gamp1 Tamp3( h) Gamp5 en.i.mc( h) 656.468 10 V

20000Hz
1 2
en.o.mc( f) df
B1
20Hz
SNriaa.o.amp3.5.mc 20 log SNriaa.o.amp3.5.mc 78.201 [dB]
1V
20000Hz
1 2 2
en.o.mc( f) B( f) df
B1
20Hz
SNariaa.o.amp3.5.mc 20 log SNariaa.o.amp3.5.mc 82.523 [dB(A)]
1V

20000Hz
1 2 2
en.i.mc( f) B( f) df
B1
20Hz
SNariaa.i.amp3.5.mc 20 log SNariaa.i.amp3.5.mc 82.523 [dB(A)]
vi.ref.mc

8. Preceding MC Amp1alone and noise calculation up to the o/p of Amp5 :

6
en.o.mc.eff( f) en.o.amp1 Tamp3( f) Gamp5 en.o.mc.eff( h) 1.31 10 V

en.o.mc.eff( f) 12
en.i.mc.eff( f) en.i.mc.eff( h) 655.039 10 V
Tamp3( h) Gamp1 Gamp5

20000Hz
1 2
en.o.mc.eff( f) df
B1
20Hz
SNriaa.o.mc 20 log SNriaa.o.mc 78.263 [dB]
1V

20000Hz
1 2
en.i.mc.eff( f) df
B1
20Hz
SNriaa.i.mc 20 log SNriaa.i.mc 78.263 [dB]
vi.ref.mc

20000Hz
1 2 2
en.o.mc.eff( f) B( f) df
B1
20Hz
SNariaa.o.mc 20 log SNariaa.o.mc 82.582 [dBA]
1V

20000Hz
1 2 2
en.i.mc.eff( f) B( f) df
B1
20Hz
SNariaa.i.mc 20 log SNariaa.i.mc 82.582 [dB(A)]
vi.ref.mc
46 3 Mathcad Worksheets Amp3

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks Page 15

95
100
105
110
20 log en.o.amp3.5( f)
115
[V/rtHz]

20 log en.o.mc( f) 120

20 log en.o.mc.eff( f) 125
130
135
140
145
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

Fig. 3.16 Comparison of the various output noise voltage densities vs. frequency
(trace 3 is nearly hidden by trace 2)

1
Dev3e ( f) 20 log R0 ( f) 20 log en.o.mc.eff( f) en.o.mc.eff( h)

0.1
0.075
0.05
0.025
[dB]

Dev3e ( f) 0 Fig. 3.17

0.025 Deviation from the
0.05 exact RIAA transfer of
0.075 Amp1+Amp3+Amp5
0.1
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

Mainly, the shown deviation comes from the chosen C44 & C45 values!

9. MC Amp1 noise worsening by additional noise from Amp3+Amp5:

NFe.mc12 SNariaa.o.amp3.5.mc SN ariaa.o.mc NFe.mc12 0.059 [dB]

10. MM Amp noise worsening by additional noise from Amp3+Amp5:

SNariaa.o.mm 80.5 [dB(A) ref. 100mVrms ]

SNariaa.o.mm SNa.i.amp3.5
10 10
SNariaa.i.amp3.mm 20 log 10 10 SNariaa.i.amp3.mm 80.453 [dB(A)]

NFe.mm SN ariaa.i.amp3.mm SN ariaa.o.mm NFe.mm 0.047 [dB]

3 Mathcad Worksheets Amp3 47

3.1 MCD-WS: The Triode Gain Stage Amp3 with RIAA Networks Page 16

11. Amp3 with i/p loaded by the noise voltage of a 1k resistor and via o/p of Amp5:

R0 1000 T 315K

9
en.R0 4 k T B1 R0 en.R0 4.171 10 V

2 2 2 2
en.o.1k ( f) en.R0 Tamp3( f) Gamp5 en.o.amp3.5( f) 9
en.o.1k ( h) 127 10 V

en.o.1k ( f)
en.i.1k( f) 9
Gamp5 Tamp3( h) en.i.1k( h) 12.7 10 V

20000Hz
1 2
en.o.1k ( f) df
B1
20Hz
SN riaa.o.amp3.5.1k 20 log SNriaa.o.amp3.5.1k 92.763 [dBV]
1V

20000Hz
1 2 2
en.o.1k ( f) B( f) df
B1
20Hz
SN ariaa.o.amp3.5.1k 20 log SNariaa.o.amp3.5.1k 99.999 [dBV(A)]
1V

20000Hz
1 2 2
en.i.1k( f) B( f) df
B1
20Hz
SN ariaa.i.amp3.5.1k 20 log SNariaa.i.amp3.5.1k 99.999 [dB(A)]
100mV
The Solid-State (Op-Amp) Driven
Central Amplifier Amp4 4

4.1 General Design of Amp4

Outside the microphone amp world I rarely found a design configuration alike the
one I have chosen for the Fig. 4.1 Amp4.1 Together with low-noise and low-THD
goals, the main development drivers are overload margin and CMRR. Low-noise
means a noise level equal to the one of Amp3; hence, an input referred noise
voltage density of appr. 10 nV/rtHz balanced should make it. The chosen input op-
amp types assure very low THD in the main gain producing input gain stage of
Amp4’s OPs 3 & 4. They have Amp3-similar input current behaviour too.
Figure 4.1 shows the central Amp4 without RIAA input and output networks.
The output stage (OPs 5 & 6) has a balanced gain of 2.2 These two op-amps set
the overload margin too. Their output voltage swing is ±12.3 Vp nominal.3 Hence,
the output voltage swing of such an amplifier becomes ±24.6 Vp = 49.2 Vpp
nominal. With an input signal level of 100 mVrms + 20 dB overload mar-
gin = 1 Vrms and a gain of appr. 16 (set by P9) we need a max. voltage swing of
16 V * 2 * √2 = 45.255 Vpp. Without selection of OPs 5 & 6 and a very exact ± 15V
power supply, I measured clipping at a 17.22 Vrms ≡ 48.71 Vpp.
High CMRR is important because neither the Amp1 & Amp2 alternatives show
excellent CMRR nor the external amps will always produce it. The calculated4
CMRR becomes appr. 16,000, hence, expressed in dB: CMRRe ≥ 20 * log
(16,000) = 84 dB. Careful selection of the R9–R18 values to a 0.01 % tolerance
level could improve CMRRe to 104 dB.
The output stage’s output resistance is very low. Thus, to get equal conditions
for the RIAA network I’ve added R31 & R32 by taking their values equal to the
calculated CF output resistances of Amp3. Replacement of Ro.cf3 in (2.1) and

1
“Schaltungstips für Vierfach Op-amps” (Hints for quadruple op-amps), Bob Atwell, Elektronik
Nr. 20, 1988, p. 110ff, German language only.
2
Further details see Sect. 4.2.1.
3
Analog Devices data sheet.
4
Further details see Sect. 4.2.3.

© Springer International Publishing Switzerland 2016 49

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_4
50 4 The Solid-State (Op-Amp) Driven Central Amplifier Amp4

Fig. 4.1 Amp4 without RIAA networks

4.1 General Design of Amp4 51

0
20
40
60
[dBV]

⎛ en.o.amp4.5(f) ⎞ 80
20⋅log ⎜ ⎟ 100
⎝ 1V ⎠ 120
140
160
180
200
1 .10 1 .10 1 .10
3 4 5
10 100
f
[Hz]

Fig. 4.2 Output noise voltage density of the amp sequence Amp4 + Amp5 + Trafo with input
shorted

0.1
0.05
[dB]

Dev2e (f) 0
0.05

0.1
1 .10 1 .10 1 .10
3 4 5
10 100
f
[Hz]

Fig. 4.3 Calculated deviation from the exact RIAA transfer

without frequency dependency by a capacitance inside the signal path we can

calculate the RIAA voltage divider GT1.3(f) (and thus the time constants T1 & T3)
the same way we’ve done it for the Amp3 case. Section 4.2.2 gives the details, incl.
the input voltage divider GT2(f).
My first design worked with 4 × OPA627. These op-amps are very expensive.
I found out that in the output stage OP27s work equally well, without changing
overall noise level, THD, IMD, etc.
With the CLIO signal generator5 and depending on Amps 1 or 2 of this
development, the 1 kHz THD-level of the Amp1/Amp2 + Amp4 + Amp5 + Trafo
chain becomes always ≤0.005 %, IMD ≤0.002 %. Via shorted external input, the
amp sequence of Amp4 + Amp5 + Trafo produces a measured (calculated)
A-weighted output referred SNa.o.amp4.5 = −100.8 dBV(A) (−100.0 dBV(A)6).
Figure 4.2 shows the calculated and corresponding output noise voltage density
curve and Fig. 4.3 the calculated deviation from the exact RIAA transfer. Chapter
12 shows the measured curves.

5
More on THD & IMD see Chaps. 12 and 15.
6
See MCD-WS 5.1.
52 4 The Solid-State (Op-Amp) Driven Central Amplifier Amp4

4.2 Gain and CMRR

4.2.1 Gain

Normally, ±15 V op-amps (also rail-to-rail) cannot produce higher output voltages
than the used supply voltages would allow. However, the well-known two-op-amp
balanced in/balanced out amplifier à la Fig. 4.4 creates higher output voltages, up to
2 times the output voltage of the single op-amp, paid for it by a disadvantage of
rather lousy CMRR.
With equal values of R1 & R2 and with the input and output voltages vi1,2 and
vo1,2 from lead to ground the differential gain equation for the Fig. 4.4 amplifier
thus becomes (without any phase shift indicated by the (+) & (−) signs):
vo
Gdif ¼ ¼ Gamp4:in
vi
vo1  vo2
¼ ð4:1Þ
vi1  vi2
R1 þ R2
¼1þ
RG
To get an increase in op-amp output voltage swing that exceeds the boundaries
set by the op-amp supply voltage we could firstly load the op-amp by an additional
high-voltage output stage with gain. However, CMRR would still become lousy.
Secondly, if only doubling of the output voltage swing is enough, we could take the
fully op-amp based solution shown in Fig. 4.5. It offers additional high common
mode rejection by OPs 5 & 6, together with their corresponding circuitry.

Fig. 4.4 Two-op-amp fully

differential amplifier
4.2 Gain and CMRR 53

Fig. 4.5 Basic Amp4 circuit with input stage around OPs 3 & 4, output stage around OPs 5 & 6,
and all relevant signal voltages

Amp4 consists of two typical three-op-amp instrumentation amplifiers (INA)

with OPs 3 & 4 as the only input stage and OPs 5 or 6 as two equal but differently
connected subtractors working as output stages, hence, producing a 180° phase shift
of the first gain stage’s output signal vo.1st. With equal valued Rs we obtain the gain
equation as follows:
1. In an instrumentation amplifier, the gain of the outputs stage is 1 as long as there
are equal valued resistors R. Hence, the amps around OPs 5 & 6 have a gain of 1
each, thus, the whole output stage has a combined differential gain Gamp4.out = 2,
as follows:
vo:tot
Gamp4:out ¼
vo:1st
vo:tot1  vo:tot2 ð4:2Þ
¼
vo:1st:1  vo:1st:2

vo:tot1 ¼ ½vo:1st:1  ðvo:1st:2 ÞGop6 ¼ vo:1st Gop6

ð4:3Þ
vo:tot2 ¼ ½vo:1st:1  ðvo:1st:2 ÞGop5 ¼ vo:1st Gop5

Gop5 ¼ Gop6 ¼ 1 ð4:4Þ

) Gamp4:out ¼ 2 ð4:5Þ
54 4 The Solid-State (Op-Amp) Driven Central Amplifier Amp4

Fig. 4.6 Symbol for a fully differential amplifier with balanced electrometer input

2. The input stage has the gain already given in (4.1) Gamp4.in = Gdif.
3. With the signal voltages shown in Fig. 4.5 the differential (=balanced) gain
Gamp4 of Amp4 thus becomes:
vo:tot
Gamp4 ¼
vi
¼ Gamp4:in Gamp4:out ð4:6Þ


R1 þ R2
¼ 1þ 2
RG

With the shown op-amp arrangement, the output becomes a true differential one.
The whole amp’s circuit could be expressed by the Fig. 4.6 symbol.
Note: the Fig. 4.6 symbol is derived from another kind of fully differential amp’s
symbol: eg the OPA1632 with a different looking input arrangement alike the one
of OPs 5 or 6 plus a fully differential output configuration.

4.2.2 RIAA Transfer Function

From Figs. 1.4 and 1.6 we can derive the frequency dependent and RIAA transfer
loaded gains GT2(f) and GT1.3(f) of the Amp4 input and output networks as follows:

M
GT2 ðf Þ ¼  1
1
M þ R25 þ R26 þ P1 þ 2 R1 o:op1 þ R23 ð4:7Þ
h i1
M ¼ 2j p f ðC9 þ Ci:dif Þ þ ðR39c þ R39dÞ1

 1 1
1 1
½2jpfC12 þ R35 þ RL:dif
GT1:3 ðf Þ ¼  1 1 ð4:8Þ
1 1
½2jpfC12 þ R35 þ RL:dif þ RT1:eff
4.2 Gain and CMRR 55

 1 1
RL:dif ¼ R1amp5 þ R2amp5 þ ½R36 þ R371 ð4:9Þ

RT1:eff ¼ R31 þ R32 þ R33 þ R34 þ P4 ð4:10Þ

Ro:op1 ¼ Ro:op2  0 X ð4:11Þ

=> With C9 and C12 carefully selected according to Figs. 1.4 and 1.6 the transfer
function Tamp4(f) of the whole Amp4, including RIAA transfer, thus becomes:

Tamp4 ðf Þ ¼ Gamp4 ðf ÞGT2 ðf ÞGT1:3 ðf Þ ð4:12Þ

4.2.3 CMRR

The common mode gain Gcm of the Fig. 4.4 amp looks a bit different. With equal
values of R1 & R2 it simply becomes:

Gcm:in ¼ 1 ð4:13Þ

With equal voltages at the input leads (= the common mode input voltage) the
signal current that flows through RG becomes 0; hence, RG’s gain effective value
looks like infinite, it has thus no effect on the amplification, and OPs 3 & 4 work as
voltage followers only.
With (4.1) and (4.7) we obtain the common mode rejection ratio CMRR of the
Fig. 4.4 amp as follows:

Gdif
CMRR ¼
Gcm:in ð4:14Þ
CMRRe ¼ 20 logðCMRRÞ½dB

However, the Fig. 4.5 CMRR situation looks very much different to the one for
Fig. 4.4. The differential gain is given by (4.6). The common mode gain of an INA
is mainly produced by its output stage. The details can be studied eg in,7 however,
expressed roughly and in short words: mainly, the tolerances of the resistors R in
Fig. 4.5 set the value of the common mode gain, eg 1 % resistors would lead to a
common mode gain of 1/200 or −46 dB, 0.1 % resistors produce a common mode
gain of 1/2000 or −66 dB and so forth. We obtain thus the CMRR for the INAs
with OPs 3 & 4 & 5 and/or OPs 3 & 4 & 6:

GINA
CMRRINA ¼ ð4:15Þ
Gcm:INA

“Electronic Circuits”, 2nd ed. Tietze/Schenk, Springer, Heidelberg, ISBN 978-3-540-00429-5.

7
56 4 The Solid-State (Op-Amp) Driven Central Amplifier Amp4

Gcm:INA ¼ Gcm:in Gcm:out

tolerance ½% ð4:16Þ
¼1
200 %

and for Amp4:

Gamp4
CMRRamp4 ¼ ð4:17Þ
Gamp4:cm

Gamp4:cm ¼ Gcm:in Gcm:out

Gcm:in ¼ 1
ð4:18Þ
tolerance ½%
Gcm:out ¼2
200 %

I must point out that, with the shown circuits, the CMRR results are best-case
results only. Many influential factors may destroy the rather good-looking picture,
eg hidden and/or unequal capacitances in the circuit lead to a decreasing CMRR
with increasing frequency. However, to further improve CMRR there is a broad
range of measures by eg adjustment of input capacitance of the subtractor and/or by
trimming the resistor R between the subtractor’s (+) input and ground, etc. Details
can be studied in the application papers of the IC manufacturers.

4.3 Noise Calculations

The calculation of the noise production of Amp4 alone, and further on of Amp4 in
conjunction with the RIAA transfer producing networks, becomes a rather complex
task. The numbers of the gain stages follows the logic of Fig. 1.2: 1st gain stage =
OPs 1 & 2, 2nd gain stage = OPs 3 & 4, 3rd gain stage = OPs 5 & 6. According to
MCD-WS 5.1 we have to split the calculation approach into five different actions:
1. Calculation of the output noise voltage density en.o.2nd of Amp4’s input stage
with OPs 3 & 4 and input shorted or loaded by a resistance R0 (Fig. 4.7).
2. Calculation of the output noise voltage density en.o.5 of the subtractor with OP5
and input of the subtractor shorted; the output noise voltage en.o.6 of the sub-
tractor OP6 equals the one with OP5, hence, by rms summing we can calculate
the output noise voltage density en.o.3rd of the 3rd gain stage with input shorted
(Fig. 4.8).
3. Calculation of Amp4’s output noise voltage density en.o.tot and its input referred
noise voltage density en.i.tot.
4.3 Noise Calculations 57

Fig. 4.7 General noise voltage and current situation of the Amp4 input (2nd) gain stage

Fig. 4.8 General noise voltage and current situation of one Amp4 subtractor of the 3rd gain stage

4. With en.i.tot and in.i on the table, we can now calculate any input load R0
dependent input and/or output referred SN(R0).
5. Integration of the Fig. 1.2 input stage (OPs 1 & 2), the 75 µs network, the output
load with the 318 µs/3180 µs network, and the Amp5 input into the calculation
course. Consequently, we obtain the output referred noise voltage density
en.o.amp4.5(f) of the amp sequence Amp4 + Amp5 + Trafo, with input shorted.
58 4 The Solid-State (Op-Amp) Driven Central Amplifier Amp4

4.3.1 Output Noise Voltage of the Amp4 Input Gain Stage

(2nd in Figs. 1.2 + 4.1) with OPs 3 & 4

According to Fig. 4.7, with a balanced input load R0, the output referred noise
voltage density en.o.2nd can be calculated by a rather practical equation without big
errors. Concerning a frequency independent version I changed Fred Floru’s8 math
approach a bit by treating the noise voltage of RG in the shown different way.
However, the calculated results do not differ much. It follows the math approach for
single-ended op-amps I’ve already shown in TSOS-1, Chap. 3.5, and TSOS-2,
Chap. 8. Hence, after many measurements with various kinds of op-amps, input
loads, and gains, with R1 = R2, en.i3 = en.i4, and in.i3 = in.i4 I could obtain the below
given frequency independent version of the output noise voltage density en.o.2nd:
with
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:i ¼ e2n:i:3 þ e2n:i:4 ð4:19Þ

and
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
!1
u
u 1
in:i ¼ t þ 2
1
ð4:20Þ
i2n:i:3 in:i:4

We obtain the R0 and RG dependent expression for en.o.2nd:

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u  2
u 2
u en:i þ i2n:i R02 þ e2n:R0 1 þ 2R1
en:o:2nd ðR0,RGÞ ¼ u
t 2 
RG
 2 ð4:21Þ
þ2 in:i:3 R1 þ en:R1 þ en:RG 2R1
2 2 2
RG

Concerning a frequency dependent version by taking 1/f-noise into account and in

cases of 1/f-noise corner frequencies fc > 1 Hz (fc.e for voltages and fc.i for currents)
it makes sense to change to the frequency dependent version en.o.2nd(f, R0, RG)
by adequate integration of the following terms into (4.19)–(4.21):

8
Improved Mic Preamp IC, Fred Floru, 2001, THAT Corp., (AES UK 103)
I found in the web the following THAT Corp. based design notes and AES lectures. They give an
additional and rather deep look into the matter of balanced in/balanced out microphone amps:
• Designing Mic Preamps, Gary K. Herbert, 2010 (AES 129)
• De-integrating Integrated Circuit Preamps, Les Tyler, 2013
• Double Balanced Microphone Amplifier Notes, Graeme John Cohen, 2008
• Perform audio line receiver impedance balancing, Wayne Kirkwood.
4.3 Noise Calculations 59

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
f c:e:3
en:i:3 ðf Þ ¼ en:i:3 þ1
f
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð4:22Þ
f c:i:3
in:i:3 ðf Þ ¼ in:i:3 þ1
f

4.3.2 Output Noise Voltage of the Amp4 Output Gain Stage

(3rd in Figs. 1.2 + 4.1) with OPs 5 & 6

The 3rd gain stage around OPs 5 & 6 is composed by two subtractors with opposite
connection of their input leads to the 2nd gain stage’s output leads. By ignoring any
noise voltage influence of the 3rd gain stage the effect of this arrangement leads to
two 100 % correlated noise voltages at the outputs of OPs 5 & 6. Their sum would
automatically lead to a stage gain factor of 2, as long as all Fig. 4.5 resistors R are
chosen of equal value.
Nevertheless and rather important for low overall gains, we have to take the fully
un-correlated noise voltage of the two halves of the 3rd gain stage into account too.
The situation of all noise sources of one branch of the 3rd gain stage is given in
Fig. 4.8.
In order to catch the equation to calculate the output noise voltage density en.o5
we shorten the input. Moreover, for general purposes we rename the Rs by
R3 = R1, R4 = R2, and RP = R1||R2. The output noise voltage density en.o5 at the
output of OP5 thus becomes9:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi



R4 2 e2n:i:5
en:o:5 ¼ 2 1þ þ e2n:RP þ i2n:i:5 RP ð4:23Þ
R3 2

In cases of 1/f-noise corner frequencies fc > 1 Hz (fc.e for voltages and fc.i for
currents) it makes sense to change to the frequency dependent version en.o5(f) by
adequate integration of the following terms into (4.23):
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
f c:e:5
en:i:5 ðf Þ ¼ en:i:5 þ1
f
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð4:24Þ
f c:i:5
in:i:5 ðf Þ ¼ in:i:5 þ1
f

9
“Low-Noise Electronic System Design”, 1993, Motchenbacher/Connelly, Wiley-Interscience,
N. Y., ISBN 0-471-57742-1.
60 4 The Solid-State (Op-Amp) Driven Central Amplifier Amp4

4.3.3 Total Input and Output Noise Voltages of Amp4

With the rms sum en.o.3rd of the un-correlated subtractor output noise voltage
densities en.o5 and en.o6

en:o:6 ¼ en:o:5 ð4:25Þ

qffiffiffiffiffiffiffiffiffiffiffiffi
) en:o:3rd ¼ 2e2n:o:5 ð4:26Þ

the output noise voltage en.o.tot of Amp4 is the nothing else but the rms sum of the
output noise voltage en.o.3rd of the 3rd gain stage plus the output noise voltage en.o.2nd
of the 2nd gain stage multiplied by the gain of the 3rd gain stage (here, with equal Rs
it’s two), hence, en.o.tot becomes
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:o:tot ¼ e2n:o:2nd G23rd þ e2n:o:3rd ð4:27Þ

Consequently, we obtain Amp4’s input referred noise voltage density en.i.tot as

follows:
en:o:tot
en:i:tot ¼ ð4:28Þ
Gamp4

4.3.4 Noise and SN Calculations According to Fig. 1.2

The calculation of the noise voltage of Amp4 alone makes no sense, as long as
there are influential factors not taken into account: the ones at its input (OPs 1 &
2 and T2(f)) and at its output (T1(f) + T3(f) + Amp5). All together, they generate a
noise voltage that can be measured at the output of Amp5, and thus be compared
with the calculated results. The calculation course follows the mathematical course
given in MCD-WS 5.1, “6. Noise and SN calculations”. However, here comes the
short version, tackling the major factors.
To calculate the output noise voltage density en.o.amp4.5(f) at the output of Amp5
with input shorted the rather extensive equation looks as follows10:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
½en:o:2nd ðf ÞG3rd 2 þ en:o:3rd ðf Þ2
en:o:amp4:5 ðf Þ ¼ ð4:29Þ
þ en:Z:T1:3 ðf Þ2 þ i2n:i:amp5 ZT1:3 ðf Þ2 þ e2n:i:amp5

10
Impedances ZT1.3(f) and ZT2(f) and their respective noise voltages: see MCD-WS 5.1, Points
6.4ff.
4.3 Noise Calculations 61

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u 
u en:o:1st ðf Þ2 GT2 ðf Þ2 þ en:Z:T2 ðf Þ2
u G2
en:o:2nd ðf Þ ¼ t þ 0:5i2n:i:2nd ZT2 ðf Þ2 þ 2en:i:2nd ðf Þ2 2nd ð4:30Þ
þ i2n:i:2nd ðR61 þ R62Þ2 þ e2n:R61 þ e2n:R62 þ e2n:RG:o

en:RG:o ¼ en:RG ðG2nd  1Þ ð4:31Þ

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
h i ffi
en:o:1st ðf Þ ¼ en:i:1 ðf Þ2 þ en:i:2 ðf Þ2 G21st ð4:32Þ

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:o:3rd ðf Þ ¼ en:o:5 ðf Þ2 þ en:o:6 ðf Þ2 ð4:33Þ

From the preceding sections, we can take the gain results as follows:

G1st ¼ 1 ð4:34Þ

G2nd ¼ Gdif ¼ Gamp4:in ð4:35Þ

G3rd ¼ Gamp4:out ð4:36Þ

According to (4.29)–(4.36), Fig. 1.2, and Fig. 4.1 en.o.amp4.5(f) includes 100 % un-
correlated noise sources only. The average input referred noise voltage density of
Amp5 equals the one of Chap. 2 (see Note in Sect. 2.2.3).
I must point out that some terms in (4.29) and (4.30) do not add significant
values to the total sum underneath the roots. Nevertheless, I keep them for universal
usage with other than the chosen components.
The resulting noise voltage density multiplied by the A-weighting function,
referenced to 1 Vrms nominal signal output voltage, and integrated over the band-
width of B20k, will lead to the A-weighted output referred SNa.o.amp4.5 in B20k,
expressed in dBV(A).
The detailed calculation approaches of the different terms in (4.29) and (4.30)
can be studied in TSOS-2 and TSOS-1.

4.3.5 A Look into the Content of MCD-WS 5.1

MCD-WS 5.1 shows some additional interesting results:

• Very important for external amplifiers Point 6.6.3 shows the calculation of the
Amp4 CMRR.
• Point 7. covers the math of an extremely low-noise input load of the Fig. 2.1
arrangement with Amp4. Here, like in Chap. 2, I’ve also chosen a pre-amp with
a gain of 200, an i/p referred noise voltage density of 0.2 nV/rtHz and noise
current density of only 2.4 pA/rtHz. The i/p load is 20 Ω. Now we can compare
62 4 The Solid-State (Op-Amp) Driven Central Amplifier Amp4

the A-weighted and RIAA equalized SN result (=−82.514 dBV(A)) with the one
of Point 8.
• Point 8. covers the math of the Point 7. low-noise Amp1, followed by a no-noise
arrangement à la Fig. 2.1. The SN result becomes −82.582 dBV(A).
• Hence, the difference is appr. 0.07 dB only. It is nothing else but the Noise
Figure. In other words: a further chase for extremely low-noise solutions makes
no sense for input loads ≥20 Ω. We will see later on in Chap. 10 what it will
mean for input loads <20 Ω.
• Points 9 and 10. show calculations of the Noise Figure (NF) of the amp chain
Amp4 + Amp5, fed by a lowest-noise input amp: 9. for MC and 10. for MM
cartridge purposes. These NFs are all <0.1 dB. Hence, together with its input
and output loads the noise impact of the here presented Amp4 is completely
ignorable.
• Point 11. gives up the shorted input and replaces the shortage by an output
resistance of a preceding gain stage, here 1 kΩ. Because of the 75 µs lp at the
input, the noise impact becomes marginal too.
• Point 12. tackles the 100 % correlated noise voltages; see also next section
below!

4.3.6 100 % Correlated Noise Voltages of Amp4

The differences between the measurement results and the calculated ones look very
good, always ≤0.5–1.0 dB rounded for the balanced outputs of Amps 4 & 5.
However, if we also measure the 2nd stage’s un-balanced output noise voltages
between the output leads and ground we’ll find surprising results: they look nearly
the same like the balanced output noise voltage results between the output leads,
thus, the SNs follow the same track. This can only be the case if we assume two
100 % correlated noise voltage densities en.o3.corr & en.o4.corr and two uncorrelated
ones en.o3 & en.o4 that form the output noise voltages from each output lead to
ground. Hence, the following simplified frequency independent equations and
Fig. 4.9 should describe this situation. The encircled “+” sign signals the phase
relationship of the correlated noise voltage parts.

e2n:vo3 ¼ e2n:o3 þ e2n:o3:corr

e2n:vo4 ¼ e2n:o4 þ e2n:o4:corr ð4:37Þ
en:vo3 ¼ en:vo4

Consequently, with (4.37) and (4.21) the remaining un-correlated output noise
voltage density en.o3 (=en.o4) becomes thus:
en:o:2nd
en:o3 ¼ pffiffiffi ð4:38Þ
2
4.3 Noise Calculations 63

Fig. 4.9 Detailed output

noise voltage situation of the
2nd gain stage

and
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:o:2nd ¼ e2n:o3 þ e2n:o4 ð4:39Þ

Hence, both rms voltages eN.vo3 = en.vo3 in B20k and eN.vo4 = en.vo4 in B20k
become always worse than the output rms voltage eN.o.2nd = en.o.2nd in B20k
(approximately 1.6 dB measured and calculated) because eN.o3.corr (=en.o3.corr in
B20k) is always bigger than eN.o3 (=en.o3 in B20k).
The way to get its rms voltage eN.o3.corr in B20k and the frequency dependent
noise voltage density en.o3.corr look as follows:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2ffi
u 
u e 2 2R1
u n:i:2nd ðf Þ þin:i:2nd R0 þ en:R0 1 þ RG
2 2 2

en:o3:corr ðf Þ ¼ u
t  2
R1 ð4:40Þ
þ i2n:i:3 R12 þ e2n:R1 þ e2n:RG RG
¼ en:o4:corr ðf Þ
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u Z
u 20kHz
u1
eN:o3:corr ¼ t ðjen:o3:corr ðf ÞjÞ2 df ð4:41Þ
B1
20Hz

The following chapter’s Mathcad Worksheet 5.1 shows the respective calcula-
tions and comparisons under Point 12.
64 4 The Solid-State (Op-Amp) Driven Central Amplifier Amp4

Note: This section’s equations are of academic interest only because of the fact
that most of the existing common mode voltages between the DIF amp’s output
leads and ground will be cancelled by the CMR of a following gain stage. They
represent a first and rough attempt to mathematically describe the measured findings
with Amp4. However, to bring them (or improved versions) in formats of general
validity further investigations must be carried out.
Mathcad Worksheets Amp4
5

Contents

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks

pffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffi
Note 1: MCD 11 has no built-in unit “rtHz” or “ Hz”. To get 1 Hz based
voltage noise and current noise densities the rms noise voltage and
current in a specific frequency range B > 1 Hz must be multiplied
pffiffiffiffiffiffiffiffiffi
by 1 Hz and divided by the root of that specific frequency range √B!
Note 2: MCD 11 offers no “dB” unit. This is available from MCD 13 on!

© Springer International Publishing Switzerland 2016 65

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_5
66 5 Mathcad Worksheets Amp4

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks Page 1

Calculations of a solid-state (op-amp) driven Amp4 with T1/T3 RIAA network at the
output, T2 RIAA network at the input, and based on data-sheet figures
:

Fig. 5.1 Op-amp driven Amp4 incl. RIAA networks

1. General data :
OP1 OPA627 OP2 OP1 OP3 OP1 OP4 OP1 OP5 OP27 OP6 OP5
23 1
k 1.38065 10 VAsK T 315K B1 1Hz B20k 19980 Hz

f 20Hz 25Hz 20000 Hz h 1000Hz

2. Gain stage component data :

3 3 3
R20 20 10 R21 10 10 R22 R21 R23 4.99 10 R24 R23
3
R25 750 R26 787 R31 274 R32 R31 R33 5.62 10
3 3 3 3
R34 6.81 10 R35 1.4 10 R36 100 10 R37 R36 R39c 100 10
3
R39d R39c R60 221 R61 2.2 10 R62 R61 R63 R61

R64 R63 R65 R63 R66 R63 R67 R63 R68 R63
3
R69 R63 R70 R63 R1amp5 100 10 R2amp5 R1amp5

R61 ... R70: 0.1%

6 9
C5 100 10 F C6 C5 C7 C5 C8 C5 C9 47 10 F
12 12
Ci.op1 8 10 F C15 13.5 10 F Ci.op2 Ci.op1 Ci.dif C15 0.5 Ci.op1

12 9
Ci.dif 17.5 10 F C12 227.143 10 F

P5 100k P6 P5 P10 20k P11 P10 P12 10k P13 P12

5 Mathcad Worksheets Amp4 67

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks Page 2

3. Gains and gain setting RG :

3.1 Gain setting :

R61 R62 R69 R70
RG 550 G 1 G 18
RG R63 R66

The idle gain G0 of the gain stage with OP3 ... OP6 equals that of the triode driven gain stage, hence,

G0.amp4 16.390411 R60 P9 RG

G1st = gain of OP1 & OP2 stage G2nd = gain of OP3 & OP4 stage G3rd = gain of OP5 & OP6 stage

R61 R69 R69

G1st 1 RG 4 RG 611.518 G3rd 2
G0.amp4 R63 2 R69 R63
Ro.op1 0.003
2 R61
Ro.op2 Ro.op1 G2nd 1 G2nd 8.195 G3rd 2
RG

3.2 Overall gain of Amp4 without RIAA network :

Gamp4 G1st G2nd G3rd Gamp4 16.390411

P9 RG R60 P9 390.518

4. RIAA time constants T and transfer gains GT :

4.1 T2 = 75 s : P1 70.975293

Fig. 5.2
T2 defining network

Succ-apps of P1 should bring in line the value of T2 with the RIAA network component values (because of the
marginal influence the voltage divider effects of R o.op 1 /Ro.op 2 and R23/R26 are ignored here!):

1
1
1
1 1 1
T2 C9 Ci.dif Ro.op1 R23 R25 P1 R39c
1
1
1
1 1 1
Ro.op2 R24 R26 R39d
68 5 Mathcad Worksheets Amp4

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks Page 3

6
T2 75 10 s

1
1
2j f C9 Ci.dif ( R39c R39d)
GT2( f)
1 1
1 1 1
2j f C9 C i.dif ( R39c R39d) R25 R26 P1 2 Ro.op1 R23

1
GT2 0.701
2 T2

1
20 log GT2 3.08 [dB]
GT2.e( f) 20 log GT2( f) 2 T2

0
2
4
6
8
[dB]

GT2.e( f) 10
12 Fig. 5.3
14
16 Bode plot of GT2(f)
18
20
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

3 9
4.2 T1 = 3180 s, T3 = 318 s : P4 1.438476 10 C12 227.142857143 10 F

Fig. 5.4
T1 & T3 defining network

Succ-apps of C12 & P4 should bring in line the values of T1 & T3 with the required RIAA network values:
6
T3 C12 R35 T3 318 10 s
3
RT1.eff R31 R32 P4 R33 R34 RT1.eff 14.416 10

1
1
1 1
1 1 1 1 1
T1 C12 R35 RT1.eff R1amp5 R36 R2amp5 R37

3
T1 3.18 10 s
5 Mathcad Worksheets Amp4 69

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks Page 4

1
1 1 3
RL.dif RL.dif 100 10
R1amp5 R2amp5 R36 R37

1
1
1 1
R35
2j f C12 RL.dif
GT1.3( f) GT1.3( h) 0.098
1
1
1 1
R35 RT1.eff
2j f C12 RL.dif
20 log GT1.3( 20Hz) 1.806

1 1
20 log GT1.3 4.137 20 log GT1.3 18.203
2 T1 2 T3

GT1.3.e ( f) 20 log GT1.3( f) 20 log GT1.3( 20kHz) 21.167

0
2
4
6
8
[dB]

GT1.3.e( f) 10
12 Fig. 5.5
14
16 Bode plot of GT1(f) & GT3(f)
18
20
22
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

4.3 RIAA transfer :

1
Griaa( f) GT2( f) GT1.3( f) Griaa.e( f) 20 log Griaa( f) Griaa( h)

20
16
12
8
4
[dB]

Griaa.e ( f) 0
4
8 Fig. 5.6
12 Normalized (1kHz at 0dB)
16 RIAA transfer function
20
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]
70 5 Mathcad Worksheets Amp4

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks Page 5

1
[times]

Griaa ( f) 0.1
Fig. 5.7
Bode plot of the combined
RIAA network
0.01
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

2
6
1 2 h 318 10 s
R1000
2 2
6 6
1 2 h 3180 10 s 1 2 h 75 10 s

2
6
1 2 f 318 10 s 1
R0 ( f) R1000
2 2 Dev1e ( f) 20 log R0 ( f) Griaa.e( f)
6 6
1 2 f 3180 10 s 1 2 f 75 10 s

0.1
0.075
0.05
0.025
[dB]

Dev1e ( f) 0
0.025 Fig. 5.8
0.05 Deviation from the exact
0.075 RIAA transfer
0.1
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

5. Transfer function T(f) of Amp4 :

General assumption: the corner frequency of the i/p hp R21, R22, C5 ... C8 is chosen that there is no influence on a flat
phase and frequency response in B20k !

Tamp4( f) Gamp4 GT2( f) GT1.3( f) Tamp4( h) 1.435729

Tamp4.e( f) 20 log Tamp4( f) Tamp4.e( h) 3.141447 [dB]

30
20
10
[dB]

Tamp4.e( f)
0
10 Fig. 5.9
20 Bode plot of Amp4
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]
5 Mathcad Worksheets Amp4 71

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks Page 6

1
Dev2e ( f) 20 log R0 ( f) 20 log Tamp4( f) Tamp4( h)

0.1
0.05
[dB]

Dev2e ( f) 0
0.05

0.1 Fig. 5.10 = Fig. 4.3

3 4 5 Fig. 5.9's deviation from the
10 100 1 10 1 10 1 10
f exact RIAA transfer
[Hz] .

6
1 2j f 318 10 s
RIAA( f) 3
6 6 RIAA( h) 101.03 10
1 2j f 3180 10 s 1 2j f 75 10 s

1
Im Tamp4( f) RIAA( f)
amp4( f) atan
Re Tamp4( f) RIAA( f)
1 amp4( h) 0

1
amp4( f)
[°]

0
deg
Fig. 5.11
1
Fig. 5.9's phase response
2
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

6. Noise and SN calculations :

3
6.1 General data : vi.ref 100 10 V
3
vo.ref vi.ref Tamp4( h) vo.ref 143.573 10 V vo.ref.m vo.ref
1
1 1 3
RP3rd.i RP3rd.i 1.1 10
R64 R67
1
1 1 3
RP3rd.o RP3rd.o 1.1 10
R63 R69

RP3rd.i RP3rd.o RP3rd RP3rd RP3rd.i

9 9
en.i.1 4.2 10 V en.i.2 en.i.1 en.i.3 en.i.1 en.i.4 en.i.1 en.i.5 3.2 10 V en.i.6 en.i.5
15 12
in.i.1 2.5 10 A in.i.2 in.i.1 in.i.3 in.i.1 in.i.4 in.i.1 in.i.5 0.4 10 A in.i.6 in.i.5

fc.e.1 40Hz fc.i.1 0.1Hz fc.e.5 2.7Hz fc.i.5 140Hz

72 5 Mathcad Worksheets Amp4

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks Page 7

fc.e.1
en.i.1( f) en.i.1 1 en.i.2( f) en.i.1( f) en.i.3( f) en.i.1( f) en.i.4( f) en.i.1( f)
f

fc.e.5 fc.i.5
en.i.5( f) en.i.5 1 en.i.6( f) en.i.5( f) in.i.5( f) in.i.5 1 in.i.6( f) in.i.5( f)
f f

en.i.1st( f) en.i.1( f) en.i.2nd( f) en.i.3( f) en.i.3rd( f) en.i.5( f)

in.i.1st in.i.1 in.i.2nd in.i.3 in.i.3rd( f) in.i.5( f)

6.2 Situation of the various noise sources :

Fig. 5.12
Noise voltage and current
situation of the 2nd gain stage
(OP3 & OP4)

Fig. 5.13
Noise voltage and current
situation of the 3rd gain stage
(OP5 & OP6)

1
1 1 9
en.R25 4 k T B1 ( R25 P1) R39c en.R25 3.764 10 V

1
1 1 9
en.R26 4 k T B1 R26 R39d en.R26 3.686 10 V

9
en.R33 4 k T B1 ( R31 R33 P4) en.R33 11.294 10 V
5 Mathcad Worksheets Amp4 73

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks Page 8

9
en.R34 4 k T B1 ( R34 R32) en.R34 11.101 10 V

9
en.R61 4 k T B1 R61 en.R61 6.186 10 V

en.R62 en.R61

9
en.R63 4 k T B1 R63 en.R63 6.186 10 V

en.R69 en.R63
9
en.RP.3rd.i 4 k T B1 RP3rd.i en.RP.3rd.i 4.374 10 V

9
en.RP.3rd.o 4 k T B1 RP3rd.o en.RP.3rd.o 4.374 10 V

en.RP.3rd.i en.RP.3rd.o en.RP.3rd en.RP.3rd en.RP.3rd.i

6.3 Output noise voltage of the 1st gain stage (OP1 & OP2) with i/p shorted :

2 2 9
en.o.1st( f) en.i.1st( f) G1st en.o.1st( h) 6.057 10 V

6.4 Output noise voltage of the 2nd gain stage (OP3 & OP4 & T2) with i/p of the
1st gain stage (OP1 & OP2) shorted :
6.4.1 Noise Voltage of the 2nd gain stage :

9
en.RG 4 k T B1 RG en.RG 3.262 10 V

en.RG.o en.RG G2nd 1 9

en.RG.o 23.468 10 V

6.4.2 Noise Voltage of the T2 network :

1
1 1 3
ZT2 ( R25 R26 P1) ( R39c R39d) ZT2 1.595 10

9
en.Z.T2( f) GT2( f) 4 k T B1 ZT2 en.Z.T2( h) 4.727 10 V

8
1 10
[V/rtHZ]

Fig. 5.14
en.Z.T2( f) 1 10 9 Frequency response of
the noise voltage density
of the T2(f) network
10
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]
74 5 Mathcad Worksheets Amp4

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks Page 9

ZT2( f) ZT2 GT2( f)

6.4.3 Output noise voltage of 1st+2nd+T2 stages :

2
2 2 2 in.i.2nd 2 2
en.o.2nd( f) en.o.1st( f) GT2( f) en.Z.T2( f) ZT2( f) G2nd
2
2
2 en.i.2nd( f)
2 2 2 2 2
in.i.2nd ( R61 R62) en.R61 en.R62 en.RG.o

9
en.o.2nd( h) 77.147 10 V

6
1 10
[V/rtHz]

Fig. 5.15
en.o.2nd( f) 1 10 7 Bode plot of the
balanced o/p noise
voltage density of the
2nd gain stage
8
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

6.5 Output noise voltage of the 3rd gain stage with i/p shorted :
6.5.1 Noise Voltages of the OP5 and OP6 stages :

2 2
R69 en.i.3rd( f) 2 2 2 9
en.o.5( f) 2 1 en.RP.3rd in.i.3rd( f) RP3rd en.o.5( h) 13.997 10 V
R63 2

en.o.6( f) en.o.5( f)

6.5.2 Noise Voltage and SN of the whole 3rd gain stage :

2 2 9
en.o.3rd( f) en.o.5( f) en.o.6( f) en.o.3rd( h) 19.795 10 V

20000Hz
1 2
en.o.3rd( f) df
B1
20Hz
SNne.o.3rd 20 log SNne.o.3rd 111.068 [dBV]
1V

measured: -110.51 dBV

5 Mathcad Worksheets Amp4 75

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks Page 10

6.6 Output noise voltage of Amp4 + Amp5 with i/p shorted :

6.6.1 Noise Voltage of the T1/T3 network at the o/p of Amp4 :
1
1 1 1
ZT1.3( f) 3
P4 R31 R32 R33 R34 1 RL.dif ZT1.3( h) 1.407 10
R35
2j f C12

12
en.Z.T1.3( f) GT1.3( f) 4 k T B1 ZT1.3( f) en.Z.T1.3( h) 482.968 10 V

7
1 10

8
Fig. 5.16
1 10
[V/rtHz]

Frequency response of the

en.Z.T1.3( f) noise voltage density of the
9 T1(f)+T3(f) network
1 10

10
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

6.6.2 Relevant Amp5 data :

2.5 15 9
in.i.amp5 10 A average noise voltage in B20k : en.i.amp5 8.103 10 V
2

1V
Gamp5 Gamp5 6.965
vo.ref

0.1 3
Gcm.amp5 12 Gcm.amp5 1 10
200
Gamp5 3
CMRRamp5 CMRRamp5 6.965 10
Gcm.amp5

CMRRamp5.e 20 log CMRRamp5 CMRRamp5.e 76.859 [dB]

6.6.3 CMRR of Amp4 :

0.1 3
Gcm.amp4 12 Gcm.amp4 1 10
200
Gamp4 3
CMRRamp4 CMRRamp4 16.39 10
Gcm.amp4

CMRRamp4.e 20 log CMRRamp4 CMRRamp4.e 84.292

76 5 Mathcad Worksheets Amp4

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks Page 11

6.6.3 Total output noise voltage at the o/p of Amp5 (amp4.5) :

2 2 2
en.o.amp4.5( f) Gamp5 en.o.2nd( f) G3rd en.o.3rd( f) GT1.3( f)
2 2 2 2
en.Z.T1.3( f) in.i.amp5 ZT1.3( f) en.i.amp5

9
en.o.amp4.5( h) 111.555 10 V

5
1 10

6
1 10
[V/rtHz]

en.o.amp4.5( f)
7
1 10

8
1 10
3 4 5
10 100 1 10 1 10 1 10
f Fig. 5.17a &
[Hz] Fig. 5.17b = Fig. 4.2
Amp4+Amp5 output
noise voltage density
0
20 with i/p shorted
40
60
en.o.amp4.5( f) 80
[dBV

20 log 100
1V 120
140
160
180
200
3 4 5
10 100 1 10 1 10 1 10
f .
[Hz]

20000Hz
1 2
eN.o.amp4.5 en.o.amp4.5( f) df 6
B1 eN.o.amp4.5 15.617 10 V
20Hz

Tamp4( h) Gamp5 10

en.o.amp4.5( f) 9
en.i.amp4.5( f) en.i.amp4.5( h) 11.156 10 V
Tamp4( h) Gamp5

6.7 SNs of the sequence of Amp4 + Amp5 :

20000Hz
1 2
en.o.amp4.5( f) df
B1
20Hz
SNo.amp4.5 20 log SNo.amp4.5 96.128 [dBV]
1V

measured: -98.06 dBV

5 Mathcad Worksheets Amp4 77

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks Page 12

B(f) = A(f) = A-weighting transfer function

1.259
B( f)
2 2 2 2
20.6Hz 107.7Hz 737.9Hz f
1 1 1 1
f f f 12200 Hz

20000Hz
1 2 2
en.o.amp4.5( f) B( f) df
B1
20Hz
SNa.o.amp4.5 20 log SN a.o.amp4.5 99.962 [dBV(A)]
1V

measured: -100.82 dBV(A)

20000Hz
1 2 2
en.i.amp4.5( f) B( f) df
B1
20Hz
SN a.i.amp4.5 20 log SN a.i.amp4.5 99.962 [dB(A)]
vi.ref

7. Amp4 + Amp5 with i/p loaded by the noise voltage of a preceding MC amp1 with extremely
low input noise voltage & current density :

12
9 2.4 10 A
Gamp1 200 R0 20 en.i.0 0.2 10 V 2 in.i.0
2
3 3
T 315K vo.ref.mc 100 10 V vi.ref.mc 0.5 10 V

12
en.R0 4 k T B1 R0 en.R0 589.851 10 V

2 2 2 2 12
en.i.0.eff en.i.0 en.R0 in.i.0 R0 en.i.0.eff 655.039 10 V

9
en.o.amp1 Gamp1 en.i.0.eff en.o.amp1 131.008 10 V

B20k 6
eN.o.amp1 en.o.amp1 eN.o.amp1 18.518 10 V
B1

2 2 2 2 6
en.o.mc( f) en.o.amp1 Tamp4( f) Gamp5 en.o.amp4.5( f) en.o.mc( h) 1.312 10 V

en.o.mc( f)
en.i.mc( f) 12
Gamp1 Tamp4( h) Gamp5 en.i.mc( h) 656.245 10 V
78 5 Mathcad Worksheets Amp4

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks Page 13

20000Hz
1 2
en.o.mc( f) df
B1
20Hz
SN riaa.o.amp4.5.mc 20 log SNriaa.o.amp4.5.mc 78.212 [dB]
1V

20000Hz
1 2 2
en.o.mc( f) B( f) df
B1
20Hz
SN ariaa.o.amp4.5.mc 20 log SNariaa.o.amp4.5.mc 82.514 [dB(A)]
1V

20000Hz
1 2 2
en.i.mc( f) B( f) df
B1
20Hz
SN ariaa.i.amp4.5.mc 20 log SNariaa.i.amp4.5.mc 82.514 [dB(A)]
vi.ref.mc

8. Preceding MC Amp1 alone and noise calculation up to the o/p of Amp5 :

6
en.o.mc.eff( f) en.o.amp1 Tamp4( f) Gamp5 en.o.mc.eff( h) 1.31 10 V

en.o.mc.eff( f) 12
en.i.mc.eff( f) en.i.mc.eff( h) 655.039 10 V
Tamp4( h) Gamp1 Gamp5

20000Hz
1 2
en.o.mc.eff( f) df
B1
20Hz
SN riaa.o.mc 20 log SNriaa.o.mc 78.256 [dB]
1V

20000Hz
1 2
en.i.mc.eff( f) df
B1
20Hz
SN riaa.i.mc 20 log SNriaa.i.mc 78.256 [dB]
vi.ref.mc

20000Hz
1 2 2
en.o.mc.eff( f) B( f) df
B1
20Hz
SN ariaa.o.mc 20 log SNariaa.o.mc 82.582 [dBA]
1V

20000Hz
1 2 2
en.i.mc.eff( f) B( f) df
B1
20Hz
SN ariaa.i.mc 20 log SN ariaa.i.mc 82.582 [dB(A)]
vi.ref.mc
5 Mathcad Worksheets Amp4 79

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks Page 14

95
100
105
110
20 log e n.o.amp4.5( f)
115
[V/rtHz]

20 log e n.o.mc( f) 120

20 log e n.o.mc.eff( f) 125
130
135
140
145
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

Fig. 5.18 Comparison of the various output noise voltage densities vs. frequency
(trace 3 is nearly hidden by trace 2)

1
Dev3e ( f) 20 log R0 ( f) 20 log en.o.mc.eff( f) en.o.mc.eff( h)

0.1
0.075
0.05
0.025
[dB]

Dev3e ( f) 0 Fig. 5.19

0.025 Deviation from the
0.05 exact RIAA transfer of
0.075 Amp1+Amp4+Amp5
0.1
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

9. MC Amp1 noise worsening by additional noise from Amp4+Amp5:

NFe.mc12 SNariaa.o.amp4.5.mc SNariaa.o.mc NFe.mc12 0.068 [dB]

10. MM Amp noise worsening by additional noise from Amp4+Amp5:

SN ariaa.o.mm 80.5 [dB(A) ref. 100mVrms ]

SNariaa.o.mm SNa.i.amp4.5
10 10
SN ariaa.i.amp4.mm 20 log 10 10 SNariaa.i.amp4.mm 80.451 [dB(A)]

NFe.mm SNariaa.i.amp4.mm SNariaa.o.mm NFe.mm 0.049 [dB]

80 5 Mathcad Worksheets Amp4

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks Page 15

11. Amp4 with i/p loaded by the noise voltage of a 1k resistor and via o/p of Amp5:

R0 1000 T 315K
9
en.R0 4 k T B1 R0 en.R0 4.171 10 V

2 2 2 2
en.o.1k ( f) en.R0 Tamp4( f) Gamp5 en.o.amp4.5( f) 9
en.o.1k ( h) 115.796 10 V

en.o.1k ( f)
en.i.1k( f) 9
Gamp5 Tamp4( h) en.i.1k( h) 11.58 10 V

20000Hz
1 2
en.o.1k ( f) df
B1
20Hz
SN riaa.o.amp4.5.1k 20 log SNriaa.o.amp4.5.1k 96.042 [dBV]
1V

20000Hz
1 2 2
en.o.1k ( f) B( f) df
B1
20Hz
SN ariaa.o.amp4.5.1k 20 log SN ariaa.o.amp4.5.1k 99.798 [dBV(A)]
1V

20000Hz
1 2 2
en.i.1k( f) B( f) df
B1
20Hz
SN ariaa.i.amp4.5.1k 20 log SN ariaa.i.amp4.5.1k 99.798 [dB(A)]
100mV

12. Amp4's 100% correlated output noise voltage density at the output leads to ground :

2
2 R61 2 2 2 2
en.o3.corr( f) en.i.2nd( f) 1 2 in.i.3 R61 en.R61 en.RG.o
RG

9
en.o4.corr( f) en.o3.corr( f) en.o3.corr( h) 55.256 10 V

en.o.2nd( f) 9
en.o3 ( f) en.o3 ( h) 54.551 10 V
2

2 2 9
en.vo3( f) en.o3 ( f) en.o3.corr( f) en.vo3( h) 76.579 10 V
5 Mathcad Worksheets Amp4 81

5.1 MCD-WS: The Solid-State Gain Stage Amp4 with RIAA Networks Page 16

20000Hz
1 2 6
eN.o3.corr en.o3.corr( f) df eN.o3.corr 7.731 10 V
B1
20Hz

20000Hz
1 2 6
eN.o3 en.o3 ( f) df eN.o3 5.637 10 V
B1
20Hz

20000Hz
1 2 6
eN.vo3 en.vo3( f) df eN.vo3 9.539 10 V
B1
20Hz

20000Hz
1 2 6
eN.o.2nd en.o.2nd( f) df eN.o.2nd 7.972 10 V
B1
20Hz

20000Hz
1 2
en.vo3( f) df
B1
20Hz
SNo3.corr 20 log SNo3.corr 94.39 [dB]
0.5V

20000Hz
1 2
en.o.2nd( f) df
B1
20Hz
SNo.2nd 20 log SNo.2nd 95.948 [dB]
0.5V

SNo4.corr SNo3.corr SNo.2nd SN o3.corr 1.558 [dB]

6
1 10

en.vo3( f)
[V/rtHz]

en.o3.corr( f) 1 10 7

en.o.2nd( f)
Fig. 5.20
Noise voltage densities
at the output of the 2nd
gain stage
8
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]
The Op-Amp and Transformer Driven
Output Stage Amp5 6

6.1 General Design and Gain of Amp5

With the T1 & T3 RIAA network included, the Amp5 gain stage must lift the
output level of Amps 3 or 4 to a nominal output level of 0 dBu or 0 dBV or +6 dBu
at 1 kHz, depending on ones own needs. Hence, we set on a nominal 0 dBV/1 kHz
based gain of 6.25 here. In addition, this gain stage should be capable to drive a 1:1
output transformer. Generally, the overload margin should not be lost out of sight.
However, it becomes less dramatic: with the +20 dB goal at the input we would get
10Vrms at the output, which is much smaller than the maximal possible 17.2Vrms.
Figures 6.1 and 6.2 show the solution. It follows the basic design of Amp4 with
only one further gain stage to get an un-balanced output via OP5. This stage
produces the same output level, but less good overload margin. P1 sets the gain of
Amp5. Of course, to get higher gains R3 could be chosen smaller. The gain Gamp5
of such an amplifying stage becomes:

Gamp5 ¼ G1st G2nd



R4 þ R5 R10 R13
¼ 1þ þ ð6:1Þ
RG R6 R9
RG ¼ P1 þ R3

With equal valued resistors R6–R13 the gain G2nd of the output stage (OPs 3 & 4)
becomes 2. Thus, for a given Gamp5 we obtain RG as follows:

R4 þ R5
RG ¼ ð6:2Þ
0.5Gamp5  1

© Springer International Publishing Switzerland 2016 83

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_6
84 6 The Op-Amp and Transformer Driven Output Stage Amp5

Fig. 6.1 Circuit of the engine’s output gain stage Amp5, also showing additional offset trim
variants

Hence, with Gamp5 = 6.25 P1 becomes:

P1 ¼ RG  R3
ð6:3Þ
¼ 1161 X

The transformer circuit follows the recommendations of Jensen Transformers

Inc. Fortunately, switching the transformer in and out of the signal path does not
lead to different signal levels. It also does not lower (or lift up) the overall frequency
and phase response at the edges of B20k.
On the Mainboard there are three female strip connectors for Amps 1 and 2 and 5.
Figure 6.2 shows the connections to the Mainboard and to the plug-in PCBs.

6.2 Power Supply

Each of the various solid-state gain stages has its own power supply, on plug-in
PCBs (Amps 1, 2, and 5) as well as on the Mainboard (Amp4). Figure 6.3 shows
the circuit. Because of tolerances of the IC reference voltages R18 and R20 may
need additional trimming to get a rather exact ±15 V output.
6.2 Power Supply 85

Fig. 6.2 Plug-in connections

between PCBs of Amps1, 2,
5 and the mainboard of
Fig. 1.17

Fig. 6.3 Solid-state gain

stage ±15 V PSU
86 6 The Op-Amp and Transformer Driven Output Stage Amp5

6.3 CMRR and Noise

6.3.1 CMRR

The CMRR reflections follow the remarks I’ve already made about Amp4, hence,
with its nominal gain Gamp5 = 6.25 CMRRe of Amp5 becomes appr. 76 dB as
follows:

Gamp5
CMRRamp5 ¼ ð6:4Þ
Gamp5:cm

tolerance½%
Gamp5:cm ¼ 2
200 %
2  0:1 ð6:5Þ
¼
200
¼ 0:001

6:25
CMRRamp5 ¼
0:001 ð6:6Þ
¼ 6250
 
CMRRamp5:e ¼ 20 log CMRRamp5
ð6:7Þ
¼ 75:92 dB

Concerning the Amp5 generated common mode noise voltage CMamp5 at the
output of its solid-state section we have the same situation already described in
Sect. 4.3.6. That’s why it is important to have a well designed following gain stage
with a CMRR that is capable to damp CMamp5 to an ignorable amount. The output
transformer switched into the signal path produces a big additional CMRRe por-
tion >60 dB. In this context, a transformer is the most effective component.
The next chapter’s Mathcad Worksheet gives the details on Amp5’s CM noise
voltage production.

6.3.2 Noise and SNs

The Mathcad Worksheet 7.1’s noise and SN calculations follow Figs. 6.4 and 6.5
and the calculation course already described for Amp4. However, I’ll repeat the
main issues. Additionally, Fig. 6.6 shows the noise situation of the un-balanced
output via OP5.
With input shorted and R4 = R5, the frequency dependent output noise voltage
density en.o.1st(f) of the 1st gain stage with OPs 1 & 2 becomes:
6.3 CMRR and Noise 87

Fig. 6.4 Noise sources of Amp5’s input section (OPs 1 & 2)

Fig. 6.5 Noise sources of Amp5’s balanced output section (OPs 3 & 4)

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:o:1st ðf Þ ¼ 2 en:i:1st ðf Þ2 G21st þ i2n:i:1st ð2R4Þ2 þ 2e2n:R4 þ e2n:RG:o ð6:8Þ

2R4
en:RG:o ¼ en:RG ð6:9Þ
RG
88 6 The Op-Amp and Transformer Driven Output Stage Amp5

Fig. 6.6 Noise sources of Amp5’s un-balanced output stage (OP5)

With input shorted and R6–R13 with equal values, we obtain the frequency
dependent output noise voltage density en.o.3(f) at the output of OP3:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
!ffi
u

u R11 e ð f Þ 2
en:o:3 ðf Þ ¼ t2 1 þ
n:i:2nd
þ e2n:RP:op3 þ i2n:i:2nd RP2op3 ð6:10Þ
R7 2

The same applies for the output noise of OP4, en.o.4(f). Thus, the output noise
voltage density of the 2nd gain stage (input shorted) becomes:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:o:2nd ðf Þ ¼ en:o:3 ðf Þ2 þ en:o:4 ðf Þ2 ð6:11Þ

Hence, Amp5’s balanced output noise voltage density en.o.amp5.bal(f) looks as

follows:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:o:amp5:bal ðf Þ ¼ ðen:o:1st G2nd Þ2 þ en:o:2nd ð f Þ2 ð6:12Þ

The noise situation of the un-balanced output looks a bit different.1 With input
shorted, RPop5 = R14||R15, and Gop5 = 1 + R15/R14 we obtain en.o.5(f):
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:o:5 ðf Þ ¼ Gop5 en:i:5 ðf Þ2 þ i2n:i:5 RP2op5 þ e2n:RP:op5 ð6:13Þ

1
Details see TSOS-1, Sect. 3.5 or TSOS-2, Chap. 8.
6.3 CMRR and Noise 89

We can now formulate the frequency dependent output noise voltage density
en.o.amp5.unbal(f) at Amp5’s un-balanced output, with input of Amp5 shorted:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  ffi
en:o:amp5:unbal ðf Þ ¼ G2op5 en:o:1st ðf Þ2 þen:o:4 ðf Þ2 þ en:o:5 ðf Þ2 ð6:14Þ

The noise level of Amp5’s output stage is very low. The output transformer does
not significantly add further noise to the active parts. With input shorted at the
balanced output the A-weighted SNa.o.amp5.bal becomes a measured (calcu-
lated) −103.50 dBV(A) (−104.01 dBV(A)) and the un-balanced output offers an
SNa.o.amp5.unbal of −102.90 dBV(A) (−103.29 dBV(A)). With these results we can
completely ignore any Amp5 impact on the noise that comes in from the preceding
gain stages.

6.4 Additional Remarks

Concerning THD and IMD Amp5, incl. or excl. output trafo, did not produce
additional artefacts that could increase the CLIO generated levels.
A simpler Amp5 configuration would work too, eg OPs 1 & 2 in the typical
balanced 2-op-amp configuration, however, without output trafo CMRR would be
too low, the one we need here to kill the CM voltage generated by the Amp3 current
sink.
Mathcad Worksheets Amp5
7

Contents
7:1 MCD-WS: The Op-Amp + Transformer Driven Output Stage Amp5

pffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffi
Note 1: MCD 11 has no built-in unit “rtHz” or “ Hz”. To get 1 Hz based
voltage noise and current noise densities the rms noise voltage and
current in a specific frequency range B > 1 Hz must be multiplied
pffiffiffiffiffiffiffiffiffiffi
by 1 Hz and divided by the root of that specific frequency range √B!
Note 2: MCD 11 offers no “dB” unit. This is available from MCD 13 on!

© Springer International Publishing Switzerland 2016 91

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_7
92 7 Mathcad Worksheets Amp5

7.1 MCD-WS: The Op-Amp + Transformer Driven Output Stage Amp5 Page 1

Calculations of the op-amp driven Amp5:

Fig. 7.1 Op-amp driven Amp5 incl. balanced & un-balanced output

1. General data :

OP1 OPA627 OP2 OP1 OP3 OPA627 OP4 OP3 OP5 OPA627

23 1
k 1.38065 10 VAsK T 300K B1 1Hz B20k 19980 Hz

f 20Hz 25Hz 20000 Hz h 1000Hz

2. Gain stage component data :

3 3 3
R1 100 10 R2 100 10 R3 910 R4 2.2 10 R5 R4
3
R6 2.2 10 R7 R6 R8 R6 R9 R8 R10 R8 R11 R8

R4 ... R15: 0.1% R12 R8 R13 R8 R14 R8 R15 R8

7 Mathcad Worksheets Amp5 93

7.1 MCD-WS: The Op-Amp + Transformer Driven Output Stage Amp5 Page 2

3. Gains and gain setting RG :

3.1 Gain setting

3
P1 25000 RG R3 P1 RG 25.91 10

R4 R5 R10 R12 3
G 1 vi.ref 143.572 10 V G 2.34
RG R6 R8

Depending on the nominal output voltage the gain Gamp 5 of the gain stage with OP1 ... OP5 can be trimmed by P1,
hence, with 1Vrms nominal output voltage and an input voltage of 143.572mV rms we'll obtain:

1.0V
Gamp5 G1st G2nd Gamp5 Gamp5 6.965
vi.ref

R10 Gamp5
G2nd 2 G2nd 2 G1st G1st 3.483
R6 G2nd

R4 R5 3
RG RG 1.772 10 P1 RG R3 P1 862.355
G1st 1

3.2 Actua overall gain of Amp5, bal & un-bal :

Gamp5 G1st G2nd Gamp5 6.965

4. Noise and SN calculations :

4.1 General data :

vo.ref vi.ref Gamp5 vo.ref 1V

1
1 1 3
RPop3 RPop3 1.1 10
R6 R10

1
1 1 3
RPop4 RPop4 1.1 10
R9 R13

1
1 1 3
RPop5 RPop5 1.1 10
R14 R15

9 9
en.i.1 4.2 10 V en.i.2 en.i.1 en.i.3 4.2 10 V en.i.4 en.i.3 en.i.5 en.i.3

fc 40Hz en.i.1 and en.i.3 at 1MHz and according to the data sheet graphs

fc fc
en.i.1( f) en.i.1 1 en.i.2( f) en.i.1( f) en.i.3( f) en.i.3 1 en.i.4( f) en.i.3( f) en.i.5( f) en.i.3( f)
f f
94 7 Mathcad Worksheets Amp5

7.1 MCD-WS: The Op-Amp + Transformer Driven Output Stage Amp5 Page 3

15 12
in.i.1 2.5 10 A in.i.2 in.i.1 in.i.3 0.05 10 A in.i.4 in.i.3 in.i.5 in.i.3

en.i.1st( f) 2 en.i.1( f) en.i.2nd( f) en.i.3( f)

in.i.1
in.i.1st in.i.2nd in.i.3
2

4.2 Situation of the various noise sources :

Fig. 7.2 = Fig. 6.4

Noise voltage and current
situation of the 1st gain stage
(OP1 & OP2)

Fig. 7.3 = Fig. 6.5

Noise voltage and current
situation of the 2nd gain stage
(OP3 stage & the equivalent
OP4 stage)

9
en.R4 4 k T B1 R4 en.R4 6.037 10 V

en.R5 en.R4

en.R7 en.R4

en.R8 en.R4

en.R11 en.R4

en.R12 en.R4

9
en.RP.op3 4 k T B1 RPop3 en.RP.op3 4.269 10 V
7 Mathcad Worksheets Amp5 95

7.1 MCD-WS: The Op-Amp + Transformer Driven Output Stage Amp5 Page 4

9
en.RP.op4 4 k T B1 RPop4 en.RP.op4 4.269 10 V

9
en.RP.op5 4 k T B1 RPop5 en.RP.op5 4.269 10 V

4.3 Output noise voltage of Amp5 with i/p shorted :

At the bal o/p of the 1st gain stage (OPs 1 & 2 with i/p shorted) we'll get the balanced noise voltage density:

9
en.RG 4 k T B1 RG en.RG 5.419 10 V

R4 R5 9
en.RG.o en.RG en.RG.o 13.453 10 V
RG

2 2 2 2 2 2 2
en.o.1st( f) en.i.1st( f) G1st in.i.1 ( R4 R5) en.R4 en.R5 en.RG.o

9
en.o.1st( h) 26.436 10 V

6
1 10
[V/rtHz]

Fig. 7.4
en.o.1st( f)1 10 7 Bode plot of the
balanced o/p noise
voltage density of the
1st gain stage
8
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

At the bal o/p of the 2nd stage (OPs 3 & 4 with inputs of the 2nd gain stage shorted!) we'll get the un-correlated output noise vo
density as follows:

2 2
R11 en.i.2nd( f) 2 2 2 9
en.o.3( f) 2 1 en.RP.op3 in.i.2nd RPop3 en.o.3( h) 14.806 10 V
R7 2

en.o.4( f) en.o.3( f)

2 2 9
en.o.2nd( f) en.o.3( f) en.o.4( f) en.o.2nd( h) 20.938 10 V

20000Hz
1 2
en.o.2nd( f) df
B1
20Hz
SNne.o.2nd 20 log SN ne.o.2nd 110.612 [dBV]
1V
measured: -110.1 [dBV]
96 7 Mathcad Worksheets Amp5

7.1 MCD-WS: The Op-Amp + Transformer Driven Output Stage Amp5 Page 5

At the un-bal o/p of the 2nd stage (OP5 with inputs of the 2nd gain stage shorted!) we'll get the un-correlated output noise
voltage density as follows:

Fig. 7.5 = Fig. 6.6

Noise voltage and current
situation of the un-balanced
output gain stage (OP5)

R15
Gop5 1 Gop5 2
R14

2 2 2 2 9
en.o.5( f) Gop5 en.i.5( f) in.i.5 RPop5 en.RP.op5 en.o.5( h) 12.095 10 V

Thus, the total o/p noise voltage density at the bal o/p becomes:

2 2 9
en.o.amp5.bal( f) en.o.1st( f) G2nd en.o.2nd( f) en.o.amp5.bal( h) 56.867 10 V

en.o.amp5.bal( f) 9
en.i.amp5.bal( f) en.i.amp5.bal( h) 8.165 10 V
Gamp5

20000Hz
1 2
en.o.amp5.bal( f) df
B20k
20Hz 9
8.103 10 V
Gamp5

70
80
90
100
en.o.amp5.bal( f) 110
[dB]

20 log 120
1V 130
140
150
160
170
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

Fig. 7.6 Amp5 output noise voltage density with i/p shorted
7 Mathcad Worksheets Amp5 97

7.1 MCD-WS: The Op-Amp + Transformer Driven Output Stage Amp5 Page 6

Hence, the total o/p noise voltage density at the un-bal o/p becomes:

2 2 2 2 9
en.o.op5 ( f) Gop5 en.o.1st( f) en.o.4( f) en.o.5( f) en.o.op5 ( h) 61.795 10 V

en.o.amp5.unbal( f) en.o.op5 ( f)

4.4 SNs :

20000Hz
1 2
en.o.amp5.bal( f) df
B1
20Hz
SNne.o.amp5.bal 20 log SNne.o.amp5.bal 101.963 [dBV]
vo.ref
measured: -101.80 [dBV]

20000Hz
1 2
en.i.amp5.bal( f) df
B1
20Hz
SNne.i.amp5.bal 20 log SNne.i.amp5.bal 101.963 [dB]
vi.ref

20000Hz
1 2
en.o.amp5.unbal( f) df
B1
20Hz
SNne.o.amp5.unbal 20 log SNne.o.amp5.unbal 101.237 [dBV]
vo.ref

measured: -101.40 [dBV]

B( f) A( f) A weightingtransferfunction

1.259
B( f)
2 2 2 2
20.6Hz 107.7Hz 737.9Hz f
1 1 1 1
f f f 12200 Hz

20000Hz
1 2 2
en.o.amp5.bal( f) B( f) df
B1
20Hz
SNa.o.amp5.bal 20 log SNa.o.amp5.bal 104.014 [dBV(A)]
vo.ref

measured: -103.50 [dBV(A)]

98 7 Mathcad Worksheets Amp5

7.1 MCD-WS: The Op-Amp + Transformer Driven Output Stage Amp5 Page 7

20000Hz
1 2 2
en.i.amp5.bal( f) B( f) df
B1
20Hz
SN a.i.amp5.bal 20 log SN a.i.amp5.bal 104.014 [dB(A)]
vi.ref

20000Hz
1 2 2
en.o.amp5.unbal( f) B( f) df
B1
20Hz
SN a.o.amp5.unbal 20 log
vo.ref

SN a.o.amp5.unbal 103.288 [dBV(A)]

measured: -102.90 [dBV(A)]

5. Amp3 & 4 noise worsening by additional noise from Amp5, i/p loaded with 20 :

SN ariaa.o.mc 82.9 [dB]

SNa.i.amp5.bal SNariaa.o.mc
10 10
SN ariaa.tot.mc 20 log 10 10 SN ariaa.tot.mc 82.867 [dBV(A)]

We.mc SN ariaa.tot.mc SN ariaa.o.mc We.mc 0.033 [dB]

7 Mathcad Worksheets Amp5 99

7.1 MCD-WS: The Op-Amp + Transformer Driven Output Stage Amp5 Page 8

6. Amp3 & 4 noise worsening by additional noise from Amp5, i/p loaded with ampMM and
Standard cartridge :

SNariaa.o.mm 80.5 [dB]

SNa.i.amp5.bal SNariaa.o.mm
10 10
SNariaa.tot.mm 20 log 10 10 SNariaa.tot.mm 80.481 [dBV(A)]

We.mm SNariaa.tot.mm SNariaa.o.mm We.mm po [dB]

The Op-Amp and Transformer Driven
Input Stage Amp1 8

8.1 General Design and Gain of Amp1

In the past, in order to get the required output level, I found very good sounding
results in putting together an input transformer followed by an active gain stage.1
However, these solutions always happened in the single ended world with balanced
or un-balanced inputs. For frequency and phase response accuracy, the transformers
I used became “handicapped” by the need to load them with an additional R-C
network.
To ensure less complexity and for the Engine II purposes I was in search of
an excellent 1:10 (nominal) input transformer that allows switching between
various MC cartridge loads, eg 1 kΩ, 500 Ω, 200 Ω, 100 Ω, 50 Ω, hence, without
inclusion of any kind of frequency and phase response flattening additional R-C
network.2
I found a real good one, the Lundahl LL9226. Its coils offer outstanding low DC
resistances too, thus, the generated noise3 becomes extremely low. There are small
disadvantages that come from the measured frequency and phase responses in B20k
and from the measured turns ratios:
• Depending on Fig. 8.1’s setting of S1–S4 we find—with 0.5 mV input and/or
5 mV output level—at 20 Hz a deviation from a flat frequency and phase
response of −0.3 to −0.5 dB/+13° to +18°,4 measured with the Fig. 8.1 circuit.
Lundahl claims −1.5 dB max./10 Hz at −10 dBu output level/100 kΩ output
load. Figure 8.6 shows what I’ve measured with transformer input (output)
voltages of 0.5 mVrms (5.5 mVrms), 1.58 mVrms (17.38 mVrms), and 5 mVrms
(55 mVrms) and 100 kΩ transformer output load.

1
See TSOS-1 Chaps. 6 and 16, TSOS-2 Chaps. 5 and 28.
2
See Sect. 8.3 for corresponding graphs.
3
More on transformer noise calculations: TSOS-1 Sect. 3.7, TSOS-2 Chap. 10.
4
See Sect. 8.3, Table 8.1.

© Springer International Publishing Switzerland 2016 101

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_8
102 8 The Op-Amp and Transformer Driven Input Stage Amp1

Fig. 8.1 Input gain stage alternative Amp1

• I’ve measured 4 transformers: the first pair offers equal turns ratios, however,
11.00 instead of nominal 10, the second pair offers turns ratios with a tiny
difference: 11.19 and 11.22. The measurement set-up looks like Fig. 8.1,
however, with an interruption at point ‘A’. Thus, the signal voltage at the
circuit’s output vs. the signal voltage at the transformer’s input gives the turns
ratio, here at 1 kHz and a transformer load of nearly 100 kΩ (99.8 kΩ).
The calculation results of the two Mathcad Worksheets of Chap. 9 don’t look
rather close. However, the described disadvantages could be tolerated, as long as
the listening tests of Chap. 12 won’t create any negative indications. Without
op-amp offset trim potentiometers and blocking Cs Fig. 8.1 shows all the details of
the chosen circuit.
Formed by two OP27 op-amps a rather simple looking balanced stage follows
the transformer. I’ve also tried LT1028 op-amps, however, with an input load of
20 Ω the improvement in SN became only 0.8 dB measured. Although offering
very low input noise voltage (0.95nV/rtHz) the rather high input noise current
(1pA/rtHz) plus the 100 % correlated noise current as result of unequal loads at the
op-amp’s “ + ” and “−” inputs5 does not create very much better RIAA- and
A-weighted results, calculated and measured 0.8 dB(A) too.

5
See Linear Technology LT1028 data sheet.
8.1 General Design and Gain of Amp1 103

To get the above-mentioned different input resistances the resistors R6–R9 of the
transformer’s output load (R1 + R2) || (R6–R9) can easily be set by the switch bank
S1–S4. Division by the turns ratio squared leads to the effective MC cartridge load
and Amp1 input resistances. Trimming of P1 sets the required gain. However, I
must point out that the before mentioned actual transformer turns ratios of appr. 11
may lead to a recalculation of the R6–R9 values to get the required input
resistances.
The gain G2nd of such a balanced stage (=2nd gain stage of Amp1) becomes:

R4 þ R5
G2nd ¼ 1 þ ð8:1Þ
R3 þ P1

The gain G1st of the transformer (=1st gain stage of Amp1) is mainly6 set by its
turns ratio tr = 1:n = 1:10. Hence, we can set:

G1st ¼ n ¼ nominal 10 ð8:2Þ

Thus, with R4 = R5 and RG = R3 + P1 we obtain the Amp1 overall gain Gamp1

as follows:

Gamp1 ¼ G1st G2nd


2 R4 ð8:3Þ
¼ 10 1 þ
RG

8.2 CMRR and Noise

8.2.1 CMRR

Because of the fact that the central amps Amp 3 and 4 offer very good CMRR this
point is not a big issue for the input stages. Lundahl gives no data sheet indication,
neither about the transformer’s CMRR nor about its capacitances that influence
CMRR most. Depending on the generator output resistance, normally very good
input transformers show CMRRe values better than 60 dB. I did not measure it, but
I guess, theoretically with the op-amp stage’s CMRRe.2nd = 26 dB the amp’s
CMRRe.amp1 becomes always >86 dB.
Concerning the common mode noise voltage CMamp1 generated by Amp1 we
have the same situation already described in Sect. 4.3.6. That’s why it is important to

6
Mainly, because there is a gain loss Gloss coming from the voltage divider effect between input
load resistance R0 and the impedance Zi, formed by the Amp1 input resistance (see the details in
the next chapter’s MCD Worksheets).
104 8 The Op-Amp and Transformer Driven Input Stage Amp1

have a well designed following gain stage with a CMRR that is capable to damp
CMamp1 to an ignorable amount. The next chapter’s Mathcad Worksheet 9.1 /Point 8.
gives the details.

8.2.2 Noise and SNs

With 20 Ω (43 Ω for my DL-103 cartridge) input load and in consideration of the
RIAA transfer, A-weighting, and the following gain stages, the outstanding input
referred SNariaa.i becomes measured −81.8 dB(A) (−79.3 dB(A)) and calculated
−80.9 dB(A) (−78.6 dB(A)).
The calculation process for the Amp1 output noise voltage density and its SNs is
given in full detail in the next chapter’s MCD-WS 9.1 and 9.2. In Chap. 9, I present
two different worksheets because, after RIAA equalization and A-weighting,
there is a considerable difference in the calculation results with real transformer data
(MCD-WS 9.1) and with data from the data sheet (MCD-WS 9.2). However, I’ll
repeat the main equations as follows with Fig. 8.2 showing the circuit with all
relevant noise sources in place and Fig. 8.3 gives the details about the input load Zi.
Rs in Fig. 8.3 is the transformer’s secondary coil resistance, Rpsec is its primary
coil resistance transferred to the secondary side, and R0sec is the input load trans-
ferred to the secondary side too.
Amp1’s frequency and R0 dependent output noise voltage density en.o.amp1(f,R0)
thus becomes:
r
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

en:o:amp1 ðf; R0Þ ¼ en:o:tr1 ðf; R0Þ2 þen:i:2nd ðf Þ2 G22nd ð8:4Þ

Fig. 8.2 Noise sources of Amp1

8.2 CMRR and Noise 105

Fig. 8.3 Components of the

Fig. 8.2 input load
impedance Zi

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:o:tr1 ðf Þ ¼ en:Zi ðR0Þ2 þin:i ðf Þ2 Zi ðR0Þ2 ð8:5Þ
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
en:i ðf Þ2 G22nd þ 2 e2n:R4 þ in:i:1 ðf Þ2 R42 þ e2n:RG:o
en:i:2nd ðf Þ ¼ ð8:6Þ
G2nd
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:RG:o ¼ ðG2nd  1Þ 4 k T B1 ðR3 þ P1Þ ð8:7Þ

Both Chap. 9 worksheets calculate the relevant SNs and Noise Figures NF in an
input load dependent format. Figures 9.6, 9.7, 9.8 and 9.15 are the relevant graphs
with the R0 dependent curves.

8.3 Measurement Results

The following graphs (Figs. 8.4 and 8.5) and tables show the Amp1 frequency and
phase responses, measured with various input resistances (nominal: 1 kΩ, 500 Ω,
250 Ω, 125 Ω, 50 Ω and actual: 800 Ω, 400 Ω, 200 Ω, 100 Ω, 40 Ω) and a generator
output resistance of 10.8 Ω. The corresponding data are summed-up in Table 8.1.
The phase results are corrected by the phase response of the used un-bal. to bal.
converter UBC7: at 20 Hz: −0.05°, at 1 kHz: −0.07°, at 20 kHz: +1.42°.
Due to voltage divider effects at the 2nd stage’s input the frequency responses
show decreasing Amp1 output levels, exactly according to the input resistance. The
differences (in [dB]) between the top frequency response (with highest Ri) and the
others indicate an SN reduction by roughly the same amount, because the 2nd gain
stage must adequately increase its gain to reach Amp1’s original gain of 200.

7
Details see Sect. 15.2.
106 8 The Op-Amp and Transformer Driven Input Stage Amp1

Fig. 8.4 Frequency responses of Amp1, based on five different Amp1 input resistances
Ri(=S1–S4 settings)

Fig. 8.5 Phase responses of Amp1, based on five Amp1 input resistances Ri(=S1–S4 settings)
8.3 Measurement Results 107

Table 8.1 Detailed frequency and phase measurement results of Amp1, based on five Amp1
input resistances Ri(= S1–S4 settings)

Table 8.2 Figure 8.6 delta data at 10 Hz

1. Delta between 1 kHz and 10 Hz at * 0dBV: −0.76 dB
2. Delta between 1 kHz and 10 Hz at *−10dBV: −1.26 dB
3. Delta between 1 kHz and 10 Hz at *−20dBV: −1.48 dB
4. Delta between 1 kHz and 10 Hz at *−30dBV: −1.52 dB
5. Delta between 1 kHz and 10 Hz at *−40dBV: −1.52 dB
Note With a generator output resistance of 10.8 Ω in Fig. 8.6 Amp1’s 0 dBV output level equals a
nominal (actual) transformer output voltage of 50 mVrms [−23.8 dBu] (55 mVrms [−23.0 dBu]), etc.

We can simulate the whole matter by changing R1 and R2 according to the

chosen input resistance Ri and input load R0. Hence, with decreasing R1 + R2 the
now additionally R1 + R2 dependent term Ge.loss(R0) in the next chapter’s work-
sheets becomes bigger, thus, worsening the SNs accordingly (Figs. 8.1 and 8.3).
Concerning the deviation from a flat frequency response in the range ≤1 kHz at
the Amp1 output I’ve measured the following Table 8.2 delta data from Fig. 8.6.
These delta data look like being in line with the before mentioned Lundahl data
for high input levels, the deltas keep their levels with decreasing input signal
voltages <0.5 mVrms (=100 mVrms [−20dBV] output level). Points 3. ff of Table 8.2
108 8 The Op-Amp and Transformer Driven Input Stage Amp1

Fig. 8.6 Output voltage frequency responses for five different Amp1 input voltages in 10 dB steps
from 50 µVrms to 5.0 mVrms in a 10 Hz–20 kHz band

indicate why the overall frequency response of the Amp1 + Amp3 + Amp5 + Trafo
chain in Fig. 12.1a shows the same low-frequency tendency of its the flatness.
Table 8.1, line 8, shows the deviations at 20 Hz: always <0.5 dB.
Fully independent from the generator output resistance the responses at the
upper end of B20k—at 20 kHz—do not show these amounts of deviations (see
Table 8.1, line 10, always <0.08 dB).

8.4 Additional Remarks

8.4.1 DC Servo

It is not necessary to think about a DC servo. No matter which type of op-amp I’ve
checked the circuit’s DC conditions look very stable. Trimming the offset voltages
of OPs 1 & 2 will sufficiently make it.
8.4 Additional Remarks 109

8.4.2 Wild Oscillation

Another point is the prevention of wild oscillation. Depending on the location of the
amp’s PCB in the case there might be a tendency of ringing at very high frequencies
>1 MHz. Without touching the overall performance, a ceramic capacitance of
approximately 100p–560p parallel to R4 or R5 or parallel to each resistor will
sufficiently damp any oscillation. It’s a trial and error approach.
Mathcad Worksheets Amp1
9

Contents

9.1 MCD-WS: The Transformer + Op-Amp Driven Amp1 (Real Data)

9.2 MCD-WS: The Transformer + Op-Amp Driven Amp1 (Data Sheet Data)

pffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffi
Note 1: MCD 11 has no built-in unit “rtHz” or “ Hz ”. To get 1 Hz based
voltage noise and current noise densities the rms noise voltage and
current in a specific frequency range B > 1 Hz must be multiplied by
pffiffiffiffiffiffiffiffiffi
1 Hz and divided by the root of that specific frequency range √B!
Note 2: MCD 11 offers no “dB” unit. This is available from MCD 13 on!

© Springer International Publishing Switzerland 2016 111

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_9
112 9 Mathcad Worksheets Amp1

9.1 MCD-WS: The Transformer + Op-Amp Driven Amp1 (Real Data) Page 1

Amp1 SN and gain calculations (Real data)

Fig. 9.1 = Fig. 8.1

1. Definition of all meaningful constants, components, etc. :

23 1 3
T 300 K k 1.38065 10 VAsK vi.nom 0.5 10 V
3
B20k 19980 Hz B1 1Hz h 1000Hz f 10Hz 15Hz 100 10 Hz

2. Gain stage component data :

3 3
R0 20 R1 49.9 10 R2 R1 R3 100 R4 1 10 R5 R4

Gtot
n 11 Gtot 200 G2nd G2nd 18.182
n

R4 R5 R4 R5
G2nd 1 RG RG 116.402 P1 RG R3 P1 16.402
RG G2nd 1
note: in this calculation course RG = R3 + P1!

3. Calculation of the total input load Zi :

Measured transformer values according to Lundahl's data sheet recommendation "C":

Rp 4.3 Rs 254
2 3
R0sec R0 n R0sec 2.42 10
9 Mathcad Worksheets Amp1 113

9.1 MCD-WS: The Transformer + Op-Amp Driven Amp1 (Real Data) Page 2

2
Rpsec Rp n Rpsec 520.3
3
Ztr1 R0sec Rpsec Rs Ztr1 3.194 10

( R1 R2) Ztr1 3
Zi Zi 3.095 10
R1 R2 Ztr1

R1 R2 Rs
Ri Rp Ri 831.193
2
n

4. Calculation of the total noise voltage density at the input of Amp1 :

Fig. 9.2 = Fig. 8.2

Noise sources of Amp1

Fig. 9.3 = Fig. 8.3

Components of the
Fig. 9.2 input load Zi

4.1 Noise voltage density at the output of the 2nd stage (OPs 1 & 2) with input shorted :
9 12
en.i.1 3.0 10 V fc.e 2.7Hz in.i.1 0.4 10 A fc.i 120Hz

en.i.2 en.i.1 in.i.2 in.i.1

fc.e fc.i
en.i.1( f) en.i.1 1 en.i.2( f) en.i.1( f) in.i.1( f) in.i.1 1 in.i.2( f) in.i.1( f)
f f
114 9 Mathcad Worksheets Amp1

9.1 MCD-WS: The Transformer + Op-Amp Driven Amp1 (Real Data) Page 3

in.i.1( f) 15
in.i( f) in.i( h) 299.333 10 A
2
9
en.i( f) 2 en.i.1( f) en.i( h) 4.248 10 V

9
en.RG.o G2nd 1 4 k T B1 RG en.RG.o 23.861 10 V

9
en.R4 4 k T B1 R4 en.R4 4.07 10 V

2 2 2 2 2 2 9
en.o.2nd( f) en.i( f) G2nd 2 en.R4 2 in.i.1( f) R4 en.RG.o en.o.2nd( h) 81.051 10 V

en.o.2nd( f) 9
en.i.2nd( f) en.i.2nd( h) 4.458 10 V
G2nd

4.2 Noise voltage density at the output of the 1st stage (Z i) with input loaded :

9
en.Zi 4 k T B1 Zi en.Zi 7.161 10 V

2 2 2 9
en.o.tr1( f) en.Zi in.i( f) Zi en.o.tr1( h) 7.221 10 V

4.3 Noise voltage density at the output of Amp 1 with input loaded :

2 2 2 9
en.o.amp1( f) en.o.tr1( f) en.i.2nd( f) G2nd en.o.amp1( h) 154.29 10 V

en.o.amp1( f) 12
en.i.amp1( f) en.i.amp1( h) 771.452 10 V
Gtot

9
1 10

Fig. 9.4
[V]

en.i.amp1( f)
Amp1 input noise
voltage density with
R0 = 20
10
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

5. Calculations of loss due to the cartridge load :

2 2
R1 R2 Rs n Rp n R0
Gloss Gloss 1.032
R1 R2

Ge.loss 20 log Gloss Ge.loss 0.274 [dB]

9 Mathcad Worksheets Amp1 115

9.1 MCD-WS: The Transformer + Op-Amp Driven Amp1 (Real Data) Page 4

6. SN calculation (R0-dependent) :

2 3
R0sec( R0) R0 n Ztr1( R0) R0sec( R0) Rpsec Rs Ztr1 20 3.194 10

( R1 R2) Ztr1( R0) 3

Zi( R0) Zi 20 3.095 10
R1 R2 Ztr1( R0)

9
en.Zi( R0) 4 k T B1 Zi( R0) en.Zi 20 7.161 10 V

2 2 2 9
en.o.tr1( f R0) en.Zi( R0) in.i( f) Zi( R0) en.o.tr1 h 20 7.221 10 V

2 2 2 9
en.o.amp1( f R0) en.o.tr1( f R0) en.i.2nd( f) G2nd en.o.amp1 h 20 154.29 10 V

en.o.amp1( f R0) 12
en.i.amp1( f R0) en.i.amp1 h 20 771.452 10 V
G2nd n
12
en.i.amp1 h 0 519.494 10 V

9
1 10
[V/rtHz]

Fig. 9.5
en.i.amp1 f 0
Amp1 input noise
voltage density with
R0 = 0
10
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

2 2
R1 R2 Rs n Rp n R0
Gloss ( R0) Gloss 20 1.032
R1 R2

Ge.loss( R0) 20 log Gloss ( R0) Ge.loss 20 0.274 [dB]

20000Hz
1 2
en.i.amp1( f R0) df
B1
20Hz
SNne.i( R0) 20 log Ge.loss( R0)
vi.nom

SNne.i 20 72.959 [dB]

116 9 Mathcad Worksheets Amp1

9.1 MCD-WS: The Transformer + Op-Amp Driven Amp1 (Real Data) Page 5

20000Hz
1 2
ein( R0) en.i.amp1( f R0) df 12
B20k ein 0 519.21 10 V
20Hz

1.259 1 1 1
A( f)
2 2 2 2
20.6Hz 107.7Hz 737.9Hz f
1 1 1 1
f f f 12200 Hz

20000Hz
1 2 2
en.i.amp1( f R0) A( f) df
B1
20Hz
SN ne.i.a( R0) 20 log Ge.loss( R0)
vi.nom

SN ne.i.a 20 75.005 [dB(A)]

1
2
3 6
1 2 10 Hz 318 10 s
R1000
2 2 R1000 9.898
3 6 3 6
1 2 10 Hz 3180 10 s 1 2 10 Hz 75 10 s

2
6
1 2 f 318 10 s
R( f) R1000
2 2
6 6
1 2 f 3180 10 s 1 2 f 75 10 s

20000Hz
1 2 2
en.i.amp1( f R0) R( f) df
B1
20Hz
SN riaa.i( R0) 20 log Ge.loss( R0)
vi.nom

SN riaa.i 20 76.477 [dB]

20000Hz
1 2 2 2
en.i.amp1( f R0) R( f) A( f) df
B1
20Hz
SN ariaa.i( R0) 20 log Ge.loss( R0)
vi.nom

SN ariaa.i 20 80.889 [dB(A)]

measured via Amps 4 & 5 & trafo: -81.78 [dBV(A)]

9 Mathcad Worksheets Amp1 117

9.1 MCD-WS: The Transformer + Op-Amp Driven Amp1 (Real Data) Page 6

R0 0 0.5 100

65
67
[dB ref. 0.5mVrms]

69
71
SNne.i( R0) 73
75
SNariaa.i ( R0)
77
79
81
83
85
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
R0
[ohm]

Fig. 9.6 Input referred and A-weighted SN vs. input load

80
80.5
81
[dB ref. 0.5mVrms]

81.5
82
SNariaa.i ( R0) 82.5
83
83.5
84
84.5
85
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
R0
[ohm]

Fig. 9.7 Zoomed Fig. 9.6

118 9 Mathcad Worksheets Amp1

9.1 MCD-WS: The Transformer + Op-Amp Driven Amp1 (Real Data) Page 7

7. NF calculation (R0-dependent) :

R0 1 2 1000 RiL(R0) = R0-dependent Amp1 input load

1
1 1
RiL( R0) Ri R0 RiL 20 19.53

12
in.i.eff( f) in.i( f) n in.i.eff( h) 3.293 10 A

en.i( f) 12
en.i.eff( f) en.i.eff( h) 386.215 10 V
n

12
en.RiL( R0) 4 k T B1 RiL( R0) en.RiL 20 568.832 10 V

2 2 2 2
en.RiL( R0) in.i.eff( h) RiL( R0) en.i.eff( h)
NFe ( R0) 20 log NFe 20 1.684 [dB]
en.RiL( R0)

10

Fig. 9.8
NF of Amp1 vs. R0
[dB]

NFe ( R0) 1

0.1 3
1 10 100 1 10
R0
[ohm]
9 Mathcad Worksheets Amp1 119

9.2 MCD-WS: The Transformer + Op-Amp Driven Amp1 (Data sheet data) Page 1

Amp1 SN and gain calculations (Data sheet data)

Fig. 9.10 = Fig. 8.1

1. Definition of all meaningful constants, components, etc. :

23 1 3
T 300 K k 1.38065 10 VAsK vi.nom 0.5 10 V
3
B20k 19980 Hz B1 1Hz h 1000Hz f 10Hz 15Hz 100 10 Hz

2. Gain stage component data :

3 3
R0 20 R1 49.9 10 R2 R1 R3 100 R4 1 10 R5 R4

Gtot
n 10 Gtot 200 G2nd G2nd 20
n

R4 R5 R4 R5
G2nd 1 RG RG 105.263 P1 RG R3 P1 5.263
RG G2nd 1
note: in this calculation course RG = R3 + P1!

3. Calculation of the total input load Zi :

Measured transformer values according to Lundahl's data sheet recommendation "C":

Rp 5 Rs 260
2 3
R0sec R0 n R0sec 2 10
120 9 Mathcad Worksheets Amp1

9.2 MCD-WS: The Transformer + Op-Amp Driven Amp1 (Data sheet data) Page 2

2
Rpsec Rp n Rpsec 500
3
Ztr1 R0sec Rpsec Rs Ztr1 2.76 10

( R1 R2) Ztr1 3
Zi Zi 2.686 10
R1 R2 Ztr1

R1 R2 Rs
Ri Rp Ri 1005.6
2
n

4. Calculation of the total noise voltage density at the input of Amp1 :

Fig. 9.11 = Fig. 8.2

Noise sources of Amp1

Fig. 9.12 = Fig. 8.3

Components of the
Fig. 9.11 input load Zi

4.1 Noise voltage density at the output of the 2nd stage (OPs 1 & 2) with input shorted :

9 12
en.i.1 3.0 10 V fc.e 2.7Hz in.i.1 0.4 10 A fc.i 120Hz

en.i.2 en.i.1 in.i.2 in.i.1

fc.e fc.i
en.i.1( f) en.i.1 1 en.i.2( f) en.i.1( f) in.i.1( f) in.i.1 1 in.i.2( f) in.i.1( f)
f f
9 Mathcad Worksheets Amp1 121

9.2 MCD-WS: The Transformer + Op-Amp Driven Amp1 (Data sheet data) Page 3

in.i.1( f) 15
in.i( f) in.i( h) 299.333 10 A
2
9
en.i( f) 2 en.i.1( f) en.i( h) 4.248 10 V

9
en.RG.o G2nd 1 4 k T B1 RG en.RG.o 25.091 10 V

9
en.R4 4 k T B1 R4 en.R4 4.07 10 V

2 2 2 2 2 2 9
en.o.2nd( f) en.i( f) G2nd 2 en.R4 2 in.i.1( f) R4 en.RG.o en.o.2nd( h) 88.784 10 V

en.o.2nd( f) 9
en.i.2nd( f) en.i.2nd( h) 4.439 10 V
G2nd

4.2 Noise voltage density at the output of the 1st stage (Zi) with input loaded :

9
en.Zi 4 k T B1 Zi en.Zi 6.671 10 V

2 2 2 9
en.o.tr1( f) en.Zi in.i( f) Zi en.o.tr1( h) 6.719 10 V

4.3 Noise voltage density at the output of Amp1with input loaded :

2 2 2 9
en.o.amp1( f) en.o.tr1( f) en.i.2nd( f) G2nd en.o.amp1( h) 161.058 10 V

en.o.amp1( f) 12
en.i.amp1( f) en.i.amp1( h) 805.29 10 V
Gtot

9
1 10

Fig. 9.13
[V]

en.i.amp1( f)
Amp1 input noise
voltage density with
R0 = 20
10
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

5. Calculations of loss due to the cartridge load :

2 2
R1 R2 Rs n Rp n R0
Gloss Gloss 1.028
R1 R2

Ge.loss 20 log Gloss Ge.loss 0.237 [dB]

122 9 Mathcad Worksheets Amp1

9.2 MCD-WS: The Transformer + Op-Amp Driven Amp1 (Data sheet data) Page 4

6. SN calculation (R0-dependent) :

2 3
R0sec( R0) R0 n Ztr1( R0) R0sec( R0) Rpsec Rs Ztr1 20 2.76 10

( R1 R2) Ztr1( R0) 3

Zi( R0) Zi 20 2.686 10
R1 R2 Ztr1( R0)

9
en.Zi( R0) 4 k T B1 Zi( R0) en.Zi 20 6.671 10 V

2 2 2 9
en.o.tr1( f R0) en.Zi( R0) in.i( f) Zi( R0) en.o.tr1 h 20 6.719 10 V

2 2 2 9
en.o.amp1( f R0) en.o.tr1( f R0) en.i.2nd( f) G2nd en.o.amp1 h 20 161.058 10 V

en.o.amp1( f R0) 12
en.i.amp1( f R0) en.i.amp1 h 20 805.29 10 V
G2nd n
12
en.i.amp1 h 0 567.923 10 V

9
1 10
[V/rtHz]

Fig. 9.14
en.i.amp1 f 0
Amp1 input noise
voltage density with
R0 = 0
10
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

2 2
R1 R2 Rs n Rp n R0
Gloss ( R0) Gloss 20 1.028
R1 R2

Ge.loss( R0) 20 log Gloss ( R0) Ge.loss 20 0.237 [dB]

20000Hz
1 2
en.i.amp1( f R0) df
B1
20Hz
SNne.i( R0) 20 log Ge.loss( R0)
vi.nom

SNne.i 0 75.827 [dB]

9 Mathcad Worksheets Amp1 123

9.2 MCD-WS: The Transformer + Op-Amp Driven Amp1 (Data sheet data) Page 5

20000Hz
1 2
ein( R0) en.i.amp1( f R0) df 12
B20k ein 0 567.61 10 V
20Hz

1.259 1 1 1
A( f)
2 2 2 2
20.6Hz 107.7Hz 737.9Hz f
1 1 1 1
f f f 12200 Hz

20000Hz
1 2 2
en.i.amp1( f R0) A( f) df
B1
20Hz
SNne.i.a( R0) 20 log Ge.loss( R0)
vi.nom

SNne.i.a 20 74.669 [dB(A)]

1
2
3 6
1 2 10 Hz 318 10 s
R1000
2 2 R1000 9.898
3 6 3 6
1 2 10 Hz 3180 10 s 1 2 10 Hz 75 10 s

2
6
1 2 f 318 10 s
R( f) R1000
2 2
6 6
1 2 f 3180 10 s 1 2 f 75 10 s

20000Hz
1 2 2
en.i.amp1( f R0) R( f) df
B1
20Hz
SNriaa.i( R0) 20 log Ge.loss( R0)
vi.nom

SNriaa.i 20 76.147 [dB]

20000Hz
1 2 2 2
en.i.amp1( f R0) R( f) A( f) df
B1
20Hz
SNariaa.i( R0) 20 log Ge.loss( R0)
vi.nom

SNariaa.i 0 83.757 [dB(A)]

measured via Amps 4 & 5 & trafo: -81.78 [dBV(A)]

124 9 Mathcad Worksheets Amp1

9.2 MCD-WS: The Transformer + Op-Amp Driven Amp1 (Data sheet data) Page 6

7. NF calculation (R0-dependent) :

R0 1 2 1000 RiL(R0) = R0-dependent Amp1 input load

1
1 1
RiL( R0) Ri R0 RiL 20 19.61

12
in.i.eff( f) in.i( f) n in.i.eff( h) 2.993 10 A

en.i( f) 12
en.i.eff( f) en.i.eff( h) 424.836 10 V
n

12
en.RiL( R0) 4 k T B1 RiL( R0) en.RiL 20 569.995 10 V

2 2 2 2
en.RiL( R0) in.i.eff( h) RiL( R0) en.i.eff( h)
NFe ( R0) 20 log NFe 20 1.948 [dB]
en.RiL( R0)

10

Fig. 9.15
NF of Amp1 vs. R0
[dB]

NFe ( R0) 1

0.1 3
1 10 100 1 10
R0
[ohm]
The BJT and Op-Amp Driven Input
Amp2 10

10.1 General Design and Gain of Amp2

The Amp2 circuit shown in Fig. 10.1 is nothing else but a kind of balanced version
of Douglas Self’s original MC pre-amp,1 however, forming a differential input,
equipped with very low-noise complementary BJTs in the input stage, the Toshiba
2SC3329BL / 2SA1316BL (hfe *500, rbb′ = 2 Ω). I used such an un-balanced gain
stage as first stage of the TSOS-2 Chap. 15 and Module 2 phono-amp (TSOS-1
Chap. 6).
The exact equations to calculate the gain of the T1-T2 or T3-T4 stage are part of
this book’s Chap. 21. However, Amp2’s overall gain Gamp2 becomes:

R10 þ R11
Gamp2 ¼ 1 þ ð10:1Þ
R5
Because of the very low resistive environment, I’ve composed R5 by the shown
arrangement. T1–T4 work with 2.1–2.2 mA collector current each, set by R4, R6,
R7 and the emitter resistors

R5 ¼ ðP3jjR5aÞ þ ðR5bjjR5cjjR5dÞ ð10:2Þ

With P3 trimmed to 0 Ω, P1 and P2 (P1 or P2 could be replaced by a 100 Ω

resistor) set equal (but with opposite polarity) DC voltages at test points TP1 and
TP2, roughly ±2 to ±2.1 mV. With P3 set to the nominal gain of 200, we should
find appr. ±3 to ±3.3 mV at the test points. Generally, P3 can set overall gains in a
range of 51–301 without significantly influencing the DC voltage and current sit-
uation of the input stage.
The feedback resistors R10 & R11 seem to be rather small. They are chosen very
much smaller than the usually recommended min. 600 Ω output load for the 5534 or
equivalent op-amp types. I guess the op-amps do not have big problems by driving

“Small Signal Audio Design”, Douglas Self, 2010, Focal Press.

1

© Springer International Publishing Switzerland 2016 125

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_10
126 10 The BJT and Op-Amp Driven Input Amp2

Fig. 10.1 Input gain stage alternative Amp2

this output load with a maximal OPs 1 & 2 output voltage of nominal 500 mVrms at
20 kHz. In this respect the only point here is the question what would happen with
the op-amps if they would be forced into a +20 dB overload situation, hence 5 Vrms
at 20 kHz.
Under normal conditions the Fig. 10.1 circuit works very well, hence, by feeding
the MC cartridge with a 1 kHz / 0 dB DIN2 signal we trim P3 to the gain that
ensures a nominal 100 mVrms at Amp2’s balanced output. According to Fig. 10.2
and in cases of massive overload threat, I recommend the insertion of two boosters
at points A + A’ and B + B’. They do not negatively touch the excellent picture of

2
See TSOS-2, Chap. 3.
10.1 General Design and Gain of Amp2 127

Fig. 10.2 Booster for

insertion into Fig. 10.1’s
points A + A’ and B + B’

the Fig. 10.1 Amp2 and they work well in the Module 2 phono-amps in both TSOS
versions.
In contrast to all other switching actions switching from Amp1 to Amp2 leads to
a tiny but audible transient phenomenon. I guess the high valued input Cs create it.
The other way around does not create such a phenomenon.

10.2 CMRR

Compared with Amp1, we find a very much lower CMRRe figure. With a common
mode gain Gcm = 1 and a differential gain Gdif = 200 it becomes only 46 dB = 20 * log
(Gdif/Gcm) = 20 * log (200). We could drastically improve it (min. +60 dB)
by integration of a gain x2 output stage à la Fig. 6.1’s OPs 3 & 4 + R6–R13.
Additionally, resistors R10 and R11 of Fig. 10.1 must then be reduced to 75 Ω.

10.3 Noise

10.3.1 General Noise Aspects

With an input load of 20 Ω, the input referred SN situation is not as good as the
Amp1 solution, however, better than the goal from Chap. 1 and always better than
most phono-amps on the market. Additionally, in contrast to the Amp1 SNs with
input loads ≤10 Ω and compared with the calculated results Amp2 shows
increasingly worse measured input referred RIAA- and A-weighted SN results.
They are measured at the output of the Amp2 + Amp4 + Amp5 + Trafo chain.
Hence, eg with i/p-load 5 Ω: Amp1 calculated SNo = −84.0 dB(A) versus Amp2
128 10 The BJT and Op-Amp Driven Input Amp2

measured SNo = −81.2 dB(A). Because of the rather complex SN calculations with
unclear flicker-noise behaviour of the chosen BJTs and in conjunction with the next
chapter’s Mathcad Worksheets the following section should give satisfying
answers.
In consideration of the RIAA transfer function, A-weighting, the subsequent
gain stages, and with 20 Ω (43 Ω) input load the measured input referred SN via
Amps 3 or 4 + Amp5 + Trafo shows a worst case SNariaa.i.m = −79.8 dB(A)
(−78.4 dB(A)), the calculated SNariaa.i.calc = −80.4 dB(A) (−78.6 dB(A)). THD and
IMD become <0.01% each.

10.3.2 The BJT Noise Model Reloaded

In the past—and in my books too—I have been handling the analysis of the BJT
noise creation in a way of hopefully easy usage, hence, mainly based on a white
noise production of the BJTs. The reason lies in the Figs. 10.3 and 10.4 charts of the
used input BJTs, the now obsolete Analog Devices SSM2210.3 They show com-
pletely ignorable 1/f-noise in the audio band B20k. The 1/f-noise corner frequencies
will not change after paralleling the devices.
Hit by a surprising noise production mechanism the Amp2 SN results forced me
to change my analysis approach from purely white noise to a flicker noise
(1/f-noise) influenced one. With input loads R0 > 45 Ω Amp2’s difference between
a calculated and a measured input referred SN became always <1 dB. The white
noise based differences with input loads ≤45 Ω became >1 dB, up to 4.8 dB at
R0 = 0 Ω. If we accept a maximal deviation of 1 dB for R0 ≥ 1 Ω we have to find a
reasonable calculation method that includes 1/f-noise effects. Thus, the BJT’s noise
currents and voltages have to be treated in a frequency dependent way.
According to the respective white noise based figures in TSOS-1 & TSOS-2 and
with the hopefully self-explanatory evolution towards a simple BJT noise model of
Figs. 10.5, 10.6 and 10.7, we can sum-up the noise sources of a BJT in a frequency
dependent equivalent input referred noise voltage density en.i(f) and in an input
referred noise current density in.i(f).
Here, I talk about noise sources in B20k only. Outside the band >20 kHz we find
another deviation from the white noise flatness. Based on most BJT’s transit
frequency fT ≫ 20 MHz it’s the current noise’s increase by *f ≡ +6 dB/oct. or
+20 dB/dec. after a 2nd corner frequency of ≥*fT/hfe. Fortunately, this behaviour
lies far outside the audio band and it does not touch the here presented cases.
With the following equations and based on given operating conditions for the
collector current IC, the collector-emitter voltage Vce, and thus the small signal

3
See Analog Devices data sheet.
10.3 Noise 129

Fig. 10.3 SSM2210 noise voltage density versus frequency and collector current

Fig. 10.4 SSM2210 noise current density versus frequency and collector current

Fig. 10.5 General simplified BJT noise model for the audio band
130 10 The BJT and Op-Amp Driven Input Amp2

Fig. 10.6 Transfer of the Fig. 10.5 model into a strictly input referred one

Fig. 10.7 Transfer of the

Fig. 10.6 noise model into the
final BJT noise model with
only two equivalent noise
sources in place

current gain hfe (≈hFE = DC current gain), the operating temperature T, and
hopefully a given base-spreading resistance rbb’, we can now calculate Fig. 10.7’s
equivalent noise sources.
In a BJT the flicker noise (=1/f-noise region) and shot noise (=white noise
region) production starts in the collector current IC, hence, the noise current in.c(f) in
density format becomes:
 
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi f c:i x
in:c ðf Þ ¼ 2qIC B1 1 þ ð10:3Þ
f

In (10.3) fc.i is the 1/f-noise corner frequency of the IC noise current and ‘x’
defines the slope of the noise current in the 1/f-noise region. Slope figure x = 0.5
means that the slope decreases according to the square root of 1/f. With that, it
equals pink noise with a slope of −10 dB/decade or −3 dB/octave. However, in the
reality of low-noise BJTs x can show values between 0 (=purely white noise) and 1
(purely red noise). To find the right one in the next section we have to dive deep
into the jungle of mostly vague or rough data-sheet data and their integration into a
useful math.
10.3 Noise 131

The collector noise current transferred to the input shows two different noise
relevant effects. It creates what I call the BJT kernel noise voltage density en1(f) and
it is the source for the input referred noise current density in.i(f):

in:c ðf Þ
en1 ðf Þ ¼ ð10:4Þ
gm

in:c ðf Þ
in:i ðf Þ ¼ ð10:5Þ
hfe

With the BJT’s mutual conductance (transconductance) gm

qIC
gm ¼ ð10:6Þ
kT
We’ll get a useful expression for the kernel noise voltage density en1(f):
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 
k 2 T2 f c:i x
en1 ðf Þ ¼ 2 B1 1 þ ð10:7Þ
qIC f

The noise voltage density en.rbb′ of the base spreading resistance rbb’ becomes:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:rbb0 ¼ 4kTrbb0 B1 ð10:8Þ

We can take en.rbb′ frequency independent in B20k.4 Thus, we obtain the general
equation to calculate the frequency dependent input referred noise voltage density
en.i(f) of Fig. 10.7 as follows:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:i ðf Þ ¼ en1 ðf Þ2 þ in:i ðf Þ2 r2bb0 þ e2n:rbb0 ð10:9Þ

The BJT’s kernel noise voltage density en1(f) is fully independent of any
resistance that might disturb the nice looking picture of its decreasing value with
increasing IC. However, the bad boy in the BJT noise production game is the base-
spreading resistance rbb’, mostly not given in data-sheets and mostly too high for
real lowest-noise purposes. The good news: in the past years more and more
information about the rbb’ value of many BJTs came on the table and—among many
alternatives—it becomes easier now to select the ones we need.

4
See page xxv for further information on physical expressions and constants, etc.
132 10 The BJT and Op-Amp Driven Input Amp2

Fig. 10.8 NF of 2SC3329 at

10 Hz

10.3.3 In Search of the Slope Figures ‘x’ (2SC3329)

and ‘y’ (2SA1316)

The BJTs in the input section of Amp2 show very interesting data for MC purposes
and Toshiba claims low 1/f-noise, low base spreading resistance of 2 Ω (at
100 MHz!!!), and—without any further explanations—rather low input referred
noise voltage density of typical 0.6 nV/rtHz.
Additional noise-concerning information come from the NF charts that show
contours of various NF values in the source resistance RG versus collector current
IC plane. The absence of charts like the ones of Figs. 2 and 3 triggers my suspicion
that, by contrasting the manufacturer’s claim, the BJTs produce a certain amount of
1/f-noise too; however, I guess with lower than −10 dB/dec. slopes, especially with
input loaded by a resistance.
Figures 10.8, 10.9, 10.10 and 10.115 show the NF charts of the two devices. The
operating current gain hfe is not indicated, however, with a flat hfe over most of the
operating collector current (100 µA–10 mA) we can expect equal NF conditions for

5
Toshiba data sheet.
10.3 Noise 133

Fig. 10.9 NF of 2SC3329 at

1 kHz

the indicated hfe range of 350–700. The chosen BJTs for Amp2 operate at a selected
hfe = 500.
The first chart is taken at 10 Hz and the second one at 1 kHz. The data sheet’s
table of Electrical Characteristics indicates NF values too, however, without any
relevance for this section, because they show NF values for input loads ≥100 Ω.
The major disadvantage of the shown charts lies in the fact that all NF solutions
are based on input loads ≥10 Ω. There is no indication for lower input loads and no
one for the shorted input case. Hence, in order to get equivalent input noise voltages
and currents I had to find an easy to use path out of the mud. Generally, my
developed approach is based on the chosen operating collector current and an
average input load of 20 Ω. Consequently, other than 20 Ω input loads will yield
calculated results less close to the measured ones, as it will be the case with 20 Ω.
The Amp2 collector current IC is set to 2.2 mA for each device. The collector-
emitter voltage VCE comes up with appr. 4.3 V. Now we must find ‘x’ and ‘y’ for
the 20 Ω input load case. This means that each half of Amp2 (one half represents
the input’s N channel, the other one represents the P channel) “sees” an input load
of 10 Ω. Figure 10.12 shows the situation of Amp2 with all noise calculation
relevant active and passive components.
134 10 The BJT and Op-Amp Driven Input Amp2

Fig. 10.10 NF of 2SA1316

at 10 Hz

The search process for the N half follows the steps given below; the P half
process looks the same. I add its results in brackets.

1. At 10 Hz: from a zoomed Fig. 10.8 (10.10) value pair 2.2 mA/10 Ω we’ll get
NFe.n.10 = 10.2 dB (9.5 dB)
2. At 1 kHz: from a zoomed Fig. 10.9 (10.11) value pair 2.2 mA/10 Ω we’ll get
NFe.n.1k = 5.0 dB (5.0 dB)
3. For the NF calculation of a gain stage input with input load R0 and according to
Fig. 10.13 the general equation looks as follows (“e” signals a noise factor NF
expressed in the noise figure mode NFe = 20 log(NF) [dB]):
0qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
en:i ðf Þ2 þ e2n:R0 þ in:i ðf Þ2 R02
NFe ¼ 20 log@ A ð10:10Þ
en:R0

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:R0 ¼ 4kTB1 R0 ð10:11Þ
10.3 Noise 135

Fig. 10.11 NF of 2SA1316

at 1 kHz

By inclusion of the above given frequency dependent Eqs. (10.5) and (10.9) for
10 Hz and 1 kHz into (10.10) we obtain the x and fc.i dependent expressions for
NFe.n.10 (at 10 Hz) and NFe.n.1k (at 1 kHz):

NFe:n:10 ðx; fc:i Þ ¼ 10:2

0sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
en:i ð10Hz; x; fc:i Þ2 þ e2n:R0
B C
B þ in:i ð10Hz; x; fc:i Þ2 R02 C ð10:12Þ
B
¼ 20 logB C
C
@ en:R0 A

NFe:n:1k ðx; fc:i Þ ¼ 5

0sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
en:i ð1kHz; x; fc:i Þ2 þ e2n:R0
B C
B þ in:i ð1kHz; x; fc:i Þ2 R02 C ð10:13Þ
¼ 20 logB
B
C
C
@ en:R0 A
136 10 The BJT and Op-Amp Driven Input Amp2

Fig. 10.12 Amp2 with all noise calculation relevant active and passive components

Fig. 10.13 Input load

situation of a BJT
10.3 Noise 137

The P channel equations for NFe.p.10(y,fc.i) and NFe.p.1k(y,fc.i) look the same,
however, with adapted NF values according to points 1 & 2 from above:

NFe:p:10 ðy,f c:i Þ ¼ 9:5

ð10:14Þ
NFe:p:1k ðy,f c:i Þ ¼ 5

4. This step of the search process needs some patience. Without big math, I found
the following successive approximation (succ-apps) approach rather easy to
handle. Starting with guessed values for x and fc.i in (10.12) and (10.13) further
changes of 0.1 < x (y) < 0.5 and 100 Hz < fc.i < 2 MHz will gradually lead to the
NF values required by the above given points 1 & 2
• The values for the N-BJTs become thus: x = 0.16965 and fc.i = 33,750 Hz;
• The values for the P-BJTs become thus: y = 0.1492 and fc.i = 55,650 Hz;
These specific 2SC3329 & 2SA1316 values of the input loaded circuit
environment (that includes R0 = 20 Ω) allow calculating Amp2’s input referred
SNs with 10 Ω < R0 < 100 Ω input load. Resulting SNs will show a < 1 dB dif-
ference between measured and calculated values. SNs for input loads R0 < 10 Ω
show deltas >1 dB, maximal 1.5 dB with R0 = 0 Ω.
Because of the BJT’s application in an input load dependent connection to get
noise figure values NFe à la Figs. 10.8, 10.9, 10.10 and 10.11, they do not show the
BJT’s stand-alone 1/f-noise corner frequency and slope of its current noise. It won’t
be possible to get them neither with the demonstrated approach nor by studying the
data sheet.
Mathcad Worksheets 11.1 and 11.2 demonstrate the ‘x’ and ‘y’ search process.
Note: Because of the uncertainty of the NF-picking in points 1 & 2 we should
not wear out the exactness of x and y and fc.i: for x, y two digits after the decimal
point is exact enough, hence, x = 1.70 and y = 1.50 will make it in our case here,
and fc.i ±2 % will work well too.

10.3.4 The SN Calculation Process

Generally, and in full length exercised on next chapter’s Mathcad Worksheets 11.3
(x, y > 0), the SN calculation process is split in many steps. It begins with the
calculation of all relevant gain-creating aspects of the amp, such as the BJT’s
reduced mutual conductance gm.red, its input resistances rbe and Ri, gain setting
resistor R5, total input load Z1, etc. The noise voltage densities en.i.n & en.i.p
represent the BJT input noise voltages, including their emitter loads. Then follow
separate calculations for the closed-loop input referred noise voltages of the two
halves of Amp2. Figure 10.14 shows the N-half of Amp2 with all relevant noise
voltage and noise current sources. Of course, with polarities inversed the P-half
looks the same.
138 10 The BJT and Op-Amp Driven Input Amp2

Fig. 10.14 Upper (N) half of Amp2 (excl. input network Z1(f,R0)), showing all relevant noise
sources (frequency independent)

In Fig. 10.14 “1.2” always indicates a relationship to the paralleled BJTs T1 and
T2. The same applies to “3.4” for the not shown BJTs T3 & T4 in the P-half. The
subscript “n” in the first place means noise, in the 2nd place means NPN. “i” in the
first place means noise current, and as subscript it means “input related”.
The input referred noise voltages of both halves en.i.amp.n(f) & en.i.amp.p(f) include
all relevant active and passive component’s noise sources, such as those of R8 (R9),
R5n (R5p), R12 (R13), OP1 (OP2), and OP3 (OP4).
These input referred noise voltages have to be rms summed-up to form the input
referred noise voltage en.i.amp(f) of the amp without input load.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:i:amp ðf Þ ¼ en:i:amp:n ðf Þ2 þ en:i:amp:p ðf Þ2 ð10:15Þ

The input referred noise currents of the halves have to be summed-up according
to the rules of the summing of two series-connected noise current sources.6 This
noise current sum in.i.amp(f) is the input referred noise current of Amp2 too.
Therefore, it needs no further treatment.
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
!1ffi
u
u
in:i:amp ðf Þ ¼ t
1 1
2
þ ð10:16Þ
in:i:n ðf Þ in:i:p ðf Þ2

6
See TSOS-1 Chap. 3 or TSOS-2 Chap. 4.
10.3 Noise 139

With the input referred noise current densities of the two pairs of BJTs (in.T.b1.2(f) &
in.T.b3.4(f)) the following two equations are given in the most practical form:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
in:i:n ðf Þ ¼ in:T:b1:2 ðf Þ2 þ i2n:R6 þ i2n:R4n ð10:17Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
in:i:p ðf Þ ¼ in:T:b3:4 ðf Þ2 þ i2n:R7 þ i2n:R4p ð10:18Þ

The additional inclusion of the frequency and R0 dependent input load impedance
ZiL(f,R0) yields the following equations:
 1 

 R01 þ R11 þ 2jpfC1 þ ðR2 þ R3Þ1 
ZiL ðf; R0Þ ¼  
 ð10:19Þ
 þð2jpfC2Þ1 þ ð2jpfC3Þ1 

If C1, C2, and C3 are chosen of sizes that do not hurt a flat frequency and phase
response in B20k (10.22) could be simply written as frequency independent, but R0
dependent, impedance ZiL(R0):
 1
ZiL ðR0Þ ¼ R01 þ R11 þ ðR2 þ R3Þ1 ð10:20Þ

The output referred frequency and R0 dependent Amp2 noise voltage density
en.o.amp(f, R0) thus becomes:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:i:amp ðf Þ2 þ in:i:amp ðf Þ2 ZiL ðf; R0Þ2
en:o:amp ðf, R0Þ ¼ Gamp ð10:21Þ
þ en:Z:iL ðf, R0Þ2

Now we can calculate the R0 dependent output referred non-equalized SNne.o(R0):

0vu
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
u Z
20kHz
Bu 1  
Bt en:o:amp ðf; R0Þ2 dfC C
B B1 C
B C
SNne:o ðR0Þ ¼ 20 logB C
20Hz
B C ð10:22Þ
B vo:nom C
B C
@ A

With vo.nom = 100 mVrms, vi.nom = 0.5 mVrms, and Gamp = 200 we can calcu-
late the input referred SNne.i(R0) too. We will obtain the same result because
vo.nom /vi.nom = Gamp.
140 10 The BJT and Op-Amp Driven Input Amp2

Table 10.1 Amp2 non-weighted SN results for two different input loads
1/A B C D E F G H I
2 R0 SNne Delta 1 Delta 2 Delta 3 Remarks
3 Calculated Measured E-D E-C D-C
4 wn based fc.i based
5 Ω dB ref. 0.5 Vrms dB dB dB
6 0 −78.592 −75.768 −75.370 0.398 3.222 2.824 See remarks for
7 20 −74.005 −72.801 −73.130 −0.329 0.875 1.204 Table 10.2

Table 10.2 Same as Table 10.1 after RIAA equalization and A-weighting
1/A B C D E F G H I
2 R0 SNariaa Delta 1 Delta 2 Delta 3 Remarks
3 Calculated Measured E-D E-C D-C
4 wn based fc.i based
5 Ω dB(A) ref. 0.5 Vrms dB dB dB
6 0 −86.440 −83.095 −81.600 1.495 4.840 3.345 x, y, fc.i calculated
7 5 −84.845 −82.283 −81.200 1.083 3.645 2.562 for R0 = 10 Ω +10 Ω
8 10 −83.637 −81.571 −80.700 0.871 2.937 2.066
9 20 −81.933 −80.431 −79.800 0.631 2.133 1.502
10 43 −79.557 −78.612 −78.400 0.212 1.157 0.945

10.3.5 Results

Based on two different input loads Table 10.1 shows the non-weighted output
referred SNne.o = SNne.i of Amp2. The calculated column is split into two versions:
wn in column C is based on white noise (x = 0 and y = 0) only, ‘fc.i based’ in
column D includes the found x, y, and fc.i values of Sect. 10.3.3, Point 4. Column E
shows the measured SN results. Delta values are given in columns F, G, and H. F
indicates that the presented calculation approach works, whereas G gives a signal
that it would be better not to set on a white noise approach only. Table 10.2’s
column G shows why: A-weighted and RIAA equalized SN result differences
increase stronger with decreasing R0s.
In Table 10.2 we find the more realistic measured and calculated A-weighted and
RIAA equalized SNs for a broader range of input R0 s, from 0 Ω to 43 Ω (my DL-103).
It shows the SNs of the outcome after adding low-noise RIAA stages à la Amps 3 & 4
& 5 and A-weighting to the Amp2 output, thus, SNariaa.o(R0) = SNariaa.i(R0).
Because of the rather vague NF picking process with a guessed error of
± 0.0–± 0.3 dB, we should set the SN exactness to one digit after the decimal point.
According to the <1 dB goal from the beginning of this chapter we can see that down
to 10 Ω input load the x, y, fc.i based math works quite well (columns D, E, & F).
10.3 Noise 141

Fig. 10.15 Input resistor alternatives (input Cs not shown)

Although the wn-based results (C) look better, the deltas (G) don’t look equally good
and worse than the Table 10.1 ones.
Figure 12.1a shows a very flat frequency and phase response of Amp2, followed
by Amp4 + Amp5 + Trafo.

10.4 Additional Remarks

10.4.1 Input Resistors R2, R3, R4

Additionally, I have checked three other alternatives with the input resistor’s
arrangements à la Fig. 10.15, however, without any audible change in sound and
SN:

a. Without connection to ground and cold ends of R2 and R3 tied together,

b. Halving of R4 and connection of the newly created cold ends of R4a = 10 kΩ
and R4b = 10 kΩ to ground,
c. Both versions.
The switching transient phenomenon, mentioned at the end of Sect. 10.1,
became worth with these alternatives in place.

10.4.2 Wild Oscillation

A measure to take care of is the prevention of wild oscillation. Depending on the

location of the amp’s PCB in the case there might be a tendency of ringing at very
high frequencies >1 MHz. It can happen without producing any audible indication
while listening to music or other signals.
Not touching the overall performance, a ceramic capacitance of approximately
100 p–1 n parallel to R10 or R11 or parallel to each resistor will sufficiently damp
any oscillation. Like in the Amp1 case, it becomes a trial and error approach.
Mathcad Worksheets Amp2
11

Contents

11:1 MCD-WS: Evaluation of the 1/f-Noise Corner Frequency of a 2SA1316 BJT

11:2 MCD-WS: Evaluation of the 1/f-Noise Corner Frequency of a 2SC3329 BJT
11:3 MCD-WS: Amp2 SN and Gain Calculations—1/f-Noise Based Version

pffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffi
Note 1: MCD 11 has no built-in unit “rtHz” or “ Hz”. To get 1 Hz based
voltage noise current noise densities the rms noise voltage current in a
pffiffiffiffiffiffiffiffiffi
specific frequency range B > 1 Hz must be multiplied by 1 Hz divided
by the root of that specific frequency range √B!
Note 2: MCD 11 offers no “dB” unit. This is available from MCD 13 on!

© Springer International Publishing Switzerland 2016 143

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_11
144 11 Mathcad Worksheets Amp2

11.1 MCD-WS: Evaluation of the 1/f-Noise Corner Frequency of a 2SA1316 BJT

Page 1

1/f-Noise Evaluation of 2SA1316 with R0 = 10 and IC = 2.2mA

1. General data :
23 1 19
k 1.38065 10 VAsK q 1.6021765 10 As T 300 K
5
B20k 19980 Hz B1 1Hz h 1000Hz f 10Hz 20Hz 10 Hz
12
rbb 2 en.rbb 4 k T B1 rbb en.rbb 182.032 10 V

12
R0 10 en.R0 4 k T B1 R0 en.R0 407.036 10 V

3 q IC
hfe 500 IC 2.2 10 A gm
kT

2. Evaluation of 'x' and and the current noise density corner frequency 'fc.i' :

succ-apps of 'x' and 'fc.i' lead to the results of (1) & (2) further down in '3.3'!

succ-apps of 'x': x 0.1492 succ-apps of 'f c.i' : fc.i 55650 Hz

3. Density curves based on results of '2.' and relevant NF equations :

3.1 The current noise density curve :

x
fc.i 12
in.c( f) 2 q IC B1 1 in.c( h) 48.491 10 A
f
x
2 q IC fc.i 12
in.i( f) B1 1 in.i( h) 2.169 10 A
hfe f

x
2 q IC 2 q IC fc.i
in.i1( f) B1 in.i2( f) B1
hfe hfe f

x
fc.i
in.c1( f) 2 q IC B1 in.c2( f) 2 q IC B1
f
11 Mathcad Worksheets Amp2 145

11.1 MCD-WS: Evaluation of the 1/f-Noise Corner Frequency of a 2SA1316 BJT

Page 2

11
1 10

in.i( f)
[A/rtHz]

in.i1( f) Fig. 11.1

2SA1316 c ur re nt noise
in.i2( f) density curve and its
tangents, after succ-apps
and the chosen input load
12
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

3.2 The voltage noise density curve :

2
2 2 2 in.c( f) 12
en.i( f) en.rbb in.i( f) rbb en.i( h) 598.196 10 V
2
gm

2 2
2 2 2 in.c1( f) 2 2 in.c2( f)
en.i1( f) en.rbb in.i1( f) rbb (3) en.i2( f) in.i2( f) rbb (4)
2 2
gm gm

8
1 10

en.i( f)
[V/rtHz]

en.i1( f) 1 10 9 Fig. 11.2

2SA1316 voltage noise
en.i2( f) density curve and its
tangents, after succ-apps
and the chosen input load
10
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]
146 11 Mathcad Worksheets Amp2

11.1 MCD-WS: Evaluation of the 1/f-Noise Corner Frequency of a 2SA1316 BJT

Page 3

3.3 NF equations for succ-apps purposes :

s ucc-apps goals : NF ( 10Hz) 9.5 [dB] NF ( 1kHz) 5 [dB]

2 2 2 2
en.R0 R0 in.i( h) en.i( h)
NFe.1k 20 log NFe.1k 5.001 [dB] (1)
en.R0

2 2 2 2
en.R0 R0 in.i( 10Hz) en.i( 10Hz)
NFe.10 20 log NFe.10 9.501 [dB] (2)
en.R0

4. Evaluation of the voltage noise density corner frequency 'fc.e' :

From 3.2 and Fig. 11.2 we can arrange (3) and (4) as follows :

2 2
2 2 2 in.c1( f) 2 2 in.c2( f)
en.rbb in.i1( f) rbb in.c2( f) rbb
2 2
gm gm

=>

2
x
2 fc.i
x 2 q IC B1
2 2 q IC 2 2 q IC B1 2 q IC fc.i 2 f
en.rbb B1 rbb B1 rbb 0 (5)
hfe 2 hfe f 2
gm gm

To get 'f' solving of (5) leads to 'f c.e = 20.85kHz'

fc.e 20850 Hz
11 Mathcad Worksheets Amp2 147

11.1 MCD-WS: Evaluation of the 1/f-Noise Corner Frequency of a 2SA1316 BJT

Page 4

5. Slope of 'fc.i' (incl. a given R0!) :

x
fc.i
10Hz
20 log 2.984 => slope i = -3.0dB/dec or -0.9dB/oct.
x
fc.i
100Hz

6. Slope of 'fc.e' (incl. a given R0!) :

x
fc.e
10Hz
20 log 2.984 => slope e = -3.0dB/dec or -0.9dB/oct.
x
fc.e
100Hz

7. Input referred SN with the chosen input load R0 and based on the evaluated corner frequencies
from above :

2 2 2 2
en.i( f) en.R0 R0 in.i( f) en.i( f)

20000Hz
1 2
en.i( f) df
B1
20Hz
SNi 20 log
3 SNi 75.02 [dB]
0.5 10 V
148 11 Mathcad Worksheets Amp2

11.2 MCD-WS: Evaluation of the 1/f-Noise Corner Frequency of a 2SC3329 BJT

Page 1

1/f-Noise Evaluation of 2SC3329 with R0 = 10 and IC = 2.2mA

1. General data :

23 1 19
k 1.38065 10 VAsK q 1.6021765 10 As T 300 K
5
B20k 19980 Hz B1 1Hz h 1000Hz f 10Hz 20Hz 10 Hz
12
rbb 2 en.rbb 4 k T B1 rbb en.rbb 182.032 10 V

12
R0 10 en.R0 4 k T B1 R0 en.R0 407.036 10 V

3 q IC
hfe 500 IC 2.2 10 A gm
kT

2. Evaluation of 'x' and and the current noise density corner frequency 'fc.i' :

succ-apps of 'x' and 'fc.i' lead to the results of (1) & (2) further down in '3.3'!

succ-apps of 'x': x 0.16965 succ-apps of 'f c.i' : fc.i 33750 Hz

3. Density curves based on results of '2.' and relevant NF equations :

3.1 The current noise density curve :

x
fc.i 12
in.c( f) 2 q IC B1 1 in.c( h) 48.474 10 A
f

x
2 q IC fc.i 12
in.i( f) B1 1 in.i( h) 2.168 10 A
hfe f

x
2 q IC 2 q IC fc.i
in.i1( f) B1 in.i2( f) B1
hfe hfe f

x
fc.i
in.c1( f) 2 q IC B1 in.c2( f) 2 q IC B1
f
11 Mathcad Worksheets Amp2 149

11.2 MCD-WS: Evaluation of the 1/f-Noise Corner Frequency of a 2SC3329 BJT

Page 2

11
1 10

in.i( f)
[A/rtHz]

Fig. 11.3
in.i1( f) 2SC3329 c ur re nt noise
density curve and its
in.i2( f)
tangents, after succ-apps
and the chosen input load

12
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

3.2 The voltage noise density curve :

2
2 2 2 in.c( f) 12
en.i( f) en.rbb in.i( f) rbb en.i( h) 598.003 10 V
2
gm

2 2
2 2 2 in.c1( f) 2 2 in.c2( f)
en.i1( f) en.rbb in.i1( f) rbb (3) en.i2( f) in.i2( f) rbb (4)
2 2
gm gm

8
1 10

en.i( f)
[V/rtHz]

en.i1( f) 1 10 9 Fig. 11.4

2SC3329 voltage noise
en.i2( f) density curve and its
tangents, after succ-apps
and the chosen input load
10
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]
150 11 Mathcad Worksheets Amp2

11.2 MCD-WS: Evaluation of the 1/f-Noise Corner Frequency of a 2SC3329 BJT

Page 3

3.3 NF equations for succ-apps purposes :

s ucc-apps goals : NF ( 10Hz) 10.2 [dB] NF ( 1kHz) 5 [dB]

2 2 2 2
en.R0 R0 in.i( h) en.i( h)
NFe.1k 20 log NFe.1k 4.999 [dB] (1)
en.R0

2 2 2 2
en.R0 R0 in.i( 10Hz) en.i( 10Hz)
NFe.10 20 log NFe.10 10.198 [dB] (2)
en.R0

4. Evaluation of the voltage noise density corner frequency 'fc.e' :

From 3.2 and Fig. 11.2 we can arrange (3) and (4) as follows :

2 2
2 2 2 in.c1( f) 2 2 in.c2( f)
en.rbb in.i1( f) rbb in.c2( f) rbb
2 2
gm gm

=>

2
x
2 fc.i
x 2 q IC B1
2 2 q IC 2 2 q IC B1 2 q IC fc.i 2 f
en.rbb B1 rbb B1 rbb 0 (5)
hfe 2 hfe f 2
gm gm

To get 'f' solving of (5) leads to 'f c.e = 14.233kHz'

fc.e 14233 Hz
11 Mathcad Worksheets Amp2 151

11.2 MCD-WS: Evaluation of the 1/f-Noise Corner Frequency of a 2SC3329 BJT

Page 4

5. Slope of 'fc.i' (incl. a given R0!) :

x
fc.i
10Hz
20 log 3.393 => slope i = -3.4dB/dec or -1.1dB/oct.
x
fc.i
100Hz

6. Slope of 'fc.e' (incl. a given R0!) :

x
fc.e
10Hz
20 log 3.393 => slope e = -3.4dB/dec or -1.1dB/oct.
x
fc.e
100Hz

7. Input referred SN with the chosen input load R0 and based on the evaluated corner frequencies
from above :

2 2 2 2
en.i( f) en.R0 R0 in.i( f) en.i( f)

20000Hz
1 2
en.i( f) df
B1
20Hz
SNi 20 log
3 SNi 75.107 [dB]
0.5 10 V
152 11 Mathcad Worksheets Amp2

11.3 MCD-WS: Amp2 SN and Gain Calculations - 1/f-Noise Based Version

Page 1

Amp2 SN and Gain Calculations - 1/f-Noise Based (x = 0.15, y = 0.17)

Fig. 11.5 = Fig. 10.12

1. Definition of all meaningful constants, components, gain setting resistance R5, etc. :

23 1 19 3
k 1.38065 10 VAsK q 1.6021765 10 As T 300 K vi.nom 0.5 10 V

B20k 19980 Hz B1 1Hz h 1000Hz vo.nom 0.1V

3 3
R0 20 R1 1.82 10 R2 1.3 10 R3 R2
3 3
R4 20 10 R5 tbd R6 47.5 10 R7 R6
3
R8 2.2 10 R9 R8 R10 150 R11 R10
6 6
R12 2.2 10 R13 R12 R14 1 10 R15 R14
9 3 12
C1 1.0 10 F C2 1 10 F C3 C2 C4 100 10 F
9
C5 C4 C6 100 10 F C7 C6
12
C8 10 10 F C9 C8
11 Mathcad Worksheets Amp2 153

11.3 MCD-WS: Amp2 SN and Gain Calculations - 1/f-Noise Based Version

Page 2

NIe
NIe 30 [dB] 20 6 9
NI 10 10 NI 31.623 10

Gamp 200

R10 R11 R10 R11

Gamp 1 R5 R5 1.508
R5 Gamp 1

3
R5n 0.5 R5 R5n 753.769 10 R5p R5n

R10
Gn 1 Gp Gn Gn 200
R5n
R0n R0 0.5 R1n R1 0.5 R2n R2 R4n R4 0.5

R0p R0n R1p R1n R3p R3 R4p R4n

3 2
IC1.2 4.4 10 A hfe1.2 500 rbb1.2 VA.n 160V
2
IC3.4 IC1.2 hfe3.4 hfe1.2 rbb3.4 rbb1.2 VA.p 77V
R5 3
VDC.C1.2 R8 IC1.2 VDC.C1.2 9.68 V VDC.E1.2 IC1.2 VDC.E1.2 3.317 10 V
2
VDC.C3.4 R9 IC3.4 VDC.C3.4 9.68 V VDC.E3.4 VDC.E1.2

2. Calculation of the amp's input resistance Ri :

VA.n 3
rce1.2n rce1.2n 36.364 10
IC1.2
q IC1.2 3
gm1.2 gm1.2 170.199 10 S
kT
hfe1.2 3
rbe1.2 rbe1.2 2.938 10
gm1.2

1
1 1
Ri.1.2
R5n 1
1 1
rbe1.2 hfe1.2 R5n 1 gm1.2 R12 R6
R10 R5n R8 rce1.2n

3
Ri.1.2 49.182 10
VA.p 3
rce3.4 rce3.4 17.5 10
IC1.2
q IC3.4 3
gm3.4 gm3.4 170.199 10 S
kT
hfe3.4 3
rbe3.4 rbe3.4 2.938 10
gm3.4 R4p R4n
154 11 Mathcad Worksheets Amp2

11.3 MCD-WS: Amp2 SN and Gain Calculations - 1/f-Noise Based Version

Page 3

1
1 1
Ri.3.4
R5p 1
1 1
rbe3.4 hfe3.4 R5p 1 gm3.4 R13 R7
R11 R5p R9 rce3.4

3
Ri.3.4 49.064 10
1
1 1 3
Ri.tot.n Ri.tot.n 8.31 10
Ri.1.2 R4n

1
1 1 3
Ri.tot.p Ri.tot.p 8.307 10
Ri.3.4 R4p

1
1 1
Ri.tot 3
R4 Ri.1.2 Ri.3.4 Ri.tot 16.617 10

3. Evaluation of the impedance of the input network :

R0 0 1 100 f 10Hz 11Hz 20000 Hz

ZiL incl. R0 is the total noise-relevant input load, composed by Z iL.n + Z iL.p

1 1 1
1 1 1
ZiL.n( f R0) 0.5 R0 2j f C1 R1 ( R2 R3) 2j f C2 2j f C3

ZiL.p( f R0) ZiL.n( f R0) ZiL( f R0) ZiL.n( f R0) ZiL.p( f R0) ZiL.n h 20 9.818

50

40
30
[ohm]

ZiL f 20
20 Fig. 11.6
Impedance of the input
10 network ZiL(f)
0
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

Zi(f) is the frequency dependent input impe dance (~ Ri from Fig. 11.5), composed by Zi.n (f) + Zi.p (f)

1
1
1
1 1 1 1 1
Zi.n( f) 0.5 2j f C1 2j f C2 2j f C3
R1 R2 R3 Ri.tot

Zi.n( h) 502.885

Zi.p( f) Zi.n( f) Zi( f) Zi.n( f) Zi.p( f) Zi( h) 1005.771

11 Mathcad Worksheets Amp2 155

11.3 MCD-WS: Amp2 SN and Gain Calculations - 1/f-Noise Based Version

Page 4

1100
960
Zi.n( f)
820
[ohm]

Zi( f) 680 Fig. 11.7

540
Input impedances
Zi(f) & Zi.n (f)
400
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

4. Gain evaluations : Gain equations for T1+T2 or T3+T4 : see Chapter 20

VA.n 3
rce1.2 rce1.2 36.364 10
IC1.2

R5n
gm1.2 rce1.2 R8 hfe1.2
rce1.2
gm1.2.red
hfe1.2 rce1.2 R5n R8 gm1.2 R5n hfe1.2 rce1.2 rce1.2 R8

3
gm1.2.red 151.795 10 S

VA.p 3
rce3.4 rce3.4 17.5 10
IC3.4

R5p
gm3.4 rce3.4 R9 hfe3.4
rce3.4
gm3.4.red
hfe3.4 rce3.4 R5p R9 gm3.4 R5p hfe3.4 rce3.4 rce3.4 R9

3
gm3.4.red 152.747 10 S

gm1.2.red R6 1
G1.2
1 1 G1.2 301.678
1 R6
R8 rce1.2

G1.2.e 20 log G1.2 G1.2.e 49.591 [dB]

gm3.4.red R7 1
G3.4
1 1 G3.4 286.679
1 R7
R9 rce3.4

G3.4.e 20 log G3.4 G3.4.e 49.148 [dB]

156 11 Mathcad Worksheets Amp2

11.3 MCD-WS: Amp2 SN and Gain Calculations - 1/f-Noise Based Version

Page 5

3
G0.n gm1.2.red R12 G0.n 333.948 10

3
G0.p gm3.4.red R13 G0.p 336.043 10

5. Calculation of the relevant noise currents and voltages of the whole amp
(n refers to the T1+T2 part,p refers to the T3+T4 part) :

Fig. 11.8 Noise situation of Amp2, transferred into the u pper half of
the amp and into the lower half, and vice versa

5.1 BJT noise currents (for white-noise based calculations set x = 0 & y = 0) :

3 3
x 0.16965 fc.i1.2 33.75 10 Hz fc.e1.2 14.233 10 Hz

3 3
y 0.1492 fc.i3.4 55.65 10 Hz fc.e3.4 20.85 10 Hz

x
fc.i1.2 12
in.T.c1.2( f) 2 q IC1.2 B1 1 in.T.c1.2( h) 68.552 10 A
f

x
2 q IC1.2 fc.i1.2 12
in.T.b1.2( f) B1 1 in.T.b1.2( h) 3.066 10 A
hfe1.2 f

y
fc.i3.4 12
in.T.c3.4( f) 2 q IC3.4 B1 1 in.T.c3.4( h) 68.576 10 A
f

y
2 q IC3.4 fc.i3.4 12
in.T.b3.4( f) B1 1 in.T.b3.4( h) 3.067 10 A
hfe3.4 f
11 Mathcad Worksheets Amp2 157

11.3 MCD-WS: Amp2 SN and Gain Calculations - 1/f-Noise Based Version

Page 6

5.2 Noise voltages and currents of resistors :

4 k T B1 12
in.R4.n in.R4.p in.R4.n in.R4.n 1.287 10 A
R4n

4 k T B1 12
in.R5.n in.R5.p in.R5.n in.R5.n 148.256 10 A
R5n

4 k T B1 15
in.R6 in.R7 in.R6 in.R6 590.589 10 A
R6
4 k T B1 15
in.R12 in.R12 86.78 10 A
R12
4 k T B1 15
in.R13 in.R13 86.78 10 A
R12
NIe
10 12 2
10 10 IC1.2 R8 B1 9
eNex.R8( f) eNex.R8( h) 6.379 10 V
ln( 10) f

9
eN.R8 4 k T R8 B20k eN.R8 853.378 10 V

2 2 9
eN.R8.eff( f) eNex.R8( f) eN.R8 eN.R8.eff( h) 853.402 10 V

B1 9
en.R8.eff( f) eN.R8.eff( f) en.R9.eff( f) en.R8.eff( f) en.R8.eff( h) 6.037 10 V
B20k
en.R8.eff( f) 12
in.R8( f) in.R9( f) in.R8( f) in.R8( h) 2.744 10 A
R8
9
en.R14 4 k T B1 R14 en.R15 en.R14 en.R14 128.716 10 V

12
en.Z.iL( f R0) 4 k T B1 ZiL( f R0) en.Z.iL h 20 570.371 10 V

12
en.R0( R0) 4 k T B1 R0 en.R0 20 575.635 10 V
1
1 1
RBE1.2 rbb1.2 RBE1.2 1.75
R5n R10
1
1 1
RBE3.4 rbb3.4 RBE3.4 1.75
R5p R11

12
en.R.BE1.2 4 k T B1 RBE1.2 en.R.BE1.2 170.275 10 V

12
en.R.BE3.4 4 k T B1 RBE3.4 en.R.BE3.4 170.275 10 V
158 11 Mathcad Worksheets Amp2

11.3 MCD-WS: Amp2 SN and Gain Calculations - 1/f-Noise Based Version

Page 7

5.3 Noise of op-amps :

3
fc.e1 6Hz fc.i1 8Hz fc.e3 0.1 10 Hz

9 fc.e1
en.i.op1 4 10 V en.i.op1( f) en.i.op1 1
f

12 fc.i1
in.i.op1 0.4 10 A in.i.op1( f) in.i.op1 1
f

en.i.op2( f) en.i.op1( f) in.i.op2( f) in.i.op1( f)

9 fc.e3
en.i.op3 25 10 V en.i.op3( f) en.i.op3 1
f
en.i.op4( f) en.i.op3( f)

12
in.i.op3 0.01 10 A in.i.op4 in.i.op3

2 2
1 1 2
2j f C6 2 2j f C6 1
en.o.op3 ( f) en.R14 en.i.op3( f) 1 in.i.op3 2j f C6
R14 R14

9
en.o.op4 ( f) en.o.op3 ( f) en.o.op3 ( h) 26.219 10 V

5.4 Input referred noise voltages of BJTs and Amp2 input :

2
in.T.c1.2( f) 12
en.i.T1.2( f) 4 k T rbb1.2 B1 en.i.T1.2( h) 422.841 10 V
2
gm1.2

2
in.T.c3.4( f) 12
en.i.T3.4( f) 4 k T rbb3.4 B1 en.i.T3.4( h) 422.978 10 V
2
gm3.4

2
in.T.c1.2( f) 2 2 2 12
en.i.n( f) en.R.BE1.2 in.T.b1.2( f) RBE1.2 en.i.n( h) 437.321 10 V
2
gm1.2

2
in.T.c3.4( h) 2 2 2
en.i.p( f) en.R.BE3.4 in.T.b3.4( h) RBE3.4 12
2 en.i.p( h) 437.452 10 V
gm3.4

2 2 2 12
in.i.n( f) in.T.b1.2( f) in.R6 in.R4.n in.i.n( h) 3.377 10 A

2 2 2 12
in.i.p( f) in.T.b3.4( f) in.R7 in.R4.p in.i.p( h) 3.378 10 A
11 Mathcad Worksheets Amp2 159

11.3 MCD-WS: Amp2 SN and Gain Calculations - 1/f-Noise Based Version

Page 8

Note : Because of the low DC voltage across R5+P1 (appr. 3 ...45 mV) any generated R5+P1-excess noise can fully be ignored.

Fig. 11.9 = Fig. 10.14

2 2
2 en.i.op1( f) en.o.op3 ( f)
en.i.amp.n( f) en.i.n( f)
2
G1.2
2 2 2 12
in.R8( f) in.i.op1( f) in.R12 en.i.amp.n( h) 446.369 10 V

2
gm1.2

2 2
2 en.i.op2( f) en.o.op4 ( f)
en.i.amp.p( f) en.i.p( f)
2
G3.4
2 2 2 12
in.R9( f) in.i.op1( f) in.R13 en.i.amp.p( h) 447.427 10 V

2
gm3.4

5.5 Amp2's total input and output referred noise voltages incl. R0 :

2 2 12
en.i.amp( f) en.i.amp.n( f) en.i.amp.p( f) en.i.amp( h) 632.01 10 V

9
en.o.amp( f) en.i.amp( f) Gamp en.o.amp( h) 126.402 10 V

1
1 1 12
in.i.amp( f) in.i.amp( h) 2.388 10 A
2 2
in.i.p( f) in.i.n( f)

2 2 2 2
en.i.amp( f R0) en.i.amp( f) in.i.amp( f) ZiL( f R0) en.Z.iL( f R0)
160 11 Mathcad Worksheets Amp2

11.3 MCD-WS: Amp2 SN and Gain Calculations - 1/f-Noise Based Version

Page 9

12
en.i.amp h 20 852.619 10 V

6 12
en.i.amp h 10 636.169 10 V

9
en.o.amp( f R0) en.i.amp( f R0) Gamp en.o.amp h 20 170.524 10 V

3
f 10Hz 20Hz 20 10 Hz R0 0 1 100

8
1 10

en.i.amp( f)
[V/rtHz]

Fig. 11.10
en.R0 20 1 10
9
Input referred noise
en.i.amp f 20 voltage densities vs.
frequency based on
two different input loads
10
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

8
1 10

en.i.amp( h R0)
[V/rtHz]

9
1 10 Fig. 11.11
en.i.amp( 20Hz R0)
Amp2 equivalent input
noise voltage densities vs.
R0 based on two
10
1 10 different frequencies
0 20 40 60 80 100
R0
[ohm]

8
1 10

Fig. 11.12
6 Amp2 equivalent input
[V/rtHz]

en.i.amp f 10
9 noise voltage densities vs.
1 10 frequency showing
en.i.amp f 20
1/f-noise based on two
different input loads

10
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]
11 Mathcad Worksheets Amp2 161

11.3 MCD-WS: Amp2 SN and Gain Calculations - 1/f-Noise Based Version

Page 10

6. NF calculation :

2 2 2 2
en.Z.iL( h R0) in.i.amp( h) ZiL( h R0) en.i.amp( h R0)
NFe ( R0) 20 log NFe 20 5.107 [dB]
en.Z.iL( h R0)

100

Fig. 11.13
[dB]

NFe ( R0) 10
Noise Figure
vs. R0

1
1 10 100
R0
[ohm]

7. Evaluation of SNs with referrence to B20k and 100mVrms /1kHz nominal output voltage:

20000Hz
1 2
en.o.amp( f R0) df
B1
20Hz
SNne( R0) 20 log
vo.nom SNne 20 72.801 [dB]

SNne.m 20 73.13 [dB]

6
SNne 10 75.768 [dB]
6
SNne.m 10 75.37 [dB]

A-weighting by A(f) :

1.259 1 1 1
A( f)
2 2 2 2
20.6Hz 107.7Hz 737.9Hz f
1 1 1 1
f f f 12200 Hz

20000Hz
1 2 2
en.o.amp( f R0) A( f) df
B1
20Hz
SNne.a( R0) 20 log
vo.nom

SNne.a 20 74.797 [dB(A)]

SNne.a.m 20 75.22 [dB(A)]

162 11 Mathcad Worksheets Amp2

11.3 MCD-WS: Amp2 SN and Gain Calculations - 1/f-Noise Based Version

Page 11

RIAA equalizing by R(f) :

1
2
3 6
1 2 10 Hz 318 10 s
R1000
2 2 R1000 9.898
3 6 3 6
1 2 10 Hz 3180 10 s 1 2 10 Hz 75 10 s

2
6
1 2 f 318 10 s
R( f) R1000
2 2
6 6
1 2 f 3180 10 s 1 2 f 75 10 s

20000Hz
1 2 2
en.o.amp( f R0) R( f) df
B1
20Hz
SNriaa( R0) 20 log
vo.nom

SNriaa 20 75.054 [dB]

20000Hz
1 2 2 2
en.o.amp( f R0) A( f) R( f) df
B1
20Hz
SNariaa( R0) 20 log
vo.nom

Calculated and measured (.m) results :

6
SN ariaa 10 83.095 [dB(A)] SNariaa 20 80.431 [dB(A)]
6
SN ariaa.m 10 81.60 [dB(A)] SNariaa.m 20 79.80 [dB(A)]

SN ariaa 5 82.283 [dB(A)] SNariaa 43 78.612 [dB(A)]

SN ariaa.m 5 81.20 [dB(A)] SNariaa.m 43 78.40 [dB(A)]

SN ariaa 10 81.571 [dB(A)]

SN ariaa.m 10 80.70 [dB(A)]

11 Mathcad Worksheets Amp2 163

11.3 MCD-WS: Amp2 SN and Gain Calculations - 1/f-Noise Based Version

Page 12

3
f 20Hz 25Hz 20 10 Hz

5
1 10

6
1 10
en.o.amp f 20

en.o.amp f 20 A ( f)
[V/rtHz]

7
1 10
en.o.amp f 20 R ( f)

en.o.amp f 20 R ( f) A ( f)
8
1 10

9
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

Fig. 11.14 Amp2 output noise voltage densities vs. frequency

RIAA-equalized + A-weighted + un-weighted

66
68
[ dB ref. 0.5mVrms ]

70
72
SNne( R0) 74
76
SNariaa ( R0)
78
80
82
84
86
0 10 20 30 40 50 60 70 80 90 100
R0
[ohm]

Fig. 11.15 Input referred un-weighted + A-weighted & RIAA-equalized SN vs. R0

Engine II Performance
12

12.1 Visible and Audible Effects

12.1.1 Visible Effects

Before we dive deeper into real listening matters let’s have a look at the worst
looking frequency and phase responses, generated by the left channel of Engine II
and Amp1 & Amp2 via Amp3 only and Amp5 (incl. output transformer).
Figure 12.1a shows the measured curves. They are flat according to the goals set in
Sect. 1.2. However, there is one exception: the low-frequency response of Amp1. It
comes from the deviation already mentioned in Sect. 8.3, Fig. 8.6, and in con-
junction with the UBC1 output resistance of 10.8 Ω.
To demonstrate the difference Fig. 12.1b shows the curves of the sequence
Amp1 via Amp3 & Amp4 and Amp5 + Trafo, however fed directly by the RIAA
encoder’s output with an output resistance of approximately 0 Ω. The curves
become much more flat.
In Fig. 12.1a the Amp1 curves are the ones that show the biggest deviations from
the flatness at 20 Hz and 20 kHz. The frequency responses are located around
0dBV/1 kHz (left ordinate) and the phase responses around 0°/1 kHz (right ordi-
nate). The Amp2 generated curves look very flat.
In Fig. 12.1b the two frequency responses and the two phase responses are
hidden by each other. At the low-end of the audio band, they look very much better
than the Amp1 ones of Fig. 12.1a. Further down Table 12.4 gives the detailed
mesurement results.
Figure 12.2 shows the output noise trace of the chain Amp1 + Amp4 +
Amp5 + Trafo, input loaded by a 20 Ω metal film resistor. Based on a typical
3-pham concept this trace should very closely follow the RIAA transfer function.
Definitely not created by the engine in this graph we find a 150 Hz spike. It
appears in many other ones of my measurement graphs too. I always thought that

1
See un-balanced-to-balanced converter in Sect. 15.2.

© Springer International Publishing Switzerland 2016 165

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_12
166 12 Engine II Performance

Fig. 12.1 a Frequency and phase response of the left channel’s Amps 1 & 2 via Amp3 and
Amp5 + Trafo, fed by a generator output resistance of 10.8 Ω. b Frequency and phase response of
the left channel’s Amp1 via Amp3 and Amp4 and Amp5 + Trafo, fed by a generator output
resistance of appr. 0 Ω
12.1 Visible and Audible Effects 167

Fig. 12.2 Engine II output noise voltage density curve of the left channel, input loaded with 20 Ω
and Amp1 + Amp4 + Amp5 + Trafo

one of the many running PSU devices in my lab could generate it. Together with a
specialist of the local mains supplier we could identify where it comes from. Only
15 m away from my lab there is a transformer house with a 10 kV/230 V/50 Hz
transformer. As long as it runs with full power, the spikes will disappear. In all
other cases they will come and go—you never know the rhythm.
I guess, with these frequency and phase response results in conjunction with the
amp’s rather low noise production the listening tests will became a real challenge.

12.1.2 Audible Effects

Studying the above shown graphs will lead to two realizations: it will be hard to
hear noise because the majority of the shown noise voltage density values becomes
<−120 dBV, and, despite the small deviation from Amp1’s frequency response
flatness <100 Hz it will also be hard to identify differences in sound between Amp1
and Amp2.
In the various listening tests, I could go through many different test arrangements
with very different source material as well as different loudspeaker and amplifier
chains. The basic procedure became always the same: for each loudspeaker situa-
tion, the source material had to go through all Engine II offered switching
168 12 Engine II Performance

possibilities. I used the following test records, test noise, headphone, and loud-
speaker arrangements in two different rooms, my 16 m2 lab and my 30 m2 library.

12.1.3 Test Records

Piano 1. “The Köln Concert”, Keith Jarrett, ECM 1064/65ST

Live grand piano
2. “Piano”, Martin Vatter, www.martin.vatter.de, DMM by Pauler
Acoustics
Studio grand piano play and direct string treatment by fingers and
other measures
Concert 1. “Piano Concert No. 1 in B flat minor”, Tchaikovsky
Svatoslav Richter/Herbert von Karajan/Vienna Symphony Orchestra
Deutsche Grammophon 138822
Recorded at the Vienna Musikverein building with its stunning
acoustic
2. “Moonlight Sonata”, L.v. Beethoven,
Wilhelm Kempff, Piano, Deutsche Grammophon LPE 17026,
Mono record
Pop etc. “Saitensprung”, Friedemann, Biber Records, www.in-akustik.com
Plugged and un-plugged percussion and string instruments: guitar &
harp
Electronic “Minimum–Maximum”, Kraftwerk, EMI 0946 3 11828 1 5
Live electronic music at its best; includes Autobahn, Tour de France,
etc.

12.1.4 Test Noise and Source Equipment

Pink Via Pink noise generator,2 RIAA encoder,3 and

un-balanced to balanced converter4
Third octave & Bark 45 rpm Deutsche Grammophon test record TM 10 99 109.
It offers wobbling frequencies in 27 third octave bands
format as well as in 24 Bark bandwidth format
Turntable Direct driven Pioneer PL-L1000 with tangential tone arm,
modified for fully balanced operation
MC Denon DL-103 No. 2798

2
TSOS-1 Chap. 10, TSOS-2 Chap. 22.
3
TSOS-1 Chap. 11, TSOS-2 Chap. 23.
4
UBC—see Sect. 15.2.
12.1 Visible and Audible Effects 169

12.1.5 Loudspeaker Situation and Headphones

The direct driven lab loudspeakers are located alike the monitoring loudspeakers in
a mastering studio (distance LS-ear: max. 1.5 m). Additionally, a center bass unit
works down to 15 Hz. It is an improved version of Russel Breden’s subwoofer.5 An
active crossover and various kinds of power amps control them all in a 2.1 basic
arrangement.
The library room’s loudspeaker situation is very much different; however, here I
use a 2.1 basic arrangement too. A pair of two small bookshelf loudspeakers can be
switched into the arrangement as well as a pair of modified BRAUN LE1 elec-
trostatic loudspeakers,6,7 (distance LS-ear: max. 2.5 m). They work in the frequency
range >100 Hz, filtered by 24 dB Linkwitz/Riley lps and hps. An active KEF B139
in a closed box plays the bass part here.
The head/ear-phones are: AKG 271 Mk II and B&O earphones.

12.1.6 Listening Tests

1st Round
No matter which of the eight different amplifier and output transformer alternatives
I’ve been choosing I couldn’t hear any difference when playing one of the records
from Sect. 12.1.3. I thought that one of the few and rather old mono records I found
in my collection could be of additional value because its higher frequencies were all
gone by many listening sessions: no chance! Even with headphones, there was no
indication of the tiniest difference.
2nd Round
Here I thought that I had to force the Engine II into a worse situation by changing
the actual t1–t2 double-triode of the right channel. I took a non-selected old
Siemens ECC88 double-triode that has very low noise levels but very different
triode characteristics. The SN values do not change but the THD and IMD values
became worse. THD = 0.043 %, IMD = 0.26 % (worst case).
Because of the unequal gains of the triodes, the only additional difference I could
measure became the fact that the overload margin became worse too.
Like in the first round, from a listening point of view I could not identify any
differences.

5
“Roaring subwoofer”, Russel Breden, Electronics (Wireless) World 02-1997, p. 104ff.
6
“Improved Electrostatic Loudspeaker Power Supply for QUAD ESL57 and BRAUN LE1”, B.
Vogel, Electronics (Wireless) World, 02-2006.
7
“Abenteuer LE1—Restaurierung der Prof. Rams’schen Design-Legende” “Adventure LE1—
Restoration of Prof. Rams’ design legend”, B. Vogel, German and English version, Design + Design
Nr. 32, 1995, Jo Klatt Design + Design Verlag Hamburg, www.design-und-design.de.
170 12 Engine II Performance

3rd Round
Two years ago, I was coaching a young student at Prof. Seelmann’s department of
Electronics & Information Technology of the Aalen University, Germany. To get
the bachelor degree he was working on a test arrangement with four different kinds
of feedback mechanisms in audio pre-amplifiers. He designed four different FET
configured three-stage pre-amplifiers,8 according to an idea I’ve been presenting in
HTGG-1.9 The aim was to identify differences in sound and to find corresponding
explanations, alike the ones Peter Schüller described in the MST, the Mayer-
Schüller-Theory.10
In an anechoic chamber, we have been listening to a broad range of source
material. It was very hard to identify the differences between the four different
feedback situations. Only one test signal allowed a clear identification of differ-
ences: pink noise.
Therefore, the “killing” test signal for the Engine II should become pink noise
via RIAA encoder and an un-balanced to balanced converter. Here, at the first time,
I could identify tiny differences, but only when switching the two input amps. There
were no differences audible when switching the two central amps 3 and 4 nor when
switching the output transformer in and out. Unfortunately, even after this proce-
dure I cannot say what the best situation for the user could be. The differences are
simply too small.
A further attack on the amp chains with wobbling frequencies followed. Again,
clear differences could not be detected.
4th Round
I thought, maybe my wife and some friends and sound specialists could help and
identify differences. However, they obtained the same listening results.

12.2 Measurement Results

12.2.1 Noise

This section explains my personal interpretation why I did not identify audible
differences between the various amplifier sequences. Two additional graphs show
the outstanding low noise voltage generation by the Engine. Practically not con-
trasting Fig. 12.2. Figure 12.3 shows the Amp2 + Amp3 + Amp5 + Trafo chain’s

8
“Audio Amplifier Feedback beats Feedback-free”, Florian Ermer, despite the English title the text
is written in German and not translated yet, www.audioexperts.de/audio-technik/554-
gegenkopplung-schlaegt-gegenkopplungsfrei.html.
9
HTGG-1, Chap. 15.
10
“Maier-Schüller-Theory MST”: see my letter to the editor of Linear Audio Vol 4-1 (www.
linearaudio.net/) and the PDF document on http://Burosch.de/audio-technik/509-high-end-2012-
klang-2-english.html.
12.2 Measurement Results 171

Fig. 12.3 Engine II output noise voltage density curve of the left channel, input loaded with 20 Ω
and Amp2 + Amp3 + Amp5 + Trafo

noise voltage density curve, also loaded with 20 Ω at its input. However, here
comes Amp2 into the game, and the triode driven Amp3 replaces Amp4.
With input shorted via external input another graph, Fig. 12.4, demonstrates the
noise performance of the chain Amp3 + Amp5 + Trafo. Now, there is no effect
>1 kHz from the 75 µs time constant because most of this chain’s noise is created
after the time constant creating section. The same applies to Amp4 instead of
Amp3.
At 1 kHz the difference of the two curves becomes approximately 20 dB. This is
roughly the same difference of the measured SNs too: Fig. 12.3: −79.9 dB(A),
Fig. 12.4: −99.0 dB(A).

12.2.2 THD and IMD Matters, Left Channel

The following graphs show the distortion effects of a 1 kHz test signal of the left
channel only. I have also gone through other test signal frequencies below and
above 1 kHz. However, the general impression of their spikes ≥d2 did not change.
Therefore, here, I will concentrate on the 1 kHz effects only. The horizontal line in
each graph represents the RIAA equalized SNriaa of −67.3 dB. No distortion spike
172 12 Engine II Performance

Fig. 12.4 Engine II output noise voltage density curve of the left channel, external input shorted
and Amp3 + Amp5 + Trafo

should cross this line. Otherwise, it becomes noticeable among the mixed noise of
the phono-amp and the LP.
I refrain from discussing the before expressed claim about the audibility of
distortion spike levels and the level difference between the spike and the −67.3 dB
line. However, my personal experience has led to the described conclusion, after the
findings of Sect. 15.2 (masking of harmonics by noise) and after many listening
tests via loudspeakers and headphones fed by white noise in B20k and mixed-up
with single sinus tones between 50 Hz and 1 kHz.
To create the 1 kHz/0 dBV output signal in Fig. 12.5 I’ve fed the balanced
1 kHz/0.00159 % generator signal11 into the triode amp chain Amp1 + Amp3 +
Amp5 + Trafo. The d2 and d3 generator levels lie at −97.662 and −114.571 dBV,
the corresponding levels of the graph show levels of −81.672 and −85.704 dBV.
Transferred into distortion percentages that means 0.0083 and 0052 %. Total dis-
tortion in B20k (THD) measures 0.010 % and the level decrease with increasing
harmonics order follows the findings of the MST.12 Theoretically, this path should
sound superb.

11
See Sect. 15.2.
12
See footnote 9.
12.2 Measurement Results 173

In contrast to Fig. 12.5 we have in Fig. 12.6 the Amp3 triode path replaced by
Amp4 with op-amps. The 1 kHz test-signal changes its d2/d3 output values to
−99.131 and −94.738 dBV, both distortion values now far below the Fig. 12.5
ones, leading to an overall THD of 0.002 %. However, it does not matter, this amp
sequence did not sound different to the Fig. 12.5 chain.
Total THD of the Fig. 12.7 amp chain with Amp2 and triode gain stage
(Amp2 + Amp3 + Amp5 + Trafo) measures 0.013 %, the d2/d3 levels become
−78.388 dBV and −86.580 dBV. Because of the triode’s tendency to create
stronger equal harmonics the Fig. 12.7 harmonics look similar to the ones of
Fig. 12.5; d2 does not affect the sound of this sequence negatively, and d3 also fully
disappears in the noise, hence it has no sound disturbing effect. That is why we
could not hear any difference to the other amp sequence alternatives.
Now, in Fig. 12.8, things become even stronger after replacement of Amp3 by
Amp4, because the d2/d3 spikes (−86.376 dBV and −94.327 dBV, THD =
0.005 %) also fully disappear in the unweighted output referred SNriaa of −67.3
dBV best-case LP noise. Again, differences to the other amp sequences became
inaudible.

Fig. 12.5 Left channel’s Amp1 + Amp3 + Amp5 + Trafo distortion measurement result of a
1 kHz signal via the central triode path
174 12 Engine II Performance

Fig. 12.6 Left channel’s Amp1 + Amp4 + Amp5 + Trafo distortion measurement result of a
1 kHz signal via the central op-amp path

Fig. 12.7 Same as Fig. 12.5, however, Amp1 is replaced by Amp2

12.2 Measurement Results 175

Fig. 12.8 Same as Fig. 12.6, however, Amp3 is replaced by Amp4

12.2.3 THD and IMD Matters, Right Channel

Concerning IMD, the measurement results become always <0.01 %. ‘Always’

because I have measured all amp sequence possibilities with different frequency
pairs with levels set to 20 %/80 %, such as 8 kHz/250 Hz (DIN), 3 kHz/60 Hz, and
3 kHz/300 Hz.
In the left channel the Amp3’s first E88CC (t1 & t2) is a selected one from JJ
with perfectly matched operating conditions at 2 mA/90 V. For right-channel test
purposes, I’ve chosen a non-selected but low-noise Siemens NOS ECC88 with
rather different operating conditions of its two triode systems. However, P5 of
Fig. 2.1 allows trimming to equal gains of t1 & t2.
Compared with the ones of Figs. 12.5 and 12.7 the THD and IMD measurement
results look worse: THD = 0.043 % (Amp1) and 0.055 % (Amp2), IMD < 0.01 %
for 8 kHz/250 Hz but up to 0.026 % with the other two frequency pairs.
Only the before mentioned pink-noise test signal (in Sect. 12.1.4) could generate
sound differences between the left and the right channel. Music material did not
produce audible differences.
176 12 Engine II Performance

12.2.4 General THD and IMD Matters

Obviously, the chosen 3-pham concept is a favourable approach to decrease THD

and IMD. The 75 µs lp filter, with increasing frequency more and more, suppresses
distortion and intermodulation artefacts created by the high-gain input stages. The
318 µs + 3180 µs arrangement creates a further step up to 1/10.

12.2.5 Summary Tables and Graphs

I’ve summed-up the measured findings in four different tables: Table 12.1 gives
relevant SNs of different amp sequences, Tables 12.2 and 12.3 play the THD &
IMD part, and Table 12.4 shows the generator output resistance dependency of the
Amps 1 and 2 frequency and phase responses.
Table 12.1 needs some explanations:
1. Shown in Figs. 12.2 and 12.3, the noise curve of RIAA equalized white noise
versus frequency looks like the RIAA transfer curve versus frequency. Thus,
when talking about the corresponding SN I always use the subscript ‘riaa’; in
case of an additional A-weighting, the subscript changes to ‘ariaa’. The here used
terms SNo (not A-weighted output referred SN) and SNo.a (A-weighted output
referred SN) indicate something between fully RIAA equalized and half
equalized, because with input of Amps 3 and 4 shorted, the 75 µs has practically
no effect on the output SN. Hence, the noise curve has no decrease above 1 kHz
à la Fig. 12.4 and the RIAA effect comes from the 318 µs/3180 µs network
below 1kHz only.
2. With the exception of the reference level, the input referred SNs equal the output
referred ones. Automatically, with a gain of 2000 an input referred SNi refer-
enced to 0.5 mVrms/1 kHz equals an output referred SNo referenced to 1 Vrms/
1 kHz. Or, in the case of Amps 3 or 4 plus 5 plus trafo the input reference level
is 100 mVrms and the gain is 10. Thus, the output referred SNo becomes the
input referred SNi too.
3. The same input SN versus output SN mechanism applies to the Sect. 13.3
example amp with a gain of 1000, leading to an overall gain of 10,000. The
input reference level becomes thus 0.1 mVrms/1 kHz.
Note 1: Table 12.4 colours are the colours of Figs. 12.9, 12.10 and 12.11.
Note 2: Further adjustment of Fig. 1.2’s P3 and an increase of the capacitance
values of Fig. 2.1’s C44a–C45b will improve the Amp3 deltas in boxes H-5, H-7,
and H-17 to ≤(±0.1 dB). However, my RIAA encoder’s13 exactness is ±0.1 dB
only!
Note 3: Concerning Amp2 there is no difference to the Table 12.4 values when
fed by the RIAA encoder directly (0 Ω).

13
See Sect. 14.1.
12.2 Measurement Results 177

Table 12.1 Relevant Engine II SN results

1/A B C D E F
2 L channel R channel
3 Amp sequence measured SN.i.ariaa SN.i.ariaa
4 Ref. 0.5 mV/1 kHz/20R
5 dB(A) dB(A)
6 Amp1 + Amp3 + Amp5 + o/p-trafo −81.71 −81.86
7 Amp1 + Amp4 + Amp5 + o/p-trafo −81.78 −81.66
8 Amp2 + Amp3 + Amp5 + o/p-trafo −79.91 −80.21
9 Amp2 + Amp4 + Amp5 + o/p-trafo −79.83 −80.01
10 SN.o SN.o.a SN.o SN.o.a
11 Ref. 1 V/1 kHz/0R
12 dBV dBV(A) dBV dBV(A)
13 Amp3 + Amp5 + o/p-trafo −90.43 −99.00 −90.43 −99.00
14 Amp4 + Amp5 + o/p-trafo −98.06 −100.82 −98.06 −100.82
15 Amp5 + o/p-trafo −101.73 −103.50 −101.73 −103.90
16 Amp5 single ended −101.30 −102.90 −101.30 −102.90
17 SN.i SN.i.a SN.i SN.i.a
18 Ref. 0.5 mV/1 kHz/20R
19 dB dB(A) dB dB(A)
20 Amp1 −74.00 −76.00 −74.00 −76.00
21 Amp2 −73.13 −75.22 −73.37 −75.34
22
23 left & right channel & amp sequence SN.i.ariaa Deltas
24 calculated Ref. 0.5 mV/1 L.meas - calc
kHz/20R
25 dB(A) dB
26 Amp1(tr = re) + Amp3 −80.89 −0.82
(ds/1 k) + Amp5 + o/p-trafo
27 Amp1(tr = re) + Amp4 + Amp5 + o/p-trafo −80.89 −0.89
28 Amp2 + Amp3 + Amp5 + o/p-trafo −80.43 0.52
29 Amp2 + Amp4 + Amp5 + o/p-trafo −80.43 0.60
30 SN.o SN.o.a
31 Ref. 1 V/1 kHz/0R
32 dBV dBV(A)
33 Amp3(ds/1 k) + Amp5 + o/p-trafo −92.87 −100.40 2.44 1.40
34 Amp4 + Amp5 + o/p-trafo −96.11 −99.94 −1.95 −0.88
35 Amp5 + o/p-trafo −101.96 −104.01 0.23 0.51
36 Amp5 single ended −101.24 −103.29 −0.06 0.39
(continued)
178 12 Engine II Performance

Table 12.1 (continued)

1/A B C D E F
37 SN.i SN.i.a
38 Ref. 0.5 mV/
1 kHz/20R
39 dB dB(A)
40 Amp1 −72.62 −74.67 −1.38 −1.33
41 Amp2 −72.80 −74.80 −0.33 −0.42
ds/1 k: data sheet figures and f.c = 1 kHz
tr = re: real data (tr = 11) for the i/p trafo

Table 12.2 Relevant Engine II THD figures

1/A B C D
2 Left channel via ext. i/p measured ref. i/p:100 mV + o/p:1 V o/p trafo
3 Incl. Excl.
4 %
5 Amp3 + Amp5 0.011 0.010
6 Amp4 + Amp5 0.002 0.001
7
8 Left channel via Amp1&2 measured ref. i/p:0.5 mV + o/p:1 V o/p trafo
9 Incl. Excl.
10 %
11 Amp1 + Amp3 + Amp5 0.010 0.010
12 Amp1 + Amp4 + Amp5 0.002 0.002
13 Amp2 + Amp3 + Amp5 0.013 0.013
14 Amp2 + Amp4 + Amp5 0.005 0.005
15
16 Right channel via ext. i/p measured ref. i/p:100 mV + o/p:1 V o/p trafo
17 Incl. Excl.
18 %
19 Amp3 + Amp5 0.040 0.039
20 Amp4 + Amp5 0.002 0.002
21
22 Right channel via Amp1&2 measured ref. i/p:0.5 mV + o/p:1 V o/p trafo
23 Incl. Excl.
24 %
25 Amp1 + Amp3 + Amp5 0.043 0.043
26 Amp1 + Amp4 + Amp5 0.002 0.002
27 Amp2 + Amp3 + Amp5 0.055 0.055
28 Amp2 + Amp4 + Amp5 0.014 0.014
12.2 Measurement Results 179

Table 12.3 Relevant Engine II IMD figures

1/A B C D
2 Left channel via Amp1&2 measured ref. i/p:0.5 mV + o/p:1 V o/p trafo
3 250 Hz/8 kHz + 80 %/20 % Incl. Excl.
4 %
5
6 Amp1 + Amp3 + Amp5 0.004 0.004
7 Amp1 + Amp4 + Amp5 0.002 0.002
8 Amp2 + Amp3 + Amp5 0.007 0.007
9 Amp2 + Amp4 + Amp5 0.004 0.004
10
11 Left channel via Amp1&2 measured ref. i/p:0.5 mV + o/p:1 V o/p trafo
12 Other frequency pairs Incl. Excl.
13 %
14
15 Amp1 + Amp3 + Amp5: 60 Hz/3 kHz + 80 %/20 % 0.006 0.005
16 Amp2 + Amp3 + Amp5: 300 Hz/3 kHz + 80 %/20 % 0.008 0.008
17 Amp2 + Amp3 + Amp5: 60 Hz/3 kHz + 80 %/20 % 0.007 0.006
18
19 Right channel via Amp1&2 measured ref. i/p:0.5 mV + o/p:1 V o/p trafo
20 250 Hz/8 kHz + 80 %/20 % Incl. Excl.
21 %
22
23 Amp1 + Amp3 + Amp5 0.008 0.006
24 Amp1 + Amp4 + Amp5 0.006 0.001
25 Amp2 + Amp3 + Amp5 0.010 0.009
26 Amp2 + Amp4 + Amp5 0.008 0.003
27
28 Right channel via Amp1&2 measured ref. i/p:0.5 mV + o/p:1 V o/p trafo
29 Other frequency pairs Incl. Excl.
30 %
31
32 Amp1 + Amp3 + Amp5: 60 Hz/3 kHz + 80 %/20 % 0.012 0.003
33 Amp2 + Amp3 + Amp5: 300 Hz/3 kHz + 80 %/20 % 0.026 0.026
34 Amp2 + Amp3 + Amp5: 60 Hz/3 kHz + 80 %/20 % 0.018 0.009
180 12 Engine II Performance

Table 12.4 Frequency and phase response measurement results of Amp1 and Amp2, fed by
different generator output resistances
1/A B C D E F G H I
2 Source Colours Unit 20 Hz 1 kHz 20 kHz Deltas
3 L-channel F–EF–G
4 Amp1 fed directly by RIAA encoder with Ro = 0R
5 +Amp3 wht [dBV] 0.544 0.817 0.793 0.273 0.024
6 +Amp3 mag [°] 10.81 −0.21 −8.63 −11.02 8.42
7 +Amp4 red [dBV] 0.334 0.463 0.478 0.129 −0.015
8 +Amp4 yel [°] 10.62 −0.21 −9.26 −10.83 9.05
9
10 Amp1 fed by RIAA encoder via un-bal/bal-converter with Ro = 10R8
11 +Amp3 wht [dBV] −0.916 0.285 0.256 1.201 0.029
12 +Amp3 mag [°] 29.95 0.15 −10.85 −29.8 11
13 +Amp4 red [dBV] −0.980 0.257 0.221 1.237 0.036
14 +Amp4 yel [°] 30.14 0.12 −11.37 −30.02 11.49
15
16 Amp2 fed by RIAA encoder via un-bal/bal-converter with Ro = 10R8
17 +Amp3 wht [dBV] 0.388 0.515 0.484 0.127 0.031
18 +Amp3 mag [°] 5.58 −0.16 −6.05 −5.74 5.89
19 +Amp4 red [dBV] 0.378 0.440 0.470 0.062 −0.03
20 +Amp4 yel [°] 5.75 −0.16 −5.49 −5.91 5.33

12.3 Conclusions and Final Remarks

Generally, it became a big surprise that by listening to music material my wife, my

friends and I could identify neither any big differences nor rather tiny ones (only
exception: with pink noise). However, taking into account the fact that many
microphone amps produce a certain amount of THD and IMD, mostly not as good
as the Engine II figures, and, additionally, power amps and loudspeakers pack
another portion of distortion and intermodulation on the complete signal mix, then,
we might have an additional explanation for not being far away from the observed
listening findings.
Provided that the whole amp chain works with the nominal signal levels between
all stages and with the volume knob setting at 14:00 = 2.00 pm in 1 m distance from
the loudspeakers there is no audible phono-amplifier generated noise and hum with
the DL-103 at the input. My conclusions can thus be summed-up by the following
statements:
1. It seems to become a hard job to design different sounding solid-state or triode
driven RIAA phono-amps with passive networks, as long as there is no devi-
ation from a flat frequency and phase response in B20k inside the goals from the
12.3 Conclusions and Final Remarks 181

Fig. 12.9 Frequency and phase responses of Amp1 via Amps 3 & 4, fed by a generator with an
output resistance of 0 Ω

beginning, a slew rate always >1.8 V/µs, 1 kHz THD < 0.05 %, and
IMD < 0.01 %.
2. Even the tiny input transformer created −1.3 dB at 20 Hz does not create
audible impacts. In addition, other influential factors suppress negative sound
impacts caused by any kind of harmonic artefacts too. Here, I’m thinking of the
disappearance of these artefacts in the noise level of vinyl records (best case
representation by a fictitious horizontal line in the shown THD diagrams at
SNriaa = −67.3 dB unweighted14), and by putting the 75 µs lp at the output of
the amp chain’s biggest THD and IMD creators, the non-equalized input Amps
1 & 2.
3. Indeed, through disturbing the flatness of the frequency and phase response by
playing around with Fig. 1.2’s trim pots P1, P3, and P4, from a certain dis-
turbance level on the results will become audible. A flatness deviation of eg
±2 dB at 4 kHz can certainly be heard. Then it sounds like changing treble,
middle, and/or bass tone controls of the pre-amp that follows the Engine II. The

14
With SNra = SNar–SNr from TSOS-1, p. 207 or TSOS-2, Chap. 16, Mathcad worksheet 16.2, and
the DMM-cut SNariaa from this book’s Sect. 1.B.2 the derivation of the worst case line of
SNriaa = −67.3 dB looks as follows: −71.6 dB(A)–(−4.3 dB) = −67.3 dB.
182 12 Engine II Performance

Fig. 12.10 Frequency and phase responses of Amp1 via Amps 3 & 4, fed by a generator with an
output resistance of 10.8 Ω

corresponding threshold level of detecting differences in sound depends on

one’s personal hearing sensitivity only.
4. Excellent broadband input and/or output transformers do not influence the sound
nor do they significantly worsen the noise level of pre-amps. Switching the
Amp5 output transformer does not create any audible effect. The corresponding
THD and IMD effects are tiny. From a CMRR point of view, transformers are
best-in-class components, always outperforming any electronic solution.
5. It seems to me that the other (maybe fundamental) sound influencing differences
could come from the basic configuration side of the story: single ended, semi
balanced, or fully balanced amplification, with or without overall or purely local
or a mixed version of negative feedback.
Principally and not only in case of Engine II, the fully balanced solution
seems to be the easiest and most successful way to overcome the many con-
nectivity problems between the various devices of an amp chain from cartridge
to loudspeaker. It is a provably better solution than a single ended one and it
stands at the beginning of the signal creating process in the recording studios.
Because of the high CMRR of the amps even non-shielded TQ2 double-relays
from Matsushita /Panasonic do not let in hum and/or other interferences. They
provide the all-time balanced switching on the main PCB and on Amp5’s PCB.
12.3 Conclusions and Final Remarks 183

Fig. 12.11 Frequency and phase responses of Amp2 via Amps 3 & 4, fed by a generator with an
output resistance of 10.8 Ω

I guess the recording studio designers really knew and still know how to
manage audio signals best. They opted for the balanced way, the one that simply
affects nasty influences most.
Not to forget: the passive RIAA networks allow rather easy calculation
approaches, frequency response trimming, and easy adaptation to other than
RIAA transfer characteristics.
6. Consequently, I think that the main drivers to cover over any differences in
sound produced by different phono-amp /pre-amp configurations are the
following ones:
• The frequency and phase response flatness in B20k,
• The very high un-weighted SN of the phono-amp with input load (for
comparison purpose I recommend 20 Ω),
• The disappearance of the harmonic and intermodulation artefacts in the mix
of LP and phono-amp noise levels.
In the end, if the designers do not follow these conditions, of course, different
phono-amps will sound differently. I checked this claim too by listening to the
Engine I phono-amps; all four modules were designed according to point 6. from
above. It became a hard job to find input signals, which produced a different
sounding output. In the period of changing the connections between Engine I and II,
184 12 Engine II Performance

a certain amount of stored listening information gets always lost. After having it
done up to ten times, sometimes I could identify tiny differences. However, I was
not able to say which result sounded best.
Nevertheless, my favourite phono-amp is the sequence of Amp2 + Amp3 +
Amp5 + Trafo—but I have no corresponding rational explanation. It’s simply
psychology according to my personal (Vogel’s) Razor: if you have two devices that
sound equal take the one you like best—for many different subjective reasons,
including design and integration into one’s living space.
 Part II
Knowledge Transfer
Selection of Draft Designs of Other
Input Stages 13

13.1 Intro

There are many phono enthusiasts having only un-balanced connection possibilities
between turntable and amplifier, some are not interested in a variable gain, and
others don not mind about a fully balanced amplification chain. In this chapter,
I want to demonstrate on how to integrate various Amp1/Amp2-type pre-amp
design examples that do not strictly follow the rules of the Engine II approach.
However, via the external input and a balanced line we can simply connect these
draft pre-amps with Engine II.
What we need here are linear amps with balanced or un-balanced input, balanced
output, and no equalization. At least they should fulfil the above-mentioned goals
concerning frequency and phase response, overload margin, and SNs. It is not my
aim to dive deep into the development of such amps. Many other authors have
tackled this issue in depth1 and their phono-amp circuits without RIAA equalization
network and low-noise input section will work well. The only new requirement is
the need of a low-noise balanced output section. Douglas Self in his ‘Small Signal
Audio Design’ handbook gives a very good overview on the many different ver-
sions. For transformer driven outputs, the Jensen Transformers website shows a
broad range of solutions. Professional recording equipment mostly uses them.
The following sections show three extremely to rather low-noise alternatives for
MC and MM purposes, two of them derived from the phono-amps already pre-
sented in TSOS-1 & TSOS-2.
The first example deals with an un-balanced i/p section and it is BJT driven; the
second version is a transformer driven amp with extremely low nominal i/p voltage
of 100 µVrms/1 kHz and shows a rather low i/p resistance of appr. 50 Ω, and the
third version shows a fully triode and transformer driven balanced in (MC)/
un-balanced in (MM) and balanced out solution.

1
See Sect. 13.5.

© Springer International Publishing Switzerland 2016 187

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_13
188 13 Selection of Draft Designs of Other Input Stages

Finally, at the end of this chapter in Sect. 13.5 I’ve added some redesigned
phono-amp solutions from other engineers together with my ideas on how we could
integrate these solutions into the Engine II concept.
Being in favour of not blowing-up the size of the chapter, the following
descriptions concentrate on a few essential and specific issues only.

13.2 BJT/Op-Amps Driven MC Input Stage

with Un-balanced Input and Balanced Output

The design in Fig. 13.1 is calculated in detail in the following chapter’s Mathcad
Worksheet MCD-WS 14.1. It is a modified version of a Douglas Self pre-pre-amp
design that I have also described in detail in my TSOS-1 and TSOS-2 books as
input configuration of the Module 2 phono-amp.
The amp has two trimming pots, P1 and P2 (multi turn). With P1 and input
shorted we set the collector current of the 4 paired BJTs to 1.7–2 mA each
(hfe ≥ 500), hence in total max. 8 mA. P2 sets the overall gain, depending on the
nominal output voltage of the MC cartridge in conjunction with the nominal output
voltage of this amp: 100 mVrms. DC-servo OP2 keeps the output of OP3 at 0.00 V.
The gain equations look as follows:

Gb ¼ G1ub G2b
ð13:1Þ
¼ 200

Fig. 13.1 BJT driven input stage alternative for MC purposes

13.2 BJT/Op-Amps Driven MC Input Stage … 189

R5 þ P2
G1ub ¼ 1 þ
R4 ð13:2Þ
¼ 100

G2b ¼ Gop3  Gop4

 
R11
¼1  ð13:3Þ
R10
¼2

To avoid hum interferences I strongly recommend designing the PCB’s ground

lines exactly as shown in Fig. 13.1. The emitter current must directly flow back to
the ground lead of the PSU.
With a gain of 2 ( = bal out/un-bal in) Op-amps 4 & 5 play the balancing output
stage. The chosen configuration and op-amps create lowest noise. The negative SN
impact of the output stage is 0.01 dB only. Nevertheless, all op-amp types create
tiny portions of additional—partly frequency dependent—noise; the worksheet
considers it. It also considers all frequency dependent resistor excess noise effects
and the frequency dependency of the BJT collector current noise.2 If needed,
additional offset voltage trimming pots (20 kΩ) could be inserted into the OP4 & 5
output circuit.
C1 is a polypropylene type and it should be directly located at the input con-
nector, C2 is a Panasonic FC/25 V type, C3 & C4 are ceramic types, C5 is a WIMA
MKS2, and C6 could be a simple MKT type.
The Mathcad worksheet calculates graphs (Figs. 14.5 and 14.6) that show a sig-
nificant increase of the amp’s input or output noise voltage density in the
low-frequency range. Thus, the calculation of the input referred noise voltage density
must be expressed as an average value in B20k: 312pV/rtHz. With input shorted and
referenced to 0.5 mVrms/1 kHz we can achieve an input referred SNariaa.i of −88.2 dB
(A). An input load of 20 Ω worsens the input referred SNariaa.i to −82.3 dB(A).
Consequently, as long as P2 is correctly set to a gain of 200, the output referred SN,
which is referenced to 100 mVrms/1 kHz, equals the one of the input.
Section 21.5 explains the input section’s fundamental equations around T1–T4
and Sect. 21.7 dives into the math of the often seen configuration of an input BJT
(or JFET) followed by an op-amp.

13.3 BJT/Op-Amps Driven MC Input Stage with Balanced

Transformer Input and Balanced Output

Figure 13.2 shows the second input amp alternative. It works with a balanced input
and with Fig. 13.1 balanced output formed by two AD797s.

2
See Chap. 11, MCD-WS 11.1 & 11.2.
190 13 Selection of Draft Designs of Other Input Stages

Fig. 13.2 Balanced transformer input and BJT/Op-Amp driven input stage alternative for low-
output MC cartridges

The JT-346-AXT input transformer is set to a turns ratio of tr = 1:12, hence

n = 12, thus, with the chosen input resistance of the stage after the transformer (R1),
a 1–5 Ω MC cartridge gets loaded by a load of approximately 49 Ω (e.g. Audio
Note loLtd). This is sufficiently enough for these very low source resistances and
low cartridge output levels.
The gain equations look as follows:

Gb ¼ G1b G2ub G3b

ð13:4Þ
¼ 1000

G1b ¼ n
ð13:5Þ
¼ 12

R5 þ P2
G2ub ¼ 1 þ
R4
¼
1000 ð13:6Þ
nG3b
¼ 41:67
13.3 BJT/Op-Amps Driven MC Input Stage … 191

G3b ¼ Gop3  Gop4

 
R11
¼1  ð13:7Þ
R10
¼2

I’ve described the basic design in TSOS-1 & -2, Module 3. The differences come
from the balanced output stage around OPs 3 & 4, the absence of the RIAA
feedback path, and the collector currents of T1 and T2. I’ve increased the collector
currents to 2.5 mA, hence, we’ll get a small input referred improvement over the
original design’s SN: now, with input shorted, A-weighting, and RIAA equalization
the output referred SNariaa.o becomes −94.3 dB(A), with input loaded by 5 Ω we’ll
get SNariaa.o = −88.7 dB(A).
Note: The input reference level of 0.5 mVrms/1 kHz is not realistic for the above-
mentioned kind of MC cartridges; 0.1 mVrms/1 kHz hits reality most. Hence, we
must decrease the fantastic looking SN figures by 14 dB! Mathcad Worksheet 14.2
considers the lower reference level. However, these decreased SN figures are still
very acceptable, because with input shorted, the average input referred noise
voltage density becomes calculated 170pV/rtHz only.
In the input shorted case the negative SN impact of the output stage with OPs 3
and 4 is 0.006 dB only.
The trimming procedure starts with P1. Its setting should produce 0.7 V across
R5. Then we feed the input with a 1 mVrms/1 kHz signal and we trim P2 to 0.5 Vrms
at OP3’s output. Automatically, the OP4 output will show the same level with a
reversed phase. This procedure yields a balanced output voltage of 1 Vrms, hence
the overall gain becomes 1000. If there would be a need for trimming the output DC
level to 0.00 V we could add offset voltage trimmers (20 kΩ) to OPs 3 & 4.
Because the JT input transformer seems to be no longer part of the Jensen
Transformer product range the Lundahl LL9226 becomes a rather good replace-
ment. By choosing the “E” termination alternative, we can set tr = 1:20 or n = 20
(see LL data sheet). With R1 = 20 kΩ, R11 = 604 Ω, no Rx and Cx, R0 = 5 Ω (0 Ω),
and P2 set for an overall gain of 1000 we have a calculated SNariaa.o = −74.7 dB(A)
(−80.3 dB(A)), ref 0.1 mVrms/1 kHz. With input shorted the average input referred
noise voltage density becomes 168.7 pV/rtHz.

13.4 Fully Triode Driven MC/MM Pre-Amp

with Transformer MC-Input and Balanced Output

The fully triode driven version’s circuit is given in Fig. 13.3.

It is composed of a low gain CCS (common cathode stage) with a double triode
in parallel operation followed by a CPS (concertina or cathodyne phase splitter)
with rather low output resistance, however not <1 Ω! Nevertheless, I want to
192 13 Selection of Draft Designs of Other Input Stages

Fig. 13.3 Fully triode driven MC/MM input stage alternative with transformer (Lundahl LL9226)
MC-input and balanced output
13.4 Fully Triode Driven MC/MM Pre-Amp with Transformer … 193

demonstrate here the possibility to create a fully triode and transformer driven
MC/MM phono-amp input stage that could also be taken as stand-alone pre-amp,
offering an excellent SN performance too.
The question might come-up why I’ve chosen the parallel operation of t1 and t2
and did not take a single triode that could produce the same gm value. First answer:
The single triode needs 1 mA more operating current (5 mA) to produce a gm of
7 mS at 90 V. It also represents a change of the operating point closer to “A”.
According to the findings of TSOS-2, Sect. 7.11, Table 17.7, 7 mS seems to be a
mutual conductance figure high enough for our purposes here, including the gain
stage application of MM amplification by replacing the transformer with a 47 kΩ
resistance and the required paralleled load capacitances, switched by S1–S4. Sec-
ond answer: I found out that the euphonic harmonics production at the operating
point “2 mA/90 V” per triode sounds best in my ears. The parallel operation of t3
and t4 follows the same operating point related arguments.
P1 trims the overall gain from the balanced input to the balanced output. P2
trimming leads to equal 1 kHz output levels to ground. Because the two output
resistances to ground differ very much (at the cathode Ro.c = 750 Ω, at the anode
Ro.a = 18.7 kΩ) the following measurement approach works quite well: we load the
balanced o/p by a 10 kΩ resistance and with two 1 MΩ (or 10 MΩ) input resistance
instruments we measure equal amplitudes from anode and cathode to ground.
The CPS is fully described in HTGG-2, Chaps. 10 and 11, and in one of my
LTEs3 to Linear Audio. Its rather low noise production mostly disappears in the
noise of the t1, t2 input stage with input load. Figure 13.4 shows the noise model of
the amp, configured as an MC phono-amp stage. In case of an MM phono-amp
configuration we have to take as input load the noise voltages4 of the admittances
that generate the two input noise voltages (en1(f)2 + en2(f)2)0.5 as result of the
voltage divider effects at the MM input.5
MCD Worksheet 14.3 dives deep into the SN calculation course for both versions:
MC and MM. The input double triode shows the following low-noise relevant fig-
ures: fc ≤ 500 Hz, gm ≥ 3.7 mS at 2 mA/90 V.6 Hence, with 20 Ω input load, and
n = 10 the calculated output referred and A-weighted SNariaa.o becomes −80.3 dB(A).
A change to the transformer configuration of Sect. 13.3 yields an output referred
SNariaa.o of −73.9 dB(A). To get it we need the following component changes:
R0 = 5 Ω, n = 20, S5–8 and R11–R14 replaced by a new resistance Rx = 25 kΩ, in
conjunction with a low-noise current generator as gain effective anode load P1 must
be set to a gain of 1000! (instead of 200). Then, with input shorted, the average
input noise voltage density becomes a calculated 221 pV/rtHz.

3
See on Linear Audio’s website ‘www.linearaudio.net/letters’ my LTE ‘Vol. 0-3’ to Stuart
Yaniger’s article in Vol. 0.
4
Details see MCD WS 14.4 of next chapter.
5
See TSOS-1, p. 158, TSOS-2, p. 293.
6
See TSOS-2 Tables 17.2 and 17.7.
194 13 Selection of Draft Designs of Other Input Stages

Fig. 13.4 Noise model of the Fig. 13.3 MC input stage

The same calculation with a Shure V15 V as the input load yields an output
referred SNariaa.o of −82.1 dB(A). For comparison reasons with the Engine I per-
formance7 I add the measured V15 V and 1 kΩ input load value too: −78.1 dB(A)
and −78.0 dB(A).
We obtain the following gain equations8 for the MC case:

Gb:mc ¼ G1ub G2ub G3b

ð13:8Þ
¼ 200

G1ub ¼ n
ð13:9Þ
¼ 10

G2ub ¼ gm1:2 ðR2 þ P1Þ

ð13:10Þ
¼ 11:1
 
1 1
2l þ
Rre 0:5RL
G3b ¼  
ra þ ð2 þ lÞ
1
þ
1 ð13:11Þ
Rre 0:5RL
¼ 1:8

7
See TSOS-2 Table 29.2.
8
Details of the G3 gain equation: see HTGG-2, Chaps. 10 and 11.
13.5 Other Development Examples 195

13.5 Other Development Examples

It might be conspicuous that I do not present my own amp solutions with input
sections driven by FETs. The reason is very simple: I never tried to develop them.
However, this is not a judgement about the quality of FET driven phono-amps. On
the contrary, I think that there are enough excellent solutions on the market. Among
them, three very different ones were presented in the past issues of the Linear Audio
bookzine. In this section, I want to demonstrate how they could be turned into linear
amplifiers, very useful as external or internal Engine II input stages. Choosing them
will lead to three main adaptation tasks: to get rid of the RIAA transfer-producing
components, to find a balanced output in two cases, and to set the right gain. The
next rather condensed sections will show how we can achieve very positive
solutions.9

13.5.1 The Joachim Gerhard Approach10

Mr. Gerhard’s phono-amp solution is a rather complex amplifier, split into two
stages, one linear for the input, and one for the RIAA transfer function performing
output. Both stages work in fully differential mode. Resistances separately set their
gains. DC servos ensure DC output voltages close to 0 V. The whole design
approach and the detailed circuit were published in Linear Audio’s Vol. 0, 2010.
Figure 13.5 (top) shows the stripped-down principal circuit without RIAA
components and some input (bottom) and other circuit components. Not shown are
the DC servos (DCS) for the input and output stages.
The overall differential gain Gtot.b of this arrangement is split into two gain
blocks the following way:
vo:b
Gtot:b ¼ 
vi:b
¼ G1b G2b ð13:12Þ
¼ 200

vo:1st:b
G1b ¼
vi:b
¼1þ
2Rf1 ð13:13Þ
RG
¼ 100

9
In the following three sections the several times used term ‘roughly calculated SNs’ means:
application of a HP21 and some rule-of-thumb figures for noise densities close to white noise
(=low fc < 1 kHz). They can be found in TSOS-1 and TSOS-2.
10
‘Down the Rabbit Hole—Adventure in the Land of Phonostages’, Linear Audio Vol. 0, 2010,
on http://www.linearaudio.net.
196 13 Selection of Draft Designs of Other Input Stages

Fig. 13.5 Principal Joachim Gerhard design with high-Z (Principally, Mr. Gerhard’s low-Z input
approach looks the same; details see footnote 11) voltage driven input, turned into a linear input
amp for Engine II purposes

vo:b
G2b ¼ 
vo:1st:b
¼
Rf2 ð13:14Þ
R
¼ 2
13.5 Other Development Examples 197

RG = 2.5 Ω = R11 in the Gerhard circuit. If we would set G2 = 2 then Rf2

becomes 2 times R. With R = 440 Ω (=R38/R39) Rf2 becomes 880 Ω. Because of
the rather high gain of the input stage (G1 = 100) the noise production of the output
stage is of minor importance.
With G1 = 100 we get Rf1 = 124 Ω (=R12/R13). If we assume a worst case
input referred noise voltage density of 0.6 nV/rtHz the proposed circuit changes do
not negatively touch the general noise behaviour of the Gerhard design, hence, the
output referred noise voltage density becomes 200 × 0.6 nV/rtHz = 120 nV/rtHz.
With input shorted and referenced to 100 mVrme/1 kHz the non-equalized
Fig. 13.5 output referred SNne.o thus becomes −77.0 dB. If we assume a flat F and P
in B20k the output referred RIAA equalized and A-weighted SNariaa.o yields roughly
−84.9dBV(A).
With 20 Ω input load the output referred SNariaa.o would then decrease to
−81.3 dBV(A). Hence, the shown arrangement would fulfil the Engine II goals.
There is another advantage of this type of JFET input stage: with 2 × 23.7 kΩ the
input resistance could be set to 47.4 kΩ, thus, MC cartridges like the Denon DL 103
R could be connected according to its data sheet specs.
This design could work without the output stage too. However, without touching
the input stage’s noise behaviour and according to (13.14) the output stage offers an
elegant possibility of changing the overall gain by simultaneously changing the two
Rf2 resistances. A bank of several jumpers or a multi-DIP-switch and a set of
selected resistors (0.1 %) will make it.

13.5.2 The Ovidiu Popa Approach11

In detail Mr. Popa’s approach, Synaesthesia HP5.1 can be studied on his website
http://www.synaesthesia.ca/LNschematics.html. It is a pre-pre-amp with a config-
uration alike the NPN-half of Fig. 13.5 input stage, however with the doubled
number of JFETs. The principal circuit is given in Fig. 13.6’s input stage. Without
increasing the excellent noise voltage situation at the input stage’s output the output
stage creates the balanced output around OPs 2 and 3 and the two resistors R. An
additional DC-servo ensures DC output voltages close to 0 V (not shown here).
The gain equations look as follows:
vo:b
Gtot:b ¼
vi:ub
¼ G1ub G2b ð13:15Þ
¼ 200

See Mr. Popa’s article in Linear Audio Vol. 1, 04-2011, ‘On the Noise Performance of Low
11

Noise Input Stage’ and the corresponding LTEs on http://www.linearaudio.net.

198 13 Selection of Draft Designs of Other Input Stages

Fig. 13.6 Principal Ovidiu Popa design with additional and new output stage, turned into a linear
input amp for Engine II purposes

vo:1st:ub
G1ub ¼
vi:ub
Rf1 ð13:16Þ
¼1þ
RS
¼ 100

vo:v
G2b ¼
vo:1stub
vo:1st:ub  ½vo:1st:ub ðR=RÞ ð13:17Þ
¼
vo:1st:ub
¼2

With input shorted, an average input noise voltage density of 275 pV/rtHz, and
referenced to 100 mVrme/1 kHz the non-equalized Fig. 13.6 output referred SNne.o
thus becomes −82.2 dB. If we assume a flat F and P in B20k the output referred
RIAA equalized and A-weighted SNariaa.o yields roughly −90.1 dBV(A).
With 20 Ω input load the output referred SNariaa.o would then decrease to
−82.8 dBV(A).12 Hence, the shown arrangement would fulfil the Engine II goals
too. Alike the Gerhard design there is an equal advantage of this type of JFET input
stage: with 47.5 kΩ as input resistance MC cartridges like the Denon DL 103 R
could be connected too.
A variable gain set-up could be performed by copying the method shown in
Fig. 13.1.

12
The complete SN calculations can be found in my LTE to Mr. Popa’s Linear Audio article in
Vol. 1, on http://www.linearaudio.net/LTE
13.5 Other Development Examples 199

Fig. 13.7 Principal Bob Cordell design with an additional and new output stage

13.5.3 The Bob Cordell Approach13

In contrast to the above-described voltage-to-voltage solutions (VV-A) with

4 × 2SK170 & 4 × 2SJ74 and 8 × BF862 JFETs, the Cordell design sets on a high-Z
current generating output of his multi-LTP-JFET-input stage with four Dual-JFETs
LSK389. Thus, a simple resistor R at the output can set the gain of this stage. Its
value depends on the required gain and the transconductance of the amp; hence, we
talk about a transconductance amplifier or VC-A which transfers its input voltage
into an output current.
The gain equation thus becomes the following general format:

G ¼ gm R ð13:18Þ

At the left side of Fig. 13.7 we find the principal circuit of this stage with a gain
of 100 at 1 kHz. On the right side, as balanced signal producing output stage, we
find a configuration that equals the one of Fig. 13.6.
The connection of the output stage to the output of the input stage is rather
simple: delete R of the i/p stage (=R40 of Mr. Cordell’s design) and connect A & B.
Now, the input R1 = R of the output stage plays the role of the deleted R and the
feedback R2 = R of OP3 sets the gain of OP 3 to “1”. Thus, we obtain the following
gain equations:

See Mr. Cordell’s article in Linear Audio Vol. 4, 09-2012, ‘VinylTrak—A full-featured MM/MC
13

phono-preamp’ or on www.cordell.com.
200 13 Selection of Draft Designs of Other Input Stages

vo:b
Gtot:b ¼
vi:b
¼ G1b G2b ð13:19Þ
¼ 200

vo:st:ub
G1b ¼
vi:b
¼ gm R1 ð13:20Þ
¼ 100

vo:b
G2b ¼
vo:1st:ub
 
vo:1st:ub  vo:1st:ub R2
R1 ð13:21Þ
¼
vo:1st:ub
¼2

By making R = R1 = R2 variable by a range of fixed resistors and jumpers, this

amp configuration allows setting a broad range of gains, eg from 100–300 with
R = 890 Ω–2.67 kΩ.
Mr. Cordell measured an input referred noise voltage density of 0.7 nV/rtHz. If
we take it as average value, then, with input shorted, the output referred SNne.o
becomes −74.1 dB (gain = 200 and reference level = 100 mVrms). If we assume a
flat F and P in B20k the output referred RIAA equalized and A-weighted SNariaa.o
yields roughly -82.0 dBV(A).
With 20 Ω input load the output referred SNariaa.o would then decrease to
−79.8 dBV(A). Hence, the shown arrangement would fulfil the Engine II goals too.

13.6 The Output Stage

The task of the output stages in Figs. 13.1, 13.2, 13.6, and 13.7 lies in the fact of
turning an un-balanced signal into a balanced one. A range of other solutions could
also do it. However, we need a very low-noise solution here, a solution that does
not add more than 0.1 dB noise to the output noise of the preceding gain stage. Each
component counts here.
The chosen two-op-amp arrangement does an excellent job. The calculated and
measured SN worsening never exceeds the set boundary; even in the extremely
low-noise Popa solution, it becomes calculated <0.06 dB.
In the Cordell case, the preceding input stage produces more than a doubled
Popa output noise voltage. Hence, the output stage’s noise contribution with
6.23 nV/rtHz (≡−101.1 dB ref. 0.1 Vrms) falls below 0.02 dB.
13.7 Summary of Results 201

13.7 Summary of Results

Table 13.1 shows the most important noise voltage and SN results of the presented
draft designs, compared with measured Engine II results.
For a better understanding Table 13.1 needs some additional remarks:
1. Column D shows values of the average noise voltage density in B20k, hence,
including 1/f-noise effects that may exist.
2. Column E transfers Column D values into non-equalized SN (dB) values by
referencing them to the nominal signal voltage of 0.1 Vrms at the output of the
input stages. The contributions of the shown two-op-amp output stages is
always <0.06 dB.
3. Column F shows MM usability of the stage, however, with the following
adaptations only:
line 9: (yes) with gain reduction to 20, high-Z input equipped with an appro-
priate MM cartridge load resistance, in conjunction with a long input cable14
(1 m ≡ 100–150 pF) the input capacitance could be the bottle-neck: Ci * 95 pF
line 10: (yes) with the same line 9 arguments; Ci * 80 pF
line 11: (yes) with Mr Cordell’s similar looking MM input stage with a different
input LTP.
4. Column G values are Column F values transferred into RIAA equalized and
A-weighted SNs, i/p shorted. Here, because of the input referred SN of the
following Engine II stages and according to Figs. 15.3 and 15.4, we can find a
noise contribution from the following stages, however, not shown in the table.
For example: The amp sequence Amp3 + Amp5 + Trafo has an input referred
SNo.a of −99 dB(A).15 It includes the contributions of the RIAA transfer net-
work. Compared with the line 5 and column G value of −88.2 dB(A) we have a
difference of only 10.8 dB, thus, with Fig. 15.3 we find an SN worsening of 0.
4 dB. If we calculate the whole amp sequence with input shorted the box 5/G
value should thus be corrected and reduced by a worsening figure We(B) = 0.
4 dB.
5. Column H shows the RIAA equalized and A-weighted results. Here, the cal-
culation of a worsening figure makes no sense, because the input load moves the
difference of the SN values into a region of >16 dB, hence, We(B) < 0.1 dB.
6. Column I shows the gains of the whole RIAA amplifier sequence from input to
output.

14
Remedy: See TSOS-1, p. 144ff, TSOS-2, p. 220ff.
15
See Table 12.1, line 13, column D.
 202

Table 13.1 Draft design SNs compared with Engine II results

1/A B C D E F G H I J K
2 Type or i/p i/p referred noise o/p referred Usability for Engine II o/p Engine II o/p Engine Remarks Usage
Fig. characteristic voltage density SNne.o (i/p MM SNariaa.o (i/p SNariaa.o (i/p II gain
shorted) purposes shorted) 20 Ω)
3 calc. calc. calc. calc.
4 nV/rtHz average dB dBV(A) dBV(A) times
5 13.1 4 × NPN BJTs 0.312 −81.1 No −88.2 −82.3 2,000 MCD-WS 14.1 ub-to-b
6 13.2 n = 12 trafo and i/p 0.170 −72.4 No −80.3 −74.7 10,000 MCD-WS 14.2 ub&b-to-b
load 5 Ω
13

7 13.3a 4 × E88CC and 0.536 −76.4 Yes −83.0 80.3 2,000 MCD-WS 14.3 ub&b-to-b
n = 10 trafo
8 13.3b 4 × E88CC and i/p 0.221 −70.1 No −78.0 −73.9 10,000 Rough calc. ub&b-to-b
like 13.2
9 13.5 4 × (NPN and PNP) 0.500 −77.0 (Yes) −84.9 −81.3 2,000 Rough calc. b-to-b
JFETs
10 13.6 8 × NPN JFETs 0.275 −82.2 (Yes) −90.1 −82.8 2,000 Rough calc. ub-to-b
11 13.7 4 × LTP NPN 0.700 −74.1 (Yes) −82.0 −79.8 2,000 Rough calc. ub&b-to-b
JFETs
12 Meas.
13 Amp1 n = 10 trafo 0.519 −76.6 No −84.5 −81.8 2,000 MCD-WS 9.1 ub&b-to-b
14 Amp2 2 × (NPN and PNP) 0.517 −76.7 No −83.3 −79.8 2,000 MCD-WS 11.3 b-to-b
BJTs
Selection of Draft Designs of Other Input Stages
13.7 Summary of Results 203

7. Column K gives some usage hints:

b-to-b balanced-in to balanced-out (i/p socket mostly XLR)
ub-to-b un-balanced-in to balanced-out (i/p socket mostly Cinch or BNC)
ub&b-to-b un-balanced-in to balanced-out (eg by grounding the XLR pin 3 input
with a Neutrik Cinch-XLR adapter), and balanced-in to balanced-out
Mathcad Worksheets Draft Designs
14

Contents

14.1 MCD-WS: BJT/Op-Amp Driven MC Input Stage with Un-Balanced Input

and Balanced Output
14.2 MCD-WS: BJT/Op-Amp Driven MC Input Stage with Balanced
Transformer Input and Balanced Output
14.3 MCD-WS: Fully Triode Driven MC/MM Pre-Amp with Transformer
MC-Input and Balanced Output

pffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffi
Note 1: MCD 11 has no built-in unit “rtHz” or “ Hz”. To get 1 Hz based
voltage noise and current noise densities the rms noise voltage and
current in a specific frequency range B > 1 Hz must be multiplied by
pffiffiffiffiffiffiffiffiffi
1 Hz and divided by the root of that specific frequency range √B!
Note 2: MCD 11 offers no “dB” unit. This is available from MCD 13 on!

© Springer International Publishing Switzerland 2016 205

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_14
206 14 Mathcad Worksheets Draft Designs

14.1 BJT/Op-Amp Driven Input Stage Page 1

BJT/Op-Amp Driven MC input Stage with Un-Balanced Input and Balanced Output :

Fig. 14.1 = Fig. 13.1

1. Definition of all meaningful constants, components, etc.:

23 1 19 3
k 1.38065 10 VAsK q 1.6021765 10 As T 300 K vi.nom 0.5 10 V

B20k 19980 Hz B1 1Hz h 1000Hz vo.nom 0.1V

3 3 3
R0 20 R1 1.14 10 R2 10 10 R3 44.2 10
3
R4 1 R5 1.3 10 R6 199 R7 180
6 3 6
R8 1 10 R9 10 R10 1 10 R11 R10 R12 2.21 10
1
1 1
RPop4 RPop4 500
R10 R11
9
NI 31.62 10 NIe 20 log( NI) 120 NIe 30.001 [dB]

Note: In the following calculation course I've set R5 = R5+P1, R6 = R6+P2 !

9 3 12 12
C1 1.2 10 F C2 1 10 F C3 100 10 F C4 18 10 F
6 9
C5 2.2 10 F C6 1 10 F
14 Mathcad Worksheets Draft Designs 207

14.1 BJT/Op-Amp Driven Input Stage Page 2

3
IC 8 10 A hfe 500 VA 160V

VCE 14.4V R5 IC VCE 4V VBE 0.7V

3
VR5 IC R5 VR5 10.4 V VR4 IC R4 VR4 8 10 V

OP1 5534 OP2 LF411 OP3 BUF634 OP4 AD797 OP5 OP4

R6
Gamp 1 Gamp 200
R4

x 0.17 fc.i 33750 Hz

q IC 3
gm gm 309.454 10 S
kT

2 12
rbb rbb 0.5 en.rbb 4 k T B1 rbb en.rbb 91.016 10 V
4

2. Evaluation of the impedances of the input network as the input load :

f 10Hz 20Hz 20000 Hz

1
1 1 1
Z1( f R0) 2j f C1 Z1 h 20 19.656
R0 R1 2j f C2

50
40
[ohm]

30
Z1 f 20 Fig. 14.2
20
Impedance of the input
10 network Z1(f)
0
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

12
en.Z1( f R0) 4 k T B1 Z1( f R0) en.Z1 h 20 570.661 10 V

3. Evaluation of the impedance of the feedback network :

1
1
Z2( h) 2j h C6 Z2( h) 199
R6

Z2( h)
Gamp( h) 1 Gamp( h) 200
R4
208 14 Mathcad Worksheets Draft Designs

14.1 BJT/Op-Amp Driven Input Stage Page 3

3
1 10
[times]

Gamp ( f)
Fig. 14.3a
Frequency dependent gain

100
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

Im Gamp( f)
amp( f) atan
Re Gamp( f)

5
4
3
2 Fig. 14.3b
amp ( f) 1 Phase response of the gain
[°]

0
deg 1
2
3
4
5
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

4. Calculation of the gain and input resistance of the T1 ... T4 stage :

VA 3
rce rce 20 10
IC

R4
gm rce R5 hfe
rce 3
gm.red gm.red 239.657 10 S
hfe rce R4 R5 gm R4 hfe rce rce R5

gm.red R3 1
GT1.4 GT1.4 284.651
1 1
1 R3
R5 rce

GT1.4.e 20 log GT1.4

GT1.4.e 49.086 [dB]

hfe 3
rbe rbe 1.616 10
gm
14 Mathcad Worksheets Draft Designs 209

14.1 BJT/Op-Amp Driven Input Stage Page 4

1
1 1 1
ri
R2 R4 1 3
rbe hfe R4 1 gm R12 1 1 ri 8.186 10
R4 R6 R3
R5 rce

1
1 1
Ri.amp
ri R1 Ri.amp 1000.653

1
1 3
Zin( f) 2j f C1 Zin( h) 1.001 10
Ri.amp

1050

1025
[ohm]

Zin ( f) 1000
Fig. 14.4
975 Input impedance Zin (f)

950
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

5. Calculation of the amp's input noise current and voltage :

5.1 Relevant input noise current densities :

x
fc.i 12
in.c( f) 2 q IC B1 1 in.c( h) 92.55 10 A
f
x
2 q IC fc.i
in.i( f) B1 1 12
hfe f in.i( h) 4.139 10 A

4 k T B1 12
in.R2 in.R2 1.287 10 A
R2

4 k T B1 15
in.R3 in.R3 612.239 10 A
R3

5.2 Relevant resistor noise voltages and currents :

NIe
10 12 2
10 10 IC R4 B1 12
eN.R4ex( f) eN.R4ex( h) 5.272 10 V
ln( 10) f

eN.R4 4 k T R4 B20k
210 14 Mathcad Worksheets Draft Designs

14.1 BJT/Op-Amp Driven Input Stage Page 5

2 2 B1 12
en.R4ex.tot( f) eN.R4ex( f) eN.R4 en.R4ex.tot( h) 128.716 10 V
B20k

NIe
10 12 2
10 10 IC R5 B1 9
eN.R5ex( f) eN.R5ex( h) 6.853 10 V
ln( 10) f

eN.R5 4 k T R5 B20k

2 2 B1 9
en.R5ex.tot( f) eN.R5ex( f) eN.R5 en.R5ex.tot( h) 4.641 10 V
B20k

9
en.R6 4 k T R6 B1 en.R6 1.816 10 V
9
en.R7 4 k T R7 B1 en.R7 1.727 10 V
9
en.R8 4 k T R8 B1 en.R8 128.716 10 V
9
en.R10 4 k T R10 B1 en.R10 4.07 10 V
9
en.R11 4 k T R11 B1 en.R11 4.07 10 V
9
en.R12 4 k T R12 B1 en.R12 191.35 10 V

en.R5ex.tot( f) 12
in.R5ex( f) in.R5ex( h) 3.57 10 A
R5
en.R4ex.tot( f) 12
in.R4ex( f) in.R4ex( h) 128.716 10 A
R4

5.3 Noise of op-amps :

OP1 :
fc.e1 6Hz fc.i1 8Hz

9 fc.e1
en.i.op1 4 10 V en.i.op1( f) en.i.op1 1
f

12 fc.i1
in.i.op1 0.4 10 A in.i.op1( f) in.i.op1 1
f

OP2 :
3
fc.e2 0.1 10 Hz fc.i2 unknown

9 fc.e2
en.i.op2 25 10 V en.i.op2( f) en.i.op2 1
f
12
in.i.op2 0.01 10 A
14 Mathcad Worksheets Draft Designs 211

14.1 BJT/Op-Amp Driven Input Stage Page 6

2 2
1 1 2
2j f C6 2 2j f C6 1
en.o.op2 ( f) en.R8 en.i.op2( f) 1 in.i.op2 2j f C6
R8 R8

9
en.o.op2 ( h) 18.184 10 V

OP3 :
fc.e3 unknown fc.i3 unknown

9 12
en.i.op3 4 10 V in.i.op3 1 10 A guessed!

OPs 4 & 5 :

fc.e4 30Hz fc.e5 fc.e4 fc.i4 unknown fc.i5 unknown

9 fc.e4
en.i.op4 0.9 10 V en.i.op4( f) en.i.op4 1
f
12
in.i.op4 2 10 A in.i.op5 in.i.op4 en.i.op5( f) en.i.op4( f)

Noise gains in the output stage : GN.op4 1 GN.op5 2

2 2 2 2 2 2 2 2
en.o.op4.5 ( f) GN.op4 en.i.op4( f) GN.op5 en.i.op5( f) en.R10 in.i.op5 R11 en.R11

9
en.o.op4.5 ( h) 6.427 10 V

5.4 Input referred noise voltage density of T1 ... T4 :

2
in.c( f) 12
en.i.T1.4( f) 4 k T rbb B1 en.i.T1.4( h) 312.619 10 V
2
gm

2
2 2 2 2 2 1 12
en.i( f) en.i.T1.4( f) R4 in.i( f) in.R4ex( f) in.R5ex( f) R4 en.i( h) 338.443 10 V
gm

2 2 2 12
in.i.amp( f) in.i( f) in.R2 in.R3 in.i.amp( h) 4.378 10 A

6. Calculation of the amp's input and output noise voltages :

3
G0.amp gm R12 G0.amp 683.892 10

2 2
2 en.i.op1( f) en.o.op2 ( f) 2 2 2 2 2 2
en.o.tot.0( f) en.i( f) G0.amp in.R5ex( f) R12 in.i.op1( f) R12 en.R12
2
GT1.4
2 2 2 2
in.i.op3 R7 en.R7 en.i.op3
212 14 Mathcad Worksheets Draft Designs

14.1 BJT/Op-Amp Driven Input Stage Page 7

6
en.o.tot.0( h) 235.877 10 V

en.o.tot.0( f) 12
en.i.tot( f) en.i.tot( h) 344.903 10 V
G0.amp

2 2 2 9
en.o.tot( f) en.i.tot( f) Gamp( f) en.R6 en.o.op4.5 ( f) en.o.tot( h) 69.303 10 V

en.o.tot( f)
en.i.amp( f) 12
Gamp( f) en.i.amp( h) 346.516 10 V

2 2 2 2 12
en.i.amp( f R0) en.i.amp( f) in.i.amp( f) Z1( f R0) en.Z1( f R0) en.i.amp h 20 673.147 10 V

8
1 10
[ V/rtHz ]

en.i.amp f 20 Fig. 14.5

1 10
9 Frequency and R0
12 dependent equivalent input
en.i.amp f 10
noise voltage density for
two different input loads
10
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

7. SN calculations :

en.o.amp( f R0) en.i.amp( f R0) Gamp( f) 9

en.o.amp h 20 134.629 10 V

6
1 10

Fig. 14.6
[ V/rtHz ]

en.o.amp f 20
Frequency and R0
7
12 1 10 dependent equivalent
en.o.amp f 10 output noise voltage density
for two different input loads
8
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]
14 Mathcad Worksheets Draft Designs 213

14.1 BJT/Op-Amp Driven Input Stage Page 8

20000Hz
1 2
en.o.amp( f R0) df
B1
20Hz
SNne.o( R0) 20 log SNne.o 20 74.662 [dB]
vo.nom

Average input referred noise voltage density (R0 dependent) :

R0 0.001 1 100 SNne.i( R0) SN ne.o( R0)

SNne.i( R0)
20 B1 12 12
en.i.amp( R0) vi.nom 10 en.i.amp 10 312.261 10 V
B20k
12
en.i.amp 20 653.966 10 V

8
1 10
[V/rtHz]

en.i.amp( R0) 1 10 9
Fig. 14.7
R0 dependent average
input noise voltage density

10
1 10
0 10 20 30 40 50 60 70 80 90 100
R0
[ohm]

A-weighting by A(f) :

1.259 1 1 1
A( f)
2 2 2 2
20.6Hz 107.7Hz 737.9Hz f
1 1 1 1
f f f 12200 Hz

20000Hz
1 2 2
en.o.amp( f R0) A( f) df
B1
20Hz
SNne.o.a( R0) 20 log
vo.nom

SNne.o.a 20 76.783 [dB(A)]

214 14 Mathcad Worksheets Draft Designs

14.1 BJT/Op-Amp Driven Input Stage Page 9

RIAA equalizing by R(f) :

1
2
3 6
1 2 10 Hz 318 10 s
R1000
2 2 R1000 9.898
3 6 3 6
1 2 10 Hz 3180 10 s 1 2 10 Hz 75 10 s

2
6
1 2 f 318 10 s
R( f) R1000
2 2
6 6
1 2 f 3180 10 s 1 2 f 75 10 s

20000Hz
1 2 2
en.o.amp( f R0) R( f) df
B1
20Hz
SN riaa.o( R0) 20 log SN riaa.o 20 71.799 [dB]
vo.nom

20000Hz
1 2 2 2
en.o.amp( f R0) A( f) R( f) df
B1
20Hz
SN ariaa.o( R0) 20 log
vo.nom

SN ariaa.o 20 82.373 [dB(A)]

75
76
[dB(A) ref. 100mVrms]

77
78
79
80
81
SNariaa.o ( R0) 82
83
84
85
86
87
88
89
90
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
R0
[ohm]

Fig. 14.8 R0 dependent, A-weighted, and RIAA equalized output referred SNs
14 Mathcad Worksheets Draft Designs 215

14.2 BJT/Op-Amp Driven MC Input Stage with Balanced Transformer Page 1

Input and Balanced Output

BJT/Op-Amp Driven MC Input Stage with Balanced transformer Input and Balanced Output

Fig. 14.9 = Fig. 13.2

1. Definition of all meaningful constants, components, etc.:

23 1 19 3
k 1.38065 10 VAsK q 1.6021765 10 As T 300 K vi.nom 0.1 10 V

B20k 19980 Hz B1 1Hz h 1000Hz vo.nom 0.1V

3
f 10Hz 15Hz 20 10 Hz
3 3
R0 5 R1 6.81 10 R2 499 R3 R2 R4 1.82 10
3 3 6 3
R6 499 R7 5.49 10 R8 9.31 10 R9 2.21 10 R10 15 10
3 3 3 6
R11 1 10 R12 33.2 R13 1 10 R14 1 10 R15 2.21 10
6 3
R16 2.21 10 Rx 4.12 10
NIe
NIe 30 [dB] 20 6 9
NI 10 10 NI 31.623 10

Transformer data sheet: n 12 RP 0.643 RS 154

216 14 Mathcad Worksheets Draft Designs

14.2 BJT/Op-Amp and Transformer Driven Input Stage Page 2

2
R0sec( R0) R0 n R0sec 5 720 optimal transformer load = 6k81
2
RPsec RP n RPsec 92.592

Rn( R0) R0sec( R0) RPsec RS Rn 5 966.592

6 6 6 6 12
C1 0.1 10 F C2 470 10 F C3 4.7 10 F C4 10 10 F C5 15 10 F
6
C6 1 10 F C7 C6

R1 R0sec( R0) RPsec RS

Input load: Ri.L( R0) Ri.L 5 846.449
R1 R0sec( R0) RPsec RS

Transistor data: T1 T2 2SC3329BL rbb1 2 rbb2 rbb1

hfe1 500 hfe2 hfe1

3
IC1 2.5 10 A IC2 IC1 VA1 160V VA2 VA1

T3 BC550C T4 T5 T6 BC560C

Op-Amps: OP1 5534 OP2 LF411 OP3 OP4 AD797

2. Calculation of the amp's gain and input resistance Ri :

succ-apps P1: P1 347.5
R11 P1
G1 n G2 1 G2 41.677 G3 2
1
1 1
R10 R12
3
Gamp G1 G2 G3 Gamp 1 10

3
Note: In the following calculation course R11p = R11+P1! R11p R11 P1 R11p 1.347 10

q IC1 3
gm1 gm1 96.704 10 S
kT
hfe1 3
rbe1 rbe1 5.17 10 transformer load:
gm1
1
R12 6 1 1 3
ri1 2 rbe1 1 gm1 R9 ri1 53.152 10 6.809 10
R11p R12 ri1 R1

1
1 1 1
Ri RPsec RS Ri 48.998
ri1 R1 2
n
14 Mathcad Worksheets Draft Designs 217

14.2 BJT/Op-Amp and Transformer Driven Input Stage Page 3

3. Calculation of the relevant noise currents and voltages of the whole amp
3.1 BJT noise currents :

x 0.5 fc.i1 10 Hz guessed! fc.i2 fc.i1

x
fc.i1 12
in.c1( f) 2 q IC1 B1 1 in.c1( h) 28.445 10 A
f

x
2 q IC1 B1 fc.i1 12
in.b1 ( f) 1 in.b1 ( h) 1.272 10 A
hfe1 f

in.c2( f) in.c1( f) in.b2 ( f) in.b1 ( f)

Note: Noise impact of T3 ... T6 ignored!

3.2 BJT noise voltages :

2
in.c1( f) 2 2 12
en.i1( f) 4 k T B1 rbb1 in.b1 ( f) rbb1 en.i1( h) 345.92 10 V
2
gm1

en.i2( f) en.i1( f)

3.3 Resistor noise voltages :

9
en.Ri.L( R0) 4 k T B1 Ri.L( R0) en.Ri.L 5 3.745 10 V
1
1 1
RP RP 33.127
R12 R10
12
en.RP 4 k T B1 RP en.RP 740.835 10 V

9
en.R11p 4 k T B1 R11p en.R11p 4.725 10 V

9
en.R13 4 k T B1 R13 en.R13 4.07 10 V

9
en.R14 4 k T B1 R14 en.R14 4.07 10 V

9
en.R15 4 k T B1 R15 en.R15 191.35 10 V

9
en.R16 4 k T B1 R16 en.R16 191.35 10 V
218 14 Mathcad Worksheets Draft Designs

14.2 BJT/Op-Amp and Transformer Driven Input Stage Page 4

3.4 Op-Amp noise voltages and currents :

OP1 :
fc.e1 6Hz fc.i1 8Hz

9 fc.e1
en.i.op1 4 10 V en.i.op1( f) en.i.op1 1
f

12 fc.i1
in.i.op1 0.4 10 A in.i.op1( f) in.i.op1 1
f
Note: Noise impact of OP1 ignored!

OP2 :
3
fc.e2 0.1 10 Hz fc.i2 unknown

9 fc.e2
en.i.op2 25 10 V en.i.op2( f) en.i.op2 1
f
12
in.i.op2 0.01 10 A

1
2j f C6
Gop2 ( f) 1 Gop2 ( h) 1
R16

Gop2 ( 0.01Hz) 7.271

2 2
1 1
2j f C7 2 2j f C6 R16 2
en.o.op2 ( f) en.R15 en.i.op2( f) in.i.op2 Gop2 ( f)
1 1
R15 2j f C7 2j f C6 R16

9
en.o.op2 ( h) 26.22 10 V

en.o.op2 ( f) 12
in.o.op2 ( f) in.o.op2 ( h) 1.748 10 A
R10

OPs 3 & 4 :

fc.e3 30Hz fc.e4 fc.e3 fc.i3 unknown fc.i4 unknkown

9 fc.e3
en.i.op3 0.9 10 V en.i.op3( f) en.i.op3 1
f
12
in.i.op3 2 10 A in.i.op4 in.i.op3 en.i.op4( f) en.i.op3( f)

Noise gains in the output stage : GN.op3 1 GN.op4 2

2 2 2 2 2 2 2 2
en.o.op3.4 ( f) GN.op3 en.i.op3( f) GN.op4 en.i.op4( f) en.R13 in.i.op4 R14 en.R14
9
en.o.op3.4 ( h) 6.427 10 V
14 Mathcad Worksheets Draft Designs 219

14.2 BJT/Op-Amp and Transformer Driven Input Stage Page 5

4. Calculation of the amp's input and output noise voltages :

2 2 2 2 2 2 2 2 2
en.o.tot( f R0) en.i1( f) in.b1 ( f) Ri.L( R0) en.Ri.L( R0) G2 G3 in.b2 ( f) in.o.op2 ( f) R11p
2 2
en.R11p en.o.op3.4 ( f)

9
en.o.tot h 5 326.183 10 V

en.o.tot( f R0) 9
en.i.tot( f R0) en.i.tot h 5 3.913 10 V
G2 G3

en.i.tot( f R0) 12
en.i.amp( f R0) en.i.amp h 5 326.101 10 V
n

9
1 10

Fig. 14.10
[V/rtHz]

en.i.amp f 5
Input referred noise
en.i.amp f 0 voltage density based on
two different input loads

10
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

9
en.o.amp( f R0) en.i.amp( f R0) Gamp en.o.amp h 5 326.183 10 V

5. SN calculations :

20000Hz
1 2
en.i.amp( f R0) df
B1
20Hz 12
SNne.i( R0) 20 log SNne.i 10 72.383 [dB]
vi.nom

SNne.i 5 66.729 [dB]

Average input referred noise voltage density (R0 dependent) :

R0 0.001 1 100

SNne.i( R0)
20 B1 12 12
en.i.amp( R0) vi.nom 10 en.i.amp 10 170.032 10 V
B20k
12
en.i.amp 5 326.014 10 V
220 14 Mathcad Worksheets Draft Designs

14.2 BJT/Op-Amp and Transformer Driven Input Stage Page 6

9
1 10
[V/rtHz]

en.i.amp( R0)
Fig. 14.11
R0 dependent average
input noise voltage density

10
1 10
0 1 2 3 4 5 6 7 8 9 10
R0
[ohm]

A-weighting by A(f) :

1.259 1 1 1
A( f)
2 2 2 2
20.6Hz 107.7Hz 737.9Hz f
1 1 1 1
f f f 12200 Hz

20000Hz
1 2 2
en.o.amp( f R0) A( f) df
B1
20Hz
SN ne.o.a( R0) 20 log
vo.nom

SNne.o.a 5 68.773 [dB(A)]

RIAA equalizing by R(f) :

1
2
3 6
1 2 10 Hz 318 10 s
R1000
2 2 R1000 9.898
3 6 3 6
1 2 10 Hz 3180 10 s 1 2 10 Hz 75 10 s

2
6
1 2 f 318 10 s
R( f) R1000
2 2
6 6
1 2 f 3180 10 s 1 2 f 75 10 s

20000Hz
1 2 2
en.o.amp( f R0) R( f) df
B1
20Hz
SN riaa.o( R0) 20 log SNriaa.o 5 70.298 [dB]
vo.nom
14 Mathcad Worksheets Draft Designs 221

14.2 BJT/Op-Amp and Transformer Driven Input Stage Page 7

20000Hz
1 2 2 2
en.o.amp( f R0) A( f) R( f) df
B1
20Hz
SNariaa.o( R0) 20 log
vo.nom

SNariaa.o 0 80.314 [dB(A)]

SNariaa.o 5 74.659 [dB(A)]

70
71
72
[dB(A) ref. 100mVrms]

73
74
75
76
SNariaa.o ( R0) 77
78
79
80
81
82
83
84
85
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10
R0
[ohm]

Fig. 14.12 R0 dependent, A-weighted, and RIAA equalized output referred SNs
222 14 Mathcad Worksheets Draft Designs

14.3 Fully Triode Driven Input Stage Page 1

Fully Triode Driven MC/MM Pre-Amp with Transformer MC-Input and Balanced Output

Fig. 14.17 = Fig. 13.6

1. Circuit and component data :

23 1 B20k 19980 Hz B1 1Hz fhp 0.5Hz

k 1.38065 10 VAsK T 315K
f 10Hz 15Hz 20000 Hz h 1000Hz
3 3 3
vo.nom 100 10 V vi.nom.mm 5 10 V vi.nom.mc 0.5 10 V

3
t1.2 E88CC Ia1.2 4 10 A Va1.2 90V Vg1.2 2.8V
3
t3.4 E88CC Ia3.4 4 10 A Va3.4 90V Vg3.4 2.8V
14 Mathcad Worksheets Draft Designs 223

14.3 Fully Triode Driven Input Stage Page 2

3 3 3 3
R1a 100 10 R1b 90 10 R2 2 10 R3 700 R4 40.1 10
3 3 3
R5 301 R6 23.9 10 R7 R6 R8 2.35 10 P1 5 10
3
R9 100 10 R10 R9

VR2.P1 Ia1.2 ( R2 P1) VR2.P1 28 V VR3 Ia1.2 R3 VR3 2.8 V

Vcc1.2 Va1.2 VR2.P1 VR3 Vcc1.2 120.8 V

VR6 Ia3.4 R6 VR6 95.6 V

VR7 Ia3.4 R7 VR7 95.6 V

Vcc3.4 Va3.4 VR6 VR7 Vcc3.4 281.2 V

Tr1 LL9229 n 10 Rp 5 Rs 260 Termination alternative "C

2. Triode valve constants ( = 29 selected) :

3 1 3
1 29 gm1 3.7 10 S ra1 ra1 7.838 10
gm1

3 2 3
2 29 gm2 4.1 10 S ra2 ra2 7.073 10
gm2

3
1.2 29 gm1.2 gm1 gm2 gm1.2 7.8 10 S

1
1 1 3
ra1.2 ra1.2 3.718 10
ra1 ra2

3 3 3
3 29 gm3 3.3 10 S ra3 ra3 8.788 10
gm3

3 4 3
4 29 gm4 3.7 10 S ra4 ra4 7.838 10
gm4

3
3.4 29 gm3.4 gm3 gm4 gm3.4 7 10 S

1
1 1 3
ra3.4 ra3.4 4.143 10
ra3 ra4

12 12 12
Cg.c1 3.1 10 F Cg.a1 1.4 10 F Ca.c1 1.75 10 F
12 12 12
Cg.c2 3.1 10 F Cg.a2 1.4 10 F Ca.c2 1.75 10 F
12
Cg.c1.2 Cg.c1 Cg.c2 Cg.c1.2 6.2 10 F
224 14 Mathcad Worksheets Draft Designs

14.3 Fully Triode Driven Input Stage Page 3

12
Cg.a1.2 Cg.a1 Cg.a2 Cg.a1.2 2.8 10 F
12 12
Ca.c1.2 Ca.c1 Ca.c2 Ca.c1.2 3.5 10 F Cstray1.2 2 10 F
12 12 12
Cg.c3 3.1 10 F Cg.a3 1.4 10 F Ca.c3 1.75 10 F
12 12 12
Cg.c4 3.1 10 F Cg.a4 1.4 10 F Ca.c4 1.75 10 F
12
Cg.c3.4 Cg.c3 Cg.c4 Cg.c3.4 6.2 10 F
12
Cg.a3.4 Cg.a3 Cg.a4 Cg.a3.4 2.8 10 F
12 12
Ca.c3.4 Ca.c3 Ca.c4 Ca.c3.4 3.5 10 F Cstray3.4 2 10 F

6 3 6
C1 100 10 F C2 C1 C3 2.2 10 F C4 10 10 F C5 C4
3
fc1 500Hz fc2 1000Hz fc3 3500Hz fc4 10 10 Hz

gm1 gm2
fc1.2 fc1 fc2
gm1 fc2 gm2 fc1 fc1.2 678.261 Hz

gm3 gm4
fc3.4 fc3 fc4
gm3 fc4 gm4 fc3 fc3.4 5331.882 Hz

3. Circuit variales, gains, and output resistances :

3
R0mc 20 Shure V15V: R0mm 793 L0mm 0.3318H RL 10 10

3.1 Gain of second stage (t3 & t4 & output load) :

1
1 1 3
R6re R6re 19.29 10
R6 R9
1
1 1 3
R7re R7re 19.29 10
R7 R10

1
1 1
2 3.4
R6re 0.5 RL
G2nd G2nd 1.81
1
1 1
ra3.4 2 3.4 R7re 0.5 RL

3.2 Output resistance of second stage (t3 & t4) at the anode and cathode :
3
R1.3.4 ra3.4 1 3.4 R7 R1.3.4 721.143 10

3
R1.3.4.re ra3.4 1 3.4 R7re R1.3.4.re 582.835 10
14 Mathcad Worksheets Draft Designs 225

14.3 Fully Triode Driven Input Stage Page 4

1
1 1 3
Ro.a3.4 Ro.a3.4 23.133 10
R1.3.4 R6

1
1 1 3
Ro.a3.4.re Ro.a3.4.re 18.672 10
R1.3.4.re R6re

ra3.4 R6
rc3.4 rc3.4 934.762
1 3.4

ra3.4 R6re
rc3.4.re rc3.4.re 781.087
1 3.4
1
1 1
Ro.c3.4 Ro.c3.4 899.578
R7 rc3.4

1
1 1
Ro.c3.4.re Ro.c3.4.re 750.69
R7re rc3.4.re

3.3 Balanced output resistance of second stage (t3 & t4) :

2 ra3.4 R6re
Ro.bal Ro.bal 265.442
ra3.4 2 3.4 R7re

3.4 Gain (MM) of first stage (t1 & t2 & load-1st = infinite) :
vo.nom
Gamp.mm Gamp.mm 20
vi.nom.mm

Gamp.mm
G1st G1st 11.049
G2nd

Note: In the following calculation course I set R2 eff = R2+P1!

ra1.2 R2eff G1st ra1.2 3

G1st gm1.2 => R2eff R2eff 2.289 10
ra1.2 R2eff gm1.2 ra1.2 G1st

P1eff R2eff R2 P1eff 288.574

ra1.2 R2eff
G1st gm1.2
ra1.2 R2eff G1st 11.049

VR2.eff Ia1.2 R2eff VR2.eff 9.154 V

226 14 Mathcad Worksheets Draft Designs

14.3 Fully Triode Driven Input Stage Page 5

3.5 Gain (MC) of first stage (t1 & t2 & load-1st = infinite & Tr1) :

Gamp.mc n Gamp.mm Gamp.mc 200

3.6 Specific capacitances :

12 12 12 12
C6 47 10 F C7a 100 10 F C7b C7a C7c 47 10 F C7d 22 10 F

12
Ci1.2 1 G1st Cg.a1.2 Cg.c1.2 Cstray1.2 Ci1.2 41.938 10 F

12
Co.bal Cg.a1.2 Ca.c1.2 Cstray1.2 Co.bal 8.3 10 F

R2eff ra1.2
rc1.2
1 rc1.2 200.217
1.2
1
1 1
Ro.c1.2
R3 rc1.2 Ro.c1.2 155.687

1 3
C3 C3 2.045 10 F
2 fhp Ro.c1.2

4. Noise voltages and SNs calculations :

Fig. 14.18 = Fig. 13.7

14 Mathcad Worksheets Draft Designs 227

14.3 Fully Triode Driven Input Stage Page 6

4.1 Triodes noise voltages :

3.06
rN1.2 rN1.2 392.308
gm1.2
fc1.2 9
en.rN1.2( f) 4 k T B1 rN1.2 1 en.rN1.2( h) 3.384 10 V
f

3.06
rN3.4 rN3.4 437.143
gm3.4
fc3.4 9
en.rN3.4( f) 4 k T B1 rN3.4 1 en.rN3.4( h) 6.939 10 V
f

4.2 Noise voltage of the passive components :

NIe
20 6 9
Vishay / Beyschlag MBB 0207 NIe 30 [dB] NI 10 10 V NI 31.623 10 V

4.2.1 Effective noise voltage of RL :

9
en.RL 4 k T B1 RL en.RL 13.189 10 V

Ro.bal 12
en.RL.eff en.RL en.RL.eff 341.051 10 V
RL Ro.bal

4.2.2 Noise voltage of R5 :

9
en.R5 4 k T B1 R5 en.R5 2.288 10 V

4.2.3 Noise voltage of R0mc and the input transformer Tr1 :

2 3
R0mc.sec R0mc n R0mc R0mc.sec R0mc 2 10

2
Rpsec Rp n Rpsec 500
3
Ztr1 R0mc R0mc.sec R0mc Rpsec Rs Ztr1 R0mc 2.76 10

1
1 1 3
Zi R0mc Zi R0mc 2.686 10
Ztr1 R0mc R1a

R1a Rs 3
Ri.mc Rp Ri.mc 1.008 10
2
n
9
en.Zi R0mc 4 k T B1 Zi R0mc en.Zi R0mc 6.835 10 V
228 14 Mathcad Worksheets Draft Designs

14.3 Fully Triode Driven Input Stage Page 7

4.2.4 Noise voltage of a V15V cartridge connected to R i.mm = R1a||R1b :

12 12
Assumption : Ccable 150 10 F C0 Ci1.2 C6 Ccable C0 238.938 10 F

1
1 1 3
Ri.mm R1a R1b Ri.mm 47.368 10

3
Z0( f) R0mm 2j f L0mm Z0( h) 2.23 10

1
1 3
Z0a ( f) 2j f C0 Z0a ( h) 2.237 10
Z0( f)

1
1 3
Z0b ( f) 2j f C0 Z0b ( h) 47.249 10
Ri.mm

9
en.R0.mm 4 k T B1 R0mm en.R0.mm 3.714 10 V

9
en.Ri.mm 4 k T B1 Ri.mm en.Ri.mm 28.706 10 V

Z0b ( f)
en1( f) en.R0.mm 9
Z0( f) Z0b ( f) en1( h) 3.661 10 V

Z0a ( f)
en2( f) en.Ri.mm 9
Z0a (f) R en2( h) 1.332 10 V
i.mm

4.2.5 Frequency dependent effective noise voltage of R2 eff , R6, & R7 :

NIe
10 12 2
10 10 VR2.eff 9
en.R2.eff.ex( f) B1 en.R2.eff.ex( h) 6.033 10 V
ln( 10) f

9
en.R2.eff 4 k T B1 R2eff en.R2.eff 6.31 10 V

2 2 9
en.R2.eff.tot( f) en.R2.eff en.R2.eff.ex(f) en.R2.eff.tot ( h) 8.73 10 V

NIe
10 12 2
10 10 VR6 9
en.R6ex( f) B1 en.R6ex( h) 63.001 10 V
ln( 10) f
9
en.R6 4 k T B1 R6 en.R6 20.39 10 V

2 2 9
en.R6.tot( f) en.R6 en.R6ex( f) en.R6.tot( h) 66.219 10 V
14 Mathcad Worksheets Draft Designs 229

14.3 Fully Triode Driven Input Stage Page 8

en.R7.tot( f) en.R6.tot( f)

9
en.R10 4 k T B1 R10 en.R9 en.R10 en.R10 41.709 10 V

2 2
2 rc3.4.re 2 Ro.c3.4
en.R7.eff( f) en.R7.tot( f) en.R10
R7 rc3.4.re R10 Ro.c3.4

9
en.R7.eff( h) 2.128 10 V

2
1
2 1 1
2 R1.3.4.re 2 R1.3.4.re R6
en.R6.eff( f) en.R6.tot( f) en.R9
R6 R1.3.4.re 1
1 1
R9 R1.3.4.re R6

9
en.R6.eff( h) 64.085 10 V

4.2.6 Correlated noise voltage of R7 at the anode output, 100% correlated with the one at the cathode :
R6re
G0.cgs3.4 1 3.4 R6 G0.cgs3.4 24.696
re ra3.4
9
en.R7.a( f) en.R7.eff( f) G0.cgs3.4 en.R7.a( h) 52.562 10 V
9
en.R7.corr( f) en.R7.a( f) en.R7.eff( f) en.R7.corr( h) 50.434 10 V

4.2.7 Balanced output noise voltage & SN - MC-case :

2
2 ra1.2 9
en.o.2nd( f) G2nd en.R2.eff.tot( f) en.o.2nd( h) 82.163 10 V
ra1.2 R2eff
2 2 2
en.R7.corr( f) en.R7.eff( f) en.R6.eff( f)

2 2 2 2 9
en.o.mc f R0mc en.Zi R0mc en.rN1.2( f) G1st G2nd en.o.mc h 20 173.318 10 V
2 2 2 2
en.R5 G2nd en.o.2nd( f) en.RL.eff

20000Hz
1 2
en.o.mc f R0mc df
B1
20Hz
SNne.o.mc R0mc 20 log
vo.nom

SNne.o.mc 20 73.028 [dB]

230 14 Mathcad Worksheets Draft Designs

14.3 Fully Triode Driven Input Stage Page 9

Input noise voltage density (average in B20k) with input shorted :

20000Hz
1 2
en.o.mc f 0 df
B20k
20Hz
en.i.avg.mc 12
Gamp.mc en.i.avg.mc 535.767 10 V

A-weighting by A(f) :

1.259 1 1 1
A( f)
2 2 2 2
20.6Hz 107.7Hz 737.9Hz f
1 1 1 1
f f f 12200 Hz

RIAA equalizing by R(f) :

1
2
3 6
1 2 10 Hz 318 10 s
R1000
2 2 R1000 9.898
3 6 3 6
1 2 10 Hz 3180 10 s 1 2 10 Hz 75 10 s

2
6
1 2 f 318 10 s
R( f) R1000
2 2
6 6
1 2 f 3180 10 s 1 2 f 75 10 s

20000Hz
1 2 2 2
en.o.mc f R0mc A( f) R( f) df
B1
20Hz
SNariaa.o.mc R0mc 20 log
vo.nom

SNariaa.o.mc 20 80.31 [dB(A]

R0mc 0 1 100

76
77
78
79
[dB(A)]

80
SNariaa.o.mc R0mc 81
82
83
84
85
86
0 10 20 30 40 50 60 70 80 90 100
R0mc
[ohm]

Fig. 14.19 Output referred A-weighted and RIAA equalized SN vs. R0 of the MC input stage
14 Mathcad Worksheets Draft Designs 231

14.3 Fully Triode Driven Input Stage Page 10

4.2.7 Balanced output noise voltage & SN - MM-case:

2 2 2 2 2
en.o.mm( f) en1( f) en2( f) en.rN1.2( f) G1st G2nd
2 2 2 2
en.R5 G2nd en.o.2nd( f) en.RL.eff

9
en.o.mm( h) 127.287 10 V

20000Hz
1 2
en.o.mm( f) df
B1
20Hz
SNne.o.mm 20 log
vo.nom

SNne.o.mm 65.958 [dB]

20000Hz
1 2 2 2
en.o.mm( f) A( f) R( f) df
B1
20Hz
SNariaa.o.mm 20 log
vo.nom

SNariaa.o.mm 82.141 [dB(A]

6
1 10
[V/rtHz]

en.o.mc f 20
7
1 10
en.o.mm( f)

8
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

Fig. 14.20 Output noise voltage densities of the two phono-amp input stage versions,
inputs loaded
Measurement Tools and Trimming
15

15.1 Computer Test Equipment

15.1.1 Intro

Before we enter into the measurement activities, we need some information about
the new measurement system I am using since the middle of the year 2013. It is
called Clio 8.5 PCI1 and it is on Audiomatica’s list of discontinued products now.
The purchase of this system followed my basic attitude towards very expensive
measurement systems like eg the ones of the Audio Precision family: find ways
with cost saving approaches to get equally good measurement results. The price of
the Clio 8.5 system is roughly a factor of 1/20 smaller than that of the AP system. It
works with 18 bit and it runs under WIN XP. It follows the elder 16 bit Clio 6.5
system I’ve already described in the TSOS books. The basic measurement set-up
did not change, however, emphasis lies on the balanced approach here.
The system consists of a small PCI card and two double cables equipped with
Cinch connectors. They transfer data between the card and an external measurement
Cinch input/Cinch output box, called signal conditioner SC-02. The software itself
runs on WIN 2 k and WIN 7 computers, the signal conditioner works with WIN XP
only. With this arrangement, the user must measure with WIN XP but could per-
form the post-measurement work on other computers with more or less modern
operating software.
The most important feature of a measurement instrument should be the fre-
quency (F) and phase (P) response. For audio purposes, Clio’s F & P is perfect:
±0.000 dB and ±0.000° in the B20k range of 20 Hz–20 kHz.

1
www.audiomatica.com.

© Springer International Publishing Switzerland 2016 233

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_15
234 15 Measurement Tools and Trimming

Fig. 15.1 Input referred noise voltage density curve of the Clio 8.5 measurement system,
balanced input shorted

15.1.2 Signal-to-Noise Ratio

With the un-balanced (.u) inputs of the two channels A & B shorted the 18 bits limit
the input referred signal-to-noise ratio SNne.i.u of each channel to appr.
6.02 dB * 18 = −108.4 dBV in B20k. A switch to a balanced (.b) input configuration
yields an SNne.i.b of −111.8 dBV in B20k. With an FFT resolution FFTres = 0.73 Hz
Fig. 15.1 shows the frequency range of 2 Hz–20 kHz and the measured noise
voltage density curve of the system with its balanced input shorted.
To show the flexibility of the Clio system I add a better looking smoothed
version in Fig. 15.2. The FFT size (fs = 65,536) and the sampling rate (FS = 48 kHz)
are not changed here! Therefore, the smoothed version looks like the one after
≫100 averages and not alike one with reduced fs.2 Remember: each halving of fs
would lead to a doubling of FFTres and it would move up the Fig. 15.1 curve by
3 dB. Practically, the 1/f noise can be ignored; with fc.e = 200 Hz it plays a role in
the region of 20 Hz–200 Hz only.
With the given SNne.i.b and based on the Chap. 16 findings the best-case dif-
ference ‘B’ between the measurement amp’s input SN end the measured output SN
of the DUT should always become ≥16 dB (≥10 dB) for an error (=Worsening

2
See Fig. 22.6 in TSOS-2.
15.1 Computer Test Equipment 235

Fig. 15.2 Same as Fig. 15.1, however, smoothed by ½ Octave

Figure We) ≤0.1 dB (≤0.5 dB).3 In other words: the DUT’s measured output
referred SNne.o.b should then become ≤ −95.8 dBV (≤101.8 dBV). Values better
than that would include certain additional portions of Clio’s noise. For example, a
measured DUT output SNo.m value of −106.8 dBV creates a B = 5 dB difference to
the Clio SNne.i.b. Thus, as of Figs. 15.3 and 15.4 it includes We = 1.2 dB Clio noise.
Consequently, the real DUT output SNo.re becomes −106.8 dBV −1.2 dB
= −108 dBV. Therefore, the Clio system needs very low-noise external measure-
ment amplifiers at its input, allowing to measure DUT output SNs up to −120 dBV
(or even better) without Clio impact >0.1 dB. With a gain of 1000, the balanced-
in/balanced-out measurement amp PMMA of Chap. 16 will be able to manage such
a challenge.
With the measured output referred SNo.dut.m [dBV] of the DUT and the input
referred SNi.ma [dBV] of the measurement system the creation of the Figs. 15.3 and
15.4 curves is based on the following general equations:
0sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
SNo:dut:m SNi:ma ðBÞ
B C
B
We ðBÞ ¼ 20 log@ 10 10 þ 10 10 C  SNo:dut:m ½dB ð15:1Þ
A

3
See Figs. 15.3 and 15.4.
236 15 Measurement Tools and Trimming

Fig. 15.3 Worsening figure 10

We(B) as function of the
difference B of two SNs 1

[dB]
We (B) 0.1

0.01

1 .10
3
0 5 10 15 20 25 30
B
[dB]

Fig. 15.4 Zoomed Fig. 15.3 3

with B ≤ 5 dB 2.75
2.5
2.25
[dB]

We (B)
2
1.75
1.5
1.25
1
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
B
[dB]

0 1
SNo:dut:m  B
SNi:ma ¼ 20 log@10 20 A ½dBV ð15:2Þ

B ¼ SNo:dut:m  SNi:ma ½dB ð15:3Þ

15.1.3 Distortion (THD)

The Clio brochure claims a THD level of 0.01 % in B20k. I have gone through many
different THD measurement sessions and I found out that—without big error—
Clio’s 1 kHz signal is clean enough for our measurement purposes here. Figure 15.5
shows the spectrum of a 0 dBV output signal fed into one un-balanced input
channel; here it is the ‘A’ input. The ‘B’ input looks the same.
Without noise N, Clio measures THD = 0.006 %. A calculation with (15.4) and
the levels taken from the Fig. 15.5 graph (vd1 = 1 kHz at 0 dBV) yields a THD level
of 0.00592 %, a result very close to the rounded three-digit-after-the-decimal-point
Clio result. In Sect. 15.2, we will see how this result is improved by passing it
through the un-balanced/balanced converter UBC.
15.1 Computer Test Equipment 237

Fig. 15.5 Clio’s 1 kHz 0 dBV signal and its distortion artefacts in B20k

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
v2d2 þ v2d3 þ    þ v2d20
THD ¼  100 % ½% ð15:4Þ
vd1
 
THD ð%Þ
THD ¼ 20 log ½dB ð15:5Þ
100 %

15.1.4 IMD

With its broad range of test signals Clio allows all kinds of IMD measurements.
I have tried some of them. Finally, I took the following approach: a high-level
low-frequency signal and a low-level high-frequency signal are simultaneously fed
into the DUT’s input and the output FFT spectrum shows the sidebands around the
high-frequency signal. Thus, the amount of sidebands and their levels indicate
the total IMD amount produced by the DUT. This approach is the basis of the
DIN IMD measurement with its 250 Hz/8 kHz frequency and 4:1 level pair. I also
tried two other pairs: 60 Hz/3 kHz and 300 Hz/3 kHz. However, because the
results4 did not drastically change the DIN version became the customary one.

4
See Table 12.3.
238 15 Measurement Tools and Trimming

Fig. 15.6 The Clio IMD measurement result with 250 Hz/8 kHz and 4:1 signal levels

Figure 15.6 shows Clio’s IMD FFT spectrum according to DIN. The SMPTE
RP120 approach results look similar to the DIN ones, however based on a different
frequency pair: 60 Hz/7 kHz. The DIN result without noise N becomes
IMD = 0.0153 %. With the below given equations the calculation result lies very
close to the measured one: 0.0132 %.
With f1 = 250 Hz, f2 = 8 kHz, A = amplitude, and n = 2–6 the equations look as
follows:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
IMD ¼ IMD22nd þ IMD23rd þ    þ IMD2nth ð15:6Þ

Aðf 2 nf 1 Þ þ Aðf 2 þnf 1 Þ

IMDnth ¼ ð15:7Þ
Af2

In Fig. 15.6 the distortion artefacts of the 250 Hz signal <6 kHz and the sideband
artefacts around 16 kHz are not part of the measurement result. Here again, the
un-balanced/balanced converter UBC will improve the results too.
There is another method to test IMD: the CCIF or ITU-R approach with two
different frequencies of equal amplitude and 1 kHz difference, placed in the higher
region of B20k, eg at 14 and 15 kHz or at 19 and 20 kHz. The CCIF IMD result was
0.0024 %.
15.2 The Un-Balanced to Balanced Converter UBC 239

Fig. 15.7 The un-balanced to balanced converter UBC

15.2 The Un-Balanced to Balanced Converter UBC

15.2.1 Circuit

Although equipped with a balanced input one of the disadvantages of the Clio
system is the fact that there is no balanced output. Roughly, 25 years ago, I needed
a balanced signal and that is why the then developed UBC in Figs. 15.7 and 15.8
looks a bit old-fashioned with its rather high number of discrete components.
Nevertheless, it works sufficiently well and it does not add negative extra signals to
the measurement signals we need here. The circuit is a modified version of a
transformer replacement for differential line drivers that I found in a 1984 Wireless
World magazine,5 and again in Douglas Self’s ‘Small Signal Audio Design’ book.6
I show a less component-burning booster alternative too; it works well in other
configurations (eg Module 2 of TSOS-1 & -2) and according to the respective
data-sheet, however, I did not yet test it in the UBC configuration.
The power supply lines of the UBC have ±15 V regulated DC voltages and they
are further cleaned by gyrators (= C-multipliers).

15.2.2 F & P and SN Performance

The frequency (F) and phase (P) response of the UBC is flat in B20k. With an output
signal of 0 dBV via output load = 10 kΩ at PL01 or of −66 dBV via output

WW 1984–2012, p 73, ‘Differential line driver replaces transformer’ by S. Whitt.

5

‘Quasi floating balanced output’ by Douglas Self, Small Signal Audio Design, 2010, p. 387.
6
240 15 Measurement Tools and Trimming

Fig. 15.8 Booster alternatives for Fig. 15.7

load = 1 kΩ at J03 the deviation from the F flatness becomes only −0.001 dB at
20 kHz in both cases.
The deviation from the P flatness shows different pictures at 20 Hz/1 kHz/
10 kHz/20 kHz:
• with 0.0 dBV output signal we have a phase of +0.12°/0.00°/−0.67°/−1.21°,
• with −66 dBV output signal we find a phase of +0.54°/0.00°/−1.90°/−3.77°.
If we would assume a −6 dB/oct. hp & lp character +0.54° and −3.77° represent
−3 dB corner frequencies of roughly 0.2 Hz and 300 kHz.
With input shorted and because of the still existing noise generator at the input
(P1 with appr. 12.25 kΩ noise-effective resistance) the output referred SN at the
PL01 output shows an SNne.o.re of −107.9 dBV and −107.7 dBV only,7 measured
with the measurements amps of Chaps. 16 (PMMA) and 17 (PFMA). At the

7
Calculation results see MCD-WS 18.1; based on the Figs. 15.3 and 15.4 process subscript “re”
refers to real results.
15.2 The Un-Balanced to Balanced Converter UBC 241

J03 output we find SNne.o.re = −116.82 dBV measured with the PFMA and SNne.o.re
= −136.16 dBV measured with the PMMA.8 These J03 results make no sense
because they only reflect the input referred SNs of the measurement amps. The J03
output offers a calculated SNo of appr. −154.4 dBV = −108.4 dBV (= calculated
PL01 output SN) –46 dB (= voltage divider effect at J03).
The PMMA offers a measured input referred SNne.i.m of −138.18 dBV
(≡ 0.875nV/rtHz). According to Fig. 15.3 and with B = 30.27 dB the original
measured result of −107.89 dBV at the PL01 output could be corrected by a tiny
We(B) = 0.004 dB, hence, we would get −107.894 dBV real SNne.o.re. At the J03
output B becomes −16.677 dB, hence, we can ignore any correction.
The PFMA offers a measured input referred SNne.i.m of −117.3 dBV
(≡ 9.7nV/rtHz). According to Fig. 15.3 and with B = 9.99 dB the original measured
result of − 107.31 dBV at the PL01 output must be corrected by 0.41 dB, hence,
we’ll get −107.72 dBV real SNne.o.re. At the J03 output, B becomes negative
(−37.037 dB), hence, we have nothing to correct.
The approximate difference of 20 dB input referred SN comes from the rather high
input referred noise voltage density of the PFMA9: calculated 10.6nV/rtHz average.

15.2.3 THD Performance

Provided that UBC’s own un-balanced THD level is lower than the one of the
preceding generator and if we turn the un-balanced signal into a balanced format the
advantages come from the fact that the generator’s THD gets damped by the UBC,
especially the even harmonics. Therefore, the Fig. 15.9 looks so much better than
the one of Fig. 15.5, and with a 1 kHz input signal at 0 dBV the THD level
improves by approximately 10 dB to 0.002 %.
Figures 15.10 and 15.11 show what happens with the harmonics by decreasing
the input signal to −46 dBV (MM case) or −66 dBV (MC case): nearly all har-
monics disappear in the noise mud formed by the noise voltage density level at
approximately −156 dBV.
At this point of the survey, it became clear to me that the effective disappearance of
harmonics in the DUT’s noise level became a major trigger of the Sect. 12.3 con-
clusions and the integration of the SNriaa line at −67.3 dBV in the Sect. 12.2.2 graphs.
Of course,
• the results at the J03 output become always better, because this output follows a
−46 dB voltage divider, formed by resistors R5, R20, R12, R13, R29,
• the spikes of the IMD measurement result (Fig. 15.6) will disappear in the noise
ground too, hence, the astonishing Table 12.3 measurement results should not
be very surprising.

8
Details & calculation results see Chaps. 16 and 18’s MCD-WS 18.2.
9
Details & calculation results see Chaps. 17 and 18’s MCD-WS 18.3.
242 15 Measurement Tools and Trimming

Fig. 15.9 THD at UBC’s PL01 output, fed by a 0 dBV/1 kHz signal

Fig. 15.10 THD at UBC’s J03 output, fed by a 0 dBV/1 kHz signal
15.2 The Un-Balanced to Balanced Converter UBC 243

Fig. 15.11 THD at UBC’s J03 output, fed by a −20 dBV/1 kHz signal

The levels of the noise voltage density grounds of Figs. 15.9, 15.10 and 15.11
show a difference of approximately 28 dB, from −125 dBV to −153 dBV. Two
sources produce this difference:
• The above-mentioned voltage divider moves every input signal 46 dB down.
• The then effective input referred noise level of the Clio system moves the noise
ground up to the balanced Clio input referred rms noise voltage level of
−111.8 dBV or −156 dBV noise voltage density level (see Figs. 15.2 and 15.3).
In conjunction with Clio’s input noise current, it includes a guessed 3 dB
penalty for the yet undefined noise production mechanisms at the Clio input.

15.2.4 Output Resistances

The measured UBC’s PL01 output resistance Ro.pl01 becomes 45.5 Ω. With the
shown components in Fig. 15.7 the output resistance at the J03 output Ro.j03
becomes 11.13 Ω. A change of R15 from 63 Ω (20 Ω + 43 Ω) to 53.2
(20 Ω + 33.2 Ω) reduces the output resistance to 10.8 Ω. All measurements were
performed with this output resistance; however, instead of 0.5 mVrms with the
shown R15 value the output voltage is 0.48 mVrms with a 0.1 Vrms/1 kHz input
signal.
244 15 Measurement Tools and Trimming

15.3 RIAA Encoder and Trimming

15.3.1 Encoder

The Fig. 15.14 encoder circuit goes back to the one I have already published in
TSOS-1 & -2,10 MC output not shown here. For trimming purposes of the RIAA
transfer function and to drive the external input we only need a 50 mVrms–100 mVrms
output level in B20k. Thus, the inclusion of a jumper across R17 ensures a 50 mVrms/
1 kHz signal at the MM output. A further increase of the input voltage to 2 Vrms will
increase the output level by 6 dB.

15.3.2 Trimming Actions

Before insertion into the main board’s amplification chain, with the help of a special
small test-board case (Figs. 15.12 and 15.13) we have to undertake separate trim-
ming processes for Amp1, Amp2, and Amp5, DC offset as well as gain. A first B20k
frequency and phase response, THD and IMD, and overload control check follows
by application of the Clio generator + UBC + UBC-J03 output. Additionally, SNs
must be measured and crosschecked with the calculation results.
Fed by the Clio frequency sweep generator with very constant output level in
B20k the Fig. 15.14 encoder solution allows the trimming of Fig. 1.2’s P1, P3, and
P4. The operating level should be set to 50 mVrms–100 mVrms at 1 kHz. The
encoder plus UBC & PL01 must be connected to the external input of the engine.
The following steps, a look at Fig. 1.2, and Clio’s frequency response diagram
(sinusoidal analysis) give a guideline through the frequency response trimming,
provided that the trimming of Amp3’s P7 ensures equal amplitudes at TPs 5 & 6 of
Fig. 2.1:
1. We start with Amp3 and P3. At the low-end of B20k any trimming action of the
T1/T3 network gives the perfect anchor point at 20 Hz and P3 moves the found
frequency response to a level where it creates a nearly horizontal line towards
1 kHz.
2. Next, P1 comes into the game: with it, we can move the high-end of B20k to a
level position equal to the one at 20 Hz. With perfectly selected component
values of the three RIAA networks, we should obtain a rather flat frequency
response in B20k.
3. If the frequency response is not flat enough we repeat the first two steps.
However, wrong values of the T1/T3 network cannot be trimmed away.
4. We switch to Amp4 and—alike the process of 1.—with P4 we trim the found
frequency response at the low-end of B20k. Automatically, the already trimmed
T2 network will move the frequency response at the high-end of B20k into a
flatness creating position.

10
TSOS-1: Chap. 12; TSOS-2: Sect. 23.3.
15.3 RIAA Encoder and Trimming 245

Fig. 15.12 Test-board case with Amp1

Fig. 15.13 Circuit of the test-board

246 15 Measurement Tools and Trimming

Fig. 15.14 RIAA encoder for RIAA transfer function trimming actions via external input of the
Engine II

We have two flat frequency responses now with different levels at 1 kHz. Hence,
the next step must be the trimming of equal gains for Amps 3 & 4. The gain of Amp3
is fixed because of the chosen components; thus, with a 50 mVrms–100 mVrms/1 kHz
signal at the external input the gain of Amp4 must be trimmed to the Amp3 gain
(P9 of Fig. 4.1).
Additionally, all DC offset trim pots of the main board must ensure an output DC
level of 0.00 V. It is highly recommended to trim the Fig. 1.2 trim pots of OPs 1 & 2
to 0.00 V output DC-level, because, if not properly trimmed to zero volts, they
would additionally influence the triode stage’s anode current.
Before the listening tests can take place the overall frequency and phase
responses, the overall THD, IMD, and SN measurements with input loaded become
the final check procedures.
The Very Low-Noise Balanced
Measurement Amp PMMA 16

16.1 Intro

The measurement of noise in the balanced world requires special measurement

amplifiers (MA). The best case would be one with balanced input and balanced
and/or un-balanced output, being capable to feed one of the many computer based
measurement solutions (CMS), like eg the Clio system. Then, in this kind of
environment, the task of the MA is to lift the DUT noise level far above the input
noise level of the CMS. Finally, the CMS reading (in −dB) minus the MA gain (in
dB) will show the output noise level of the DUT. This method allows the mea-
surement of signal and noise levels in the region of 1 µVrms balanced, a low-level
region nowadays mainly occupied by very high priced audio analyzers from eg
Audio Precision or Rohde und Schwarz.
Recently, in stereoplay 2013-03, I read an article on the “NAD M51” DA con-
verter. For me the most astonishing thing was the measurement result of the
A-weighted output referred SNne.o.a = −122 dBV(A)—and hum-free. Based on the
FFT diagram in the review, the noise curve shows white noise behaviour with a rather
low 1/f noise corner frequency of approximately 350 Hz. I have transferred the Audio
Precision SYS 2722 measurement graph into the rough representation of Fig. 16.1.
The 350 Hz corner frequency 1/f-noise effect worsens the otherwise nearly pure
white noise signal in B20k by appr. 0.5 dB. Additionally, the A-weighting effect
counts another 2.05 dB. Theoretically, the corresponding purely white noise output
signal would thus become −122 dB(A) + 2.05 dB − 0.5 dB = −120.45 dBV (NAD
data sheet spec: −125 dB ref. 2 Vrms ≡ −119 dBV, incl. 1/f-noise, of course).
Measured in B20k this extremely low SN level stands for an average output noise
voltage density of 6.72 nV/rtHz. The balanced output resistance shows a measured
141 Ω (110 Ω = NAD spec).

A shorter version of this chapter was firstly published in Jan Didden’s Linear Audio Vol. 7, 2014-03,
‘A poor man’s measurement amp (PMMA)’.

© Springer International Publishing Switzerland 2016 247

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_16
248 16 The Very Low-Noise Balanced Measurement Amp PMMA

Fig. 16.1 Output voltage noise density curve of the NAD M51

Nevertheless, the measurement result is not the truth! The above-mentioned

SN of −122 dBV(A) includes a certain portion of AP’s unweighted input referred
SNi.AP = −122.4 dBV of the channel that was used for the measurement. With
balanced inputs shorted, stereoplay’s four channels in two AP SYS 2722 analyzers,
all equipped with 20 kHz brick wall S-AES17 low pass filters, show input referred
SNs between −122.0 and −122.4 dBV (≡ −119.0 dBu to −119.4 dBu in
B22k = 22 Hz to 22 kHz with 3rd order lp à la IEC 468).
The identification of a 1 µVrms signal spike in an FFT diagram seems not to be a
problem, as long as the analyzer’s FFT resolution can be set to eg ≤5 Hz.
Depending on the sample rate and FFT size setting and as long as the analyzer’s
input referred noise voltage density is low enough (practically without impact on
the measured signal) the representation of a B20k 1 µVrms white noise signal on the
screen will show a noise level far lower than −120 dBV. In any case, it should be
lower than an average level of 7.07 nV/rtHz. Why? See further down in Sect. 16.2.
I do not want to debate what is the ‘right’ lowest noise level and thus the highest
possible SN value of HiFi equipment. In Douglas Self’s LA Vol. 5 article,1 Table 4,
we can easily study what we can achieve in that field. Because of their extremely low
value all measured figures had to be corrected according to the AP noise impact. To
get dBV in B20k simply subtract 2.22 dB + 0.81 dB = 3.03 dB from the table’s dBu
values; here, 0.81 dB comes from the reduction of the bandwidth (0.41 dB) plus the
lp change (0.4 dB) from 3rd ord. Butterworth to the brick wall version).
Thus, I guess an amplifier output referred SN benchmark of −120 dBV in B20k is
still an extreme value and as music listener, I can live with it—even in the quietest
moments. However, seen from a development point of view, we have some open
questions on an MA that should be capable to measure such extremely low SNs
with lowest possible impact on the measurement result. With that in mind, in this
chapter I will try to answer at least the following questions:
1. What should be the level of input noise voltage density of an MA with balanced
input that causes the low output noise voltage density of a DUT with a measured
SNo.dut.m = −120 dBV/B20k to be off eg 0.1 dB maximum only?

1
Linear Audio Vol. 5, 2013-04, ‘A low-noise preamplifier with variable-frequency tone controls’,
Douglas Self.
16.1 Intro 249

2. How does the DUT’s output resistance additionally worsens the measurement
result and what are the roles of the MA’s input resistance Rin and its input noise
current density in.i?
3. Based on the following three subgoals how could a rather low budget and easy
to built measurement-amp look like?
3:1. Flat frequency and phase response in B20k (±0.02 dB/±10°)
3:2. Overall gain: 1000
3:3. Balanced input and balanced/un-balanced outputs
4. How can we calculate the noise performance of that new MA?
We will see further down that the chosen PMMA concept fulfils all
above-mentioned goals.

16.2 The Input Noise Voltage Density Question

Before we start developing the MA and as consequence of the first question and its
−120 dBV goal and in order to find the MA’s input referred noise voltage density
allowed (en.i) we have to go through some math first.2 The equations for the
required MA input referred SNi.ma look as follows:
With the DUT output SNo.dut.m [dBV] < SNi.ma [dBV]3 and with

B = SNo:dut:m
SNi:ma ½dB ð16:1Þ

we obtain the MA’s allowed input referred SNi.ma as function of B as follows:

0 1
SNo:dut:m
B
SNi:ma ðBÞ = 20log@10 20 A ½dBV ð16:2Þ

Application of (16.2) in the following equation will lead to the expression for the
allowed impact We(B) [dB] of the MA on the DUT. It should become—in our case
here—0.1 dB.
0sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
SNo:dut:m SNi:ma ðBÞ
B C
We ðBÞ = 20logB
@ 10 10 þ 10 10 C
SNo:dut:m ½dB
A ð16:3Þ

Figures 16.2a, b show the corresponding graphs.

2
Just to keep the overall picture in one chapter, I repeat Eqs. (15.1)– (15.3) and Figs. 15.3 and 15.4
here.
3
Here, “<” means worse! The other way around (>) makes no sense!
250 16 The Very Low-Noise Balanced Measurement Amp PMMA

10
(a)

1
[dB]

We ( B) 0.1

0.01

3
1.10
0 5 10 15 20 25 30
B
[dB]

(b) 3
2.75
2.5
2.25
[dB]

We ( B) 2
1.75
1.5
1.25
1
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
B
[dB]

Fig. 16.2 a Worsening Figure We(B) as function of the difference B of two SNs. b Zoomed
version of Fig. 16.2a for correction purposes of two SNs with values that are close together
(B ≤ 5 dB)

150
155
160
en.Rc.e ( f) 165
170
[dBV]

en.T1.e ( f) 175
en.T2.e ( f) 180
185
190
195
200
3 4 5
10 100 1.10 1.10 1.10
f
[Hz]

Fig. 16.3 Noise voltage density situation with a corner frequency of 1 kHz (incl. tangents)
16.2 The Input Noise Voltage Density Question 251

Based on the calculated range of Worsening Figures (y-ordinate), at the allowed

Fig. 16.2a value of We(B) = 0.1 dB we can now pick-out the delta B (x-ordinate)
between the DUT’s output referred SNo.dut and the SNi.ma of the MA: it becomes
16.33 dB. Hence and in other words: to ensure a ≤ 0.1 dB MA impact down to a
DUT output SN level of −120 dBV on all kinds of SN measurement in B20k we need
an optimal MA input referred SNi.ma of −136.33 dBV = −120 dBV − 16.33 dB.
Thus, transferred into a white noise voltage density level flat over the complete audio
band we need an MA that offers at least a balanced average input noise voltage
density ≤1.08 nV/rtHz, hence, without 1/f-noise corner frequency in B20k.
If we would, for example, allow a 1/f-noise corner frequency of 1 kHz, the
average noise voltage density in B20k would grow to −135.05 dBV (see Fig. 16.3)
and the MA’s impact on the DUT’s SNo.dut would also grow to We(B) = 0.134 dB.
If we would keep the original −136.33 dBV level and we would accept the example
corner frequency of 1 kHz, then, to keep an average of 1.08 nV/rtHz in B20k the
white noise region >1 kHz of the MA’s input noise voltage density has to drop from
1.08 nV/rtHz to a balanced 0.93 nV/rtHz.
Thus, to keep things on the safe side and as derived sub-goal for our purposes
here the PMMA should offer an input noise voltage density of ≤ 1 nV/rtHz average
in B20k with no 1/f-noise corner frequency. The permitted amount of input noise
current density is part of the next section’s discussion.

16.3 The Roles of the MA Input Resistance and Input

Noise Current Density

16.3.1 Influence of the DUT Output Resistance

I think it is obvious that a voltage divider effect between the DUT’s total output
resistance Ro.tot and the MA’s total input resistance Rin.tot will lead to an improved
meter reading of the output referred SNo.dut.m. This would happen if we would keep
the reference level like the one without voltage divider effect.
However, if we would reduce the idle case reference level according to the
voltage divider ratio, the measured SNo.m of the DUT would become the DUT’s
real one. To get the real SNo.r with reference to the original reference level (in dBu
or dBV) we have to compensate this kind of erroneous improvement by simply
adding the magnitude of the voltage divider gain–loss Gloss1.e = 20log(Gloss1) to the
measured SNo.m. Hence, we obtain the real SNo.r as

SNo:r = SNo:m þ jGloss1:e j ð16:4Þ

In Fig. 16.4 the DUT has two kinds of output resistances; Ro1 and Ro2 represent
the real internal output resistance of the DUT. Not taking into account any external
noise producing influences, we assume that the whole noise generation of the DUT
already covers their internal electronic noise. A typical example for this kind of
252 16 The Very Low-Noise Balanced Measurement Amp PMMA

Fig. 16.4 General output situation of the DUT (left) and balanced input of measurement
instrument (right)

output comes from the valve world: the anode output of a common cathode (or
common grid) gain stage. In most cases, op-amp outputs offer an output resistance
close to 0 Ω, I guess in most cases < 10 Ω in B20k.
By bridging the two input capacitances Cin1 & Cin2, the gain–loss Gloss1.e thus
becomes:
 
Rin:tot
Gloss1:e = 20log ð16:5Þ
Ro:tot þ Rin:tot

Rin:tot = Rin jj ðR1 þ R2Þ

ð16:6Þ
Ro:tot = Ro1 þ Ro2 þ Rout1 þ Rout2

Together with the following cable capacitance, the two resistors Rout1 and Rout2
play an additional low-pass role in cases of very low Ro1 + Ro2. Many manufac-
turers put them in place in order to ensure a high-frequency cut far outside B20k. In
conjunction with the input noise current of the next stage, these DUT internal
resistances may produce significant portions of additional noise.
Thus, both kinds of resistances play a role in the gain–loss but can also be
significant producers of additional noise.

16.3.2 The Input Noise Current Density Question

Figure 16.5 shows the situation of the Fig. 16.4 based noise sources. Here, en.o.dut.eff
is the DUT’s noise voltage after the voltage divider, however without noise voltage
impact of the MA input.
16.3 The Roles of the MA Input Resistance and Input Noise … 253

Fig. 16.5 Effective output

noise voltage of the DUT

Fig. 16.6 The creation of

the total input noise voltage of
the measurement amp

The noise voltage impact of the MA comes into the game in Fig. 16.6. It is
composed of several sources, such as the Fig. 16.5 ones and those of a typical
balanced input stage. Here, en.i1 and en.i2 are the input noise voltage densities of the
MA’s input stage, both of equal value en.i, series connected, thus increasing their
combined value by √2. The input noise current densities in.i1 and in.i2 are both of
equal value in.i. Also connected in series they decrease their combined value by √2.
Figure 16.6 shows the result after connecting the DUT output with the MA
input: all noise sources form what I call the total input referred noise voltage density
en.i.tot at the (+) and (−) input of the MA.
For calculation purposes, we need the following equations:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2
in:i
en:i:tot ¼ 2en:i þ
2 p ffiffi
ffi Rin:rel þ e2n:R:in:eff þ e2n:o:dut:eff ð16:7Þ
2

en:R:in:eff = en:R:in:tot Gloss2

Rin:rel = Rin:tot jjRo:tot
ð16:8Þ
Ro:tot
Gloss2 ¼
Ro:tot þ Rin:tot
254 16 The Very Low-Noise Balanced Measurement Amp PMMA

In (16.7), 2nd term, we can see that, as a result of a noise voltage, the input noise
current density flows also through the DUT output resistors Rout1 and Rout2 and
resistances Ro1 and Ro2. Here, their sum plays by far the major role because of the
relatively high value of Rin.tot in Rin.rel [see (16.6) and (16.8)]. According to
Fig. 16.5 and with a noiseless Rin.tot we can calculate the 4th term.
The inclusion of Ro1 and Ro2 into the 2nd term is questionable. Imagine an
anode output of a triode with an anode load resistor Ra. Thus, in case of a bridged
cathode resistor (by a capacitance) the output resistance becomes Ra || ra
(ra = internal triode anode resistance). I doubt that the input noise current of the MA
will also flow through the triode against the triode’s electron current from cathode
to anode, hence, from anode to cathode. Consequently, in (16.5) we should take Ra
as the value for Ro1 and Ro2 (in case of a balanced output stage). As long as the
value of Ra is unknown, we can take the measured or data sheet indicated output
resistances of the DUT.
I have calculated a graph that demonstrates the respective relationships.
Figure 16.7a shows the additional noise voltage creation if we measure a DUT with
output resistances in the range up to 10 kΩ. Starting point is 1 nV/rtHz and I took three
different input noise current density versions: BJT input devices with 1 pA/rtHz (solid
and dotted traces and subscript a), 2 pA/rtHz (dashed and dashed-dotted traces and
subscript b), and FET input devices with 50 fA/rtHz (dotted and solid traces and
subscript c). Additionally, I have chosen two different MA balanced input resistances:
18.182 kΩ (PMMA and subscript 1) and 200 kΩ (AP and R&S and subscript 2).
At the right-hand scale of Fig. 16.7a I have added the trace values for Ro = 10 kΩ too.
It should allow easier identification of the traces in the left graph.
Figure 16.7b shows the zoomed version for the 18.182 kΩ MA input resistance
case, up to Ro = 2 kΩ versus the three different input noise current densities.

16.3.3 Conclusions

With an MA input resistance of 18.182 kΩ a look at Fig. 16.7b signals that there is
no big difference up to 2 kΩ output resistance between nearly no input noise current
and a rather high one of 2 pA/rtHz. Additionally, for all shown Fig. 16.7a cases up
to 2 kΩ the input resistance difference does not play a major role either. However,
we can see that 180 Ω output resistance doubles the 1 nV/rtHz (= −180 dBV) input
noise voltage density goal. With that we no longer can maintain the 0.1 dB
worsening goal for the measurement of a 1 µVrms (= −163 dBV average noise
voltage density level) DUT output noise signal. With 2 nV/rtHz (= −174 dBV) the
difference B to −163 dBV is only 11 dB; hence, as of Fig. 16.2, the worsening
figure grows to 0.33 dB. At eg 1 kΩ output resistance we obtain thus We = 1.2 dB.
To keep both the complexity as well as the required budget of the whole MA
circuit as low as possible (Occam’s Razor!) I will adequately take the We effects into
account and my input noise current density goal for the MA should be set to
in.i ≤ 1 pA/rtHz. In such a high gain and low-noise configuration FET input op-amps
do not offer any advantage over their BJT counterparts. Although their input noise
16.3 The Roles of the MA Input Resistance and Input Noise … 255

7
(a) 1.10

( 4 )
en.i.b2 10 Ω = 18.24 × 10
−9
V

( )
en.i.a1( R0)
4 −9
en.i.a2 10 Ω = 14.022 × 10 V
en.i.b1( R0)

en.i.a2( R0) ( 4 )
en.i.b1 10 Ω = 12.378 × 10
−9
[V/rtHz]

V
8
1.10
( )
en.i.b2( R0)
4 −9
en.i.c2 10 Ω = 12.304 × 10 V
en.i.c1( R0)

en.i.c2( R0) ( 4 )
en.i.a1 10 Ω = 9.527 × 10
−9
V

( 4 )
en.i.c1 10 Ω = 8.367 × 10
−9
V

9
1.10 4
0 5000 1.10
R0
[ohm]

(b) 1.10
8

en.i.a1( R0)
[V/rtHz]

en.i.b1( R0)

en.i.c1( R0)

9
1.10
0 200 400 600 800 1000 1200 1400 1600 1800 2000
R0
[ohm]

Fig. 16.7 a DUT output resistance versus various MA total input noise voltage variants.
b Zoomed version of Fig. 16.7a (for Rin.tot = 18.182 kΩ only)

current impact is marginal, the chance for a very much higher 1/f-noise corner fre-
quency is a given fact. With that in mind the two chosen input noise density values are
thus in line with the same values of the un-balanced MA that is already described in
Fig. 11.1 (TSOS-1) and Fig. 22.1 (TSOS-2), however, now in a fully balanced mode.
Nevertheless, as input devices for MAs with gains ≤100 and input noise voltage
densities in the range of 3–6 nV/rtHz, FET input op-amps can very well be used,
like eg the ones of the OPA627 or OPA827 family. Using those, noise voltage
loaded outputs of valve gain stages with high output resistances >10 kΩ can eas-
ily be measured, with a very low MA noise impact on the measured signal.
256 16 The Very Low-Noise Balanced Measurement Amp PMMA

By changing the MA input resistance to 1 MΩ, or even higher, such a gain stage
could also be fit into the PMMA configuration presented here, with only one
balanced input stage.

16.4 The Final PMMA

16.4.1 Principal Circuit Approach

Low complexity means application of rather easy to handle circuits with op-amps or
special measurement ICs. I do not know a single IC that offers the specs I want to
have; hence, I took the traditional approach by arranging two input op-amps in a
balanced in/out configuration and a third one that creates the un-balanced o/p. This
is nothing else but the well-know instrumentation amplifier (INA), its principal
circuit is given in Fig. 16.8.
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Unfortunately, to create 1 nV/rtHz = 2  en.i average input noise voltage
density in B20k I also do not know an op-amp that offers an average input noise
voltage density of en.i = 0.707 nV/rtHz in B20k. However, the only one I know that
offers an input noise current density of 1 pA/rtHz also offers 0.85 nV/rtHz input
noise voltage density in B20k, the LT1028. To get a lower noise voltage density we
must thus work with a paralleled input stage of four op-amps and two additional
summing op-amps. Hence, the final MA consists of seven op-amps à la Fig. 16.9.
The resistors Rf1, Rf2, and R should have 0.1 % tolerance. Variable RG1 and
RG2 allow the setting of equal gains for each sub input stage up to the output via
the summing OPs 5 and 6. We obtain thus a gain GM.bal = 1000 ≡ +60 dB at the
balanced output. The common mode gain GCM.bal of that stage becomes 2, 1 for the
input stage with Amp A and Amp B and 2 for the output stage (see Fig. 16.10),
hence, the common mode rejection ration CMRR of the PMMA becomes (at least)
54 dB = 20 * log(GM.bal/GCM.bal).

Fig. 16.8 Principal INA circuit

16.4 The Final PMMA 257

Fig. 16.9 Principal final MA. (+) and (−) at the input and outputs indicate the phase relationship

Fig. 16.10 Relevant circuits for the gain calculation process

258 16 The Very Low-Noise Balanced Measurement Amp PMMA

OP7 serves as subtractor with a gain of 1. Consequently, the gain GM.unbal for the
PMMA’s balanced to un-balanced conversion becomes +60 dB too. Its CMRR
offers an additional 60 dB.

16.4.2 Gain Calculations

The calculation of the various PMMA gains looks a bit complex. Derived from the
Fig. 16.9 principal circuit Fig. 16.10 shows all gain-calculation relevant stages and
voltages with the arrows in the right phase direction. To get the balanced gain GM
from input to output we have to consider several different balanced and un-balanced
gains of all gain stages. Additionally, in Fig. 16.10, the first gain stage is composed
of the two sub gain stages A and B. Their un-balanced output voltages are
summed-up by the second gain stages.
The following equations are the result of the derivation process given in detail in
the Mathcad Worksheet MCD-WS 18.2, Point 6:
• The balanced PMMA gain GM:
The input gain stage (sub gain stages Amps A and B = 1st gain stage with
OP1–OP4) together with the summing output gain stage (=2nd gain stage with OP5
and OP6) form the differential (balanced in/balanced out) PMMA.
• With RG1 = RG2 = RGM the differential overall gain GM becomes:
vo:b
GM ¼ ¼
1000
vi:b
  ð16:9Þ
2 Rf1 Rf2
¼
2 1þ
RGM R

• With Rf2 = R (recommended) we obtain:

 
2 Rf1
GM ¼
2 1 þ ð16:10Þ
RGM

According to Fig. 16.10, the “−” sign represents a 180° phase inversion.
• The two differential gains GA and GB of the sub gain stages Amps A and B
become:

2 Rf1
GA ¼ 1 þ
RGM
¼ 500
ð16:11Þ
2 Rf1
GB ¼ 1 þ
RGM
¼ 500
16.4 The Final PMMA 259

• The two gains GA.ub.1 and GA.ub.2 of Amp A’s balanced in/un-balanced out sub
gain stage path become:
vo1
GA:ub:1 ¼ ¼ 250
vi:b ð16:12Þ
¼ 0:5 GA

vo2
GA:ub:2 ¼ ¼
250
vi:b ð16:13Þ
¼
0:5 GA

• The two gains GB.ub.3 and GB.ub.4 of Amp B’s balanced in/un-balanced out sub
gain stage path become:
vo3
GB:ub:3 ¼ ¼ 250
vi:b ð16:14Þ
¼ 0:5 GB

vo4
GB:ub:4 ¼ ¼
250
vi:b ð16:15Þ
¼
0:5 GB

• The gain of the second gain stage wth its un-balanced in/un-balanced out
becomes:
The second gain stage consists of two inverting and summing op-amps (OP5
and OP6). With Rf2 = R their gains Gop5 and Gop6 become −1 for each
un-balanced input path. Hence, we obtain the following set of equations:

vo:b ¼ vo5
vo6 ð16:16Þ

vo5 ¼
vi:b ðGA:ub:1 þ GB:ub:3 Þ

ð16:17Þ
vo6 ¼
vi:b ðGA:ub:2 þ GB:ub:4 Þ

=>
vo:b
GM ¼
vi:b ð16:18Þ
¼
ðG A + G B Þ
260 16 The Very Low-Noise Balanced Measurement Amp PMMA

16.4.3 Noise Calculations—Rule-of-Thumb SN Calculation

Approach

Chapter 18’s MCD-WS 18.2 gives the full details of the calculation course.
However, I’ll present the basics of the two noise calculation approaches (rot and
detailed) here.
To get a feeling on the output referred SN we will start with a rule-of-thumb
(rot) approach. It is based on a single op-amp gain stage in series configuration with
a gain of 1000 and input shorted. Figure 16.11 gives the details.
To get SNo.rot in B20k and with GM = 1000 the rot math goes as follows:
e 
n:o:rot
SNo:rot = 20log þ 43 dB ½dBV ð16:19Þ
1V
With the op-amp’s en.i = 0.85 nV/rtHz and RG’s en.RG.rot we obtain the output
noise voltage density en.o.rot:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:o:rot = GM e2n:i þ e2n:RG:rot ð16:20Þ

Rf
RG = ð16:21Þ
GM
1
rffiffiffiffiffiffiffih
RG pffiffiffiffiffiffii
en:RG:rot ¼ 0:13 nV/ Hz
X ð16:22Þ
h pffiffiffiffiffiffii
¼ 0:273 nV/ Hz

en:o:rot ¼ 893 nV/rtHz ð16:23Þ

Hence, SNo.rot and the gain stage input referred noise voltage density en.i.rot
become:

Fig. 16.11 Circuit for

rule-of-thumb SN calculation
16.4 The Final PMMA 261

SNo:rot ¼
78:0 dBV ð16:24Þ

en:o:rot
en:i:rot ¼
GM ð16:25Þ
pffiffiffiffiffiffi
¼ 0:893 nV/ Hz

We have now found the two most important noise based values that show where
we should end-up with the PMMA performance.

16.4.4 Noise Calculations—Detailed SN Calculation Approach

The path through the noise calculation jungle of such a gain stage looks rather
difficult. However, I will present it in this book because gain stages with lower gain
than ×1000 and other than the shown active and passive components bear many
traps. I hope I could avoid most of them.
According to the Fig. 16.9 MA configuration, we have to split the calculation
course into three different parts:
• Input stage with OP1–OP4 and composed by differential Amps A and B,
• 2nd and balanced output stage with the two summing stages OP5 and OP6,
• 3rd stage with balanced to un-balanced converter OP7.
Normally, in a ×1000 MA the influence of the noise of the 2nd and 3rd gain
stage can fully be ignored; the 1st stage creates far more noise than the two other
stages. However, in configurations with lower gain, the noise impact of the output
stage becomes significant. We should know the noise producing mechanism behind
this fact. It is also responsible for the increase of the input referred noise voltage
density of commercially available INAs, if we set them to low gains.
In contrast to the before mentioned Linear Audio article and the noise calculation
approach described there the chosen noise model of one-half of the 1st stage with
input shorted is given in Fig. 16.12. The mathematical treatment of the noise
voltage of RG and its transfer to the output follows the approach already described
in Sect. 4.3. Thus, compared with Table 1 of the original article the calculated
results of Tables 16.1 and 16.2 yield a tiny change.
The handling (suppression) of the 100 % correlated noise voltages between the
two output leads and ground will be the task of a following amp with balanced
input and rather good CMRR.4 This point is not a problem for the signal path via
OP7; with 0.1 % resistances, CMRR of this stage becomes always ≥60 dB.
Additionally we add an input load R0, and, as a reminder, we show the con-
nections to the parallel operating stage with OP3 and 4.

4
See Sect. 16.5.1.
262 16 The Very Low-Noise Balanced Measurement Amp PMMA

Fig. 16.12 Noise model of one-half input stage (Amp A with OP1 and OP2)

The noise model of the other half of the input stage (Amp B) looks absolutely
the same. By paralleling these two halves, the resultant output noise voltage
becomes factor √2 higher.
Paralleling goes via the summing stages OP5 and 6. Figure 16.13 shows the
corresponding noise model.
In Fig. 16.15, we find jumpers jmp1–jmp4. They can produce short-circuits at
the inputs of the output stage; hence, we can separately check their calculated
output noise voltage. Here, RP is the parallel configuration of the input resistors at
the (−) input (R11||R12, R13||R14 in Fig. 16.15).
Finally, the well-known subtraction stage in an INA shows its noise model in
Fig. 16.14. Here, RP is the parallel configuration of the resistors at the (+) input
(R17||R19 in Fig. 16.15).

Fig. 16.13 Noise model of one of the two summing stages (OP5 and OP6)
16.4 The Final PMMA 263

Fig. 16.14 Noise model of the balanced to un-balanced converter (OP7)

60
62
64
66
68
[dBV]

SNo.tot( R0) 70
72
74
76
78
80
0 200 400 600 800 1000 1200 1400 1600 1800 2000
R0
[ohm]

Fig. 16.15 Input load dependency of the output referred SN

I refrain from filling the pages with a huge amount of equations. The chosen
process of output noise voltage calculations follows the block concept: each gain
stage’s output noise voltage density becomes multiplied by the gain of the fol-
lowing stages. Then, to get the total output referred noise voltage density en.o.tot, we
sum-up the un-correlated output noise voltage densities. Division by the gain of the
MA will lead to the input referred noise voltage density en.i.tot. In addition, the
calculation course allows playing around with various values of input load resis-
tances R0. To meet the goal from the beginning it should become ≤1 nV/rtHz with
R0 = 0 Ω.
The calculated Fig. 16.15 (= MCD − WS 18.2’s Fig. 18.10) graph of the input
load dependent output referred SN shows rather drastically that it would be always
better to measure DUT output SNs with DUT output resistances close to 0 Ω.
To keep the PMMA’s average input noise voltage density in any
case ≤1 nV/rtHz the output resistance should be ≤10 Ω. For consumer products,
this becomes rather seldom the case. The final SN determination should always
consider these effects.
264 16 The Very Low-Noise Balanced Measurement Amp PMMA

16.5 The Complete PMMA Circuit

Figure 16.16 shows the final and complete built-up PMMA circuit. For a better
understanding, the following list gives some hints:
1. The main balanced i/p is J01; it should be used for balanced purpose only. For
other purposes, I added J02 and J03; they allow un-balanced application by eg
switching S2 and nulling J03. Usage of these inputs may lead to unstable
conditions.
2. S1 plays the ground lift, thus, in most cases solving the pin1 problem; based on
a Jensen Transformer recommendation, an open S1 and the sequence of R1 and
C1 ensure grounding for very high frequencies.
3. R3 and R4 parallel to R2 set the input resistance Rin of 18.182 kΩ. An attempt
to increase R3 and R4 to 100 kΩ each works too, however the offset-nulling
process becomes a nightmare and after each input load change the amp’s output
shows a tendency to run into the power supply line level. This effect disappears
after switching the psu off and on a few seconds later.

Fig. 16.16 Complete MA circuit

16.5 The Complete PMMA Circuit 265

Fig. 16.17 RG alternatives for gains of a ×10 and b ×100 (for only one input stage)

4. To allow direct amplification jumpers 5 and 6 can bridge the input Cs

(C2a + C2b and C3a + C3b). I do not recommend it. It would require a new
offset trimming process for each new input load.
5. For easy change to another configuration than the one shown, the two input
stages OP1 and OP2 and OP3 and OP4 (including their feedback resistors) are
built-up on two separate plug-in PCBs; RG examples for gains of × 10
and × 100 are given in Fig. 16.15.
6. A DC voltmeter between TP1 and TP2 or TP3 and TP4 measures the alternately
performed offset-nulling of the input stages by P3–P6, hence, after replacing
jumpers1–4 there should be no offset voltage left at outputs PL01 and J04.
I must point out that offset-nulling makes sense only after 15 min heating-up
the amp with shorted input, and without jumpers 5 and 6. It takes some
additional time to get 0.00 V at the outputs.
7. C4, C5, Cx, Cy may be a necessary addition to the circuit; it depends on the
PCB layout as well as on the shielding effect of the case; however, in my
configuration I only had to pack Cx and Cy on R7–R10, each 22 pF ceramic
and selected 1 %.
8. Setting jumpers 1–4 to ground allows easy offset-nulling of OPs 5–7 before the
offset-nulling process of the input stages should take place.
9. We feed the PMMA with a 1 kHz/1 mVrms balanced input signal. To create a
level of 0.5 Vrms between testpoints T1 and T2 and between T3 and T4 the
gains of the two input stages must then be trimmed by P1 and P2; automati-
cally, there will be an output signal of 1 Vrms at both outputs PL01 and J04.
10. Alternatively, we can measure and trim 0.25 Vrms between T1 and ground plus
T2 and ground and 0.25 Vrms between T3 and ground plus T4 and ground.
11. In case of a ×10 or ×100 gain solution only one input stage will be necessary.
12. Replacement of R3 and R4 by eg 1 MΩ each and R2 by 2 MΩ leads to an input
resistance of 1 MΩ, very useful for 1st gain stages with FET op-amps for the
measurement of high resistive valve output stages.
266 16 The Very Low-Noise Balanced Measurement Amp PMMA

13. I’ve set the input resistance to 18.182 k, a value high enough to measure DUTs
with output resistances up to 2 kΩ balanced. Of course, the measured values
have to be corrected for the gain-loss of the voltage to obtain the DUT’s actual
output referred SN.

16.6 PMMA Performance

Table 16.1 shows a summary of the main measurement versus calculation results of
the MA with gain of ×1000. In addition, further down, I show some graphs of the
output noise voltage. We see that hum and computer induced spikes will fully
disappear if we supply the input with a noise voltage that equals the noise voltage of
a resistor ≥1 kΩ, which will always be the case with DUT output noise volt-
ages ≥ 1µVrms in B20k.
Notes on Figs. 16.18 and 16.19:
• On the y-ordinate the gain (+40.00 dB) of an additional Extra Measurement
Amp PFMA5 and the gain of the measurement filter section (+6.02 dB6) are
taken into account.
• FFT resolution is 0.73 Hz = sample rate divided by FFT size (length)
• Averaging in all cases: 50
• No computer-induced spikes could be detected. I could not suppress the tiny
150 Hz hum spike that comes in via (my suspicion) the unshielded part of the
input XLR plug. It was created by vagabonding hum interferences in my lab7;
however, with a 1 kΩ input load the upper traces show that an input noise
voltage density signal of 4 nV/rtHz fully hides the spike. The chosen input load
simulates a very low-noise amp output (≡ −123.63 dBV measured), lower than
the original −120 dBV goal. Nevertheless, it is really challenging to find the
right shielding.
• The heavy lp slope at 20 kHz comes from the 6th order 0.1 dB Chebyshev
measurement lp filter. Together with its hp counterpart at 20 Hz, they form a
band-pass filter8 with nearly brick wall character in B20k.
Figure 16.20 shows the frequency and phase response. At 20 kHz, the frequency
response is 0.02 dB lower than at 1 kHz, at 20 Hz there is no difference. Taking into
account the phase shift of the signal generator’s un-bal to bal converter, we find a
−5.0° phase at 20 kHz versus −0.25° at 1 kHz; at 20 Hz the phase becomes 0.15°.
The un-bal output shows clipping at 9.9 Vrms/1 kHz, the bal output at 19.4 Vrms.

5
see Chap. 17.
6
see Fig. 11.2 + OP13 (TSOS-1) or Fig. 22.2 + OP13 (TSOS-2).
7
See my corresponding remarks on Fig. 12.2.
8
Full calculation in TSOS-2, Chap. 26, MCD-WS 26.2.
16.6 PMMA Performance 267

Table 16.1 PMMA performance results

1/A B C D E F G
2 Item Unit Rot Measured Calculated Remarks
3 en.o.tot nV/rtHz 893.0 874.38 892.564 R0 = 0 Ω
4 SNo.tot dBV −78.0 −78.14 −77.981 R0 = 0 Ω
5 EIN dBV −138.0 −138.16 −137.976 EIN = SNo.tot − GM
6 en.o.1st.tot nV/rtHz 892.35 R0 = 0 Ω
7 SNo.2nd dBV −110.23 −111.168 R0 = 0 Ω
8 SNo.3rd dBV −113.82 −114.811 R0 = 0 Ω
9 en.i.tot pV/rtHz 893.0 874.38 892.564 goal ≤ 1nV/rtHz,
R0 = 0 Ω
9 ein dBV −181.0 −181.17 −180.978 ein = EIN − 43.006 dB
Noise voltages in column F are average values in B20k

Fig. 16.18 Output noise voltage density curves, including input Cs lower trace input shorted,
upper trace input loaded with 1 kΩ
268 16 The Very Low-Noise Balanced Measurement Amp PMMA

Fig. 16.19 Same as Fig. 16.18 with linear frequency range

16.7 Practical Issues

16.7.1 Ground Loop Avoidance and CMRR of a Following Amp

An interruption of the connecting ground or earth or case line between two

amplifying balanced devices seems to be one of the most successful measures to
avoid ground loops and to increase common-mode rejection.9 In Chap. 17, the
galvanically isolated measurement amplifier PFMA shows how we could fight
ground loops, not only for balanced lines, and how we could increase CMRR of the
whole measurement set-up by application of an input transformer.

9
Inter alia: Bill Whitlock on the Jensen Transformers website with
1. ‘Balanced Lines in Audio Systems: Fact, Fiction, and Transformers’.
2. ‘Interconnection of balanced and un-balanced equipment’.
3. ‘A new balanced audio input circuit for maximum common-mode rejection in real-world
environments’.
16.7 Practical Issues 269

Fig. 16.20 Frequency (top) and phase response (bottom) (bal in/bal out)

16.7.2 Enclosure

Figures 16.21 and 16.22 demonstrate the final PMMA solution, housed in an
Al-box with the following dimensions: 170 × 120 × 55 mm3.
In Fig. 16.21, at the right edge of the PCB’s top we find the internal PSU
(±15 V), made-up by integrated voltage regulators, fed by a main PSU unit that also
could feed all other MAs of my collection by a stable ±18 V, as long as there won’t
be no additional ground loop effect. In this case, a separate ±18 to ±25 V psu should
be used. The op-amp’s PSRRs sufficiently suppress any ripple and noise from the
supply lines that might enter the op-amp’s internal circuits.
On the right PCB of Fig. 16.22 we can see how I packed Cx and Cy on the
feedback resistors.

16.7.3 Room for Improvements

1. I’ve already mentioned that the MA has a tendency to run very quickly into
overload. In such a case switching the psu off-on will lead to a new stable
situation. Further efforts are needed to develop a better behaviour.
2. The MA’s input stages have no support from a DC servo. Therefore, the DC
output level slowly moves up and down, roughly ±30 mV.
3. The offset nulling process needs a lot of patience—once performed we will get
at least the result of the previous point.
270 16 The Very Low-Noise Balanced Measurement Amp PMMA

Fig. 16.21 The PMMA in its enclosure

Fig. 16.22 Plug-in input stages of the PMMA

16.7 Practical Issues 271

Table 16.2 Increasing input loads yield increasing differences between measured and calculated
output referred SNs
∙ SNo(0 Ω) = −77.981 dBV calc. and −78.14 dBV measured; D = −0.16 dB
∙ SNo(1 kΩ) = −64.570 dBV calc. and −63.63 dBV measured; D = 0.94 dB
∙ SNo(2 kΩ) = −61.65 7dBV calc. and −59.72 dBV measured; D = 1.94 dB

4. Strong shielding of the MA is essential. The chosen Al boxes work quite well in
a rather hum-free environment. This is not the case everywhere, especially if the
DUT itself produces heavy vagabonding hum interferences, like eg valve driven
(power) amps. The MA’s balanced input allows rather long lines between the
DUT and the MA. The MA could thus be placed a distance away—2 m make no
problem.
5. Increasing values of input load resistances lead to increasing differences between
the calculated and measured noise production of the MA, the output referred SNs
become eg with Rin.rel = 0 Ω or 1 or 2 kΩ the values shown in Table 16.2) .
My interpretation after having checked the data sheet: at the (+) and (−) inputs of
each LT1028 the growing imbalance of the source resistances forces its bias current
cancellation mechanism to produce additional noise voltage. It has less
common-mode behaviour than it would have in the matched case. Therefore, the
op-amp’s CMRR becomes increasingly less effective. Hence, this additional
un-correlated noise voltage sums-up with the data-sheet spec based one. We can
study this effect on the two corresponding LT data sheet graphs.
Unfortunately, there is no indication on how to calculate it. Only interpolation of
the above shown Ds will lead to calculation results close to the measured ones.
MCD-WS 18.2, Sect. 16.4 gives a rather simple derivation of D as function of the
input load; it will ease any required interpolation.

16.8 Recommendations

In a lowest-noise measurement environment, the PMMA works quite well, as long

as we consider the above-described points. The following paragraphs show how we
can use it in general.
We can work through the measurement process and the clarification of potential
measurement amp impact according to the following example guidelines:
272 16 The Very Low-Noise Balanced Measurement Amp PMMA

16.8.1 DUT Output Resistance ≤10 Ω

1. Generally, the determination of the whole measurement arrangement’s EIN

works by shorting the input, by measurement of the output SN, and by sub-
traction of the measurement arrangement’s gain in dB; here, the PMMA’s
measured input referred rms noise voltage is: EINm = −138.16 dBV

2. Determination of the DUT’s output resistance; here,

we assume a very low one: Ro ≤ 10 Ω

3. We measure eg a DUT output referred SNo.m: SNo.m = −130.64 dBV

4. We calculate the difference B: B = SNo.m − EIN

= 7.52 dB

5. Fig. 16.2a shows the impact We(B): We(B) = 0.71 dB

6. The DUT’s real output SNo.r thus becomes: SNo.r = SNo.m − We(B)
= −131.35dBV

16.8.2 DUT Output Resistance >10 Ω

Depending on the amount of PMMA input noise current in.i any Ro > 10 Ω will add
more or less significant portions of noise voltage, thus, worsening the measurement
amp’s original EIN to EINcorr by an equivalent amount in dB, hence, we’ll get
|EINcorr| < |EIN|.
Let’s check a worst case example with a DUT’s total output resistance
Ro = 2247 Ω, in.i = 1 pA/rtHz, Rin.eff = Ro||Rin = 2 kΩ:
1. Determination of EINcorr: by worsening the original EIN (−138.16 dBV) the
PMMA’s total input load Rin.eff and input current and voltage noise effect on Ro
lead to an additional input noise voltage density en.i.add. With it we can calculate
EINcorr as follows:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:i:add ¼ ðin:i Rin:eff Þ2 þe2n:Rin:eff
ð16:26Þ
) EINadd ¼
164:757 dBV þ 43:006 dB
0sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
EINadd EIN
B C
EINcorr B
= 20log@ 10 10 þ10 10 C A
ð16:27Þ
¼
121:697 dBV
pffiffiffiffiffiffi
) en:i:corr ¼ 5:819 nV/ Hz

Crosscheck for (16.27): See Fig. 18.9 on the MCD-WS 18.2. It shows the
PMMA’s output referred SNo.tot versus the total input load R0. We can pick
16.8 Recommendations 273

SNo.tot for R0 = 2 kΩ and we’ll get −61.693 dBV. To get EINcorr we have to
subtract the PMMA gain: EINcorr =
61:693dBV
60dB
=
121:693dBV:
2. We measure eg a DUT output referred SNo.m: SNo:m =
104:38 dBV

3. We calculate Gloss1.e: Gloss1:e = 20log½Rin =ðRin + Ro Þ

= 20logð0:89Þ
=
1:012 dB

4. Interpolation term according to Table 2: D = 2.0 dB

SNo:r = SNo:m þ jGloss1:e j
D

5. We obtain SNo.r:
=
105:37 dBV

B = SNo:r
EINcorr
6. We calculate the difference B:
= 16:32 dB

7. Fig. 16.2a shows the impact We(B): We(B) = 0.1 dB

8. Without PMMA impact the DUT’s real output

SNo.dut thus would become: SNo:dut = SNo:r
We ðB)
=
105:47 dBV

16.8.3 Summary of Recommendations

Now, we can sum-up all the above given findings into the following main
recommendations:
a. As long as the DUT output resistance is very low (≤10 Ω) the PMMA measures
output referred SNs of DUTs up to −121.8 dBV (−129.0 dBV) in B20k with an
error of ≤0.1 dB (≤0.5 dB) only.
b. To keep the maximum allowed 0.1 dB (0.5 dB) border line and by usage of LT1028
op-amp types any DUT output resistances from 0 Ω up to 2.25 kΩ (equals 2 kΩ with
the paralleled PMMA input resistances) will shift the minimal output referred SNs
from measured −121.8dBV (−129.0dBV) to −105.5dBV (−112.7dBV) rounded.

16.9 Final Notes

• The LT1028 works quite well in an input load environment with nearly equal loads
at both inputs. With RGs ≤ 10 Ω DUT output resistances with values of RG ± 20
−50 % do not produce much harm. All cases of larger DUT output resistances
require additional calculation efforts, making the calculation results less precise.
274 16 The Very Low-Noise Balanced Measurement Amp PMMA

• Although handicapped by a doubled input noise current, a future project will

check the AD797 usability.
• Further CMRR improvement could come from Bruno Putzey’s “optimal input
biasing”, given in Fig. 20 of his very interesting article in LA Vol. 5. I have to
check this too.
• The nasty DC drift at the outputs (TP1–TP4) of the input stages should be
controlled by a well-working DC servo. Maybe it could also be the right
measure to suppress the tendency of these stages of sometimes running into
overload after an input load change. Then, I guess these measures won’t prevent
the MA from getting the RMMA status (Rich Man’s MA).
• Occasionally, in the above given equations we find the term 43.006 dB or its
pffiffiffiffiffiffiffiffiffi
rounded version 43 dB. It is nothing else but B20k expressed in [dB] = 20 log
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
( 19; 980 Hz), and it stands for the difference between the rms voltage of a
white noise signal and its equivalent noise voltage density in B20k.
• Of course, the chosen expression “worsening figure We(B)” is nothing else but
another approach to express the noise figure NFe [dB] of the PMMA. However,
in this chapter I use it the chosen way to better underline its influential and
negative effect in the measurement process of tiny noise signals.
• Finally, here comes the calculation to find the real A-weighted output referred
SNo.a.NAD.r = −124.2 dBV(A) ± 0.5 dB of the NAD M51, without AP analyzer
impact, however, in the S-AES17 filter given equivalent noise bandwidth of
ENB = B20k, rounded to one digit after the decimal point:

! SNo:a:NAD:m ¼
122:0 dBV(A) = measured A-weighted SN of the NAD M51

þ2:05 dB = A = A-weighting based improvement
) SNo:NAD:m ¼
119:95 dBV = output referred SN excl: 1/f-noise,

0:5 dB = WFe ðfc Þ ¼ worseningfigurebasedonthe
1=f-noise corner frequency fc = 350Hz
ðsee LA Vol: 4, p: 19, Fig:5
or TSOS-2, Fig:7:6a in [dB]!)
) SNo:NAD:m:wn ¼
120:45 dBV = output referred SN excl: 1/f-noise,
hence, theoretically white noise only

! SNi:AP ¼
122:4 dBV = AP SYS 2722 bal: i/p referred

SN with i/p shorted
)B ¼ SNo:NAD:m:wn
SNi:AP
¼
120:45 dBV
ð
122:4 dBV)
¼ 1:95 dB
) We(B) ¼ 2:2 dB taken from Fig. 16.2b

! SNo:a:NAD:r ¼ SNo:a:NAD:m
We(B)

¼
122:0 dBV(A)
2:2 dB
¼
124:2 dBV(A)

Note: The AP’s SYS 2722 input referred noise voltage is free of 1/f-noise,
hence, if we would calculate B in the A-weighted domain we would create an error
that grows for fc > 1 kHz from appr. 0.5 dB to >5 dB at 20 kHz
The Galvanically Isolated
Measurement Amp PFMA 17

17.1 Intro

The measurement of extremely low noise levels creates problems at many corners
of every measurement project. In most cases, the nastiest ones of these come on
board with the connections to the CMS. Balanced lines improve things; however,
only a galvanically isolated connection helps avoiding ground loops and the pen-
etration of hum interferences into the measurement set-up. Therefore, I strongly
recommend the inclusion of a broadband transformer and associated amplification
electronic between MA and CMS.

17.2 Ground Loop Avoidance

For my own measurement purposes, I have developed such an extra measurement

amp PFMA around the Jensen Transformers 4:1 type JT-10KB-D. In addition to the
galvanic isolation it also offers an outstanding CMRR. The electronic configuration
consists of an LT1028C op-amp (gains from transformer input to op-amp output x5
—trimmed by P1, or x50 with Jmp1 set and trimmed by P3), followed by an un-bal
to bal and an un-bal to un-bal converter, each with a gain of 2 (SSM2142 or
THAT1646 and OP27 plus a trimmed P2).
Figure 17.1 shows the details. This MA has a flat frequency and phase response
in B20k (see Fig. 17.3) and its input noise voltage density is flat in B20k (see
Fig. 17.2). To keep the input referred SN as good as possible and with its gain of
200 (or 20) OP1 must have outstanding noise voltage and noise current figures. The
LT1028C sufficiently fulfils these requirements The measured PFMA input referred
noise production looks as follow: gain x100: 9.641 nV/rtHz (calculated:
10.571 nV/rtHz; gain x10: 10.945 nV/rtHz (calculated 11.647 nV/rtHz). These
values are very much smaller than the appr. 900 nV/rtHz of a preceding MA alike
the PMMA; hence, there will not be any significant impact on the MA’s output
noise. Next chapter’s MCD-WS 18.3 gives the details.

© Springer International Publishing Switzerland 2016 275

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_17
276 17 The Galvanically Isolated Measurement Amp PFMA

Fig. 17.1 PFMA circuit

Fig. 17.2 Output noise voltage density of the PFMA, input shorted
17.2 Ground Loop Avoidance 277

The advantage of the MA–PFMA sequence is the fact that the MA can be
located very close to the DUT while the PFMA might be located far away, closer to
the CMS. The balanced lines and the very high CMRR of the PFMA ensure high
immunity against any line induced interference.

17.3 Additional PFMA Data

The input XLR socket J01 feeds the input transformer Tr1. The tr = 4:1 relationship
of the transformer windings makes n = 1/tr, hence, 0.25. The input impedance thus
becomes approximately (R2a + R2b)/(n)2 = 100 kΩ. Including the coil resistances,
Zi increases to a calculated 107 kΩ.
Referenced to 1 kHz the Fig. 17.3 deviation from a flat F response becomes
−0.003 dB at 20 Hz and −0.003 dB at 20 kHz. We obtain the deviation from a flat P
response as follows: +2° at 20 Hz and −13° at 20 kHz.
Inputs J02 & J03 ensure direct connection to the transformer input leads. In case
of J02 usage we must ground the hot pin of J03, and vice versa. S2 switched into
the right position allows an access to the transformer via J02 without any
grounding. In this case, J03 must be kept open.

Fig. 17.3 F & P of the PFMA

278 17 The Galvanically Isolated Measurement Amp PFMA

S1 plays the ground lift, thus, in most cases solving the pin1 problem; based on a
Jensen Transformer recommendation, an open S1 and the sequence of R1 and C1
ensure grounding for very high frequencies.
The transformer’s CMRR is extremely high; the data sheet claims 120 dB at
60 Hz and 85 dB at 3 kHz.

17.4 Gain and SN Calculations

17.4.1 Gains

The gain equations of the PFMA are spilt into three different stages: the first gain
stage with the transformer Tr1 and its gain GTr1, the second one with the LT1028
OP1 and its gain G1, and the third one with the two output gain stages OPs 2 & 3
with equal gains G2. Hence, we obtain the effective gain GMA—from input to
output—as follows:

GMA ¼ GTr1 G1G2

ð17:1Þ
¼ 100 or 10

tr ¼ 1 : n
¼4:1 ð17:2Þ
¼ 1 : 0:25

GTr1 ¼ n ð17:3Þ

R4 100
G1 ¼ 1 þ ¼ ¼ 200
R3a n  2 ð17:4Þ
R4 10
¼1þ ¼ ¼ 20
R3 n  2

R3 ¼ R3a þ R3b þ P1 ð17:5Þ

R4 ¼ R4a þ ðR4bjjR4cÞ þ P2 ð17:6Þ

G2 ¼ 2 ðOP3Þ
R6 þ P2 ð17:7Þ
¼1þ ðOP2Þ
R5
17.4 Gain and SN Calculations 279

17.4.2 Evaluation of Noise Voltages and SNs

The basic situation of the various noise voltage and noise current sources is given in
Fig. 17.4. Figure 17.5 shows the circuit that leads to the input load Ri.tot of OP1.
Thus, we’ll obtain the effective frequency and R0 dependent input referred noise
voltage density en.i.MA(f, R0) of the Fig. 17.1 MA as follows:

en:o ðf, R0Þ

en:i:MA ðf; R0Þ ¼ ð17:8Þ
GMA

en:o ðf, R0Þ ¼ en:i ðf, R0ÞGamp ð17:9Þ

Here, we need the auxiliary gain Gamp. It becomes:

Gamp ¼ G1G2
ð17:10Þ
¼ 400 or 40

The input referred noise voltage density en.i1(f, R0) at the input of OP1 in
Fig. 17.4 becomes thus:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
u
u en:i1 ðf Þ2 þ in:i1 ðf Þ2 RF2 þ Ri:tot ðR0Þ2
en:i ðf; R0Þ ¼ t ð17:11Þ
þe2n:RF þ en:Ri:tot ðR0Þ2

Fig. 17.4 Noise sources of the PFMA

280 17 The Galvanically Isolated Measurement Amp PFMA

Fig. 17.5 Evaluation of the OP1 input load

Fig. 17.6 R0 dependency of 110

111
the PFMA’s input referred SN 112
113
114
[dBV]

SN ne.i ( R0) 115

116
117
118
119
120
1000 2000 3000 4000 5000 6000 7000 8000 9000 1 .10
4
0
R0
[ohm]

Finally, with Fig. 17.6 showing its R0 dependency the R0 dependent input
referred SNne.i looks as follows (in [dBV] with vi.ref = 1 Vrms):
0vu
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
u Z
20kHz
But 1 jen:i:MA j2 dfC
B C
B B1 C
B C
B
SNne:i ðR0Þ ¼ 20 logB
20Hz
C ð17:12Þ
C
B v i:ref C
B C
@ A
17.5 Enclosure 281

17.5 Enclosure

Figure 17.7 demonstrates the final PFMA solution, housed in an Al-box

(170 × 120 × 55 mm3). We find the internal PSU (±15 V) at the right of the
transformer, made-up by integrated voltage regulators, fed by a main PSU unit that
also could feed all other MAs of my collection by a stable ±18 V, as long as there
won’t be any additional ground loop effects. In this case, an additional ±18 to ±25 V
PSU should be used. The op-amp’s PSRRs sufficiently suppress any ripple and
noise from the supply lines that might enter the op-amp’s internal circuits.

Fig. 17.7 PFMA and its

enclosure
Mathcad Worksheets of Measurement
Tools 18

Contents

18.1 MCD-WS: The UBC

18.2 MCD-WS: The PMMA
18.3 MCD-WS: The PFMA

pffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffi
Note 1: MCD 11 has no built-in unit “rtHz” or Hz. To get 1 Hz based
voltage noise and current noise densities the rms noise voltage and
current in a specific frequency range B > 1 Hz must be multiplied by
pffiffiffiffiffiffiffiffiffi
1 Hz and divided by the root of that specific frequency range √B!
Note 2: MCD 11 offers no “dB” unit. This is available from MCD 13 on!

© Springer International Publishing Switzerland 2016 283

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_18
284 18 Mathcad Worksheets of Measurement Tools

18.1 MCD-WS: The UBC Page 1

Gain and SN calculation of the UBC

Fig. 18.1 = Fig 15.7

1. Definition of all meaningful constants, components, etc.:

− 23 −1 − 19
k := 1.38065 ⋅10 ⋅ V ⋅ A ⋅s ⋅ K q := 1.6021765 ⋅10 A ⋅s T := 300 ⋅K vi.nom := 1V

B20k := 19980 ⋅Hz B1 := 1Hz h := 1000Hz vo.nom := 1V

3
f := 10Hz, 15Hz.. 20 ⋅10 Hz
3 3 3
R1 := 47.5 ⋅10 Ω R2 := 2.21 ⋅10 Ω R3 := R2 R4 := 3.92 ⋅10 Ω R5 := R4 R6 := R2
R7 := R2 R8 := R4 R9 := R4 R10 := 47.5Ω R11 := R10
3 3
R12 := 1 ⋅10 Ω R13 := R12 R14 := 1.1 ⋅10 Ω R15 := 63Ω R16 := R14
R17 := 6.8Ω R18 := R17
OP1 OP27 OP2 OP1 OP3 OP1 OP4 BUF634 OP5 OP4
3 3 −6
P1 := 50 ⋅10 Ω P2 := 10 ⋅10 Ω C1 := 10 ⋅10 F
3 3
RL.pl01 := 10 ⋅10 Ω RL.j03 := 1 ⋅10 Ω

measured o/p resistance at PL01 : Ro.pl01 := 45.5 ⋅Ω

−1
⎛ 1 +
Ro.j03 := 2 ⋅⎜
1
+
1 ⎞
R17 0.5 ⋅R15 R14 + 0.5 ⋅Ro.pl01 ⎟
calculated o/p resistance at J03 :
⎝ ⎠
Ro.j03 = 11.13 Ω
18 Mathcad Worksheets of Measurement Tools 285

18.1 MCD-WS: The UBC Page 2

2. Gain and input impedance

R4 R4 1
G2nd G2nd 1.75252 G1st G1st 0.571
R2 R10 R4 G2nd

via PL01 : Gpl01 G1st G2nd Gpl01 1

3 3
P1b P1 G1st P1b 28.53 10 P1a P1 P1b P1a 21.47 10

Ro.j03 3
via J03 : Gj03 Gj03 5.034 10
Ro.j03 R14 R16

Gj03.e 20 log Gj03 Gj03.e 45.962 [dB]

1
1 1
Zi( f) 3
R1 1 Zi( h) 24.359 10
P1 2j f C1

4
2.437 10

4
2.4365 10
[ohm]

Zi( f)
Fig. 18.2
UBC input impedance
4
2.436 10

4
2.4355 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

3
=> Ri Zi( h) Ri 24.359 10

Input load with input shorted:

1
1 1 3
Ri.L Ri.L 12.251 10
P1a P1b

3. SN calculations
1 1
1 1 3 1 1 3
RP1 R4 R2 RP1 1.413 10 RP2 R5 R6 RP2 1.413 10
9 9
en.R4 4 k T B1 R4 en.R4 8.059 10 V en.R2 4 k T B1 R7 en.R2 6.051 10 V

9
en.RP1 4 k T B1 RP1 en.RP2 en.RP1 en.RP1 4.839 10 V

9 9
en.i1 3.0 10 V en.i2 en.i1 en.i3 en.i1 en.i4 4 10 V en.i5 en.i4
286 18 Mathcad Worksheets of Measurement Tools

18.1 MCD-WS: The UBC Page 3

12 12
in.i1 0.4 10 A in.i2 in.i1 in.i3 in.i1 in.i4 1 10 A in.i5 in.i4

fc.e1 2.7Hz fc.i1 120Hz

fc.e1
en.i1( f) en.i1 1 en.i2( f) en.i1( f) en.i3( f) en.i1( f)
f

fc.i1
in.i1( f) in.i1 1 in.i2( f) in.i1( f) in.i3( f) in.i1( f)
f
9
en.RiL 4 k T B1 Ri.L en.RiL 14.247 10 V

3.1 Output referred SN via PL01 :

2 2 2 2 9
en.o.1st( f) en.RiL en.i1( f) in.i1( f) Ri.L en.o.1st( h) 15.456 10 V

By ignoring the noise production of the stages around OPs 2 & 3 & Boosters 1 & 2 we'll obtain the noise voltage at the
PL01 output as follows:

9
en.o.pl01 ( f) en.o.1st( f) G2nd en.o.pl01 ( h) 27.087 10 V

7
1 10
[V/rtHz]

en.o.pl01( f)
Fig. 18.3
Output noise voltage density
at PL01

8
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

20000Hz
1 2
en.o.pl01 ( f) df
B1
20Hz
SNne.o.pl01 20 log SNne.o.pl01 108.374 [dBV]
vo.nom

3.2 Output referred SN via J03 :

12
en.o.j03 ( f) en.o.pl01 ( f) Gj03 en.o.j03 ( h) 136.344 10 V

20000Hz
1 2
en.o.j03 ( f) df
B1
20Hz
SNne.o.j03 20 log SNne.o.j03 154.336 [dBV]
vo.nom
18 Mathcad Worksheets of Measurement Tools 287

18.1 MCD-WS: The UBC Page 4

4. Output referred SN calculated and measured with PMMA from MCD-WS 18.2 (lt) :

4.1 SN via PL01 :

12 12 3
From MCD-WS 18.2: en.i.lt.avg 892.564 10 V in.i.lt.avg 1.042 10 A Ri.lt 18.182 10

SNi.PMMA.m 138.16 [dBV]

Glt 1000 Glt.e 20 log Glt Glt.e 60 [dB]

=>

1
1 1
RiL.lt1 Ri.lt Ro.pl01 RiL.lt1 45.386

12
en.RiL.lt1 4 k T B1 RiL.lt1 en.RiL.lt1 867.152 10 V

2 2 2 2 2 6
en.o.lt1( f) Glt en.o.pl01 ( f) en.i.lt.avg in.i.lt.avg RiL.lt1 en.RiL.lt1 en.o.lt1( h) 27.115 10 V

20000Hz
1 2 3
eN.o.lt1 en.o.lt1( f) df eN.o.lt1 3.817 10 V
B1
20Hz

eN.o.lt1
SNne.o.lt1 20 log SNne.o.lt1 48.364 [dBV]
vo.nom

SNne.o.pl01.lt1 SNne.o.lt1 Glt.e SNne.o.pl01.lt1 108.364 [dBV]

measured: SNne.o.pl01.lt1.m 107.89 [dBV]

B SNne.o.pl01.lt1.m SN i.PMMA.m B 30.27 [dB] => We ( B) 0.004 [dB]

SNne.o.pl01.lt1.re SNne.o.pl01.lt1.m We ( B) SNne.o.pl01.lt1.re 107.894 [dBV]

4.2 SN via J03 :

1
1 1
RiL.lt2 RiL.lt2 11.123
Ro.j03 Ri.lt

12
en.RiL.lt2 4 k T B1 RiL.lt2 en.RiL.lt2 429.285 10 V

2 2 2 2 2
en.i.lt2( f) en.o.j03 ( f) en.i.lt.avg in.i.lt.avg RiL.lt2 en.RiL.lt2 12
en.i.lt2( h) 999.84 10 V
288 18 Mathcad Worksheets of Measurement Tools

18.1 MCD-WS: The UBC Page 5

9
en.o.lt2( f) Glt en.i.lt2( f) en.o.lt2( h) 999.84 10 V

20000Hz
1 2 6
eN.o.lt2 en.o.lt2( f) df eN.o.lt2 141.318 10 V
B1
20Hz

eN.o.lt2
SNne.o.lt2 20 log SNne.o.lt2 76.996 [dBV]
vo.nom

SNne.o.j03.lt2 SNne.o.lt2 Glt.e SNne.o.j03.lt2 136.996 [dBV]

measured: SNne.o.j03.lt2.m 136.02 [dBV]

No correction because B becomes negative:

B SNne.o.j03 SNi.PMMA.m B 16.176 [dB]

5. Output referred SN calculated and measured with PFMA from MCD-WS 18.3 (tr) :

5.1 SN via PL01 :

9 15 3
From MCD-WS 18.3: en.i.tr.avg 10.571 10 V in.i.tr.avg 260.521 10 A Ri.tr 106.9 10

SNi.PFMA.m 117.30 [dBV]

Gtr 100 Gtr.e 20 log Gtr Gtr.e 40 [dB]

=>

1
1 1
RiL.tr1 Ri.tr Ro.pl01 RiL.tr1 45.481

12
en.RiL.tr1 4 k T B1 RiL.tr1 en.RiL.lt1 867.152 10 V

2 2 2 2 2 6
en.o.tr1( f) Gtr en.o.pl01 ( f) en.i.tr.avg in.i.tr.avg RiL.tr1 en.RiL.tr1 en.o.tr1( h) 2.909 10 V

20000Hz
1 2 6
eN.o.tr1 en.o.tr1( f) df eN.o.tr1 409.756 10 V
B1
20Hz

eN.o.tr1
SNne.o.tr1 20 log SNne.o.tr1 67.749 [dBV]
vo.nom
18 Mathcad Worksheets of Measurement Tools 289

18.1 MCD-WS: The UBC Page 6

SNne.o.pl01.tr1 SN ne.o.tr1 Gtr.e SNne.o.pl01.tr1 107.749 [dBV]

measured: SNne.o.pl01.tr1.m 107.31 [dBV]

B SNne.o.pl01.tr1.m SNi.PFMA.m B 9.99 [dB] => We ( B) 0.41 [dB]

SNne.o.pl01.tr1.re SNne.o.pl01.tr1.m We ( B) SNne.o.pl01.tr1.re 107.72 [dBV]

5.2 SN via J03 :

1
1 1
RiL.tr2 RiL.tr2 11.129
Ro.j03 Ri.tr

en.RiL.tr2 4 k T B1 RiL.tr2

2 2 2 2 2 2 9
en.i.tr2( f) en.o.pl01 ( f) Gj03 en.i.tr.avg in.i.tr.avg RiL.tr2 en.RiL.tr2 en.i.tr2( h) 10.581 10 V

6
en.o.tr2( f) Gtr en.i.tr2( f) en.o.tr2( h) 1.058 10 V

20000Hz
1 2 6
eN.o.tr2 en.o.tr2( f) df eN.o.tr2 149.557 10 V
B1
20Hz

eN.o.tr2
SNne.o.tr2 20 log SNne.o.tr2 76.504 [dBV]
vo.nom

SNne.o.j03.tr2 SNne.o.tr2 Gtr.e SNne.o.j03.tr2 116.504 [dBV]

measured: SNne.o.j03.tr2.m 116.82 [dBV]

No correction because B becomes negative:

B SNne.o.j03 SNi.PFMA.m B 37.036 [dB]

290 18 Mathcad Worksheets of Measurement Tools

18.2 MCD-WS: The PMMA Page 1

The Very Low-Noise Balanced Measurement Amp PMMA

Fig. 18.4 = Fig. 16.15

1. Components :

OP1 LT1028C OP2 OP3 OP4 OP1 OP5 OP27G OP6 OP7 OP5

R0 Ro.tot Ro1 Ro2 Rout1 Rout2 R0 0 10 2000

3 3 3
R1 51 R2 200 10 R3 10 10 R4 R3 R7 2.2 10 R8 R7 R9 R7
R2 2 R3 3
Rin.tot Rin.tot 18.182 10 R7 ... R20: 0.1% R R7
R2 2 R3
Rin.tot R0 3
The real input load: Rin.re( R0) Rin.re 2.247 10 1999.849
Rin.tot R0
3
R10 R7 R11 2.2 10 R12 R11 R13 R11 R14 R11 R15 R11 R16 R11
6
R17 R11 R18 R11 R19 R11 R20 R11 R21 10 C2 C3 220 10 F
18 Mathcad Worksheets of Measurement Tools 291

18.2 MCD-WS: The PMMA Page 2

2. Gain :
Note: The complete derivation of the gain equations is shown in 6.!

GM 1000 vo.ref 1V

Memo for further calculations: RG1 R5 P1 RG2 R6 P2

With RG1 = RG2 = RGM and equal Rs R7 ... R16 and R = R7 we'll get:

2R 4R 2R
GM 2 1 GA GB RGM
RGM GM 2 GA 1

R 2R
RG1 4 GA 1 GA 500
GM 2 RG1

RG1 8.818

R 2R
RG2 4 GB 1 GB 500
GM 2 RG2

RG2 8.818

G2nd 1 => Gtot GA GB G2nd Gtot 1000

RGM RG1

3. Noise and SN :

Fig. 18.5 = Fig. 16.12

292 18 Mathcad Worksheets of Measurement Tools

18.2 MCD-WS: The PMMA Page 3

3.1 Noise relevant data :

23 1
k 1.38065 10 VAsK T 300K B1 1Hz B20k 19980 Hz h 1000Hz
3
f 10Hz 15Hz 20 10 Hz
9 9
en.i1 0.85 10 V en.i2 en.i1 en.i3 en.i1 en.i4 en.i1 en.i5 3.0 10 V en.i6 en.i5
12 12 12
in.i1 1 10 A in.i2 in.i1 in.i3 1 10 A in.i4 in.i1 in.i5 0.4 10 A in.i6 in.i5

fc.e1 3.5Hz fc.e2 fc.e1 fc.e3 fc.e1 fc.e4 fc.e1 fc.e5 2.7Hz fc.e6 fc.e5

fc.i1 250Hz fc.i2 fc.i1 fc.i3 fc.i1 fc.i4 fc.i1 fc.i5 120Hz fc.i6 fc.i5

fc.e1
en.i1( f) en.i1 1 en.i2( f) en.i1( f) en.i3( f) en.i1( f) en.i4( f) en.i1( f)
f

fc.i1
in.i1( f) in.i1 1 in.i2( f) in.i1( f) in.i3( f) in.i1( f) in.i4( f) in.i1( f)
f
fc.e5
en.i5( f) en.i5 1 en.i6( f) en.i5( f)
f

fc.i5
in.i5( f) in.i5 1 in.i6( f) in.i5( f)
f

2 2
en.i.A( f) en.i1( f) en.i2( f) 9
en.i.A( h) 1.204 10 V
en.i.B( f) en.i.A( f)

en.i.2nd( f) en.i5( f) in.i.2nd( f) in.i5( f)

1
1 1 3
RP2nd.1 R11 R12 RP2nd.1 1.1 10

9
en.R7 4 k T B1 R7 en.R7 6.037 10 V

en.R8 en.R7

12
en.RG1 4 k T B1 RG1 en.RG1 382.216 10 V

9
en.RP.2nd.1 4 k T B1 RP2nd.1 en.RP.2nd.1 4.269 10 V

9
en.R11 4 k T B1 R8 en.R11 6.037 10 V

9
en.R16 4 k T B1 R16 en.R16 6.037 10 V

12
en.Rin.re( R0) 4 k T B1 Rin.re( R0) en.Rin.re 10 406.924 10 V
18 Mathcad Worksheets of Measurement Tools 293

18.2 MCD-WS: The PMMA Page 4

3.2 Noise voltage density at the bal o/p of one of the two 1st stages (i/p shorted) :
R7 R8 9
en.RG1.o en.RG1 en.RG1.o 190.726 10 V
RG1

2 2 2 2 2 2 2 2 2
en.o.A( f) en.i.A( f) GA in.i1( f) R7 in.i2( f) R8 en.R7 en.R8 en.RG1.o

9
en.o.A( h) 631.645 10 V

en.o.B( f) en.o.A( f)

3.3 Paralleling of the two i/p stages and the inclusion of an input load R0 :

Fig. 18.6
Noise situation after
paralleling of two i/p
stages

en.o.A( f) 9
en.i.A( f) en.i.A( h) 1.263 10 V
GA
9
en.i.B( f) en.i.A( f) en.i.B( h) 1.263 10 V

en.o1 ( f) en.i.A( f) GM 6
en.o1 ( h) 1.263 10 V
6
en.o2 ( f) en.i.B( f) GM en.o2 ( h) 1.263 10 V

1
1 1 15
in.i.A( f) in.i.A( h) 790.569 10 A
2 2
in.i1( f) in.i2( f)

1
1 1 15
in.i.B( f) in.i.B( h) 790.569 10 A
2 2
in.i3( f) in.i4( f)

2 2 12
in.i( f) in.i.A( f) in.i.B( f) in.i( h) 1.118 10 A

20000Hz
1 2 12
in.i.avg in.i( f) df in.i.avg 1.042 10 A
B20k
20Hz
294 18 Mathcad Worksheets of Measurement Tools

18.2 MCD-WS: The PMMA Page 5

1
1 1 2 2 2 2
en.o.1st( f R0) in.i( f) Rin.re( R0) en.Rin.re( R0) GM
2 2
en.o1 ( f) en.o2 ( f)
9
en.o.1st h 0 893.281 10 V

5
1 10

Fig. 18.7
Output noise voltage
[V/rtHz]

density of the whole

en.o.1st f 0 6 1st stage
1 10

7
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

Average noise voltage density in B20k :

20000Hz
1 2 9
en.o.1st.avg( R0) en.o.1st( f R0) df en.o.1st.avg 0 892.35 10 V
B20k
20Hz

3.4 Output noise voltage density and SN at the bal o/p of the 2nd stage (OP5 & OP6), i/p shorted

Fig. 18.8 = Fig. 16.13

2 2
2 R16 2 2 2 2 R16
en.o5 ( f) en.i.2nd( f) 1 en.R16 in.i.2nd( f) R16 en.RP.2nd.1
RP2nd.1 RP2nd.1

9
en.o5 ( h) 13.836 10 V
en.o6 ( f) en.o5 ( f)

2 2 9
en.o.2nd( f) en.o5 ( f) en.o6 ( f) en.o.2nd( h) 19.567 10 V
18 Mathcad Worksheets of Measurement Tools 295

18.2 MCD-WS: The PMMA Page 6

20000Hz
1 2
en.o.2nd( f) df
B1
20Hz
SN o.2nd 20 log SNo.2nd 111.168 [dBV]
1V

measured with R0 = 0 : SNo.2nd.m 110.23 [dBV]

3.5 Total input load dependent o/p noise voltage density and the corresponding
input and output referred SNs :

2 2
en.o.tot( f R0) en.o.1st( f R0) en.o.2nd( f) 9
en.o.tot h 0 893.495 10 V

Average noise voltage density in B20k :

20000Hz
1 2 9
en.o.tot.avg( R0) en.o.tot( f R0) df en.o.tot.avg 0 892.564 10 V
B20k
20Hz

3.5.1 Balanced i/p and o/p referred SNs :

en.o.tot( f R0)
en.i.tot( f R0) 12
GM en.i.tot h 0 893.495 10 V

Average noise voltage density in B20k :

20000Hz
1 2 12
en.i.tot.avg( R0) en.i.tot( f R0) df en.i.tot.avg 0 892.564 10 V
B20k
20Hz

ein and SN evaluation :

en.i.tot.avg 0
ein 20 log ein 180.987 [dBV]
1V

20000Hz
1 2 6
eN.o.tot ( R0) en.o.tot( f R0) df eN.o.tot 0 126.165 10 V
B1
20Hz

20000Hz
1 2 9
eN.i.tot( R0) en.i.tot( f R0) df eN.i.tot 0 126.239 10 V
B1
Hz
296 18 Mathcad Worksheets of Measurement Tools

18.2 MCD-WS: The PMMA Page 7

eN.o.tot ( R0)
SNo.tot ( R0) 20 log SNo.tot 0 77.981 [dBV]
vo.ref

measured (R0 = 0 ): SNo.tot.m 78.14 [dBV]

eN.i.tot( R0)
SNi.tot( R0) 20 log SNi.tot 0 137.976 [dBV]
vo.ref

measured (R0 = 0 ): SNi.tot.m 138.16 [dBV]

3.5.2 Equivalent input noise voltage EIN :

EIN SN i.tot 0 EIN 137.976 [dBV]

EINm 138.16 [dBV]

EINm 43.006
20 12
en.i.tot.m 10 V en.i.tot.m 874.38 10 V

3.6 The un-balanced o/p referred SN at OP7's J04 :

Fig. 18.9 = Fig. 16.14

1
1 1
en.i7( f) en.i5( f) in.i7( f) in.i5( f) RP7 RP7 1100
R17 R19
9
en.RP7 4 k T B1 RP7 en.RP7 4.269 10 V
9
en.R18 4 k T B1 R18 en.R18 6.037 10 V
9
en.R20 4 k T B1 R20 en.R20 6.037 10 V

3.6.1 Output noise voltage density and SN (i/p shorted) :

2 2
2 2 2 2 R20 2 R20 2 2 2
en.o7 ( f) en.i7( f) in.i7( f) RP7 en.RP7 1 en.R18 en.R20 in.i7( f) R20
R18 R18

9
en.o7 ( h) 13.551 10 V
18 Mathcad Worksheets of Measurement Tools 297

18.2 MCD-WS: The PMMA Page 8

20000Hz
1 2
en.o7 ( f) df
B1
20Hz
SN o7.sh 20 log SNo7.sh 114.359 [dBV]
vo.ref

measured with R0 = 0 : SNo7.sh.m 113.82 [dBV]

3.6.2 Output noise voltage and SN with i/p loaded by the noise of the preceding stages :

2 2 9
en.o.ub( f R0) en.o.tot( f R0) en.o7 ( f) en.o.ub h 0 893.598 10 V

20000Hz
1 2
en.o.ub( f R0) df
B1
20Hz
SN o.ub ( R0) 20 log SNo.ub 0 77.98 [dBV]
vo.ref

measured : SNo.ub.m 78.14 [dBV]

3.7 Graph of the input load dependent output referred SN:

60
62
64
66
68
[dBV]

SNo.tot( R0) 70
72
74
76
78
80
0 200 400 600 800 1000 1200 1400 1600 1800 2000
R0
[ohm] .

Fig. 18.10 = Fig. 16.15

3
SN o.tot 0 77.981 [dBV] SNo.tot 2 10 SN o.tot 0 16.324 [dB]

GM.e 20 log GM GM.e 60 [dB]

3
EIN( R0) SNo.tot ( R0) GM.e EIN 2 10 121.657 [dBV]
298 18 Mathcad Worksheets of Measurement Tools

18.2 MCD-WS: The PMMA Page 9

3.8. SN calculation by rule-of-thumb (according to Fig. 16.11) :

We assume only one un-balanced LT1028 input stage in series configuration with a gain of G M .0 = 1000!

Fig. 18.11 = Fig. 16.11

3 9
GM.0 1000 Rf 2 2.2 10 en.i 0.85 10 V in.i ignored!
Rf
RG RG 4.404
GM.0 1

RG 9 12
en.RG.rot 0.13 10 V en.RG.rot 272.827 10 V

2 2 9
en.o.rot GM.0 en.i en.RG.rot en.o.rot 893 10 V

en.o.rot 12
en.i.rot en.i.rot 893 10 V
GM.0

en.o.rot
SNo.rot 20 log 43 SNo.rot 78 [dBV]
1V

EINrot SNo.rot 20 log GM.0 EINrot 138 [dBV]

einrot EINrot 43 einrot 181 [dBV]

4. The measurement correction figure D :

3
SNo.tot 0 10 77.981 [dBV] SNo.0R.m 78.14 [dBV]
3
SNo.tot 1 10 64.57 [dBV] SNo.1k.m 63.63 [dBV]
3
SNo.tot 2 10 61.657 [dBV] SNo.2k.m 59.72 [dBV]

3
D1 SNo.0R.m SNo.tot 0 10 D1 0.16 [dB]
3
D2 SNo.1k.m SNo.tot 1 10 D2 0.94 [dB]
3
D3 SNo.2k.m SNo.tot 2 10 D3 1.94 [dB]

R0
=> D( R0) [dB]
1000
18 Mathcad Worksheets of Measurement Tools 299

18.2 MCD-WS: The PMMA Page 10

2
1.8
1.6
1.4
1.2
[dB]

D ( R0) 1
0.8
0.6
0.4
0.2
0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
R0
[ohm]

Fig. 18.12 Graph for interpolation purposes of the measurement correction figure D

6. Derivation of the PMMA gain :

Fig. 18.13 = Fig. 16.10

300 18 Mathcad Worksheets of Measurement Tools

18.2 MCD-WS: The PMMA Page 11

vo.b vo5 vo6

Differential (balanced) gain of the MA: GM
vi.b vi.b

vi.b vo1 vo2 vo3 vo4 vo.b vo5 vo6

Un-balanced gain and voltages of the 2nd gain stage with OP5 & OP6:
Rf2
vo5 Gop5 vo1 vo3 Gop5 Rf2 R => Gop5 1 => vo5 vo1 vo3
R

Rf2
vo6 Gop6 vo2 vo4 Gop6 => Gop6 1 => vo6 vo2 vo4
R

Differential gains of the two input amps Amp A & Amp B and the gain-setting resistance RG:

vo1 vo2 vo1 vo2 2 Rf1 vo3 vo4 vo3 vo4 2Rf1
GA 1 GB 1
vi.b vi1 vi2 RG1 vi.b vi3 vi4 RG2

2Rf1
GA GB => RG1 RG2 RGM => RGM
GA 1

Differential (G A , GB) and balanced / un-balanced gains (G A.ub , GB.ub ) and voltages of the 1st gain stage with its sub
gain stages Amp A and Amp B (OP1 & OP2 and OP3 & OP4):

vo1 vi.b GA.ub.1 vo2 vi.b GA.ub.2 vo3 vi.b GB.ub.3 vo4 vi.b GB.ub.4

2 vo1 Rf1
vo2 vo1 => GA => vo1 0.5 GA vib => GA.ub.1 0.5 GA 0.5
vi.b RGM

2 vo2 Rf1
vo1 vo2 => GA => vo2 0.5 GA vib => GA.ub.2 0.5 GA 0.5
vi.b RGM

2 vo3 Rf1
vo4 vo3 => GB => vo3 0.5 GB vib => GB.ub.3 0.5 GB 0.5
vi.b RGM

2 vo4 Rf1
vo3 vo4 => GB => vo4 0.5 GB vib => GA.ub.4 0.5 GB 0.5
vi.b RGM
18 Mathcad Worksheets of Measurement Tools 301

18.2 MCD-WS: The PMMA Page 12

Consequences for the output voltages and the differential gain GM:

vo5 vi.b GA.ub.1 GB.ub.3 vo.6 vi.b GA.ub.2 GB.ub.4

vo5 vo6 vi.b GA.ub.1 GB.ub.3 GA.ub.2 GB.ub.4 vi.b 0.5GA 0.5GB 0.5GA 0.5GB

=> vo5 vo6 vi.b GA GB

vo5 vo6
GM GA GB 2GA 2GB
vi.b

3
Proof with Rf1 = Rf2 = R R 2.2 10 Rf1 R Rf2 R and GA 500

Rf1
RG 2 RG 8.818
GA 1

GB GA G GA GB G 1000

GA.ub.1 0.5 GA GA.ub.1 250

GA.ub.2 0.5 GA GA.ub.2 250

302 18 Mathcad Worksheets of Measurement Tools

18.3 MCD-WS: The PFMA Page 1

The Measurement Amp PFMA with Galvanically Isolated Input (gain = 100)

Fig. 18.14 = Fig. 17.1

1. Definition of all meaningful constants, components, etc. :

1.1 Constants, etc. :
23 1
k 1.38065 10 VAsK T 300K B1 1Hz B20k 19980 Hz
3
vi.ref 1V f 10Hz 15Hz 20 10 Hz h 1000Hz

1.2 Transformer : Tr1 = JT-10KB-D

1
n n 0.25
4
GTr1.e
20
GTr1.e 20 log( n) [dB] GTr1.e 12.041 [dB] GTr1 10 GTr1 0.25

1.3 Gain from i/p to o/p :

GMA GTr1 Gamp Gamp G1 G2 G2 2

GMA
GMA 100 G1 G1 200
2n

G2 2 Gamp G1 G2 Gamp 400

18 Mathcad Worksheets of Measurement Tools 303

18.3 MCD-WS: The PFMA Page 2

1.4 Components and resistances :

12 3 3 3
R0 10 Rp 2.5 10 Rs 225 R2a 5.1 10 R2b 1.2 10

3
R2 R2a R2b R2 6.3 10 R3a 10

1
1 1 3
R4 P1 R4a R4 ( G1 1) R3a R4 1.99 10
R4b R4c

3 3 3
R4a 1 10 R4b 1.1 10 R4c 8.25 10

1
1 1
P1 R4 R4a P1 19.412
R4b R4c

Proof: GM GTr1 G1 G2 GM 100

1.5 Noise calculation relevant feedback and input resistances :

Fig. 18.15 = Fig. 17.4

Fig. 18.16 = Fig. 17.5

304 18 Mathcad Worksheets of Measurement Tools

18.3 MCD-WS: The PFMA Page 3

1
1 1
RF RF 9.95
R3a R4

1
1 1
Ri.tot ( R0) Ri.tot ( R0) 359.495
2 R2
n ( R0 Rp) Rs

( R2a R2b Rs) 3

Zi Rp Zi 106.9 10
2
n

2. Noise and SN calculations :

9 12
en.i1 0.85 10 V fc.e1 3.5Hz in.i1 1.0 10 A fc.i1 250Hz

fc.e1 fc.i1
en.i1( f) en.i1 1 in.i1( f) in.i1 1
f f

in.i1.eff( f) in.i1( f) n

20000Hz
1 2 15
in.i1.eff.avg in.i1.eff( f) df in.i1.eff.avg 260.521 10 A
B20k
20Hz

12
en.RF 4 k T B1 RF en.RF 406.017 10 V

9
en.Ri.tot( R0) 4 k T B1 Ri.tot ( R0) en.Ri.tot( R0) 2.44 10 V

Noise of OP2 & OP3 ignored!

2 2 2 2 2 2
en.i( f R0) en.i1( f) in.i1( f) RF Ri.tot( R0) en.RF en.Ri.tot( R0)

9
en.i( h R0) 2.647 10 V

6
en.o( f R0) en.i( f R0) Gamp en.o( h R0) 1.059 10 V

20000Hz
1 2
eN.o ( R0) en.o( f R0) df 6
B1 eN.o ( R0) 149.428 10 V
20Hz
18 Mathcad Worksheets of Measurement Tools 305

18.3 MCD-WS: The PFMA Page 4

en.o( f R0) 9
en.i.MA( f R0) en.i.MA h 0 10.589 10 V
GMA

20000Hz
1 2 9
en.i.MA.avg( R0) en.i.MA( f R0) df en.i.MA.avg 0 10.571 10 V
B20k
20Hz

7
1 10
[V/rtHz]

Fig. 18.17
en.i.MA f 0
PFMA input noise
voltage density vs.
frequency
8
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

20000Hz
1 2
en.i.MA( f R0) df
B1
20Hz
SNne.i( R0) 20 log SNne.i( R0) 116.511 [dBV]
vi.ref

measured with i/p shorted: SNne.i.m 117.31 [dBV]

SNne.o( R0) SNne.i( R0) 20 log GMA SNne.o( R0) 76.511 [dBV]

3
R0 0.001 50 10 10

110
111
112
113
114
[dBV]

SNne.i( R0) 115

116
117
118
119
120
4
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 1 10
R0 .
[ohm]

Fig. 18.18 Input referred R0 dependent SN

A Unique MM Phono-Amp Noise
Reduction Method 19

19.1 Intro

Attached to a phono-amp input not many different methods allow decreasing the
noise production of an MM cartridge. Normally, the following ones help: low input
referred noise voltage and noise current of the phono-amp, low values of the
cartridge’s internal resistance, high internal inductance, etc. However, an obstacle
cannot be pushed away by simple means. It’s resistor R1 in Fig. 19.1. This figure
stands for the principal input situation of the MM cartridge with the input of the
phono-amp; Fig. 19.2 moves the situation into its noise model.
In Fig. 19.1 R0 and L0 represent the cartridges impedances, here with the values
of a so-called standard model. C1 stands for the phono-amp’s input capacitance
plus the cable capacitance; the value should hit the cartridge’s required capacitance
load. R1 is the standard phono-amp input resistance of 47 kΩ.
Figures 19.3 and 19.4 show the same situation in a balanced input environment.
Because of the complexity of the whole process, we’ll start with the noise
reduction for the un-balanced amplification first.

19.2 The Un-Balanced Noise Reduction Approach

19.2.1 Basics

The approach goes back to a JAES article,1 published in 1978. Principally, the
authors attack the noise production of a resistor like R1 in Fig. 19.1 by replacing its
noise current in.R1 with a very much lower noise current in.R1.red. According to
(19.1) a lower resistor noise current would require a higher resistor value R1new.
Without changing the load-effect of R1 by its original value this could be achieved
by an adequate resistor synthesis.

1
“Improvement of the Noise Characteristics of Amplifiers for Magnetic Transducers”, Jean M.
Hoeffelman & René P. Meys 1978, JAES Vol 28, Nr 12, p. 935 ff.

© Springer International Publishing Switzerland 2016 307

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_19
308 19 A Unique MM Phono-Amp Noise Reduction Method

Fig. 19.1 Principal situation of an MM cartridge attached to an un-balanced phono-amp input

Fig. 19.2 Noise model of Fig. 19.1

Fig. 19.3 Principal situation of an MM cartridge attached to a balanced phono-amp input

19.2 The Un-Balanced Noise Reduction Approach 309

Fig. 19.4 Noise model of Fig. 19.3

rffiffiffiffiffiffiffiffiffiffiffiffiffi
4kTB1
in:R1 ¼ ð19:1Þ
R1

In a 2003 Electronics World2 article this new method was firstly adapted to
improve the MM phono-amp input referred SN. In this article, Mr van de Gevel
named the method ‘resistor cooling’, because in (19.1) a reduction of T would
automatically decrease the noise current, and thus we would get an SN improve-
ment effect.
Later on, in 2010, Mr. Douglas Self3 named the method ‘load synthesis’4 by
electronic means, and I guess this new name in conjunction with his recommended
new circuit hits the approach best. In addition, it is rather easy to implement.
What’s missing is a mathematical course with a real MM phono-amp. It should
show the expected SN improvement based on the described methods. In the fol-
lowing sections, I will demonstrate two different calculation methods based on the
two different synthesis solutions M1 and M2 and I will compare the results with the
M3 version, the one with R1 = 47.5 kΩ. The M1 version represents Mr Self’s
original idea; M2 shows a variant of M1 by shifting the gain producing part from
the output to the input.
Figure 19.5 shows my test MM phono-amp with the input impedance Ztot(f) and the
selectable cartridge loads M1–M3. Jmp1–3 allow the selection of the different input
load situations M1–M3 of the MM phono-amp with its circuitry around OPs 3 & 4.

2
“Noise and Moving-Magnet Cartridges”, Marcel van de Gevel, EW 10-2003, p. 38 ff.
3
“Small Signal Audio Design”, Chap. 7 p. 197 ff, ‘Electronic Cartridge Loading for Lower Noise’.
4
Abbreviation for electronic load synthesis: ELS.
310 19 A Unique MM Phono-Amp Noise Reduction Method

Fig. 19.5 Test MM phono-amp à la Fig. 19.1

1 .10
7

en.i( f)
[V/rtHz]

en.R0( f) 1 .10 8

en.R1( f)

1 .10
9

1 .10 1 .10 1 .10

3 4 5
10 100
f
[Hz]

Fig. 19.6 Standard Cartridge noise voltages at the input of the test phono-amp à la Fig. 19.5

With the Standard Cartridge connected to the input Fig. 19.6 shows the various
input noise voltages before implementation of an ELS network, hence, M3 with
Jmp3 set. The solid curve is the actual one at point A of Fig. 19.5. The dashed trace
represents the noise voltage density of R0 and R1’s noise voltage density is the
dotted trace. The goal is to move the dotted trace in direction of the dashed one,
thus, drastically decreasing the solid trace >1 kHz.

19.2.2 The M1 ELS

Figure 19.7 shows the circuit situation with its replacement of the 47 kΩ by an
electronic M1 solution around OPs 1.1 & 2.1.
19.2 The Un-Balanced Noise Reduction Approach 311

Fig. 19.7 R1 replaced by an

M1 ELS

Fig. 19.8 R10 split into two

sections to produce a virtual
ground in-between them

The circuit works as follows5:

In Fig. 19.5 Jmp1 must be set. It connects A with the A in Fig. 19.7. Because of
its very high input resistance and with a gain of “1” OP1.1 has no load effect for the
cartridge and the phono-amp input. With a gain of GM1 at its output the following
op amp OP2.1 produces a signal voltage phase shift of 180°. This inverted signal is
100 % correlated to the input signal. It hits the input signal in R10 at point C. This
mixing effect is the key effect of the whole circuit. It leads to the Fig. 19.8

5
Next chapter’s Mathcad worksheet MCD-WS 20.1 gives all details of the example calculations
for versions M1–M3, plus, in its Sect. 20.10, the derivation of the here shown equations.
312 19 A Unique MM Phono-Amp Noise Reduction Method

explanation of a new virtual ground at point C where both signals eliminate each
other to 0 V. Hence, by choosing the right gain GM1 we can split R10 between A
and B into two sections: section 1 between A and C with a resistance RA = 47 kΩ
and in section 2 RC between C and D according to (19.2).

R10 ¼ ðjGM1 j þ 1ÞRA

ð19:2Þ
¼ RA þ RC

Hence, we obtain GM1 as follows:

R10
jGM1 j ¼ 1 ð19:3Þ
RA

And the new R10 based noise current looks like:

rffiffiffiffiffiffiffiffiffiffiffiffiffi
4kTB1
in:R1:M1 ¼
R10 ð19:4Þ
 in:R1

Expressed in numbers:
• in.R1.M1 = 128.72 fA/rtHz instead of 593.72 fA/rtHz with M3
• en.R.A = 6.05 nV/rtHz instead of 27.90 nV/rtHz with M3
If we would cool R1 to get the same improvement, we would need a temperature
of TM1 = TM2 = 14.1 K ≡ − 259.1 °C.
If we would choose R10 > 1 MΩ, of course, (19.4) would become even better.
However, a bigger R10 automatically leads to a bigger gain GM1. Then, the
amplified signal voltage at the output of OP2.1 could run into overload at high
frequencies. A rough calculation indicates that a nominal signal input voltage of
5 mVrms/1 kHz would become roughly 50 mVrms at 20 kHz. Then, a further
amplification by GM1 = 20.277 would increase it to appr. 1 Vrms. Thus, we are appr.
17.5 dB far from overload. Consequently, 1 MΩ is a reasonable compromise.

19.2.3 The M2 ELS

The M2 ELS version’s circuit is given in Fig. 19.9.

Principally, M2 works like M1, without change of the before given equations.
P10 sets the arrangement’s gain to GM2 = 20.277 and R10 is split into two sections,
alike Fig. 19.8. In contrast to M1, the M2 version produces a tiny portion less noise.
19.2 The Un-Balanced Noise Reduction Approach 313

Fig. 19.9 R1 replaced by an

M2 ELS

19.2.4 Results

With a Standard cartridge we gain a calculated difference between the input referred SNs
of 0.12 dB only, A-weighted and RIAA equalized, however, the difference to the M3
version becomes 2.66 dB (M1)/2.78 dB (M2). I must point out that the input capacitance
C1 plays a significant role too. My standard value is 250 pF, however, with eg 450 pF the
differences from above change to 0.13 and 2.78 dB (M1)/2.91 dB (M2).

19.2.5 Consequences

The calculations demonstrate that the difference between M1 and M2 becomes

ignorable. According to Occam’s Razor the M1 approach looks favourable (smaller
number of components with practically the same improvement effect).
When thinking of an M1 or M2 implementation we should consider the
following remarks concerning the selection of components.
Compared with the Figs. 19.5 and 19.7 components and their values the M1
noise reduction effect will increase with
• decreasing input noise voltage density of the phono-amp
• decreasing R0 and L0 values
• increasing C1 value
• decreasing noise voltage and noise current of OPs 1 & 2
• increasing R10 value
The M1 noise reduction effect will decrease with
• the opposite actions of the before recommended ones, however, on a lower
SNariaa basis even a phono-amp with 12 nV/rtHz input noise voltage density
shows an appr. 1 dB SN improvement
314 19 A Unique MM Phono-Amp Noise Reduction Method

Fig. 19.10 Input situation of a balanced phono-amp

It makes sense to chose the M2 noise reduction circuit for OPs1 & 2 input noise
voltage densities >5 nV/rtHz.

19.3 The Balanced Noise Reduction Approach

19.3.1 Basics

To check the effectiveness of the chosen noise reduction measure6 we need a

balanced input configuration. For the here presented exercise I’ve chosen an
adapted active input section à la Amp1, shown in Fig. 19.10, however, without
RIAA network components. Nevertheless, the RIAA and A-weighting effects will
be part of the calculation course. Another simplification comes from the synthe-
sizing parts. I’ll concentrate on the M1 mode because I don’t see very big
advantages of M2 and M1 works with a smaller number of components.
If we set in Fig. 19.10 two jumpers Jmp1a & Jmp1b simultaneously or the two
Jmp3a & Jmp3b simultaneously, then, we can simply select between the M1 or M3
mode of the input circuit. Figure 19.11 shows the internal circuits of M1a and M1b.
Basically, they look like two versions of the Fig. 19.7 one. However, because of the
halving of the input voltage between A and ground (between B and ground too) and
without touching the overload performance of the un-balanced version the gain of
the OP2a and OP2b stage can nearly be doubled.

6
See detailed calculation course in MCD-WS 20.2.
19.3 The Balanced Noise Reduction Approach 315

Fig. 19.11 ELS of

R1a + R1b

The following equations are essential to understand how the Figs. 19.10 and
19.11 circuits work.7

7
See Chaps. 22 and 23 for further details.
316 19 A Unique MM Phono-Amp Noise Reduction Method

• Gain of a DIF amp with equal op-amps and equal Rf values:

vo
Gdif ¼
vi
ð19:5Þ
2Rf
¼1þ
RG

• Gain of M1a or M1b:

R9
GM1a ¼  ð19:6Þ
R8

GM1b ¼ GM1a ð19:7Þ

• The value of R10a or R10b:

R10a ¼ ðjGM1a j þ 1ÞR1a ð19:8Þ

R10b ¼ R10a ð19:9Þ

• The gain GM1 can thus be obtained as follows:

R10a
jGM1a j ¼ 1 ð19:10Þ
R1a
• And the new R10a or R10b based noise current looks like:
rffiffiffiffiffiffiffiffiffiffiffiffiffi
4kTB1
in:R1a:M1a ¼
R10a ð19:11Þ
¼ in:R1b:M1b
 2
1 1
in:R1:M1a:b ¼ þ ð19:12Þ
i2n:R1a:M1a i2n:R1b:M1b

• Hence, the new noise current becomes very much smaller than the original one:

in:R1:M1a:b  in:R1a ¼ in:R1b ð19:13Þ

• Thus, the new noise voltage densities become very much smaller too:

en:R1:M1ab: ¼ in:R1:M1a:b ðR1a þ R1bÞ ð19:14Þ

19.3 The Balanced Noise Reduction Approach 317

• Improvements expressed in numbers8:

– in.R1.M1a.b = 91.01 fA/rtHz instead of 593.72 fA/rtHz with M3
– en.R1.M1a.b = 4.28 nV/rtHz instead of 27.90 nV/rtHz with M3
The left numbers look smaller than the ones of Sect. 19.2.2. However, compared
with the un-balanced case the DIF amps higher input referred noise voltage density
will worsen the resultant SNs a bit.

19.3.2 Results

With a Standard cartridge we gain a calculated difference of 3.23 dB between the

input referred SN of the M1 version and the M3 version, A-weighted and RIAA
equalized. I must point out that the input capacitance C1 plays a significant role
here too. My standard value is 250 pF, however, with eg 450 pF the differences
from above change to 3.5 dB.

19.3.3 Consequences

Generally and calculated, the balanced approach yields a better SN improvement

than the un-balanced approach, provided that the active devices are of the same type
at the same place. There’s nothing new apart from the findings already mentioned in
Sect. 19.2.5 and, concerning overload, in Sect. 19.2.2.

19.4 SN Calculations

Compared with the corresponding content of TSOS-1 & TSOS-2 concerning SN

calculations there’s nothing new in the un-balanced environment. Concerning SN
calculations of DIF amps I recommend studying Chaps. 22 and 23.

8
According to MCD-WS 20.2.
Mathcad Worksheets of the MM Noise
Reduction 20

Contents
20.1 MCD-WS: The Un-Balanced Version
20.2 MCD-WS: The Balanced Version

pffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffi
Note 1: MCD 11 has no built-in unit “rtHz” or “ Hz”. To get 1 Hz based
voltage noise and current noise densities the rms noise voltage and
current in a specific frequency range B > 1 Hz must be multiplied by
pffiffiffiffiffiffiffiffiffi
1 Hz and divided by the root of that specific frequency range √B!
Note 2: MCD 11 offers no “dB” unit. This is available from MCD 13 on!

© Springer International Publishing Switzerland 2016 319

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_20
320 20 Mathcad Worksheets of the MM Noise Reduction

20.1 MCD-WS: The Un-Balanced Version Page 1

Electronic Load Synthesis ELS - un-balanced version

(+ Standard Cartridge + 3 x OP27+1 x OP37)

20.1 Definition of physical constants and frequency ranges :

23 1 19
T 300 K k 1.38065 10 VAsK q 1.6021765 10 As
4
B20k 19980 Hz B1 1Hz h 1000Hz g 10 Hz f 20Hz 25Hz 20000 Hz

20.2 Definition of components, nominal input voltage, and calculation of the input impedance :

Fig. 20.1 = Fig. 19.5

Fig. 20.2 = Fig. 19.7 Fig. 20.3 = Fig. 19.9

20 Mathcad Worksheets of the MM Noise Reduction 321

20.1 MCD-WS: The Un-Balanced Version Page 2

3 12 3
R0 10 L0 0.5H C1 250 10 F R1 47 10
3 3 3
R8 1 10 R11 1 10 R12 47.5 R13 4.75 10 R14 R13
3
RA.M1 47 10 RA.M2 RA.M1

succ-apps of P9 and P1 0 will lead to R10= 1.000M

3 P9
P9 20.2766 10 GM1 GM1 20.277
R8
6
R10M1 RA.M1 GM1 1 R10M1 1 10

R11 R13
P10 4.3763 GM2 1 GM2 20.277
R12 P10 R14

6
P10 R12 51.876 R10M2 RA.M2 GM2 1 R10M2 1 10

3
=> RC in both cases : RC R10M1 RA.M1 RC 953 10

1
1 1
Ztot1 ( f) 2j f C1 3
R0 2j f L0 RA.M1 Ztot1 ( h) 3.237 10

1
1 1 3
Ztot2 ( f) 2j f C1 Ztot2 ( h) 3.237 10
R0 2j f L0 RA.M2

1
1 1 3
Ztot3 ( f) 2j f C1 Ztot3 ( h) 3.237 10
R0 2j f L0 R1

3
vi.nom 5 10 V

OP1 OP27 OP2 OP27 OP3 OP37 OP4 OP27

fc.e1 2.7Hz fc.e2 fc.e1 fc.e3 2.7Hz fc.e4 fc.e3

9 9
en.op1 3 10 V en.op2 en.op1 en.op3 3 10 V en.op4 en.op3

fc.e1 fc.e3
en.op1 ( f) en.op1 1 en.op2 ( f) en.op1 ( f) en.op3 ( f) en.op3 1 en.op4 ( f) en.op3 ( f)
f f

fc.i1 140Hz fc.i2 fc.i1 fc.i3 140Hz fc.i4 fc.i3

12 12
in.op1 0.4 10 A in.op2 in.op1 in.op3 0.4 10 A in.op4 in.op3

fc.i1 fc.i3
in.op1 ( f) in.op1 1 in.op2 ( f) in.op1 ( f) in.op3 ( f) in.op3 1 in.op4 ( f) in.op3 ( f)
f f
322 20 Mathcad Worksheets of the MM Noise Reduction

20.1 MCD-WS: The Un-Balanced Version Page 3

20.3 Graphs of input impedance (Ztot1(f) = Ztot2(f) = Ztot3(f) if RA = R1) :

5
1 10

4
1 10
[ohm]

Ztot1( f)
Fig. 20.4
Input impedance of the Fig. 20.1
3
1 10 phono-amp input load

100
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

90
80
70
60
50
[degrees]

40
arg Ztot1( f) 30 Fig. 20.5
20
deg 10 Phase of the Fig. 20.1 phono-amp
0 input load
10
20
30
40
50
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

20.4 Calculation of the noise voltages, currents, and the virtual temperature of the three different
input impedance components M1 ... M3 :
9
en.R0 4 k T R0 B1 en.R0 4.07 10 V

Version M1 :

4 k T B1
in.R1.M1 15
GM1 1 RA.M1 in.R1.M1 128.716 10 A

9
en.R1.M1 in.R1.M1 RA.M1 en.R1.M1 6.05 10 V

2
en.R1.M1
TM1 TM1 14.1 K
4 k B1 RA.M1

Version M2 :

4 k T B1
in.R1.M2 15
GM2 1 RA.M2 in.R1.M2 128.716 10 A
20 Mathcad Worksheets of the MM Noise Reduction 323

20.1 MCD-WS: The Un-Balanced Version Page 4

9
en.R1.M2 in.R1.M2 RA.M2 en.R1.M2 6.05 10 V

2
en.R1.M2
TM2 TM2 14.1 K
4 k B1 RA.M2

Version M3 :

9
en.R1.M3 4 k T R1 B1 en.R1.M3 27.905 10 V

4 k T B1 15
in.R1.M3 in.R1.M3 593.722 10 A
R1

20.5 The input voltage deviders (see TSOS-1 Ch. 4 and TSOS-2 Ch. 13):
1
1
Z0( f) R0 2j f L0 Z1a ( f) 2j f C1
Z0( f)

with RA = synthesized R1:

1
1
ZM1 ( f) 2j f C1
RA.M1 ZM2 ( f) ZM1 ( f)

ZM1 ( f)
en1.M1 ( f) en.R0 9
Z0( f) ZM1 ( f) en1.M1 ( h) 3.996 10 V

en1.M2 ( f) en1.M1 ( f)

Z1a ( f) 12
en2.M1 ( f) en.R1.M1 en2.M1 ( h) 416.599 10 V
Z1a ( f) RA.M1

en2.M2 ( f) en2.M1 ( f)

2 2 9
en.i.M1( f) en1.M1 ( f) en2.M1 ( f) en.i.M1( h) 4.018 10 V

en.i.M2( f) en.i.M1( f)

with R1:
1
1
ZM3 ( f) 2j f C1
R1
324 20 Mathcad Worksheets of the MM Noise Reduction

20.1 MCD-WS: The Un-Balanced Version Page 5

ZM3 ( f) 9
en1.M3 ( f) en.R0 en1.M3 ( h) 3.996 10 V
Z0( f) ZM3 ( f)

Z1a ( f) 9
en2.M3 ( f) en.R1.M3 en2.M3 ( h) 1.922 10 V
Z1a ( f) R1

2 2 9
en.i.M3( f) en1.M3 ( f) en2.M3 ( f) en.i.M3( h) 4.434 10 V

7
1 10

8
1 10
en1.M1( f)
[V/rtHz]

en2.M1( f) 1 10 9

en2.M3( f) Fig. 20.6

10 Noise voltage density
1 10 of the three different
input voltage dividers
11
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

en.R0( f) en.R0 en.R1( f) en.R1.M3

7
1 10

en.i.M3( f)
[V/rtHz]

en.R0( f) 1 10
8

en.R1( f)

Fig. 20.7 = Fig. 19.6

9
1 10
3 4 5
10 100 1 10 1 10 1 10
f . .
[Hz]

6. Calculation of the noise effect of RIAA feedback network and the input noise voltage of the
phono-amp (noise of OP4 + components = ignorable) :
3 3 9
R2 100 R3 26.5 10 R4 2.847 10 C3 120 10 F
20 Mathcad Worksheets of the MM Noise Reduction 325

20.1 MCD-WS: The Un-Balanced Version Page 6

1 1
1 1 1
Z3( f) R4 2j f C3 Z4( f) en.Z4( f) 4 k T Z4( f) B1
R3 R2 Z3( f)

2 2 2 2 9
en.i.riaa( f) en.op3 ( f) in.op3 ( f) Z4( f) en.Z4( f) en.i.riaa( h) 3.261 10 V

20.7 Calculation of the noise effects of OP1.x and OP2.y :

P9
GN.op2.1 1 GN.op2.1 21.277
R8
9
en.R8 4 k T B1 R8 en.R8 4.07 10 V

9
en.P9 4 k T B1 P9 en.P9 18.329 10 V

2
2 2 P9 2 2 2 2
en.o.op2.1.tot( f) GN.op2.1 en.op2 ( f) en.R8 en.P9 in.op2 ( f) P9
R8
9
en.o.op2.1.tot ( h) 106.338 10 V

2 2 2 9
en.o.M1( f) en.op1 ( f) GM1 en.o.op2.1.tot( f) en.o.M1( h) 122.548 10 V

Ztot1 ( f) 12
en.M1.eff( f) en.o.M1( f) en.M1.eff( h) 396.165 10 V
R10M1 Ztot1 ( f)

9
en.R11 4 k T B1 R11 en.R11 4.07 10 V

12
en.P10 4 k T B1 ( R12 P10) en.P10 927.079 10 V

2 1
2 2 1 1 1 1
en.i.op1.2 ( f) en.op1 ( f) in.op1 ( f) R11 ( R12 P10)
2 2
en.R11 en.P10

9
en.i.op1.2 ( h) 3.137 10 V
R14
GN.op2.2 1 GN.op2.2 2
R13
9
en.R13 4 k T B1 R13 en.R13 8.871 10 V en.R14 en.R13

2
2 2 R14 2 2 2 2
en.o.op2.2.tot( f) GN.op2.2 en.op2 ( f) en.R13 en.R14 in.op2 ( f) R14
R13
9
en.o.op2.2.tot ( h) 14.057 10 V
326 20 Mathcad Worksheets of the MM Noise Reduction

20.1 MCD-WS: The Un-Balanced Version Page 7

2 2 2 9
en.o.M2( f) en.i.op1.2 ( f) GM2 en.o.op2.2.tot ( f) en.o.M2( h) 65.146 10 V

Ztot2 ( f) 12
en.M2.eff( f) en.o.M2( f) en.M2.eff( h) 210.598 10 V
R10M2 Ztot2 ( f)

8. Graph and calculation of the different total input referred noise voltages :

2 2 2 2 9
en.i.tot.M1( f) en.i.riaa( f) en.i.M1( f) en.M1.eff( f) in.op3 ( f) Ztot1 ( f) en.i.tot.M1( g) 16.974 10 V

2 2 2 2 9
en.i.tot.M2( f) en.i.riaa( f) en.i.M2( f) en.M2.eff( f) in.op3 ( f) Ztot2 ( f) en.i.tot.M2( g) 16.574 10 V

2 2 2 9
en.i.tot.M3( f) en.i.riaa( f) en.i.M3( f) in.op3 ( f) Ztot3 ( f) en.i.tot.M3( g) 26.72 10 V

8
4 10
8
3.5 10
8
3 10
en.i.tot.M1( f) 8
2.5 10
[V/rtHz]

en.i.tot.M2( f) 8
2 10
8
en.i.tot.M3( f) 1.5 10
8
1 10
9
5 10
0
4 4 4
0 5000 1 10 1.5 10 2 10
f .
[Hz]

Fig. 20.8 Total input noise voltage densities at the input of the phono-amp (A)

9. Calculation of SNs :

9.1 Non-equalized :

20000Hz
1 2
en.i.tot.M1( f) df
B1
20Hz
SN ne.M1 20 log
vi.nom SN ne.M1 67.133 [dB]
20 Mathcad Worksheets of the MM Noise Reduction 327

20.1 MCD-WS: The Un-Balanced Version Page 8

20000Hz
1 2
en.i.tot.M2( f) df
B1
20Hz
SN ne.M2 20 log
vi.nom SN ne.M2 67.338 [dB]

20000Hz
1 2
en.i.tot.M3( f) df
B1
20Hz
SN ne.M3 20 log
vi.nom SN ne.M3 63.181 [dB]

SN ne.D1 SN ne.M1 SN ne.M3 SN ne.D1 3.952 [dB]

SN ne.D2 SN ne.M2 SN ne.M3 SN ne.D2 4.157 [dB]

SN ne.D3 SN ne.M2 SN ne.M1 SN ne.D3 0.205 [dB]

9.2 RIAA-equalized :

1
2
3 6
1 2 10 Hz 318 10 s
R1000
2 2 R1000 9.898
3 6 3 6
1 2 10 Hz 3180 10 s 1 2 10 Hz 75 10 s

2
6
1 2 f 318 10 s
R( f) R1000
2 2
6 6
1 2 f 3180 10 s 1 2 f 75 10 s

20000Hz
1 2 2
en.i.tot.M1( f) R( f) df
B1
20Hz
SN riaa.M1 20 log SN riaa.M1 78.114 [dB]
vi.nom

20000Hz
1 2 2
en.i.tot.M2( f) R( f) df
B1
20Hz
SN riaa.M2 20 log SN riaa.M2 78.202 [dB]
vi.nom

20000Hz
1 2 2
en.i.tot.M3( f) R( f) df
B1
20Hz
SN riaa.M3 20 log SN riaa.M3 75.979 [dB]
vi.nom
328 20 Mathcad Worksheets of the MM Noise Reduction

20.1 MCD-WS: The Un-Balanced Version Page 9

SN riaa.D1 SNriaa.M1 SN riaa.M3 SN riaa.D1 2.135 [dB]

SNriaa.D2 SNriaa.M2 SNriaa.M3 SN riaa.D2 2.223 [dB]

SNriaa.D3 SNriaa.M2 SN riaa.M1 SNriaa.D3 0.088 [dB]

9.3 Non-equalized but A-weighted :

Definition of A-filter frequencies and transfer function:

f1 20.6Hz f2 f1 f3 107.7Hz f4 737.9Hz f5 12200 Hz f6 f5 fG 1000Hz

2 2
2 2 2 2
f1 f3 f4 fG
v1000 1 1 1 1 v1000 1.259
fG fG fG f5

2 2
1 1 1 1
A( f) v1000
2 2 2 2
f1 f3 f4 f
1 1 1 1
f f f f5

20000Hz
1 2 2
en.i.tot.M1( f) A( f) df
B1
20Hz
SN a.M1 20 log SNa.M1 71.028 [dB(A)]
vi.nom

20000Hz
1 2 2
en.i.tot.M2( f) A( f) df
B1
20Hz
SNa.M2 20 log SNa.M2 71.219 [dB(A)]
vi.nom

20000Hz
1 2 2
en.i.tot.M3( f) A( f) df
B1
20Hz
SNa.M3 20 log SNa.M3 67.289 [dB(A)]
vi.nom

SNa.D1 SNa.M1 SNa.M3 SNa.D1 3.739 [dB]

SN a.D2 SNa.M2 SNa.M3 SNa.D2 3.93 [dB]

SN a.D3 SNa.M2 SNa.M1 SNa.D3 0.191 [dB]

20 Mathcad Worksheets of the MM Noise Reduction 329

20.1 MCD-WS: The Un-Balanced Version Page 10

9.4 RIAA-equalized and A-weighted :

20000Hz
1 2 2 2
en.i.tot.M1( f) A( f) R( f) df
B1
20Hz
SNariaa.M1 20 log
vi.nom

SNariaa.M1 81.405 [dB(A)]

20000Hz
1 2 2 2
en.i.tot.M2( f) A( f) R( f) df
B1
20Hz
SNariaa.M2 20 log
vi.nom

SNariaa.M2 81.525 [dB(A)]

20000Hz
1 2 2 2
en.i.tot.M3( f) A( f) R( f) df
B1
20Hz
SNariaa.M3 20 log
vi.nom

SNariaa.M3 78.745 [dB(A)]

SNariaa.D1 SNariaa.M1 SNariaa.M3 SNariaa.D1 2.66 [dB]

SNariaa.D2 SNariaa.M2 SNariaa.M3 SNariaa.D2 2.78 [dB]

SNariaa.D3 SNariaa.M2 SNariaa.M1 SNariaa.D3 0.12 [dB]

330 20 Mathcad Worksheets of the MM Noise Reduction

20.1 MCD-WS: The Un-Balanced Version Page 11

10. Derivation of the reduced noise voltage of the nominal input load of 47k :

Fig. 20.9 Derivation sequence

Fig. 20.9 a)
3 9
vi vi.nom R1 47 10 en.R1 4 k T B1 R1 en.R1 27.905 10 V

4 k T B1 15
in.R1 in.R1 593.722 10 A
R1
vi
ii.nom 9
R1 ii.nom 106.383 10 A

Fig. 20.9 b)
3
we choose Ri = 1k and an overload-friendly gain |G M 1| ~ 20 : Ri 1 10 GM1 20

Fig. 20.9 c)
3 3
If Rx 47 10 than Ry becomes Ry GM1 Rx Ry 940 10

3
Fig. 20.9 d) Rxy Rx Ry Rxy 987 10

3
succ-apps of RF should lead to an easier-to-handle Rxy = 1M and the final value for GM 1 : RF 20.2766 10

RF 6
GM1 GM1 20.277 Rxy Rx GM1 Rx Rxy 1 10
Ri

3
=> Ry Rxy Rx Ry 953 10
20 Mathcad Worksheets of the MM Noise Reduction 331

20.1 MCD-WS: The Un-Balanced Version Page 12

The value of G M 1 must equal the ration of i o / i i , hence:

3
vo vi GM1 vo 101.383 10 V

vi 9 vo 9
ii ii 5 10 A io io 101.383 10 A
GM1 1 Rx Rx 1 GM1

io 9
=> 20.277 ires1 ii io ires1 96.383 10 A
ii

vi vi GM1
ires2 vi GM1 1 9
GM1 1 Rx Rx 1 GM1 ires2 ires2 96.383 10 A
Rx GM1 1

=> ires1 ires2

Fig. 20.9 e) & d)

3
vR.xy ires1 Rx GM1 1 vR.xy 96.383 10 V

3
vR.x ires1 Rx vR.x 4.53 10 V

vR.x Rx vR.x 3
Rx Rxy Rx RA.M1 => RA.M1 47 10
vR.xy Rxy vR.xy

9
en.R.xy 4 k T B1 Rxy en.R.xy 128.716 10 V

=> noise voltage approach to get the noise voltage of the synthesized 47k resistor :

en.R.x Rx Rx 9
en.R.x e en.R.A.M1 en.R.x => en.R.A.M1 6.05 10 V
en.R.xy Rxy Rxy n.R.xy

=> noise current approach to get the noise voltage of the synthesized 47k resistor :

4 k T B1 15 9
in.xy in.xy 128.716 10 A en.R.x in.xy Rx => en.R.x 6.05 10 V
Rxy

9
vs. en.R1 29.7 10 V
332 20 Mathcad Worksheets of the MM Noise Reduction

20.2 MCD-WS: The Balanced Version Page 1

Electronic Load Synthesis ELS - balanced version

(+ Standard Cartridge + 2 x OP27)

1. Definition of physical constants and frequency ranges :

23 1 19
T 300 K k 1.38065 10 VAsK q 1.6021765 10 As
4
B20k 19980 Hz B1 1Hz h 1000Hz g 10 Hz f 20Hz 25Hz 20000 Hz

2. Definition of components, nominal input voltage, and calculation of the input impedance :

Fig. 20.10 = Fig. 19.10

3
R0 10 L0 0.5H

12
C1 250 10 F

3
R1a 23.5 10 R1b R1a
3
R8a 1 10 R8b R8a

succ-apps of G leads to R10a = 1M

GM1a 41.56 GM1b GM1a

3
R9a R8a GM1a R9a 41.56 10

R9b R9a

R10a GM1a 1 R1a

Fig. 20.11
= Fig. 19.11
6
R10a 1 10 R10b R10a
20 Mathcad Worksheets of the MM Noise Reduction 333

20.2 MCD-WS: The Balanced Version Page 2

1
1 1
Ztot1 ( f) 2j f C1 3
R0 2j f L0 R1a R1b Ztot1 ( h) 3.237 10

3
OP1b OP1a OP2b OP2a vi.nom 5 10 V

OP1a OP27 OP2a OP27 OP3 OP37 OP4 OP27

fc.e1 2.7Hz fc.e2 fc.e1 fc.e3 2.7Hz fc.e4 fc.e3

9 9
en.op1 3 10 V en.op2 en.op1 en.op3 3 10 V en.op4 en.op3

fc.e1 fc.e3
en.op1 ( f) en.op1 1 en.op2 ( f) en.op1 ( f) en.op3 ( f) en.op3 1 en.op4 ( f) en.op3 ( f)
f f

fc.i1 140Hz fc.i2 fc.i1 fc.i3 140Hz fc.i4 fc.i3

12 12
in.op1 0.4 10 A in.op2 in.op1 in.op3 0.05 10 A in.op4 in.op3

fc.i1 fc.i3
in.op1 ( f) in.op1 1 in.op2 ( f) in.op1 ( f) in.op3 ( f) in.op3 1 in.op4 ( f) in.op3 ( f)
f f

3. Graphs of input impedance Ztot1(f) :

5
1 10

4
1 10
[ohm]

Ztot1( f)
Fig. 20.12
Input impedance of the Fig. 20.1
3
1 10 phono-amp input load

100
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

90
80
70
60
50
[degrees]

40
arg Ztot1( f) 30 Fig. 20.13
20
deg 10 Phase of the Fig. 20.12 phono-amp
0 input load
10
20
30
40
50
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]
334 20 Mathcad Worksheets of the MM Noise Reduction

20.2 MCD-WS: The Balanced Version Page 3

4. Calculation of the noise voltages & currents of the input impedance components M1 & M3 :

9
en.R0 4 k T R0 B1 en.R0 4.07 10 V

Version M1 :

4 k T B1
in.R1a.M1a 15
GM1a 1 R1a in.R1b.M1b in.R1a.M1a in.R1a.M1a 128.706 10 A

1
1 1 15
in.R1.M1a.b in.R1.M1a.b 91.009 10 A
2 2
in.R1a.M1a in.R1b.M1b

9
en.R1a.M1a in.R1a.M1a R1a en.R1b.M1b en.R1a.M1a en.R1a.M1a 3.025 10 V

2 2 9
en.R1.M1a.b en.R1a.M1a en.R1b.M1b en.R1.M1a.b 4.277 10 V

Version M3 :

9
en.R1.M3a.b 4 k T ( R1a R1b) B1 en.R1.M3a.b 27.905 10 V

4 k T B1 15
in.R1.M3a.b in.R1.M3a.b 593.722 10 A
R1a R1b

5. The input voltage deviders (see TSOS-1 Ch. 4 and TSOS-2 Ch. 13):
1
1
Z0( f) R0 2j f L0 Z1a ( f) 2j f C1
Z0( f)

with R1a & R1b= synthesized:

1
1
ZM1 ( f) 2j f C1
R1a R1b

ZM1 ( f)
en1.M1 ( f) en.R0 9
Z0( f) ZM1 ( f) en1.M1 ( h) 3.996 10 V

Z1a ( f) 12
en2.M1 ( f) en.R1.M1a.b en2.M1 ( h) 294.556 10 V
Z1a ( f) R1a R1b

2 2 9
en.i.M1( f) en1.M1 ( f) en2.M1 ( f) en.i.M1( h) 4.007 10 V
20 Mathcad Worksheets of the MM Noise Reduction 335

20.2 MCD-WS: The Balanced Version Page 4

with R1a+R1b:
1
1
ZM3 ( f) 2j f C1
R1a R1b

ZM3 ( f) 9
en1.M3 ( f) en.R0 en1.M3 ( h) 3.996 10 V
Z0( f) ZM3 ( f)

Z1a ( f) 9
en2.M3 ( f) en.R1.M3a.b en2.M3 ( h) 1.922 10 V
Z1a ( f) R1a R1b

2 2 9
en.i.M3( f) en1.M3 ( f) en2.M3 ( f) en.i.M3( h) 4.434 10 V

7
1 10

8
1 10
en1.M1( f)
[V/rtHz]

en2.M1( f) 1 10 9

en2.M3( f) Fig. 20.14

10 Noise voltage density
1 10 of the three different
input voltage dividers
11
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

en.R0( f) en.R0 en.R1.M3a.b( f) en.R1.M3a.b

7
1 10

en.i.M3( f)
[V/rtHz]

en.R0( f) 1 10
8

en.R1.M3a.b( f)

Fig. 20.15 = Fig. 19.12

9
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]
336 20 Mathcad Worksheets of the MM Noise Reduction

20.2 MCD-WS: The Balanced Version Page 5

6. Calculation of the input noise voltage of the phono-amp

(noise of RIAA section + components = ignorable) :
3 Rf
RG 100 Rf 1 10 GDIF 1 2 GDIF 21
RG
1
1 1
RGf RGf 90.909
RG Rf

9
en.RG.f 4 k T B1 RGf en.RG.f 1.227 10 V

2 2 9
en.i3.4( f) en.op3 ( f) en.op4 ( f) en.i3.4( h) 4.248 10 V

1
1 1 15
in.i3.4( f) in.i3.4( h) 37.749 10 A
2 2
in.op3 ( f) in.op4 ( f)

2 2 2 2 9
en.i.tot3.4( f) en.i3.4( f) in.i3.4( f) RGf en.RG.f en.i.tot3.4( h) 4.422 10 V

7. Calculation of the noise effects of OP1a,b and OP2a,b :

R9a
GN.op2a 1 GN.op2a 42.56
R8a
9
en.R8a 4 k T B1 R8a en.R8a 4.07 10 V

8
en.R9a 4 k T B1 R9a en.R9a 2.624 10 V

2
2 2 R9a 2 2 2 2
en.o.op2a.tot ( f) GN.op2a en.op2 ( f) en.R8a en.R9a in.op2 ( f) R9a
R8a
9
en.o.op2a.tot ( h) 214.397 10 V

en.o.op2b.tot ( f) en.o.op2a.tot ( f)

2 2 2 9
en.o.M1a ( f) en.op1 ( f) GM1a en.o.op2a.tot ( f) en.o.M1a ( h) 248.099 10 V

en.o.M1b ( f) en.o.M1a ( f)

9
en.o.M1( f) 2 en.o.M1a ( f) en.o.M1( h) 350.866 10 V

Ztot1 ( f) 12
en.M1.eff( f) en.o.M1( f) en.M1.eff( h) 567.373 10 V
R10a R10b Ztot1 ( f)
20 Mathcad Worksheets of the MM Noise Reduction 337

20.2 MCD-WS: The Balanced Version Page 6

8. Graph and calculation of the different total input referred noise voltages :

2 2 2 2
en.i.tot.M1( f) en.i.tot3.4( f) en.i.M1( f) en.M1.eff( f) in.i3.4( f) Ztot1 ( f)

9
en.i.tot.M1( g) 9.685 10 V

2 2 2 9
en.i.tot.M3( f) en.i.tot3.4( f) en.i.M3( f) in.i3.4( f) Ztot1 ( f) en.i.tot.M3( g) 22.58 10 V

8
4 10
8
3.5 10
8
3 10
8
en.i.tot.M1( f) 2.5 10
[V/rtHz]

8
2 10
en.i.tot.M3( f)
8
1.5 10
8
1 10
9
5 10
0
4 4 4
0 5000 1 10 1.5 10 2 10
f .
[Hz]

Fig. 20.16 Total input noise voltage densities at the input of the phono-amp (points A-B)

9. Calculation of SNs :

9.1 Non-equalized :

20000Hz
1 2
en.i.tot.M1( f) df
B1
20Hz
SN ne.M1 20 log
vi.nom SN ne.M1 71.94 [dB]

20000Hz
1 2
en.i.tot.M3( f) df
B1
20Hz
SN ne.M3 20 log
vi.nom SN ne.M3 64.627 [dB]

SN ne.D1 SN ne.M1 SN ne.M3 SN ne.D1 7.313 [dB]

338 20 Mathcad Worksheets of the MM Noise Reduction

20.2 MCD-WS: The Balanced Version Page 7

9.2 RIAA-equalized :

1
2
3 6
1 2 10 Hz 318 10 s
R1000
2 2 R1000 9.898
3 6 3 6
1 2 10 Hz 3180 10 s 1 2 10 Hz 75 10 s

2
6
1 2 f 318 10 s
R( f) R1000
2 2
6 6
1 2 f 3180 10 s 1 2 f 75 10 s

20000Hz
1 2 2
en.i.tot.M1( f) R( f) df
B1
20Hz
SN riaa.M1 20 log SN riaa.M1 78.593 [dB]
vi.nom

20000Hz
1 2 2
en.i.tot.M3( f) R( f) df
B1
20Hz
SN riaa.M3 20 log SN riaa.M3 76.307 [dB]
vi.nom

SN riaa.D1 SN riaa.M1 SN riaa.M3 SN riaa.D1 2.285 [dB]

9.3 Non-equalized but A-weighted :

Definition of A-filter frequencies and transfer function:

f1 20.6Hz f2 f1 f3 107.7Hz f4 737.9Hz f5 12200 Hz f6 f5 fG 1000Hz

2 2
2 2 2 2
f1 f3 f4 fG
v1000 1 1 1 1 v1000 1.259
fG fG fG f5

2 2
1 1 1 1
A( f) v1000
2 2 2 2
f1 f3 f4 f
1 1 1 1
f f f f5
20 Mathcad Worksheets of the MM Noise Reduction 339

20.2 MCD-WS: The Balanced Version Page 8

20000Hz
1 2 2
en.i.tot.M1( f) A( f) df
B1
20Hz
SN a.M1 20 log SNa.M1 74.965 [dB(A)]
vi.nom

20000Hz
1 2 2
en.i.tot.M3( f) A( f) df
B1
20Hz
SNa.M3 20 log SNa.M3 68.628 [dB(A)]
vi.nom

SN a.D1 SN a.M1 SNa.M3 SNa.D1 6.337 [dB]

9.4 RIAA-equalized and A-weighted :

20000Hz
1 2 2 2
en.i.tot.M1( f) A( f) R( f) df
B1
20Hz
SNariaa.M1 20 log
vi.nom

SNariaa.M1 82.659 [dB(A)]

20000Hz
1 2 2 2
en.i.tot.M3( f) A( f) R( f) df
B1
20Hz
SN ariaa.M3 20 log
vi.nom

SN ariaa.M3 79.433 [dB(A)]

SN ariaa.D1 SN ariaa.M1 SN ariaa.M3 SN ariaa.D1 3.226 [dB]

BJT Circuits in CE Configuration
21

21.1 Intro

In my TSOS books, I’ve demonstrated the usage of a certain kind of MC

phono-amp input stage, composed by a CE configured BJT that is followed by an
op-amp, principally according to Fig. 5.6b in TSOS-2 and Fig. 3.23b in TSOS-1.
However, I didn’t show an exact method to calculate the gain of the chosen BJT
input stage that drives the Module 2 phono-amps (four BJTs in parallel operation).
In addition, the given noise calculation equations were based on white noise only.
Now, this chapter will give equations
• to calculate the gain of a range of different CE configured BJT input stages,
including the one of Module 2,
• to calculate frequency dependent BJT noise voltages for 1/f-noise corner fre-
quencies fc > 0 Hz.
Before we enter into the details of the input stages, and to prepare the handling
of the frequency dependency, we have to repeat some BJT basics in Sect. 21.2.1

21.2 BJT—Bipolar Junction Transistor—Basics

21.2.1 Equations for Low Frequency Small Signal Calculations

All given equations are based on the Fig. 21.1 NPN BJT device model and a chosen
operating point. The use of PNP devices requires inversed supply voltages and the
turn over of electrolytic capacitors only.

1
T/S, Chap. 2, and my own derivations are the main sources of this chapter’s equations (see
Appendix 2).

© Springer International Publishing Switzerland 2016 341

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_21
342 21 BJT Circuits in CE Configuration

Fig. 21.1 BJT model for low-frequency small signal calculation purposes

• Mutual conductance gm expressed in terms of DC collector current IC, Bolz-

mann’s constant k, absolute temperature T and the electron charge q:

qIC
gm ¼ ð21:1Þ
kT
• Small signal current gain hfe2 expressed in terms of the DC current gain hFE3
(=suitable approach on the small signal field):

hfe ffi hFE ð21:2Þ

• Base-emitter resistance rbe expressed in terms of small signal current gain hfe
and mutual conductance gm:

hfe
rbe ¼ ð21:3Þ
gm

• Collector-emitter resistance rce expressed in terms of Early voltage VA4 and the
operating collector current IC:

VA þ VCE
rce ¼ ð21:4Þ
IC

VA:npn ¼ 30 V to 150 V
ð21:5Þ
VA:pnp ¼ 30 V to  75 V

2
Also called β in some regions of the world.
3
Also called B in some regions of the world.
4
Usually, VA is not indicated in data sheets; it must be guessed or determined with the help of the
output characteristics chart (see T/S, p. 36). VCE can be ignored in cases of VA >> VCE.
21.2 BJT—Bipolar Junction Transistor—Basics 343

21.2.2 Circuit Parameter Based Formulae

All given formulae are based on the common emitter configuration (e).
• Mutual conductance gm:

h21;e hfe
gm ¼ ¼ y21;e ¼ ð21:6Þ
h11;e hie

• Small signal current gain hfe:

y21;e
hfe ¼ h21;e ¼ ð21:7Þ
y11;e

• Base-emitter Resistance rbe:

1
rbe ¼ h11;e ¼ ¼ hie ð21:8Þ
y11;e

• Collector-emitter resistance rce:

h11;e 1 hie
rce ¼ ¼ ¼ ð21:9Þ
h11;e h22;e  h12;e h21;e y22;e hie hoe  hre hfe

21.2.3 Noise of a BJT—Frequency Independent Version

• Equivalent noise model (Fig. 21.2):

Fig. 21.2 BJT noise model

344 21 BJT Circuits in CE Configuration

• Equivalent input noise voltage density en.T1:

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:T1 ¼ e2n1:T1 þ i2n:T1 r2bb0 þ e2n:rbb0 ð21:10Þ

sffiffiffiffiffiffiffiffiffiffiffiffiffi
2
en1:T1 ¼ kT B1 ð21:11Þ
qIC
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:rbb0 ¼ 4kTrbb0 B1 ð21:12Þ

B1 ¼ 1 Hz ð21:13Þ

• Equivalent input noise current density in.T1:

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
IC
in:T1 ¼ 2q B1 ð21:14Þ
hfe

21.2.4 Noise of a BJT—Frequency Dependent Version

Many BJTs show significant low-frequency noise (flicker or 1/f-noise). It is part of

the collector current, hence, part of the base current too. With that in mind, the key
equations of Sect. 21.2.3 change to the following ones:
• Equivalent input noise voltage density en.T1(f):
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:T1 ðf Þ ¼ e2n1:T1 þ in:T1 ðf Þ2 r2bb0 þ e2n:rbb0 ð21:15Þ

• Equivalent input noise current density in.T1(f) with the 1/f-noise corner fre-
quency fc.i:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffirffiffiffiffiffiffiffiffiffiffiffiffiffiffi
IC f c:i
in:T1 ðf Þ ¼ 2q B1 1 þ ð21:16Þ
hfe f

Note: fc.i is not given in data sheets. However, some manufacturers show noise
current versus frequency noise charts that allow a graphical evaluation of fc.i. Others
give frequency-based charts with information about a range of noise figures NF versus
resistor input load and collector current. Here, the evaluation of fc.i follows the rather
complex math approach that is described in detail in this book’s Chaps. 10 and 11.
21.2 BJT—Bipolar Junction Transistor—Basics 345

21.2.5 Noise of a Resistor R

• Noise voltage density:

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:R ¼ 4kTRB1 ð21:17Þ

• Noise current density:

rffiffiffiffiffiffiffiffiffiffiffiffiffiffi
4kT
in:R ¼ B1 ð21:18Þ
R

• Excess noise—average and thus frequency independent version for a chosen

frequency bandwidth of interest:
With NI = resistor noise index in [µV/decade/1VR] or in [dB] with NIe = 20log
(NI) + 120 (see resistor data sheet), VR = DC voltage across R = IC*R, and
d = number of decades in B = frequency bandwidth of interest we’ll obtain the
average excess noise voltage density en.Rex in B:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:R:tot ¼ e2n:R þ e2n:Rex ð21:19Þ

rffiffiffiffiffiffi
pffiffiffi B1
en:Rex ¼ NI d VR ð21:20Þ
B

• Excess noise—frequency dependent version:

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:R:tot ðf Þ ¼ e2n:R þ en:Rex ðf Þ2 ð21:21Þ
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u0 NI 1
u e
u  
uB10 10 1012 C V2R
en:Rex ðf Þ ¼ u B C ð21:22Þ
t@ lnð10Þ A f B1

21.2.6 Noise Factor & Noise Figure

• Noise Factor NF:

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
e2n:T1 þ i2n:T1 R02 þ e2n:R0
NF ¼ ð21:23Þ
en:R0
346 21 BJT Circuits in CE Configuration

• Noise Figure NFe:

NFe ¼ 20 logðNFÞ ð21:24Þ

Note: In case of frequency dependent calculations, the frequency of interest

should be 1 kHz.

21.2.7 Signal-to-Noise Ratios SN

With the input and output referred reference voltages vi.ref and vo.ref we obtain the
SNs in B = fhi − flo. Here, B1 = 1 Hz and en.i(f) stands for the frequency dependent
input noise voltage density of a gain stage. The same applies to the SN of the output
noise voltage density en.o(f):
• Input referred SNi:
0sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Z ffi1
1 f hi 2
B jen:i ðf Þj df C
B B1 f lo C
B
SNi ¼ 20 logB C ð21:25Þ
C
@ v i:ref A

• Output referred SNo:

0sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Z ffi1
1 f hi 2
B jen:o ðf Þj df C
B B1 f lo C
B
SNo ¼ 20 logB C ð21:26Þ
C
@ v o:ref A

21.3 Basic (b) CEb Circuit

See Fig. 21.3.

Fig. 21.3 Basic common
emitter circuit CE
21.3 Basic (b) CEb Circuit 347

21.3.1 Idle Gains Gb and Gb.rot

vo
Gb ¼  ð21:27Þ
vi
 
RC k rce hfe 1 1 1
Gb ¼ hfe ¼ þ
r rbe RC rce
 be    ð21:28Þ
1 1 1 qIC 1 1 1
¼ gm þ ¼ þ
RC rce kT RC rce

Gb.rot (RC ≪ rce):

Gb:rot  gm RC ð21:29Þ

21.3.2 Gb(RL) = RL Dependent Gain Gb

 1
1 1 1
Gb ðRLÞ ¼ gm þ þ ð21:30Þ
rce RC RL

Cout without negative frequency and phase response effect!

21.3.3 Input Resistance ri (O/P Open)

ri ¼ rbe ð21:31Þ

21.3.4 Output Resistance ro.o (I/P = Open)

ro:o  RC ð21:32Þ
348 21 BJT Circuits in CE Configuration

21.3.5 Output Resistances ro.s and ro.s.rot (I/P = Shorted)

ro:s ¼ rce k RC ð21:33Þ

ro.s.rot (rce ≫ RC):

ro:s:rot  RC ð21:34Þ

21.3.6 Operating Gains Gop(R0, RL) and Gops(f, R0, RL)

As of Fig. 21.4 inclusion of the bias setting input resistors R1 & R2 leads to the
operating input resistance ri.ops and output resistance ro.ops:

ri:ops ¼ ri k R1 k R2 ð21:35Þ

ro:ops ¼ ro:s ð21:36Þ

If Cin and Cout do not hurt the flat frequency and phase response in the bandwidth
of interest, the corresponding frequency independent equations look as follows:

vo ðR0; RLÞ
Gop ðR0; RL) ¼
v0
ð21:37Þ
= Gi ðR0ÞGb Go ðRLÞ
= Gi ðR0ÞGb ðRLÞ

Fig. 21.4 CEb’s operational model

21.3 Basic (b) CEb Circuit 349

Gb ðRLÞ ¼ Gb Go ðRLÞ ð21:38Þ

ri:ops
Gi ðR0Þ ¼ ð21:39Þ
ri:ops + R0

RL
Go ðRLÞ ¼ ð21:40Þ
ro:ops þ RL

Via inclusion of the Miller-capacitance-effect for Ci and Co, the frequency

dependent version becomes thus:

ri:ops ðf Þ ¼ ðzi ðf Þ k R1 k R2Þ þ ð2jpfCin Þ1

ð21:41Þ
zi ðf Þ ¼ r i k C i

ro:ops ðf Þ ¼ zo:s ðf Þ þ ð2jpfCout Þ 1

ð21:42Þ
zo:s ðf Þ ¼ ro:s k Co

vo ðf, R0, RLÞ

Gops ðf, R0, RLÞ ¼
v0
¼ Gi ðf, R0ÞGb Go ðf, RLÞ
ð21:43Þ
ri:ops ðf Þ RL
¼ Gb
ri:ops ðf Þ þ R0 ro:ops ðf Þ þ RL
¼ Gi ðf, R0ÞGb ðf, RL)

Ci ¼ Cbe + Cbc ð1  Gb ðRLÞÞ ð21:44Þ

Co ¼ Cce þ Cbc ð21:45Þ

21.3.7 Noise—Frequency Independent Version

Figure 21.5 is Fig. 21.4 transferred into a circuit that shows all noise relevant
sources:
With input and output not loaded, the input noise voltage density en.i becomes:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
e2
en:i ¼ e2n:T1 þ n:RC:tot ð21:46Þ
G2b

TSOS-1 & TSOS-2 explain the calculation of the BJTs input referred noise
voltage density en.T1. The frequency independent calculation of the noise voltage
density of RC looks as follows:
350 21 BJT Circuits in CE Configuration

Fig. 21.5 Fig. 21.4 with all relevant noise sources

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:RC:tot ¼ e2n:RC + e2n:RCex ð21:47Þ

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:RC ¼ 4kTRCB1 ð21:48Þ

With NI = resistor noise index (see resistor data sheet), VRC = DC voltage across
RC, and d = number of decades in B = frequency bandwidth of interest we’ll obtain
the average excess noise voltage density en.RCex:
pffiffiffi
NI dVRC
en:RCex ¼ pffiffiffiffi ð21:49Þ
B

The input referred noise current density becomes:

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
in:i ¼ i2n:T1 þ i2n:RA þ i2n:RB ð21:50Þ
rffiffiffiffiffiffiffiffiffiffiffiffiffiffi
4kT
in:R ¼ B1 ð21:51Þ
R

With all shown noise sources included, the output referred noise voltage density
en.o thus becomes:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u 2 2 2 2
u en:i Gb ðRLÞ þ en:R0:tot ðR0Þ Gop ðR0; RLÞ
u  2
en:o:tot ðR0; RLÞ ¼ t RL ð21:52Þ
þ e2n:RL
ro:s þ RL
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:R0:tot ðR0Þ ¼ en:R0 ðR0Þ2 þ i2n:i R02 ð21:53Þ
21.3 Basic (b) CEb Circuit 351

21.3.8 Noise—Frequency Dependent Version

Based on Fig. 21.5, the above given equations, and the frequency related
Sects. 21.2.4 and 21.2.5, the frequency dependent output referred noise voltage
density en.o.tot(f) becomes:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u 2
u en:i ðf; RLÞ Gb ðRLÞ
u
u þ en:R0:tot ðf; R0ÞGops ðf; R0; RLÞ
en:o:tot ðf; R0; RLÞ ¼ u  2 ð21:54Þ
t RL
þ en:RL
2
zo:s ðf Þ þ RL
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:RC:tot ðf Þ2
en:i ðf, RLÞ ¼ en:T1 ðf Þ2 þ ð21:55Þ
Gb ðRLÞ2

The calculation of en.RC.tot(f) follows the rules given in Sect. 21.2.5. Sec-
tion 21.2.7 gives the equations to calculate SNs.

21.4 CE Circuit CEcf with Current Feedback

See Fig. 21.6.

21.4.1 Gains Gcf and Gcf.rot

vo
Gcf ¼  ð21:56Þ
vi

Fig. 21.6 Common emitter

circuit CEcf with cf via RE
352 21 BJT Circuits in CE Configuration

gm RCðhfe rce  REÞ

Gcf ¼ 
hfe ðrce þ RE þ RC þ gm RErce Þ þ gm REðrce þ RCÞ
RCðhfe rce  REÞ ð21:57Þ
¼
kT h ðr þ RE þ RCÞ þ REðh r þ r þ RCÞ
qIC fe ce fe ce ce

Gcf and application of the reduced mutual conductance gm.red:

Gcf
gm:red ¼ gm
Gb
 
RE ð21:58Þ
gm ðrce þ RCÞ hfe 
rce
¼
hfe ðrce þ RE þ RCÞ þ gm REðhfe rce þ rce þ RCÞ

Hence:

Gcf ¼ gm:red RC ð21:59Þ

With RE ≪ RC ≪ rce we obtain gm.red.rot and Gcf.rot.1:

gm
gm:red:rot ffi ð21:60Þ
1 þ gm RE

Gcf:rot:1  gm:red:rot ðRC k rce Þ ð21:61Þ

Gcf.rot.2 (hfe ≫ 1):

gm RC
Gcf:rot:2   ð21:62Þ
1 þ gm RE

Gcf.rot.3 (like Gcf.rot.2 plus gmRE ≫ 1):

RC
Gcf:rot:3   ð21:63Þ
RE

21.4.2 Input Resistances ri and ri.rot (O/P Open)

ð1 þ hfe Þrce þ RC
ri ¼ rbe þ RE ð21:64Þ
rce þ RE þ RC

ri.rot (rce ≫ RC, RE; hfe ≫ 1):

21.4 CE Circuit CEcf with Current Feedback 353

ri:rot  rbe þ hfe RE ð21:65Þ

21.4.3 Output Resistances ro.o and ro.o.rot (I/P = Open)

ro:o ¼ ðrce þ REÞ k RC ð21:66Þ

ro.o.rot (rce ≫ RC):

ro:o:rot  RC ð21:67Þ

21.4.4 Output Resistances ro.s and ro.s.rot (I/P = Shorted)

2 0 rbe 13
hfe þ
6 B rce C7
ro:s ¼ 4rce @1 þ rbe A5 k RC ð21:68Þ
1þ
RE

ro.s.rot.1 (rce ≫ rbe; hfe ≫ 1):

 
hfe RE þ rbe
ro:s:rot:1  rce k RC ð21:69Þ
RE þ rbe

ro.s.rot.2 (rce ≫ RC):

ro:s:rot  RC ð21:70Þ

21.4.5 Operating Gains Gop(R0, RL) and Gops(f, R0, RL)

As of Fig. 21.7 inclusion of the bias setting input resistors R1 & R2 leads to the
operating input resistance ri.ops and output resistance ro.ops:

ri:ops ¼ R1 k R2 k ri ð21:71Þ

ro:ops ¼ ro:o ð21:72Þ

354 21 BJT Circuits in CE Configuration

Fig. 21.7 CEcf ’s operational model

If Cin and Cout do not hurt the flat frequency and phase response in the band-
width of interest, the corresponding frequency independent equations look as
follows:

vo ðR0, RL)
Gop ðR0, RL) ¼
v0
ð21:73Þ
¼ Gi ðR0ÞGcf Go ðRLÞ
¼ Gi ðR0ÞGcf ðRLÞ

Gcf ðRLÞ ¼ Gcf Go ðRLÞ ð21:74Þ

ri:ops
Gi ðR0Þ ¼ ð21:75Þ
ri:ops þ R0

RL
Go ðRLÞ ¼ ð21:76Þ
ro:ops þ RL

Via inclusion of the Miller-capacitance-effect for Ci and Co, the frequency

dependent version becomes thus:

ri:ops ðf Þ ¼ ðzi ðf Þ k R1 k R2Þ þ ð2jpfCin Þ1

ð21:77Þ
zi ðf Þ ¼ r i k C i

ro:ops ðf Þ ¼ zo:s ðf Þ þ ð2jpfCout Þ1

ð21:78Þ
zo:s ðf Þ ¼ ro:s k Co
21.4 CE Circuit CEcf with Current Feedback 355

vo ðf, R0, RLÞ

Gops ðf, R0, RLÞ ¼
v0
¼ Gi ðf; R0ÞGcf Go ðf; RLÞ
ð21:79Þ
ri:ops ðf Þ RL
¼ Gcf
ri:ops ðf Þ þ R0 ro:ops ðf Þ þ RL
¼ Gi ðf; R0ÞGcf ðf; RL)

Cbe
Ci ¼ þ Cbc ð1  Gcf ðRLÞÞ ð21:80Þ
1 þ gm RE

Cce
Co ¼ þ Cbc ð21:81Þ
1 þ gm RE

21.4.6 Noise and SN

According to Fig. 21.8 the noise voltage and SN calculations follow the rules given
in Sects. 21.3.7 and 21.3.8. Additionally, the following exception has to be taken
into account: the emitter resistance RE creates an additional noise voltage. Thus, in
(21.10) and (21.15) it will increase the noise voltage of T1 as follows:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:T1 ðf Þ ¼ e2n1:T1 þ in:T1 ðf Þ2 R2BE þ e2n:R:BE ð21:82Þ

RBE ¼ rbb0 þ RE ð21:83Þ

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:R:BE ¼ 4kTB1 RBE ð21:84Þ

Fig. 21.8 Noise model of Fig. 21.7 with all relevant noise sources
356 21 BJT Circuits in CE Configuration

If the excess noise of RE becomes significant we have to consider it too. By

addition of RE Fig. 21.5 changes to Fig. 21.8.

21.5 CE Type 2 Circuit CEvcf2 with Voltage Feedback

and Current Feedback

21.5.1 Gain Gvcf2

Systematically, with the reduced mutual conductance gm.red of Sect. 21.4.1 the gain
equation of Fig. 21.9 can be derived from the evolution of the Fig. 21.10 circuits
and their gain equations.
Figure 21.10a shows a CEvf1 gain stage Type1 with a voltage feedback via R2.
In Fig. 21.10b we reduce R1 to 0Ω and add R3; hence, we’ll have a CEvf2 Type 2
now. The next step is the inclusion of RE. It leads to a CEvcf2 Type 2. The CEvcf1
Type 1 is Fig. 21.9 including R1, however, not shown here.
The gain Gvf1 of Fig. 21.10a becomes:

g R2  1
Gvf1 ¼     m   ð21:85Þ
1 g 1 1
1 þ gm R1 1 þ þ R1 þ R2 þ R1R2 m þ
hfe hfe RC rce

As of Fig. 21.10b we set R1 = 0 Ω in the equation above. The gain Gvf2 thus
becomes:

gm R2  1
Gvf2 ¼    ð21:86Þ
1 1
1 þ R2 þ
RC rce

Fig. 21.9 Common emitter

circuit CEvcf2 with vf via R2
and cf via RE
21.5 CE Type 2 Circuit CEvcf2 with Voltage Feedback and Current Feedback 357

Fig. 21.10 Derivation of the gain equation for Fig. 21.9

Here, R3 plays no role in the gain equation. A further inclusion of RE into the
Fig. 21.10b circuit and application of gm.red leads to the gain Gvcf2 of Fig. 21.9 as
follows:

gm:red R2  1
Gvcf2 ¼    ð21:87Þ
1 1
1 þ R2 þ
RC rce

21.5.2 Input Resistance ri (O/P Open)

2 31
61 7
6 1 1 7
ri ¼ 6 þ þ  1 7 ð21:88Þ
4R3 rbe þ hfe RE 1 1 5
R2 þ þ
RC rce

21.5.3 Output Resistance ro.s (I/P Shorted)

ro:s ¼ ðrce þ REÞ k RC k R2 ð21:89Þ

358 21 BJT Circuits in CE Configuration

21.5.4 Output Resistance ro.o (I/P Open)

rbe ðR3 þ R2Þ þ R1R2

ro:o ¼ ðrce þ REÞ k RC k ð21:90Þ
rbe þ R3ð1 þ hfe Þ

21.5.5 Other Equations

Al other equations, already similarly given in the previous sections, can be derived
by application of the following additions:

R3eff ¼ R3 k R0 ð21:91Þ

RCeff ¼ RC k RL ð21:92Þ

21.5.6 Noise and SN

Figure 21.11 shows the Fig. 21.9 noise model.

The major difference to the Fig. 21.8 noise model comes from the treatment of
R2. However, because of the rather difficult math T/S recommends the application
of the equations for en.i and in.i according to TSOS-1, p. 51f and TSOS-2, p. 70.
These equations lead to the input referred noise voltage density and noise current
density equations of Sect. 21.3.7 plus the ones of Sect. 21.4.6. Thus, we obtain for
the here presented gain stage the input referred noise voltage density in the R0, RL,
and frequency dependent format as follows:

Fig. 21.11 Noise model of the Fig. 21.9 CEvcf2 gain stage
21.5 CE Type 2 Circuit CEvcf2 with Voltage Feedback and Current Feedback 359

en:o:tot ðf; R0; RLÞ

en:i:tot ðf; R0; RLÞ ¼ ð21:93Þ
Gvcf2 ðRLÞ
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u 2
u en:i ðf Þ Gvcf2 ðRLÞ
u
u þ en:R0:tot ðf; R0ÞGops ðf; R0; RLÞ
en:o:tot ðf; R0; RLÞ ¼ u  2 ð21:94Þ
t RL
þ en:RL
2
zo:s ðf Þ þ RL
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:RC:tot ðf Þ2
en:i ðf; RLÞ ¼ en:T1 ðf Þ2 þ ð21:95Þ
Gvcf2 ðRLÞ2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:T1 ðf Þ ¼ e2n1:T1 þ in:T1 ðf Þ2 R2BE þ e2n:R:BE ð21:96Þ

RBE ¼ rbb0 þ RE ð21:97Þ

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:R0:tot ðf; R0Þ ¼ en:R0 ðR0Þ2 þin:i ðf Þ2 R02 ð21:98Þ

zo:s ðf Þ ¼ ro:s k Co ð21:99Þ

Cce
Co ¼ þ Cbc ð21:100Þ
1 þ gm RE
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
in:i ðf Þ ¼ in:T1 ðf Þ2 þ i2n:RA þ i2n:RB ð21:101Þ

The calculation of the frequency dependent en.RC.tot(f) follows the rules shown in
Sect. 21.2.5. Section 21.2.7 gives the equations to calculate SNs.

21.6 Correction of a TSOS-1 and TSOS-2 Gain Result

Now, with the shown equations, we can calculate the gain of the example gain stage
of TSOS-1, Fig. 3.26 and TSOS-2 Fig. 5.9: We obtain

Gvcf ¼ 46:97 dB ð21:102Þ

The result is very close to the simulated one of (3.91) and (5.23) in the TSOS
books:

Gs ¼ 47:0 dB ð21:103Þ
360 21 BJT Circuits in CE Configuration

21.7 The CE in Series Configuration with an Op-Amp

21.7.1 Basics

The headline’s closed loop circuit arrangement is shown in the principal circuit of
Fig. 21.12. The working gains of this arrangement look as follows (we ignore the
value of R.DCS here):
vo
Gamp ¼
vi
ð21:104Þ
Zf1
¼1þ
RE

Gops ¼ Gi Gamp ð21:105Þ

Rin
Gi ¼ ð21:106Þ
R0 þ Rin

Because of the very low output resistance, RL does not influence the gains. In
the RIAA amp case Zf1 could be one or all RIAA time constants producing
network(s). In most cases, Zf2 should be chosen as high valued resistor
Rf2 ≥ 100 kΩ with a small capacitor Cf2 (10pF–47pF) parallel to it. The values
heavily depend on the stability and the bandwidth of the whole arrangement. To
find the right values for each circuit we have to go through a trial and error process
or via pSpice. All my designs work well with 2.2 MΩ ∥ 12pF–22pF and a 5534A as
OP1. DCS in Fig. 21.12 means DC servo voltage, or a fixed DC voltage that sets

Fig. 21.12 Principal circuit of a closed loop arrangement with a BJT followed by an op-amp
21.7 The CE in Series Configuration with an Op-Amp 361

the collector DC voltage of T1. If we chose a fixed voltage at the (+) input of OP1
the DCS input works via RDCS and a corresponding servo circuit.
With the well know general gain equation for op-amps in mind the exactness of
the working gain equations depend on the idle gain G0 of the T1 + OP1 pair.
Therefore, with a gain of 1000 and an error of less than 1 % it should become
G0 > 100,000.5 This gain G0 is composed by the mutual conductance of T1 and Zf2
as follows:

G0 ðf Þ ¼ j  gm1 Zf2ðf Þj ð21:107Þ

Simplified and not frequency dependent version:

G0 ¼ j  gm1 Rf2j ð21:108Þ

With ß = RE/(RE + Zf1(f)) the working frequency dependent and closed loop
gain G(f) thus becomes:

G 0 ðf Þ
Gamp ðf Þ ¼ ð21:109Þ
1 þ G0 ðf Þðf Þ

Hence, for a gain G0 ≥ 105 and Rf2 = 2.2 MΩ we need gm ≥ 45.5 mS. The
Module 2 phono-amp with 4 paralleled BJTs and a collector current IC = 6.7 mA
creates gm = 259 mS, Amp2’s gm becomes 170.2 mS with IC = 4.4 mA.
Only the input stage and the input load should generate the counting noise of the
phono-amp. Hence, at least the input stage’s gain should have a value that its input
referred noise voltage times the gain makes the noise of the following gain stages
ignorable. This contribution allowed6 discussion has led to my recommendation of
an input stage gain ≥ 37.5 dB.
Nevertheless, the next section explains the calculation process of the input
referred noise voltage in full detail, no matter if the op-amp stage does add noise or
not.

21.7.2 Noise and SN

We assume a CEvcf2 input à la Fig. 21.9, no noise from the DCS variants and
resistive feedback paths of the amp and OP1, hence, Rf1 and Rf2 instead of Zf1 and
Zf2. In case of an RIAA network, we take the value of the magnitude of the Zf1
impedance at 1 kHz. According to Rf2, there should be no difference between Zf2’s
magnitude at 1 kHz and the Rf2 resistor value.

5
‘Intuitive IC OP Amps’, 1984, Thomas M. Frederiksen, National’s Semiconductor Technology
Series.
6
Details see TSOS-1, Chapter 3.2, TSOS-2, Chapter 5.4.
362 21 BJT Circuits in CE Configuration

Fig. 21.13 Frequency independent noise model of the modified Fig. 21.12 amp

Figure 21.13 shows the frequency independent noise model of the modified
Fig. 21.12 amp. Nevertheless, the frequency dependency will be part of the
equation paragraphs.
We obtain the general equations for the input referred noise voltage density en.i(f)
as follows:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u
u en:i1 ðf Þ2 þ e2 2 2
u n:R:BE þ in:i1 ðf Þ RBE
en:i ðf Þ ¼ u 2 2
t þ en:i:op1 ðf Þ þ in:RC ðf Þ þ in:i:op1 ðf Þ þ in:Rf2
2 2
ð21:110Þ
2 2
jGvcf2:1 j gm1

in:C1 ðf Þ
en:i1 ðf Þ ¼ ð21:111Þ
gm1

in:C1 ðf Þ
in:i1 ðf Þ ¼ ð21:112Þ
hfe1
 1
1 1
RBE ¼ rbb10 þ þ ð21:113Þ
RE Rf1

Thus, the output referred noise voltage density looks like:

en:o ðf Þ ¼ en:i ðf ÞGamp ð21:114Þ

21.7 The CE in Series Configuration with an Op-Amp 363

With an input load R0 in.i(f) comes into the game. We obtain the amp’s input
noise voltage density as follows:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
in:i ðf Þ ¼ in:i1 ðf Þ2 þ i2n:R2 þ i2n:R3 ð21:115Þ

Hence, the input loaded amp increases the output referred noise voltage density
by the noise of R0 the following way:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:o:amp ðf; R0Þ ¼ Gamp en:i ðf Þ2 þ in:i ðf Þ2 R02 þ en:R0 ðR0Þ2 ð21:116Þ

Provided that one of the two DCS variants is chosen and their output noise
voltage might influence the amp’s noise production in a countable way we can add
this noise voltage density to the one of OP1 in (21.110) (=variant 1) or we take
variant 2 and change the above shown equations as follows (dcs2):
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:i ðf Þ2 þ in:i ðf Þ2 R02 þ en:R0 ðR0Þ2
en:o:amp:dcs2 ðf; R0Þ ¼ Gamp ð21:117Þ
þ en:o:dcs2 ðf Þ2 ðRf1RDCS Þ2

Here, en.o.dsc2(f) must include the noise of RDCS.

en:o:amp:dcs2
en:i:amp:dcs2 ðf Þ ¼ ð21:118Þ
Gamp

Normally, DCSs have an lp transfer character with very low corner frequencies.
That’s why we can assume ignorable tiny influences.
The SN calculations follow the equations of Sect. 21.2.7.
Differential (DIF) Amps
22

22.1 Intro

I use the terminus technicus DIF-amp or DIFA for amplifiers with a differential
input and a differential output. Their differential gain Gdif follows the rule:
vo1  vo2
Gdif ¼ ð22:1Þ
vi1  vi2

Another expression for Gdif could be:

vdif:o
Gdif ¼ ð22:2Þ
vdif:i

Here, according to Fig. 22.1 the signal voltages on the right side of (22.1) are the
ones between the input or output leads and ground. In addition, the ones with
subscript 2 have a 180° phase shift to the ones with subscript 1.
The voltages of (22.2) are the ones between the input leads and between the
output leads. With equal values for the input voltages vi1 and vi2 but opposite phase
we obtain:

vi2 ¼ vi1
ð22:3Þ
vo2 ¼ vo1

Hence,

2vo1 vo1 vo2

Gdif ¼ ¼ ¼ ð22:4Þ
2vi1 vi1 vi2

In the following sections I’ll concentrate on two different DIFA types only: the
DIFA-OPA and the DIFA-IC. These types serve phono-amp needs best.

© Springer International Publishing Switzerland 2016 365

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_22
366 22 Differential (DIF) Amps

Fig. 22.1 Situation of the signal voltages in a DIFA

Fig. 22.2 General DIF-OPA

circuit with input load

22.2 The DIFA-OPA1

22.2.1 Basics

Shown in Fig. 22.2 the DIFA-OPA consists of two op-amps and a handful of
passive components. There is no phase shift between input and output signal
voltages and the general appearance is alike the one of the input stage of an
instrumentation amplifier type 2.2
With equal valued resistors Rf and equal valued resistors R1 in Fig. 22.2 the gain
equations for the differential gain Gdif and the operational gain Gdif.ops look as
follows:

1
eg Amp1 of Chaps. 8 and 9.
2
See TSOS-1 Sect. 3.6 or TSOS-2 Chap. 9.
22.2 The DIFA-OPA 367

vo:dif
Gdif ¼
vi:dif
ð22:5Þ
2Rf
¼1þ
RG
vo:dif
Gdif:ops ¼
v0 ð22:6Þ
¼ Gdif Gi

R1a þ R1b
Gi ¼ v0 ð22:7Þ
R0 þ R1a þ R1b

The CMRR thus becomes:

Gdif
CMRR ¼ ð22:8Þ
Gcm

Gcm ¼ 1 ð22:9Þ

22.2.2 Noise Calculations Version 1

Without R1a and R1b Fig. 22.3 gives all relevant DIF-OPA noise sources. The
noise effects of the two input resistors R1a + R1b = R1 can be combined with the
noise of R0 as follows:

R0eff ¼ R0jjR1 ð22:10Þ

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
s  ffi
1 1 1
en:R0:eff ¼ þ ð22:11Þ
e2n:R0 e2n:R1

There are two versions to get the Fig. 22.2 input referred noise voltage density
en.i.tot.
Version 1: According to Fig. 22.3 we obtain the frequency independent output
noise voltage density en.o.tot1 as follows:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u  1 !
u 2 1 1
u Gdif e2n:i1 þ e2n:i2 þ e2n:R0:eff þ R02eff þ
u i2n:i1 i2n:i2
en:o:tot1 ¼u
u   ð22:12Þ
t 2Rf 2  
þ e2n:Rf þ 2e2n:Rf þ i2n:i1 þ i2n:i2 Rf 2
RG
368 22 Differential (DIF) Amps

Fig. 22.3 Relevant noise sources of Fig. 22.2

Thus, en.i.tot1 becomes:

en:o:tot1
en:i:tot1 ¼ ð22:13Þ
Gdif

22.2.3 Noise Calculations Version 2

Version 2 requires an adapted Fig. 22.3, shown in Fig. 22.4.

Without detour via the output, with eni1 = en.i2 and in.i1 = in.i2 the input referred
noise voltage density can thus be calculated directly:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi
en:i:tot2 ¼ e2n:i þ e2n:R0:eff þ e2n:RG:f þ i2n:i R02eff þ RG2f ð22:14Þ
22.2 The DIFA-OPA 369

Fig. 22.4 Adapted Fig. 22.3

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:i ¼ e2n:i1 þ e2n:i2
pffiffiffi ð22:15Þ
¼ 2 in:i1
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
1 1 1
in:i ¼ þ
i2n:i1 i2n:i2 ð22:16Þ
in:i1
¼ pffiffiffi
2

RGf ¼ RGjjð2Rf Þ ð22:17Þ

A short math exercise yields:

en:i:tot1 ¼ en:i:tot2 ð22:18Þ

370 22 Differential (DIF) Amps

Fig. 22.5 Typical amplifier

circuit with a DIFA-IC

22.2.4 Noise Calculations Version with One Input Lead

Grounded

If we ground one input lead the noise current through R0eff changes to in.i1 and
(22.14) changes to:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:i:tot2:grd ¼ e2n:i þ e2n:R0:eff þ e2n:RG:f þ i2n:i1 R02eff þ i2n:i RG2f ð22:19Þ

Hence, we’ll always get:

en:i:tot2 \en:i:tot2:grd ð22:20Þ

Because of the always existing 1/f-noise of op-amps next chapter’s MCD-WS

23.1 gives frequency dependent example calculations for all versions.

22.3 The DIFA-IC3

22.3.1 Basics

A completely different DIF approach works with an intelligent arrangement of two

shunt configured op-amps in such a way that the differential output signal voltage
has a 180° phase shift to the differential input signal voltage (see Fig. 22.5).
With equal valued resistors R1a = R1b and R2a = R2b the gain Gdif of this
arrangement becomes:

R2a
Gdif ¼  ð22:21Þ
R1a

3
eg in Sect. 13.5.1 Joachim Gerhard's solution.
22.3 The DIFA-IC 371

Fig. 22.6 DIFA-IC with input load

And we obtain its noise gain GN.dif as follows:

GN:dif ¼ jGdif j þ 1 ð22:22Þ

This type of amplifier IC topology creates one specific problem: any change of
its input resistors R1a and R1b will lead to other than the calculated gain results.
A typical change will occur with output resistances of preceding gain stages >0 Ω.
The ideal situation is given in Fig. 22.6 and R0a and R0b both equal but >0 Ω will
change (22.21) as follows:

R2a
Gdif2 ¼  ð22:23Þ
R1a þ R0a

In any case, |Gdif2| < |Gdif |!

Concerning CMRR the data sheet specs are the only source.

22.3.2 Noise Calculations Version 1

Basically, in Version1 the noise calculations follow the recommendations of Texas

Instruments.4 However, with a lot of math rearrangements5 the equation to calculate
the output referred noise voltage density en.o.dif of Fig. 22.7 could be simplified. We
obtain thus en.o.dif in the frequency dependent format:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u
u en:i ðf Þ2 ð1 þ jGdif jÞ2
en:o:dif1 ðfÞ ¼ t h i ð22:24Þ
þ2 in:i ðf Þ2 R2a2 þ e2n:R1a jGdif j2 þ e2n:R2a

“Fully-Differential Amplifiers”, TI Application Report SLOA054D.

4

5
See Chap. 23 and the derivation in MCD-WS 23.3.
372 22 Differential (DIF) Amps

Fig. 22.7 Noise sources of a DIFA-IC

The input referred noise voltage density becomes:

en:o:dif1 ðf Þ
en:i:dif1 ðf Þ ¼ ð22:25Þ
GN:dif

en.i, in.i, and their 1/f-noise corner frequencies must be picked from the data sheets.
By application of the specific noise voltage summing method any noise voltage
en.prec(f) from a preceding gain stage must be multiplied with |Gdif | and be added to
en.o.dif1(f) to get the new output noise voltage density en.o.tot.dif1(f), hence,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:o:tot:dif1 ðf Þ ¼ jGdif j2 en:prec ðf Þ2 þ en:o:dif1 ðf Þ2 ð22:26Þ

In the above given equations there is no term concerning the noise voltage
en.cm(f) of the common mode input. It can be ignored as long as it is grounded.
However, any DC servo that works via the vocm input adds the noise voltage of the
servo too, hence, it must be multiplied with GN.dif and be added by an additional
term squared in (22.24).

22.3.3 Noise Calculations Version 2

A look back to TSOS-2’s MCD-WS 12.16 and Sect. 8.7 (see en.i.shu) allows
calculating the input referred noise voltage density of such a double shunt con-
figured op-amp arrangement by an equivalent approach. Thus, en.i.dif2(f) becomes:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
h iffi
en:i:dif 2 ðf Þ ¼ en:i ðf Þ2 þ 2 ðin:i ðf ÞRP1Þ2 þ e2n:RP1 ð22:27Þ
22.3 The DIFA-IC 373

RP1 ¼ R1ajjR2a ð22:28Þ

We obtain the output referred noise voltage density en.o.dif2(f) as follows:

en:o:dif2 ðf Þ ¼ GN:dif en:i:dif1 ðf Þ ð22:29Þ

Thus,6

en:o:dif1 ðf Þ ¼ en:o:dif2 ðf Þ
ð22:30Þ
en:i:dif1 ðf Þ ¼ en:i:dif2 ðf Þ

and
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en:o:tot:dif2 ðf Þ ¼ jGdif j2 en:prec ðf Þ2 þ en:o:dif2 ðf Þ2 ð22:31Þ

6
See MCD-WS 23.2.
Mathcad Worksheets of DIF Amps
23

Contents

23.1 MCD-WS: DIFA-OPA

23.2 MCD-WS: DIFA-IC

pffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffi
Note 1: MCD 11 has no built-in unit “rtHz” or “ Hz”. To get 1 Hz based
voltage noise and current noise densities the rms noise voltage and
current in a specific frequency range B > 1 Hz must be multiplied by
pffiffiffiffiffiffiffiffiffi
1 Hz and divided by the root of that specific frequency range √B!
Note 2: MCD 11 offers no “dB” unit. This is available from MCD 13 on!

© Springer International Publishing Switzerland 2016 375

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_23
376 23 Mathcad Worksheets of DIF Amps

23.1 MCD-WS: DIFA-OPA Page 1

DIF-Amp with Op-Amps (DIFA-OPA) and input loaded with R0

1. Gain and Noise Calculations :

Fig. 23.1 = Fig. 22.2

23 1
k 1.38065 10 VAsK T 300 K B1 1Hz B20k 19980 Hz h 1000Hz

3 3
f 20Hz 25Hz 20 10 Hz vi.nom 5 10 V

OP1 OP27 OP2 OP1

9 fc.e1
en.i1 3.0 10 V en.i2 en.i1 fc.e1 2.7Hz en.i1( f) en.i1 1
f

en.i2( f) en.i1( f)

12 fc.i1
in.i1 0.4 10 A in.i2 in.i1 fc.i1 140Hz in.i1( f) in.i1 1
f

in.i2( f) in.i1( f)

3 3
R0 12 10 Rf 10 10 RG 100

9
en.RG 4 k T B1 RG en.RG 1.287 10 V

9
en.Rf 4 k T B1 Rf en.Rf 12.872 10 V

9
en.R0 4 k T B1 R0 en.R0 14.1 10 V

Rf
Gdif 1 2 Gdif 201
RG
23 Mathcad Worksheets of DIF Amps 377

23.1 MCD-WS: DIFA-OPA Page 2

1.1 Output referred approach (= Version 1) :

Fig. 23.2 = 22.3

1
2 2 2 2 1 1 2
en.o.tot1( f) Gdif en.i1( f) en.i2( f) en.R0 R0
2 2
in.i1( f) in.i2( f)
2
2 2 Rf 2 2 2
en.RG 2 en.Rf 2 in.i1( f) Rf
RG
6
en.o.tot1( h) 3.059 10 V

en.o.tot1( f) 9
en.i.tot1( f) en.i.tot1( h) 15.22 10 V
Gdif

1.2 Input referred approach (= Version 2) :

Fig. 23.3 = Fig. 22.4

378 23 Mathcad Worksheets of DIF Amps

23.1 MCD-WS: DIFA-OPA Page 3

in.i1( f)
en.i( f) en.i1( f) 2 in.i( f)
2

1
1 1
RGf
RG 2 Rf RGf 99.502

9
en.RG.f 4 k T B1 RGf en.RG.f 1.284 10 V

2 2 2 2 2 2 9
en.i.tot2( f) en.i( f) in.i( f) RGf R0 en.RG.f en.R0 en.i.tot2( h) 15.22 10 V

6
en.o.tot2( f) Gdif en.i.tot2( f) en.o.tot2( h) 3.059 10 V

Hence, en.o.tot2( f) en.o.tot1( f) en.i.tot2( f) en.i.tot1( f)

1.3 One input lead grounded :

2 2 2 2 2 2 2 9
en.i.tot2.grd( f) en.i( f) in.i1( f) R0 in.i( f) RGf en.RG.f en.R0 en.i.tot2.grd( h) 15.645 10 V

6
en.o.tot2.grd( f) Gdif en.i.tot2.grd( f) en.o.tot2.grd( h) 3.145 10 V

7
1 10

en.i.tot2( f)
[V/rtHz]

en.i.tot2.grd( f)

8
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

Fig. 23.4 Input referred and input loaded noise voltage densities of the fully
differential input and the grounded input

Hence, en.o.tot2.grd( f) en.o.tot2( f)

23 Mathcad Worksheets of DIF Amps 379

23.1 MCD-WS: DIFA-OPA Page 4

2. SN calculations :

20000Hz
1 2
en.i.tot2( f) df
B1
20Hz
SNne2 20 log SNne2 67.346 [dB]
vi.nom

20000Hz
1 2
en.i.tot2.grd( f) df
B1
20Hz
SNne2.grd 20 log SNne2.grd 67.124 [dB]
vi.nom

Hence, SN ne2 SNne2.grd

380 23 Mathcad Worksheets of DIF Amps

23.2 MCD-WS: DIFA-IC Page 1

DIF-Amps with ICs (DIFA-IC)

1. Gain and Noise Calculations : Example DIFA-IC = OPA1632

Fig. 23.5 = 22.5

23 1 3
k 1.38065 10 VAsK T 300 K B1 1Hz B20k 19980 Hz h 10 Hz
3
f 20Hz 25Hz 20 10 Hz vo.nom 1V

9 3 fc.e 9
en.i 1.3 10 V fc.e 0.6 10 Hz en.i( f) en.i 1 en.i( h) 1.644 10 V
f

12 3 fc.i 15
in.i 0.4 10 A fc.i 2 10 Hz in.i( f) in.i 1 in.i( h) 692.82 10 A
f
3 3
R1a 1 10 R2a 200 10 R1b R1a R2b R2a

R2a
Gdif Gdif 200
R1a

GN.dif Gdif 1 GN.dif 201

Fig. 23.6 = Fig. 22.6

23 Mathcad Worksheets of DIF Amps 381

23.2 MCD-WS: DIFA-IC Page 2

1.1 Output referred approach (= Version 1) :

9
en.R1a 4 k T R1a B1 en.R1b en.R1a en.R1a 4.07 10 V
9
en.R2a 4 k T R2a B1 en.R2b en.R2a en.R2a 57.564 10 V

2 2 2 2 2 2 2
en.o.dif.o( f) en.i( f) GN.dif 2 in.i( f) R2a en.R1a Gdif en.R2a ( 1)

6
en.o.dif.o( h) 1.216 10 V

en.o.dif.o( f) 9
en.i.dif.o( f) en.i.dif.o( h) 6.052 10 V
GN.dif

7
1 10
[V/rtHz]

en.i.dif.o( f)1 10 8

9
1 10
3 4 5
10 100 1 10 1 10 1 10
f
[Hz]

Fig. 23.7 Input referred differential noise voltage density, input shorted
or output resistance of preceding gain stage ~0R

1.2 Input referred approach (= Version 2) :

R1a R2a
RP1 RP2 RP1 RP1 995.025
R1a R2a
9
en.RP1 4 k T B1 RP1 en.RP2 en.RP1 en.RP1 4.06 10 V

2 2 2 2 9
en.i.dif.i( f) en.i( f) 2 in.i( f) RP1 en.RP1 en.i.dif.i( h) 6.052 10 V

6
en.o.dif.i( f) en.i.dif.i( f) GN.dif en.o.dif.i( h) 1.216 10 V

=> en.o.dif.o( f) en.o.dif.i( f) and en.i.dif.o( f) en.i.dif.i( f)

2. SN calculations :

20000Hz
1 2
en.o.dif.o( f) df
B1
20Hz
SN ne 20 log SNne 75.422 [dBV]
vo.nom
382 23 Mathcad Worksheets of DIF Amps

23.2 MCD-WS: DIFA-IC Page 3

3. Derivation of (1) via TI-SLOA054D :

vo.dif
Gdif vi.dif vi1 vi2 vo.dif vo1 vo2 vo.cm 0
vi.dif

R1a 3 R1b 3
ß1 ß1 4.975 10 ß2 ß2 4.975 10
R1a R2a R1b R2b

ß1 ß2 => ß ß1

1 ß 1 1 ß
Gdif a(f)>>1 => Gdif Gdif 200
ß 1 ß
1
a ( f) ß

R1a R2a R1a R2b

RA RA 995.025 RB RB 995.025
R1a R2a R1a R2b

=> RA RB

9
en.R1a 4 k T R1a B1 en.R1b en.R1a en.R1a 4.07 10 V

9
en.R2a 4 k T R2a B1 en.R2b en.R2a en.R2a 57.564 10 V

9 12
en.i 1.3 10 V in.i 0.4 10 A ecm 0V

in case of ß1 ß2 any noise voltage e cm > 0V a t t he c ommon mode input would play a role, however, with 1%
resistances the factor (ß1-ß2) makes it still ignorable and the respective term in the equation below
becomes ignorable too. But, the noise voltage of DC servos that work via the vocm input must be
added by an additional term.

2 2 2 2 2 2
2 en.i 2 in.i RA 2 in.i RB 2 en.R1a ( 1 ß) 2 en.R1a ( 1 ß) 2 ecm ( ß1 ß2)
en.o.dif
2
( 2 ß)
2 2
en.R2a en.R2b

2 2 2 2 2
2 en.i 2 in.i RA 2 in.i RB 2 en.R1a ( 1 ß) 2 en.R1a ( 1 ß) 2 2 ( 2)
en.o.dif1 en.R2a en.R2b
2
( 2 ß)

6
en.o.dif1 1.189 10 V

1 en.o.dif1 9
GN.dif1 GN.dif1 201 en.i.dif1 en.i.dif1 5.944 10 V
ß Gdif
23 Mathcad Worksheets of DIF Amps 383

23.2 MCD-WS: DIFA-IC Page 4

3.1 First symplification of (2) :

2 2 2
en.i 2 in.i RA 2 en.R1a ( 1 ß) 2 6
en.o.dif2 2 en.R2a en.o.dif2 1.189 10 V ( 3)
2
ß

B20k
en.o.dif2
B1
SN ne2 20 log SNne2 75.492 [dBV]
vo.nom

3.2 Frequency dependent version of (3) :

3 fc.e 9
fc.e 0.6 10 Hz en.i( f) en.i 1 en.i( h) 1.644 10 V
f

3 fc.i 15
fc.i 2 10 Hz in.i( f) in.i 1 in.i( h) 692.82 10 A
f

2 2 2
en.i( f) 2 in.i( f) RA 2 en.R1a ( 1 ß) 2 6 ( 4)
en.o.dif3( f) 2 en.R2a en.o.dif3( h) 1.216 10 V
2
ß

en.o.dif3( f) 9
en.i.dif3( f) en.i.dif3( h) 6.082 10 V
Gdif

Second simplification of (2) :

Second simplification and usage of the term Gdif = R2a/R1a leads to equation (1) at the beginning of this worksheet
Old Stuff?
24

24.1 Intro

I guess this chapter stands for the end of my journey through the jungle of
low-noise phono-amps and associated issues. Rather often, I’ve been asked about
my personal ranking of things that ensure optimal sound reproduction of vinyl
records. The today’s answer is not easy; however, I would like to rank the following
two tools among the top positions:
• I know people still running rather old turntables (made in the sixties of last
century), old tape recorders (eg Revox A77 or one of the BRAUN TG series)
and old FM tuners from manufacturers around the world. Moreover, they are
happy, they restore, and they collect these things. I also own an old
BRAUN PCS 52 E turntable with SME II tonearm and a Shure V15 IV car-
tridge. Sometimes, just to check its correct working, I listen to it and I’m always
surprised about its still excellent sound reproduction. Of course, a change to one
of the other cartridges I own needs the application of a tracking force mea-
surement instrument. Section 24.2 shows the one I use.1
• The quality of the sources that create the sound we’re listening to heavily
depends on the quality of the making process. Hence, we’re talking about the
groove-making or cutting technology which is—after the mix is hopefully well
done—the most sensitive part of the whole LP making process.
From time to time, a cutting lathe from eg Neumann needs a quality check and
re-trimming of its parameters, especially those of the cutting head. This is the
moment where special test and calibration records (TCRs) enter the scene. In
TSOS-1 and TSOS-2, I gave some remarks on the issue and I gave a sad look into
the future of the availability of modern test records for cutting lathe calibration

1
Pictures: Courtesy Jo Klatt, Design+Design, Hamburg, Germany.

© Springer International Publishing Switzerland 2016 385

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8_24
386 24 Old Stuff?

Fig. 24.1 BRAUN tracking

force measurement instrument

purposes. Later on, I became very surprised about the fact that there are collectors
of test records, especially of those excellent ones produced by the former DDR
(German Democratic Republic—GDR). For collector and potential duplication
purposes, in Sect. 24.3 I add tables of all known test records of the ex-GDR, DIN
and Deutsche Grammophon Gesellschaft.
24.2 The BRAUN Tracking Force Measurement Instrument 387

Fig. 24.2 The BTFMI in

action

24.2 The BRAUN Tracking Force Measurement Instrument

The BRAUN Tracking Force Measurement Instrument (BTFMI) is a rather rare

thing to get and—very surprising—we cannot find it in the Braun+Design Tax
booklet.2 Otherwise, this booklet covers everything BRAUN ever produced, incl.
actual collector prices.
The BTFMI is a typical 1962 Prof. Rams design and it is impressive because of
his simple but very effective usage. In addition, compared with an electronic
instrument, the pond scale shows very exact values of the cartridge’s tracking force.
This instrument is a must-have for every design-oriented LP enthusiast and—
because of its eye-catching appearance, exactness, and clever simplicity—it
exceeds all modern battery powered electronic. Of course, that’s why—despite its
yesterday look—it should not be regarded as old stuff (Figs. 24.1 and 24.2).

24.3 Professional Test and Calibration Records

By accident, some years ago (2013) I stumbled over a stereoplay advertisement

about the search for test records. I thought this could be a good chance to learn
more about test records in general—and I asked the owner of the advertisement for
information about his goals and his collection. I knew very well that the western

2
Braun+Design Tax, Edition 2013, ISBN 978-3-9811106-6-1.
388 24 Old Stuff?

Table 24.1 TCRs of the Deutsche Grammophon Gesellschafta

A: 17 cm TCRsb
• 10 01 941 Turntable check and adjustments
• 10 01 942 Wow and Flutter tests with 5 kHz
• 10 01 944 Wow and Flutter tests with 3 kHz
• 11 01 495 Tests of stereo replay equipment
• 11 01 496 Disque de contrôle pour appareils stéréophoniques
• 11 01 497 Tests of setting up stereo equipment
B: 33 cm TCRsc
• 10 99 008 Wow and Flutter tests with 5 kHz
• 10 99 010 Wow and Flutter tests with 3 kHz
• 10 99 011 IMD tests with 400 Hz/4 kHz, 4:1 level ratio
• 10 99 014 Demonstration of the audibility of linear distortions
• 10 99 103 1 kHz test tone 0 dB/−20 dB—stereo and mono—33 1/3 and 45 rpm
• 10 99 106 Frequency sweep à la DIN 45 547 (16 kHz–20 Hz)
• 10 99 108 Difference tone measurements à la CCIR—sweep between 1 and 20 kHz with
400 Hz difference
• 10 99 109 Loudspeaker measurements with wobbled frequencies in 1/3 octave and Bark
bands
• 10 99 111 General frequency response in 30 Hz–20 kHz and reference tone measurements
• 10 99 112 Reference, trackability and frequency
• 16 41 001 Test signals to check consumer stereo equipment
a
Old version’s numbers: without the first two digits
b
German versions only, except 11 01 496 and 11 01 497
c
Also in English: 10 99 112

world produced a huge range of different TCRs, eg Deutsche Grammophon

Gesellschaft in Table 24.1 and the ones based on DIN in Table 24.2. Quite often, in
discussions about it I was confronted with the opinion that theTCRs of the ex-GDR
must have been the best ones, if not so, at least they showed equal quality compared
with the best ones of the western world (incl. Japan).
A German collector has sent me Table 24.3 listing of his still not complete
collection. For further completion, he is searching for a handful of missing TCRs. It
is thrilling to read his additional—yet not published—extensive explanation of
real-life test record applications.
Old stuff? Of course not! Regular testing and calibration of the cutting lathe
parameters requires new, unused, or seldom used TCRs. They are essential for
keeping the LP production quality as high as possible. As long as there are
increasing LP market shares the production of high quality test and calibration
records for private usage becomes essential too.
24.3 Professional Test and Calibration Records 389

Table 24.2 DIN TCRsa

• DIN 45 541 Frequency response—stereo and mono 33 1/3 rpm
• DIN 45 542 Distortion measurements 33 1/3 and 45 rpm
• DIN 45 543 Frequency sweep 20 Hz–20 kHz and crosstalk tests
• DIN 45 544 Rumble measurement à la DIN 45 539—stereo and mono 33 1/3 rpm
• DIN 45 545 Wow and Flutter—33 1/3 and 45 rpm
• DIN 45 549 Trackability
a
German versions only

Table 24.3 TCRs of the VEB-Deutsche Schallplattena

• LB 13 TCRb for wow and flutter tests with 3150 Hz for 16 2/3, 33 1/3, 45, 78 rpm
• LB 21 TCR—single record to adjust go and stop of an automatic tone arm drive
• LB 22 TCR for controlling the starting point on 25 cm records
• LB 23 TCR for controlling the starting point on 30 cm records
• LB 24 =LB 21
• LB 27 TCR for skating compensation—without grooves
• LB 41 TCR with sweep frequency 20 Hz–20 kHz
• LB 42 TCR with sweep frequency 31.5 Hz–16 kHz
• LB 48 TCR with different music examples and test tones
• LB 49 TCR for production purposes of all kinds of turntables, music and single test tones
• LB 107 TCR with music and single test tones for early stereophonic tests
• LB 108 TCR with different music examples and test tones
• LB 138 TCR for trackability tests, 315 Hz, 94 µm lateral
• LB 202 TCR for distortion tests with 315 Hz and 3150 Hz difference tone recordings
• LB 203 TCR for production of cartridge purposes: crosstalk measurement, cartridge
transfer characteristic, frequency response
• LB 207 TCR for crosstalk and level measurements with a 1 kHz single tone
• LB 208 TCR for the error of the vertical tracking angle
• LB 209 TCR to check the correct adjustment of the headshell/cartridge installation
• LB 210 TCR for frequency response and crosstalk tests in the range of 20 Hz–20 kHz
• LB 211 TCR for frequency response and crosstalk tests in the range of 20 Hz–20 kHz
• LB 212 TCR for frequency response tests in the range of 20 Hz–20 kHz
• LB 237 TCR for trackability tests with a 10 kHz burst signal
• LB 238 TCR for trackability tests with 315 Hz and 100 µm lateral
• LB 239 TCR for rumble tests
a
Listing: courtesy Ulrich Neef, Plauen, Germany
b
TCR Test and calibration record
390 24 Old Stuff?

According to the vinyl-cutting specialist Mr Daniel Krieger from SST Brügge-

mann,3 Frankfurt, Germany, the quality of cutting lathe TCRs is still a major issue
and their availability yet not satisfyingly solved. The problem: There is no
light-pattern measurement instrument4 available allowing to test fresh master cut
signals according to their velocity specs, eg DIN 0 dB at 1 kHz with 8 cm/s/1 kHz.
Obviously, all instruments disappeared after the CD conquered the music industry.
All other frequencies on a TCR are referenced to this signal.
Today, it’s not a problem to cut test signals and frequency sweeps on test
records. There are many different ones available. However, without an exact ref-
erence signal they can be used for customer purposes only. It’s nothing for cutting
lathe calibrations.

24.4 Final Note

When talking about good sounding audio equipment for reproduction purposes I
think the design of audio equipment devices and their integration into an existing
living environment plays a major satisfying and personal quality-defining role too.
But, this is a question of subjectivity and—concerning expenses—courage for the
gap, and, in my mind, not a question of the objectively measured quality of
products. In this respect, our personal Razor should guide us and not third parties!

3
www.sst-ffm.de.
Details see JAES Vol. 5 1957, ‘The calibration of disc recordings with light-pattern
4

measurement’, P.E. Axon & W.K.E. Geddes.

Appendix 1
List of Mathcad Worksheets

3.1 MCD-WS The Triode Gain Stage Amp3 with RIAA Networks
5.1 MCD-WS The Solid-State Gain Stage Amp4 with RIAA Networks
7.1 MCD-WS The Op-Amp + Transformer Driven Output Stage Amp5
9.1 MCD-WS The Transformer + OP-Amp Driven Amp1 (Real Data)
9.2 MCD-WS The Transformer + OP-Amp Driven Amp1 (Data Sheet Data)
11.1 MCD-WS Evaluation of the 1/f-Noise Corner Frequency of a 2SA1316 BJT
11.2 MCD-WS Evaluation of the 1/f-Noise Corner Frequency of a 2SC3329 BJT
11.3 MCD-WS Amp2 SN and Gain Calculations—1/f-Noise Based Version
14.1 MCD-WS BJT/Op-Amps Driven MC Input Stage with Un-Balanced Input
and Balanced Output
14.2 MCD-WS BJT/Op-Amps Driven MC Input Stage with Balanced Trans-
former Input and Balanced Output
14.3 MCD-WS Fully Triode Driven MC/MM Input Stage with Transformer
MC-Input and Balanced Output
18.1 MCD-WS The UBC (Un-Balanced-to-Balanced Converter)
18.2 MCD-WS The PMMA (Fully-Differential Measurement Amp)
18.3 MCD-WS The PFMA (Galvanically Isolated Measurement Amp)
20.1 MCD-WS The Un-Balanced Version (MM Noise Reduction)
20.2 MCD-WS The Balanced Version (MM Noise Reduction)
23.1 MCD-WS DIFA-OPA (Differential Amps)
23.2 MCD-WS DIFA-IC (Differential Amps

© Springer International Publishing Switzerland 2016 391

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8
Appendix 2
Useful Literature and Web Sites

Books
“Electronic Circuits, Handbook for Design and Application”
U. Tietze, C. Schenk, 2nd Edition, Springer 2008
ISBN 978-3-540-00429-5
Abbreviation in this book: T/S
It is the translated version of the 12th German edition of “Halbleiter-
Schaltungstechnik”, 2002, ISBN 3-540-42849-6 (13th edition in 2010)
The accompanying CD-ROM also covers data sheets and simulation softwares like
MicroSim V8.0

“Low-Noise Electronic System Design”

C. D. Motchenbacher, J. A. Connelly, John Wiley & Sons 1993
ISBN 0-471-57742-1
Abbreviation in this book: M/C

“Self on Audio”
Douglas Self, Newnes 2000
ISBN 0 7506 4765 5
Abbreviation in this book: D/S

“Small Signal Audio Design”

Douglas Self, Focal Press, Elsevier 2010
ISBN 978-0-240-52177-0

“Intuitive IC OP Amps”
Thomas M. Frederiksen, National’s Semiconductor Technology Series 1984

© Springer International Publishing Switzerland 2016 393

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8
394 Appendix 2: Useful Literature and Web Sites

“How to Gain Gain”

Burkhard Vogel, Springer 2013, 2nd edition
ISBN 978-3-642-33032-2
Abbreviation in this book: HTGG-2

“The Sound of Silence”

Burkhard Vogel, Springer 2008, 1st edition
ISBN 978-3-540-76884-5
Abbreviation in this book: TSOS-1

“The Sound of Silence”

Burkhard Vogel, Springer 2011, 2nd edition
ISBN 978-3-642-19773-4
Abbreviation in this book: TSOS-2

Web Sites
www.tubedata.info/
This web site covers nearly all valve data sheets

www.sengpielaudio.com/
This web site covers sound studio and audio calculations in English and German

www.7a.biglobe.ne.jp/yosh/recspecs.htm
“Personal notes on record specifications”
Website with a great collection on IEC/DIN audio specs

Magazines
Electronics World (ex Wireless World, ex Electronics and Wireless World)
www.electronicsworld.co.uk

Elektor Electronics
www.elektor.de (D)
www.elektor.com (UK)
Appendix 2: Useful Literature and Web Sites 395

Linear Audio
www.linearaudio.net (NL)

Tube CAD Journal

www.tubecad.com (US)
Index TSOS-1 A-weighting filter NAB 297ff
(Subjects and Personality's Names) A-weighting filters 182, 183

B
balanced 86
balanced (b) solutions 250
Symbols balanced cable connections 299
µA723 197, 311 balanced in/balanced out 251
1/f-noise 19, 59, 92 balanced in/un-balanced out 251
2N2905A 197 base spreading resistance 36,54,159ff
2N3055 197 Bateman, Cyril 200
2N4403 193 Baxandall, P. J. 159
2SC2546E 29, 38, 124, 158ff BC212B 135, 144
2SJ74 56 benchmark 182
2SK170 56ff BFW16A 124, 193
2SK289 67 BJT 36, 119
3rd octave 126 BJT noise model 36
3rd octave band measurement 153 BNC 168, 305
Boltzmann's constant 21, 157, 183
Brüggemann, Albert 82, 126, 324
Numbers BUF603 297
5532 187ff BUVO 187
5534 186ff
7308 299 C
Cale, JJ 324
A cartridge equivalent model 154
absolute room temperature 183 cartridge loading capacitor 168
active solution 241ff cascoded 66
active-passive solution 238ff cathode input resistance 76
AD536 155, 279ff CCIR 291, 295
AD797 123, 314 CE gain stage noise model 51
Adam, Wilfried 54, 155, 291 CE stage 48
admittances 156 Chebyshev 155, 292
AEG-Telefunken 75 cinch 305
A-filter transfer function 164 Clapton, Eric 324
AMP 155, 165 CLIO 40 279
amplifier noise model 18 CLIO 6.5, 6.5 151, 155, 279ff
AN-104 135, 149, 157 CLIO AD converter 291
AN-222 159, 160 Connelly, J. A. 17
AN-346 240 contribution allowed 42ff
ANSI 291 cooling 169
anti-aliasing filter 291 correction factor 140
anti-RIAA transfer 9,11, 279, 295 CS gain stage 63ff
anti-RIAA transfer function 10, 135 Cu 126
argument 157 Cu master 128 ff
A-weighting 112 cutting technologies 125
A-weighting filter CCIR 279ff

© Springer International Publishing Switzerland 2016 397

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8
398 Index TSOS-1

D I
Dael, J. W. van 164, 245 IEC 11, 182, 314
degrees 157 impedance measurement 151
Denon DL-103 3, 121, 181, 197 impedance network 156
Deutsche Grammophon Gesellschaft 5, 168 impedance transfer 108
deviation 13, 233, 246ff in-amp 93ff
DMM (technology) 125, 140, 141, 325 in-amp IC circuitry topologies 102ff
DOSE 186 in-amp IC topology type 1 102
draft design 104 in-amp IC topology type 2 103
in-amp noise model 95ff
in-amp type 2 topology 251
E
INA103 102
E188CC 299 inductance 54,181
Early voltage 69 insertion loss 234 ff
ELC-131D 155
Elector Electronics 81, 151, 155, 245
ELMA 311 J
Emerick, Geoff 299 J113A 144
engine diagram 304 J37 (Studer tape recorder) 299
Epcos 200 JAES 7
equivalent transformer circuit 110, 189 Jensen 187 ff
example calculations 52, 70, 76, 89, Jensen Transformers 107
91, 99, 100, 111 JFET noise model 58
excess noise 34, 52 JFETs 55,119,120
experience electronics 187 JLH 229
Johnson noise 21
Jones, Morgan 81
F
JT-346-AX 123ff
fab 4 (The Beatles) 299 JT-346-AXT 314
FETs 55ff JT-347-AXT 124
FFT 190, 281, 319ff jumpers 314
filter bank 291ff
fisher electronic 305
flicker noise 19 K
formulae method 247ff Kay, Sharon x
Frederiksen, Th. M. 86, 291 Kruithof, J. A. 164, 245
Friedemann 324
full speed 324
L
L-Com 305
G lacquer (technology) 125
gain loss of the transformer 109ff LF356 144
Gevel, Marcel van 150, 154, 161 Linn Linto 182 ff
groove 3 LM317/337 197
H LM394 32,161,193
logarithmic converter 285 ff
half speed 324 low-noise BJTs 45 ff
headers 314 low-noise measurement pre-amplifier 154
heaters 85 low-noise valves 74
HM 412 155 low-noise vinyl records 324
Hood, John Linsley 229 low-noise in-amp 104
HP 331 A 155 LT1028 155, 285
hum 83ff, 298, 305, 319 Lundahl 187
Index TSOS-1 399

M noise voltage 18, 59, 72, 73

M44G 150, 168 noise voltage approach 111
magnitude 156 noise voltage densities 201
manufacturer's data 150 noise voltage sources parallel-connected 23
Massey, Howard 299 noise voltage sources series-connected 22
MAT02 44,124,193 Nordholt, E. H. 7
MathCad 149 ff Nyquist 21, 183
MC cartridge 3, 181
MC cartridge noise 182ff O
MC phono-amp 304
Okhams's Razor 49, 154
MC phono-amp noise 182ff
op-amp noise model 88, 90
McCormick, Tom 55
op-amps 86
MCD 149ff
OP27 19ff, 31, 89, 91, 197
measurement amp 158, 165, 279, 285
OPA604 151, 296
measurement filters 279
OPA627 313, 314
microphones 299
optimal source resistance 30, 62, 116
Miller capacitance 64
Ortofon RMA-297 127
Miller-C 122, 230ff, 236ff
Ortofon Rondo 197
MM cartridge 3, 149ff
Ortofon Samba 123
MM cartridge data 150
overload 236
MM cartridge noise 154ff
MM phono-amp 190, 304
module 1 304, 309, 321ff P
module 2 304, 311, 320, 324 Panasonic (FC/25V) 197
module 3 304, 313 paralleling 24
Mogami 193 passive solution 228ff
Motchenbacher, C. D. 17 Pauler, Günter 126, 325
MOTHER 126, 136 peak velocity 168
Mu-metal 86 phase angle 156
mutual conductance 38, 68 phase measurement 151
phono-amps 6ff, 140ff
N Pikatron 187ff
potential-free 299
NAB 291, 295
power supply 192, 197, 289, 304, 311
NAB-A-Filter 155
pre-amps 3, 313
Neumann 127
pre-pre-amp 3
Neumann phono-amp PUE 74 135
primary 110
Neumann PUE 7 144, 169
pSpice 13,153
Neumann VMS-80/DMM 127
PSU 1 … 3 304
Neutrik 193
PSU-4 305
noise contribution 42, 314
purpose 3
noise current 18, 59
noise current sources parallel-connected 23
noise current sources series connected 23 Q
noise factor 28, 43 Quad 84
noise figure 28, 60, 61
noise figure approach 112
noise gain 92 R
noise index 34 radians 157
noise measurement system 279 RCA 305
noise model 79 rectifier 285ff
noise resistance 72 reduced mutual conductance 63
noise spectrum 85
400 Index TSOS-1

relay control 318 succ-apps method 245ff

resistor noise 21 sum of two SNs 139
results 167, 196, 200, 235, 303, 319
RIAA 5
T
RIAA decoder 295
RIAA encoder 295 Talema toroidal transformer 197
RIAA equalization 155, 165 tape recorder 299
RIAA networks 227 Taylor, E. F. 190
RIAA phono amp engine 303ff Taylor, F. 294
RIAA time constant 227 Teldec 126, 140
RIAA transfer 5, 9, 11, 183, 190, 193, Telefunken 71, 135ff
197, 227, 279, 295ff temperature 169
RIAA transfer function 10, 163, 227 test circuit for i/p capacitances 196
Ricker, Stan 324 test record 5, 168
RTA 190 test terminal 282
THAT 300 104
THAT 320 104
S Tietze, U. 17
Schenk, C. 17 transfer factor 138
secondary 110 transformer driven MC phono-amp 190ff
Self, Douglas 17,184ff, 196 transformer equations 108ff
sensitivity 168 transformer noise model 108ff
sequence 26 transformer solution 187
Sergeant Pepper's 299 transformers 106ff
series configuration 87 TSD15 (EMT) 127
series mode 245 Tube CAD Journal 82
S-filter 185ff, 285, 294 tubes 71
Sheingold, D. H. 86, 161 turns ratio 106
Sherwin, Jim 149, 157 Twin-BNC 193, 305
shot noise 291
shunt configuration 87
U
shunt mode 243
Shure V15 V MR 3, 5, 314 U87 (microphone) 299
Signal-to-Noise Ratio 33 un-balanced (ub) solutions 228ff
SME 3012 127
Smith, L. 86, 161 V
SN by simple means 183
SN calculations 162 V15II 127
SN-factors 184 V15III 150
solid-state approach 193 V15IV 150
SON 126 V15V 136
sound 197, 200, 324 valve noise model 71
Sowter187ff valve power supplies 83
spectral noise voltage density 290, 323ff valves 71, 119
SRPP 66,81ff, 236 VALVO 75
SSM2017 103, 313ff velocity 125, 181
SSM2210 32, 105, 193 Vierzen, R. M. van 7
stereoplay 84ff vinyl record 3, 125ff
Stockfisch Records 126, 325 vinyl record noise 126ff
stray-C 230ff Vitilec 305
Studer 299 VMS-80/DMM 127
Index TSOS-1 401

W wiring 193
Walker, H. P. 149, 243 wobble speed 298
weighted 69 worst case 139
weighting filters 33 X
Whitlock, Bill 299 XLR 305
Williams, A. 294
WIMA MKP4/10 200
Index TSOS-2 5534 ( " ) 325
(Subjects and Personality's Names) 6276 (pentode) 101
6922 (double-triode) 97, 108ff, 372
7308 (double-triode) 601
10 99 112 (test record) 5, 303

Symbols A

π-model (BJT) 55 absolute temperature 292, 322

π-network (power supplies) 144 AC/DC-conversion 613
µA723 337, 680 active anode load 368
µ-F 368ff active devices 40
µ-follower 122, 131ff, 193, 368ff, 375 active-passive 2-step solution 519
1/f-corner-frequency 33, 102ff active 1-step solution 522
1/f-noise 31ff, 66, 77ff, 97ff, 160, Adam, Wilfried 59, 74, 289, 573, 615
293, 396, 621 admittances 290
2N3055 337 Adventure: Noise 283
2N2905A 337 AD53 290, 560, 576
2N4403 63ff, 333 AD745 6
2-pham 368ff, 383 AD797 198, 622, 677, 687
2-pham (example) 391ff A-DIN certified 586
2SA1083E, F 63ff AEG 106
2SA1316BL 65ff, 690 A-filter 102, 367
2SB737S 63ff A-filter transfer 299
2SC2546E, F 44, 57ff, 63ff, 73, 82ff, 195, A-filter-weighting 83
293ff, 689 Al-box 300
2SC3329BL 65ff, 690 Al plug-in box 8
2SC550 33 AMP 300ff
2SJ74 77, 83 amps 3ff
2SK170 77ff, 82ff amplifier chain 3
2SK389 83, 89, 194 amplifier noise model 32
3-pham 368ff amplifying stages 41
3-pham (example) 397ff AN-104 (Nat. Sem.) 204, 283, 292
3rd octave 202, 286ff, 383 AN-222 (Nat. Sem.) 293, 332, 680
26dB boundary 414 AN-346 (Nat. Sem.) 520
45° modulation, modulated 5, 18ff Analog Devices 34ff, 175, 290
45°/45° modulation 18ff analog meter 576
6BK8 101 analog solution 643
6SN7 108 Analogue Dialogue 296
6J5 108 anode current 690
6KX8 140 anode load 122
7DJ6 97 anode load resistance 110
12AU7 108, 367, 372 anode power lines 141f
12AT7 108, 372 ANSI 50, 102, 299, 559
12AX7 108, 372 arm (circuits) 560, 613
20Hz hp 687 anti-alias filter 588, 591, 602
anti-alias filter (CLIO) 595
anti-phase 18
Numbers anti-RIAA (ARIAA) 341, 634, 642
2549 (Mogami) 332, 677 anti RIAA transfer 9, 204, 597ff
5532 (op-amp) 327 anti RIAA transfer function 9ff

© Springer International Publishing Switzerland 2016 403

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8
404 Index TSOS-2

Application hints (National Semic.) 605 Brüggemann 144

argument 291 btb-elektronik (internet) 738
arm circuits 613f BUF634 336
ARIAA 634, 644 buffer 336
ARIAA network 642 Burr-Brown 176, 220
AS040 186, 329 Butterworth 559ff, 591, 605ff, 687
audio analyzer 118 Butterworth band-pass 602, 604
audio band 33 BUVO 327ff
AUDIOMATICA Srl 286, 595, 643 bypassed 86, 109ff
Audio Precision 584, 619, 628, 643 bypassed cathode resistor 121
audio spectrum 51 bypassing capacitance 102
Audio.TST 619
average noise voltage density 7
C
averaging 329
average 35 cable capacitance 93
average noise resistance 139 calculated deviation (2- & 3-pham) 397, 399
average spectral noise voltage density 102 calculation approach (2-& 3-pham) 385ff
average reading meter 560, 570 calculation approachA (2-& 3-pham
average single frequency component 581 ­examples) 401ff
A-weighting 50, 152, 566, 617ff calculation by blocks (2-pham) 402ff
A-weighting term 186 calculation by blocks (3-pham) 404ff
A-weighting application (silly) 609ff calculation method(s) (valves) 379, 407
A-weighting filters 320ff, 593 Cale, JJ 703
A-weighted SN 299 calibration process 220
calibration record 20
camouflage(s) 583
B capacitance load of MM cartridge 88
B32652 (capacitor) 340 capacitors 39
balanced amplifier 122 cartridge equivalent circuit 18
balanced cable connections 600 cartridge equivalent model 287f
balanced gain stage 134f cartridge impedance 285
balanced input 7 cascoded gain stage 89ff
balanced solutions (RIAA network) 531 cascoded SRPP 91
band-pass filter (hum figure) 636 cathode current sink 134
base 55 cathode follower 21, 368ff
base spreading resistance 55ff, 74f cathode input resistance 107
Bateman, C. 340 cathode resistance noise 107
Baxandall, P. J. 55, 293 cartridge 3
BC109C 579 CCA 122, 374ff, 402
BC212B 205, 220 CCIR 468-(Filter) 290, 559, 586, 633, 642
BD 679, 680 300 CCIR-1kHz 566
benchmarks 320 CCIR-2kHz 566
Berliner, Emil 17 CCIR-A-Filter 593
BFW16A 64, 198f, 333 CCG 120ff
bin(s) 582 CCS 368ff
bin width 588 CCSCF 368ff, 409
bipolar junction transistor 6, 55ff CE stage 67
BJT 6, 32, 55ff CF 128ff, 368ff
BJT noise model 55 CFA 128ff, 591
blocks 402ff Chebyshev 559ff, 605ff, 23, 641
Bloehbaum, F. 95 Chebyshev hp 289, 324, 622, 641
BNC connectors 300, 676 cinch 676
Boltzmann's constant 6, 100, 292, 322 circuit worsening factor 140
BRAUN 621 circuitry topology (in-amp) 174ff
brick wall filter 602ff Clapton. E. 703
Index TSOS-2 405

CLIO 286ff, 560ff Darlington transistors 300

CLIO noise performance 576 data sheet 19, 33
coil inductance 45, 319 DC power supply 576
coil resistance 183, 327 DC resistance 19
coil turns 328 decade 50
coil windings 183 Decca 203
collector 55 decode 9
collector current 32ff, 45, 58f, 82 decoder 559
common anode 95 decoding 10
common base 60 decoding mode 12
common cathode 95 decoding process 9
common cathode stage 95, 108, 368ff decoding transfer function 11
common collector 60 degrees (converted from rad) 291
common drain 77 definite frequency bandwidth 297
common emitter 60 Denon 3, 20ff
common gate 77 Denon GmbH 21, 26
common grid 95 derivation (arm vs. rms) 570
common plate 95 Deutsche Grammophon Gesellschaft 5, 303
common source 77 deviation 14ff, 404, 512ff
comparison (measurement results) 610 deviation area 637
Connelly, J. A. 31, 66, 737 deviation (CCIR transfer) 597
constant current generator (valve) 120ff deviation tolerance 637
contribution allowed 61ff, 88, 560 differential amplifier 122
confusion 565ff differential gain stage 134f
cooling 304 differentiation 75
corner frequency 81, 85, 88, 97ff, 367ff, 602ff differentiator 12
corner frequency changing factor 104 digital meter 576
correction factor (cartridges) 217 digitisation 319
correction factor(s) 602ff DIN 5
correlated 36, 134, 631 DIN-IEC 651 559, 586
cross-section 18ff DIN EN 61672-1 586
CS-configuration 86 DIN 0dB (level) 22, 25, 185, 303
CSRPP 91 DIN 45405 559, 586ff
CSV60 621 DIN 45535/6 11
Cu layer 214 DIN standard 8
Cu master 202 direct metal mastering (DMM) 201
Cu material 214 distortion analyzer 290
current feedback 67ff distortion artefacts 142, 631
current feedback op-amps 153 DL-103 3, 20ff, 185, 319, 337, 690
current noise index 50, 372, 409f DL-103R 25f
current mirror 329 DL series (Neutrik) 676
current sink 134 DMM 201, 220ff, 703
customer confusion 586ff DMM process 202
cutting (vinyl) 9 Dolby arm approach 612ff
cutting modulations 17ff Dolby, R. 566, 593
cutting possibilities vii DOSE 324ff
cutting process 9 draft design (in-amp) 177f
cutting technologies 201 draft design (JFET) 93
draft design (valve) 151
D/S 31
D
double-triode(s) 110ff, 368ff
Dael, J. W. van 299, 526, 528 drain 77
Dangerous Music 632 drain current 32
406 Index TSOS-2

E equivalent noise voltage density BJT 56

E88C 375 equivalent noise voltage density JFET 78
E88CC 97, 105, 108ff, 372 equivalent noise voltage density (opa) 153
E188CC 601 equivalent noise voltage density triode 95
ear 3 equivalent noise resistance (triode) 96f
Early voltage 91 ESCORT 290
EC8010 414 essentials 1
EC92 375 Etna E1 418
ECC81 108, 372 EU standard 8
ECC82 108, 367, 372 Everest C3 418
ECC83S 108, 372 example calculation 71ff, 75, 92ff, 108ff
ECC88 97, 7f, 17 example calculation (in-amp) 171ff
ECC808 140 example calculation series config. 157
ECF80 621 example calculation shunt config. 159
Edison, Alva 17 example calculation (transformer) 184ff
EF86 101 example 2-pham 391ff
EL84T 134 example 3-pham 397ff
ELC-131-D 290 example phono-amps (calc. approach) 391ff
electron (elementary) charge 100 example phono-amp (hum figure) 634ff
Elector Electronics 101, 176, 286, 289, 739 EXCEL 14, 287
Electronics World x, 101, 739 excess noise 50ff, 71ff, 86, 107ff, 409
Elektor 176, 299, 526, 739 experience electronics 327
ELMA 679 exponent change 582
Emerick, G. 601
Emerson 332 F
emitter 55
fab 4 (Beatles) 601
emitter bulk resistor 55
Fairchild 680
emitter resistor 55
FEG viii, 332
EMT TSD15 203, 214
FET(s) 6, 32, 77ff
ENB 138, 322
FFT analyzer 118
encode 9
FFT approach 622, 627
encoder 559
FFT diagram(s) 367f, 583ff
encoding mode 12
FFT measurement 329, 573, 579ff
engine 671ff
FFT measurement methods viii
engine diagram 672f
FFT noise measurement 579ff
engine functions 672ff
FFT resolution 579
engine performance 693ff
FFT size(s) 579, 627
Engineering Note 1.2 565
field effect transistor 6, 77ff
Epcos (capacitor) 340
filter approach 622, 624
equalization 297
filter bank 1 591ff
equalization factor 322
filter bank 2 604ff
equalized 49, 299
fischer elektronik 676
equalizing effects 5, 49
flank modulation 18ff, 215, 303
equalizing network 559
flicker noise 33, 97, 329
equivalent circuit(s) 35ff, 182, 327
FM stations 31
equivalent gate noise current density 80
formulae method 528
equivalent grid noise current density 95f
forward transfer admittance 80
equivalent noise bandwidth viii, 138, 322,
fourth time constant (20Hz) 376
566ff, 602ff
fourth time constant (>20kHz) 526
equivalent noise current density BJT 56
Frederiksen, Th. M. 153, 593, 738
equivalent noise current density triode 96
frequency 32
equivalent input noise sources 32
frequency response 10, 180, 693ff
equivalent input rms noise voltage EIN 99
frequency test record 5
Index TSOS-2 407

Friedemann, W. 703 How to Gain Gain (HTTG) 107, 738

full-speed mastering 201, 703 HP 331A 290, 560
fully active solution (RIAA) 506 HR2000 560
fully active-passive solution (RIAA) 506 HTTG 107f
fully passive solution (RIAA) 506 hum vii, 118, 141ff, 617ff
Future Equipment Service viii hum artefact(s) 138, 617ff, 639
hum content 144
hum figure vii, 144, 559, 589,
G
hum figure (linear amps) 617ff
gain 86 hum figure (phono-amps) 631ff
gain loss 183 hum-free 6, 586
gain rate 560 hum-infected 586, 631
gain setting options (in-amps) 165f hum interference(s) 6, 600, 639
gain stage calculation 71ff, 85f hum spectrum 624
gate 77 hum spikes 589
gate cut-off current 80
gate-drain-capacitance 87
general noise effects 31 I
Gerhard, J. 689 ideal filter 602
Gevel, M. van 284, 288, 296 ideal situation 14
Glowing GainMaker 109, 134 idle gain 136
golden ears 414 idle output voltage 144
grain size 214 idle voltage 19
graphs (o/p noise voltage density) 342f IEC 11, 297, 320, 376
graphs (performance) 693ff IEC 20Hz hp 671
Green, L. 287 IEC 651 586
grid load resistance 112 IEC/CD 1672 299
grid resistor(s) 691 impedances 39
grooves 3 INA103 174
groove depth 18ff in-akustik 703
groove width 18ff, 203 in-amp(s) 165ff
guarantee 139 in-amp noise model 165ff
Gundry, K. 566, 593 in-amp IC topology 174ff
gyrators 300 indicator(s) 560
inductance(s) 19, 39
inductance of MC cartridge 195, 319
H
influence of temperature 304
half-speed mastering 201, 703 input load 320
Hameg 290 input noise current 77
Handbook of Linear Applications 220 input noise resistance 79, 285
hard disk 562 input network 291
head shell 3 input network impedance 414
headers 687 input possibilities 411
heating effect, equivalent 32 input referred 33
heater(s) 141, 144 input referred noise voltage density 92, 121
HF measurement amp 622ff input resistance 45, 65, 285, 300, 324ff
HF measurement instrument 639ff input resistance (cathode) 107
high-end 31, 319 input section 677
high-price 31 input sensitivity 6
high-Z 131 insertion loss 511ff, 560, 576, 579
Highlights 15 - 18 217 integrated amplifier 617
Hitachi 44 internal resistance 20ff
HM 412 290 inverse RIAA transfer 634
Hood, J. L. 508, 737 ISCE 565
408 Index TSOS-2

ITU 566, 612 locations (hum) 639

ITU-T J.16 633, 25 lowest-noise 31, 95
ITU-R 486 586 low-frequency noise 31, 367
log-converter 560, 576
long-tailed pair 60, 66, 134, 329
J
loudness pot 337
J113A 220 lowest-noise 92, 95, 100, 146, 151, 176f
J37 (Studer) 601 lowest-noise in-amps 177
JAES 6, 221, 566, 593, 601 low-frequency noise 367, 375, 381, 397, 413ff
JAN double-triodes 418 low-noise 6
Jensen Transformers 152, 180, 185f, 197f, low-noise BJTs 192
327ff, 412, 601, 677, 679, 687f low-noise measurement pre-amp 289
JFET 77ff low-noise triodes 150, 379, 415
JFET gain stage 87 low-noise valves 105, 121, 139
JFET noise model 79 low-noise vinyl records 703
JFETs parallel connected 194 low-pass filter (30Hz) 593
JJ electronics viii, 108ff, 372 low-pass filter sections 611
jogis-roehrenbude 101 low-Z 131
Jones, Morgan 108, 122, 131, 328, 369, 375, LP records 201
417, 738 LT1028 289, 573, 622
Johnson noise 35, 107 Lundahl 180, 327f
JT-10KB-DPC 677
JT-123-SPLC 679
JT-346-AX or AXT 152, 197, 337, 687 M
JT-347-AXT 180, 186, 198 M44G 304
JT-44K-DX 185ff magnetic shielding 144
junction field effect transistors 77ff magnitude 12
jumper settings 686 mains interferences 142, 332
mains transformers 142
manufacturer 17
K
manufacturer selected triodes 407
Kay, Sharon x Massey, H. 601
Kelvin 292 master (Cu) 202
kernel (BJT) 55 mastering 201
Kruithof, J. A. 299, 526, 528 MAT02/03 61, 63ff, 333, 337
Mathcad ix, 14, 283
L mathematical model 290, 302
Mathew, D. 619
lacquer technology 201, 703 Mathsoft 283
lateral modulated, modulation 5, 17 Matsushita 679
L-com 332, 676 maxi records 201
LF356 220 maximal SNs (records) 216
linear amps 591, 617 maximal stage gain allowed 88
Linear Applications Databook 204, 283, 292f maximum deviation 591
Linear Audio 95, 689, 739 maxi ix
line input transformer 677 MC cartridge 3ff, 19, 81ff, 411
lin/log-converter 614 MC cartridge classification 181
Linn Linto 320ff McCormick, T. 59, 74
LL9226 329 MC encoding situation 598
LM317 337 MC phono-amp 330, 336
LM337 337 M/C 31
LM394 61ff, 71, 296, 333 measurement amp(s) 300, 559ff, 573ff
load capacitance 19, 284, 304 measurement confusion 565ff
load resistance 19 measurement equipment
Index TSOS-2 409

(HF in phono-amps) 633 Neumann (GmbH) 152, 202, 220

measurement filter(s) 559, 591ff Neumann PUE 74 203ff, 220ff, 304
measurement methods 566 Neumann U87 601
measurement results (relationships) 568f Neutrik 332, 676
measurement results (comparisons) 610 NF (calculations) 62, 145, 410
measurement set-up 561, 624 noise (cathode resistance) 107
measurement standards 566 noise (general effects) 31ff
measurement system noise basics 29ff
noise performance 576ff noise calculations (triodes) 101ff
measurement tools 289 noise contribution 61
measurement set-up 75, 561 noise currents 33
measurement set-up (hum figure) 624 noise current density 34ff, 56ff, 80, ff, 96,
mechanic arrangement 378 120, 153, 173ff, 199
metal film resistor(s) 51, 691 noise current source(s) 32ff, 55
Metzler, B. 628, 737 noise density 33
microphone pre-amps 591 noise density frequency band 292
Microsoft 287 noise factor 43ff
MicroSim ix noise figure 43ff, 82ff, 6, 7
Miller capacitance 87f, 152 noise figure approach (trafo) 184ff
MKP (capacitor) 340, 372 noise figure calculation process 17
MKT (capacitor) 340, 372 noise floor (records) 202
MM cartridge 3ff, 19, 65, 83ff, 92, 284ff, 411 noise floor (FFT diagrams) 583
MM encoding situation 598 noise-free 32
MM phono-amp 330, 411 noise gain 163
modulation possibilities vii noise generator(s) 559, 573, 579ff
modules 1 - 4 672ff, 679ff, 694ff noise index NI 50
module 3 viii noise in BJTs 55ff
Mogami 332, 677 noise in instrumentation amps 165ff
mono 17f noise in JFETs 77ff
monophonic purposes 25 noise in MC phono-amps 319ff
monophonic records 18 noise in MM cartridges 283ff
monophonic signal 22 noise in op-amps 153ff
morphological box 381 noise in transformers 179ff
Motchenbacher, C. D. 31, 66, 737 noise in valves 95ff
MOTHER 202 noise issues 9, 31
Motorola 205 noiseless amplifier 291
mud effects 583 noise level (records) 202
Mu-metal 144 noise measurement methods vii
Musical Fidelity XLP 326, 333 noise measurement results
mutual conductance of BJT 69 (different methods) 566ff
mutual conductance of JFET 91 noise measurement system 559ff
mutual conductance (anode) 100 noise model 32, 56ff, 59, 70, 95f, 118f
mutual conductance (cathode) 100 noise model (in-amps) 165ff
mutual conductance (minim. required) 410 noise model (measurement amp) 292ff
mutual conductance of triode 97ff, 690 noise model (op-amp) 155ff
mutual conductance of pentode 100 noise model series configuration 157f
noise model shunt configuration 158f
noise of capacitors 39
N
noise of inductances 39
NAB 50, 559 noise of triode driven phono-amps 367ff
NAB/ANSI 299 noise of vinyl records 201ff
NAB-A-filter 289 noise performance 46, 343
National Semiconductor 48, 204, 605 noise power 43, 49
negative valve effects 148f noise resistance 96f
410 Index TSOS-2

Noise Specs Confusing? 204 output section 687

noise voltages 33 output voltage 19ff
noise voltage approach (trafo) 184ff overload(ing) ix, 9, 89, 174, 283, 289, 328,
noise voltage density 36ff, 57ff, 77ff, 113ff 332, 507, 516, 520, 615
noise voltage density plots (valves) 146ff
noise voltage of the MOTHER 214 P
noise voltage source 32ff, 55
nominal data 21 Panasonic (FC) 336ff
nominal input voltage 179 paralleling 40f, 105, 373, 375
nominal internal resistance 21 parallel-connected 37f
nominal output load 21 parallel operation 108ff
nominal output voltage 21, 179 Parametric Technology Corp. 283
non-equalized 297, 633ff passive 1-step solution 507
non-inverting configuration 154 passive 2-step solution 515
non-selected 117 passive RIAA (transfer) network 89, 369ff
non-weighted 633 Pauler Acoustics 202f, 703
Nordholt, E. H. 66, 221 Pauler, Günter 202f
noval 108 PCC88 97
Nyquist, M. 36, 292, 627 peak value 32
Nyquist frequency 582, 588 peak velocity 5, 22ff, 201
pentode 95, 100, 109
pentode noise resistance 100f
O pentode noise voltage 101f
octal 108 phase angle(s) 286, 341
octave-band analysis 283 phase measurement 285
Okham's Razor x, 67, 131, 288 phase relationship 109
op-amp 6, 33 phase response 180, 693ff
OP27 33ff, 61f, 337, 634 phase shift 69
OPA604 285 phono-amp 3
OPA627 683 phono-amp noise 6
operating frequency bandwidth 292 phono amplification 3
operating gain 42 physical constants 32
operating output level 333 Pikatron 180, 327f
operating point 372 pin-1 problem 677
operational amplifier 6 pink noise 33, 117, 367, 559, 579ff
operational SN 333 pink noise generator 579ff
optimal (optimum) collector current 75, 336, PL504 621
688f platter 3
optimal input load 75 plug-in box 8
optimal (optimum) source resistance viii, 46, ponderé 587
75, 84, 161 pop-corn-noise 367
optimisation process 141 Potchinkov, A. 738
optimal (optimum) collector current 75, 688 potential-free 677
optimum source resistance 46f, 84f, 161f power amplifier 617
Ortofon RMA-297 203 power supply(s) 74, 84, 141ff, 300, 331f,
Ortofon Rondo 337 337ff, 369f, 520, 576, 671f, 676, 679ff
Ortofon Samba 198 power-supply-unit 329, 369ff
oscillation prevention resistor 112, 372 plug-in box 8
output capacitance 87 pre-amps 3ff
output current noise 134 pre-pre-amp(s) 3ff, 319, 411
output impedance 42 primary 179, 183, 327
output load 109, 385 professional audio (magazine) 632
output power 49 pSpice simulation 16, 67, 220, 287
output resistance 110, 136 PSU 690
Index TSOS-2 411

PSU 1 - 5 672ff RIAA decoder 597ff

PUE 74 203ff RIAA decoding function 505
purpose 3, 6f RIAA encoder 597ff
RIAA equalized 7, 92, 633
RIAA equalization 83, 300
Q
RIAA equalizating term 186
Q & A (transformer) 190ff RIAA feedback network 336
Q & A (2- & 3-pham) 389 RIAA-Isierung 299, 526, 528
Quad 24P 142ff RIAA network(s) 151, 340, 369, 372, 376, ff,
quadratic equation 57, 283, 294f 505ff, 691
quasi-peak 290, 393, 560, 566ff, 586, RIAA transfer 5, 9ff, 14, 220, 283, 298,
596, 612, 633, 642 322, 505ff
RIAA transfer circuit 299
R RIAA weighting 9ff, 2
Ricker, S. 703
rabbit hole 689 ripple rejection 680
radians 291 rms velocity 201
random 31 RMS voltmeter 290
Rauschmessungen 101 Robinson, D. 566, 593
rbb' calculation 295 Roehren 101
RCA 676 Roehren Taschen Tabelle 106
real situation 14 Rohm 66
Record Industry Ass. of America 9 role of 2nd gain-stage 414
record reference levels vii rotary switch 573, 679
rectifier 560, 576 RTA measurement 329
red noise 102, 367 rule(s) of thumb 143, 300414
reduced mutual conductance 86 rule of thumb (FFT diagrams) 579
reference level(s) vii, 5, 17ff203 rumble 204ff, 220, 289
reference point 12
reference test record 5
reference voltage 102 S
regulated power supply 144 sample frequency 581
relay control 688 sample rate 581, 627
relay drivers 688 schematic 16
relationships (measurement results) 568f Schenk, C. 737
relationship (SNs) 417 scope 290
resistor-capacitor network 194 secondary 179, 183, 327ff
resistor excess noise 50ff sensitivity 6, 220, 289, 303,
resistor noise 35ff 319, 622, 628, 631, 642, 672
results 302ff, 327, 333, 341, sequence 12, 36ff, 61, 143, 164, 169, 177,
379, 413, 610 195, 287, 327, 368, 372, 569, 587,
results (µ-F 2- & 3-pham calculated) 386f 611f, 634f
results (example triode phams) 407f selection 62ff, 106, 217ff
results (hum figure) 622f selected low-noise 116
results (hum figure in phono-amps) 633, 639, Self, Douglas ix, 31, 67, 322ff, 620, 737
642 Semiconduct. Handbook (Telefunken) 205
results (module 3) 689 sengpielaudio (internet) 738
results (test gain-stages) 382 Sennheiser electronic 202
reverse-transfer-capacitance 87 Sennheiser UPM 550 203
RG58 677 sequence-connected 36ff
RG176 332 sequence of two amplifying stages 41ff
Rhode & Schwarz 290, 560, 602, 643 Sergeant Pepper's …. 601
RIAA x, 5, 9, 283ff series configuration 153ff
RIAA curve 367 series mode (RIAA feedback) 526
412 Index TSOS-2

S-filter 323ff, 367, 573, 593 spectral noise density 153, 202
Shannon 588 spectral noise current density 34
Sheingold, D. H. 153, 296 spectral noise voltage density 34, 52, 108, 114
Sherwin, J. 204, 283, 292 spot noise 367
shielded cable 332 Springer web site www.springer.com
shot noise 102, 152, 388, 395 SRPP gain stage (JFET) 88f
shunt configuration 154ff SRPP gain stage (valve) 120, 126ff, 374ff
shunt regulated push-pull see SRPP SRPP (valve) 120f, 368
shunt mode (RIAA feedback) 524 SSM2210 63ff, 198f, 333, 337
Shure 3, 284 SSM2017 174f, 683
Shure M44G 284 SSM2220 63ff
Shure V15 III 203, 284 SST Brüggemann 144, 202ff, 703
Shure V15IV 284 stamper 202
Shure V15V (MR) 203, 205ff, 284ff, 687 standard cartridge 303
sidebands 631 step-up transformer 7, 151, 180ff, 319
Siemens (capacitor) 340, 372 stereo 18
signal-to-noise ratio (SN) ix, 5, 32ff, 49, 66, stereo equalizer 418
72, 114, 139, stereoplay 142, 217ff, 320, 341, 586,
158f, 186f, 283, 297ff, 586, 617ff, 633ff 631f, 739
signal modulation 27 Stockfish Records 202f, 703
signal track 20 stray-C 508, 511, 517
silly (A-weighting) 609ff Studer 601
single ix succ-apps 375, 404, 410
single records 201 succ-apps approach (RIAA) 512
singleton 77, 83f successive approximation (RIAA) 512
skin effect 287 subsonic 289
SME 3009 203 sum of A-weighted SNs 216f
SME 3012 203 sum of two SNs 217, 683
SME connector 300 summary (on FFT mesurement) 581f
Smith, L. 153, 297 summary (on hum figure) 628
smoothing effect 300, 388, 413 summary (on measurement methods) 571
SN by simple means 321ff summary (on MM cartridges) 305
SN calculation (vinyl record) 204 summary (on valves) 415ff
SN calculation rules 321ff summary tables 703f
SN (calculations) 49, 145, 297 suppression of noise 134
SN-delta 321
SN guessing (via FFT) 590
T
SN measurements 102
SN of the MOTHER 213 Talema 337
SN relationship(s) 417 tangents 98
socket (valve) 378 Taylor, E. F. 329
soft start 689 Taylor, F. 595, 738
solid-state devices 6 Teldec 203, 206ff, 217, 220, 222
Solid Tube Audio 417, 738 Telefunken (+ Decca) 203, 205
solving approach 57 Telefunken Labor Buch 95, 738
SON(S) 202 Texas Instruments 176
sound (& measurement & pressure) 332, 337, temperature 32, 62, 332, 340
340, 505, 586f, 601, temperature (influence of) 304
609, 618, 629, 676703 test and calibration record 20ff
source 77 test circuit (capacitances) 336
source resistor 85 TESTfactory 586, 631
source resistance (cartridge) 319 test gain-stage(s) 372
Sowter 180, 327f test magazines 333
specifications (MM cartridges) 284 test record 5, 19ff, 303
Index TSOS-2 413

test terminal 560ff U

Texas Instruments 66 un-balanced - balanced conversion 681
That Corporation 66 un-balanced input 7
That 300 63ff, 177 un-balanced solutions (RIAA network) 506
That 320 63ff, 177 un-bypassed 86, 107, 109ff, 369
That 1648 687 un-bypassed cathode resistor 107, 121
thermal noise 31 un-loaded 109
Thorens TEP3800 632 uncorrelated 36, 291
threshold voltage 89ff un-equalized 634
third octave 202ff UPGR 290, 560ff, 602
Tietze, U. 737 US standard 8
time constants 9ff, 14, 340, 405, 505,
519, 526
toroidal transformer 300, 337 V
Toshiba 65 valve control figure 97
TQ2-12V relay 679 valve constants 375, 379
trackability test record 5, 303 valve manufacturers 139
transfer factor (cartridges) 214 valve manufacturing industry 105
transfer function 49 valve noise model 95f
transfer function producing networks 591 valve power supplies 141ff
transfer of impedances 181 valves 6, 95ff
transfer plot 9 Valves Pocket Book 106
transformer 7 Valvo 106
transformer capacitance 183 van den Hul 201
transformer classification 181 velocity 19ff, 201ff, 214
transformer equation (ideal) 181ff vertical modulation 17
transformer equation (real) 183ff Vierzen, R. M. van 66, 221
transformer frequency response 180 vinyl record ix, 201ff, 520, 617, 703
transformer inductance 183 vinyl record noise 201ff
transformer input situation 412 vinyl record reference levels vii, 17ff
transformer phase response 180 Vishay 51, 410
transformer solution 327ff Vitelec 332, 676
triode constants 110, 123, 135, 402, 416 VMS-80/DMM 202, 212
triode constants data 135 voltage divider(s) 42, 74, 112f, 118f, 189, 196,
triode driven 151, 367ff, 672, 691 221, 292f, 327, 391, 399, 416, 511ff, 576, 597
triode equation 111 voltage feedback 67ff
triode noise model 95f voltage feedback op-amps 153
triode noise production 135 voltage potential 50, 369
triode noise resistance 96, 410 volume knob 31
triode noise voltage 97ff V15III 203
triodes 95ff, 367ff V15V (MR) 3, 204, 396, 414, 687ff
T/S 31
TSD15 (EMT) 203, 214
TSR-1005 21 W
TSR-1007 23 Walker, H. P. 219, 283, 524
Tube CAD Journal 101, 739 Warner Music Group 203
tube data (internet) 738 weighted 49
tubes 95ff weighted SN calculations 92
turns ratio 179 weighting filters 49, 596
turn-table 3 weighting factor 322
Twin-BNC 332, 676 Weinzierl, S. 738
414 Index TSOS-2

WEKA Media Publishing 586 worst case SNs (records) 215ff

white-noise 33, 103 wyciwym viii
white-noise based (meas. results) 568 wyciwym approach (final) 607ff
white-noise character 214
white-noise generator 580
X
white-noise level 620
white-noise region 80, 96 XLR 332, 676
Whitlock, B. 601
Williams, A. 595, 738 Y
Wilson current mirror 329
WIMA (polypropylene capacitor) 340, 372 Yageo 372
WIN98 560
wiring 332 Z
Wireless World x
wobble speed 600f Zwicky, F. 381
Wonneberg, F. 738 Zwicky approach 381ff
worsening factor 102, 140 Zwicky matrix 382
worsening figure 140
Index TSOS-E - noise calculation 104f
(Subjects and Personality's Names) - SN calculation 105
Amp2 6
- CMRR 127
- gain calculation 125ff
- noise calculation 127ff
- SN calculation 137ff
Numbers & Symbols Amp3 6, 21ff
1:1 83 - CMRR 21
1:10 101 - gain calculation 25f
1/f-noise - noise calculation 27f
- Amp1 23, - SN calculation 27f
- Amp2 128ff, 132ff Amp4 6, 49ff
- Amp4 58 - CMRR 49, 55f
- BJT 341, 344 - gain calculation 52ff
- CMS 234 - noise calculation 56ff
- DIFA 370, 372 - SN calculation 60f
- other input stages 201 Amp5 6, 25, 28, 83ff
- PMMA 247, 251, 274 - CMRR 86
180° 53 - gain calculation 83f
19" 5 - noise calculation 86f
19" 3 UH-42 HP 11ff - SN calculation 89
19" 3 UH-84 HP 11ff amp
2SA1316 BL 125ff - central 7ff
2SC3329 BL 125ff, 190 - differential 365ff
2SJ74 199 - input 7
2SK170 199 - microphone 49
2-pham 6 - sequence 51
3-pham 7, 165 - -stage, no-noise 23
317 (LM) 15 amplifier
337 (LM) 15 - chain(s) 167f
2798 168 - differential 52
5534 125, 188ff - instrumentation 53
6922 21 - measurement 247, 275
7308 21 - transconductance 199
amplifying devices 6
anode 193
A anode output 252, 254
Aalen (university) 170 Analog Devices 49, 128
Abbreviations xvii anechoic chamber 170
absolute temperature 342 Appendices xv
AD797 188ff, 274 AP 248, 254, 274
admittance(s) 193 AP system 233
AES (UK) 58 arrangement(s), loudspeaker 168
AKG 169 artefacts
Al-box 269ff, 281 - distortion 238
alternatives (input resistors) 141 - harmonic 183
Amp1 6, 101ff, 101ff - intermodulation 183
- CMRR 103f - sideband 238
- gain calculation 102f Atwell, Bob 49

© Springer International Publishing Switzerland 2016 415

B. Vogel, Balanced Phono-Amps, DOI 10.1007/978-3-319-18524-8
416 Index TSOS-E

audible effects 167ff Breden, Russel 169

audio band 128, 155 brick wall 248
Audiomatica 233 broadband 6
Audio Note 190 BTFMI 387f
Audio Precision 233, 247 budget, low 249
auxiliary (PFMA gain) 279 BUF634 188
average Burosch, Klaus 23
- (noise voltage) 61 Butterworth 248
- density value 28
- level 248
C
- value 189
averaging 266 cable capacitance 252, 307
A-weighted 5, 28, 51, 61, 89, 140, 247, calibration records 385ff
274, 312, 317 capacitance, input 307, 313, 317
A-weighting 104 calculation, NF 134ff
A-weighting function 28, 61 capacitance, Miller 349ff
capacitance(s) 9, 201, 307
capacitance(s), balanced input 8
B capacitance, cable 252, 307
balanced cartridge, standard 310ff, 317
- fully vii, 5 case(s) 5, 11ff
- gain 26, 54 case, insertion 11
- input capacitance(s) 8 cathode 193
- input load 58 - follower 9
- inputs 5 - load 26
- line 277 - output 252
- output noise 88 - resistor 254
- outputs 5 cathodyne phase splitter
Bark 168 CCIF 238
base-emitter resistance 342 CCS 191
base-spreading resistance 130ff CCSCF 21
basic consideration(s) 6 CCSi 25, 28
bass control 181 CE (BJT) 341ff
BC550C 190 center section 6
BC560C 190 central amp 7ff
Beethoven, Ludwig van 168 ceramic 8, 141
benchmark 248 CF 21
BF862 199 CF gain 26f
Biber Records 168 CGS 25, 28
BJT driven 4, 187f channel, left 11
BJT noise model 128ff channel, right 11
BJT(s) 21, 28, 125ff, 254, 341ff Chebyshev 266
block, gain 195 Cinch 203
block diagram 5 CLIO 30, 51, 89, 233ff, 243
BNC 203 clipping 23, 49
B & O 169 closed-loop 137
board, main 6 CMR 64
Boltzmann 342 CMRR 181
booster(s) 126f, 239 - Amp1 103f
BRAUN 169, 385ff - Amp2 127
BRAUN+Design Tax 385, 387 - Amp3 21, 25
BRAUN PCS 52 E 385 - Amp4 49, 55f, 61
BRAUN TG 385 - Amp5 86
Index TSOS-E 417

- PMMA 256, 258, 261, 268 D

- PFMA 275, 278 DA converter (NAD) 247
- DIFA 366, 371 data sheet data 130ff
CMRR, lousy 51 DC current 21, 125
CMS 247, 275 DC current gain 130, 342
CM voltage 89, 103 DC resistance 101
C-multiplier(s) 239 DCS 195
Cohen, Graeme John 58 DC servo 108, 195ff, 269, 274, 372
coil(s) 101 DC voltage 125
coil resistance(s) 277 decade(s) 345
collector DDR 386
- current 25, 342 Denon 104, 168
- current, operating (DC) 132 density, noise current all chapters
- -emitter resistance 342 density, noise voltage all chapters
- noise current 130, 189 dependent, frequency 10
common Design + Design 169, 385
- cathode stage 191, 252 designer(s) 183
- emitter (CE) 341ff Deutsche Grammophon 168
- grid stage 28, 191, 252 Deutsche Grammophon (TCRs) 386ff
- mode gain 256 development examples 195ff
- mode input 372 deviation 24
- mode rejection ratio 6 deviation (calculated) 51
comparisons 3, 6 devices, amplifying 6
complementary (BJTs) 125ff Didden, Jan 23, 247
concertina phase splitter 191 DIF (amp) 21, 25, 64, 316f, 365ff
condition(s), equal 6 DIFA 365ff
connection, input 7 DIFA-IC 365
Connelly, J. A. 59, 393 DIFA-OPA 365ff
consideration(s), basic 6 DIFCF 26
concept, general 5 DIFCF, noise calculation 27f
configuration 5 DIFCF, SN calculation 27f
constant current sink 25 DIF gain 25ff
constants, physical xxv differential
Contents ix - amplifier 52, 365ff
converter, DA (NAD) 247 - gain 52f, 195, 258, 365ff, 365ff, 370
converter, un-balanced to balanced 105 - gain stage 21
cooling, resistor 309 - mode (fully) 195
corner frequencies (1/f-noise) 25, 58, 128ff - output 25. 54
correlated, noise voltage(s) 25, 28, 59, - o/p resistance 24
61ff, 102, 261 DIN 5, 126, 237f
Cordell, Bob 199ff DIN (TCRs) 386ff
CPS 191, 193 distortion 236ff
C-R-C chain 20 - artefacts 238
current, collector 25, 342 - level 5f
current, DC 21 - low- 6
current, emitter 189 - spike level 23
current gain, DC 130, 342 DL-103 (Denon) 104, 168
current gain, operating 132 DL-103 R 197ff
current generator 21 DMM 186
current sink (constant) 21, 25 DMM cut 5
cutting head 385 double-relay(s) 181
cutting lathe 385, 390
418 Index TSOS-E

double triode 14, 20f, 28, 169 flatness 24

draft design(s) 187ff flexibility 7
dual JFET 199 flicker-noise 128ff, 344
DUT 234ff, 241, 247ff, 252ff, 263, 277 Floru, Fred 58
FM tuners 385
Focal Press 125
E
Frederiksen, Thomas M. 393
E188CC 21 frequencies, high 7
E88CC 21ff, 175, 192 frequency dependent 10, all chapters
Early voltage 342 frequency independent 10, all chapters
ECC88 169, 175 frequency response(s) 5, 105ff, 165ff
ECM 168 Friedemann 168
effects, audible 167ff front 13f
effects, visible 165ff fully balanced vii
EIN 272f fully passive 5, 7
Elektor 394
electron charge 342
Electronics (Wireless) World 169, 394 G
Elektronik 49 gain
ELS 309ff - Amp1 102f
emitter current 189 - Amp2 125ff
emitter load(s) 137 - Amp3 25f
ENB 274 - Amp4 52ff
enclosure (PMMA) 269, 281 - Amp5 83f
encoder, RIAA 165, 168, 243 - balanced 26, 54
Engine I vii, 6, 21, 183 - BJT 347ff
Engine II viii, 3ff, 6, 183 - (BJT driven) 188f, 341ff
Engine II results, summary 176ff - block(s) 195
environment, low resistive 125 - CF 26f
equal condition(s) 6 - common mode 256
equalized, non- 23 - current 342
equivalent noise model (BJT) 343 - (Cordell) 199f
equivalent noise source(s) 130 - DIF 25ff
Ermer, Florian 170 - differential 52f, 195, 258, 365ff, 370
ESL 57 169 - (Gerhard) 195
euphonic harmonics 193 - idle (BJT) 347ff
EW 309 - loss 103
examples, development 195ff - -loss 251f
excess noise (resistor) 345 - noise 371
extension, TSOS vii - (noise reduction) 312, 316
external input viii, 5, 187 - nominal 5, 125
- operating (BJT) 348ff
- (PFMA) 275, 278
F
- (PMMA) 258
FC 63V 24 - (Popa) 197f
female, strip connectors 84 - overall 125, 249
FET(s) 170, 195ff, 254 - transformer 103
FFT diagram 20, 23, 248 - (transformer driven) 190
FFT resolution 234, 248, 266 - (triode driven) 194
FFT size 234, 248, 266 galvanically isolated 7, 275ff
Figures (listing) xxvii galvanic isolation 275
filter, measurement 266 GDR 386
fisher (company) 11 German Democratic Republic 386
flat 5 generator, current 21
Index TSOS-E 419

generator output resistance 103, 105ff input

Gerhard, Joachim 195ff, 370 - alternative 7
Gevel, Marcel van de 309 - amp 7
goals 3f - capacitance(s) 8f, 307, 313, 317
goal, worsening 254 - common mode 372
ground - connection 7
- lead, PSU 189 - current 49
- lift(s) 15, 278 - external viii, 5
- line 189 - load 23, 104
- loop 268, 275 - load, balanced 58
grounded grid 25 - load impedance 139
grounding 15 - reference level 191
gyrator(s) 239 - resistance(s) BJT 347ff
- resistance(s) 7, 103, 137, 249, 251ff
- resistors (alternatives) 141
H
- section 6
half (N, P) 133ff - stage, paralleled 256
harmonic(s) 241 - transformer 101, 181
harmonic artefacts 183 - voltage divider 51
harmonics, euphonic 193 - voltage, transformer 101
headphone 168f insertion case 11
heater PSU, triode 18 instrumentation amplifier(s) 53, 256, 357
heater supplies 20 integrated stabilizing circuit 20
Herbert, Gary K. 58 interference(s), hum 189, 266
HiFi equipment 248 intermodulation artefacts 183
high frequencies 7 intermodulation level 6
high-frequency (cut) 252 isolated, galvanically 7
high-quality 6 isolation, galvanic 275
high-Z 196, 199, 201 ITU-R 238
hints (PMMA) 264ff
Hoeffelman, Jean M. 307
HP21 195 J
HP5.1 197 JAES 307
HTGG-2 394 Jarrett, Keith 168
hum 6, 266 Jensen Transformers Inc. 84, 187,
hum-free 247 268, 275, 278
hum interference(s) 188, 266 JFET(s) 197ff
JFET driven 4, 197
JFET, dual 199
I
JJ 175
JT-10-KB-D 275
IC, measurement 256 JT-346-AXT 190
idle gain (DIF) 26 jumper(s) 262, 314
idle gain(s) BJT 347ff
IEC 468 248
K
IMD 6, 23, 51, 89, 169, 171ff, 181, 237f
impedance, input load 139 Karajan, Herbert von 168
improvements 269 Kempff, Wilhelm 168
INA 53, 256, 261 kernel (BJT) 130f
independent, frequency 10 Kirkwood, Wayne 58
Index TSOS-1 397 Klatt, Jo 169, 385
Index TSOS-2 403 Kraftwerk 168
Indices xv Krieger, Daniel 390
420 Index TSOS-E

L mains interferences 20
LA 274 main PCB 15ff
lathe, cutting 385, 390 margin, overload 5
LE 1 169 Mathcad worksheet(s) viii
LED 11 mathematical sizes xxv
left channel 11 Matsushita 181
level, average 248 Mayer-Schüller-Theory 170
level, output 83 MC cartridge load(s) 101ff
light-pattern measurement instrument 390 MC cartridge(s) viii, 3
linear amps 5 MC pre-amp 125
linear amps 187 MC purposes 5ff, 29, 187
linear amp stage(s) 5 measurement amplifier 247, 275
Linear Audio 15, 23, 170, 193, 195ff, measurement filter 266
247f, 395 measurement IC 256
Linkwitz 169 measurement instrument, light-pattern 390
listening test(s) 167ff measurement instrument, tracking force 385ff
list of figures xxvii Measurement tools 233ff
list of tables xxxvii Meys, René P. 307
LF411 188ff microphone amp(s) 49, 58, 180
LL9226, Lundahl 101, 191 middle control 181
LSK389 199 Miller capacitance 349ff
load, cathode 26 mix (noise levels) 183
load-effect 307 MKS 189
load resistor 254 MKT 189
load(s), emitter 137 MM cartridge(s) viii 4, 307ff
load synthesis 309 MM purposes 5, 29, 187
loss, gain 013 MM, standard model 307f
loudspeaker 169 mode, differential (fully) 195
loudspeaker arrangement(s) 168 model, operational (BJT) 354ff
loudspeaker chain(s) 167f model(s), noise (BJT) 356ff
low budget 249 Module 4 21
low-distortion 6 Module 2 125ff, 188, 341
low-end (audio band) 165 Mono record 168
low-hum 20 MST 170
low-noise 6, 20, 49 Motchenbacher, C. D. 59, 393
low-pass role 252 Musikverein (Vienna) 168
low resistive environment 125 mutual conductance 131, 193, 342
low-THD 49 mutual conductance, reduced 137
low-Z 196
LP vinyl record 5 N
LT783KC 20
N (noise) 236ff
LT1028 102, 256ff, 271, 275, 278
NAD M51 247, 274
LTE(s) 193, 198
Neef, Ulrich 389
LTP 201
network, R-C 101
Lundahl 103
network(s), RIAA 6
Lundahl LL9226 101, 191
Neumann 385
Neutrik 203
M NF(s) 29, 62, 104f, 274, 344
MA 247ff, 252ff, 275 NF charts 132ff
magnitude 9, 251 NF calculation 134ff
main board 6, 12, 20, 84 NF-picking 137, 140
main board wiring 19 NF values 132ff
mains connections 15 N-half 133ff
Index TSOS-E 421

noise operational gain (DIFA) 366f

- (BJT) 349ff operational model (BJT) 354ff
- calculation(s) 56ff, 86f, 127ff o/p resistance, differential 24
- calculation, DIFCF 27f o/p resistance(s) 24
- current all chapters oscillation, wild 109, 141
- current, collector 130 output
- factor 345 - anode 252, 254
- figure 29, 62, 344, 346 - cathode 252
- flicker- 128ff - differential 25, 54
- gain 371 - level 83
- index (resistor) 345ff - level, nominal 83
- low- 6 - load, transformer 101
- model(s), BJT 128ff, 343, 355ff - op-amp 252
- model (PMMA) 261ff - resistance 7f, 10, 29, 49, 243,
- model (triode driven) 194 247, 251ff, 272f
- pink 168ff, 175 - resistance, generator 103, 105ff,
- red 130 - resistance(s) (BJT) 347ff
- reduction 307ff - resistance(s), triode 7, 193
- results 170f - stage(s) 49, 200
- shot 130 - transformer 6, 83, 88, 165
- sources (DIFA-IC) 372 - un-balanced 83
- sources (DIFA-OPA) 368f -voltage swing 49, 52
- source(s), equivalent 130 - voltage, transformer 101
- test 168 overall gain(s) 125, 249
- voltage(s) 27f, 279 overload 312
- voltage density all chapters - (PMMA) 269
- voltage, correlated 25, 28, 61ff, - goal 7, 23
102, 261 - margin 5, 23, 49, 83
- voltage, un-correlated 28, 59f - question 7
- white 23, 128 - requirement 6
nominal gain 5, 125 - situation 126
nominal output level 83 - threat 126
nominal signal level 6 overview, Engine II 3
non-equalized 23
no-noise amp-stage 23
P
no-noise arrangement 62
non-shielded 181 pair, value 134
NOS 23, 175 Panasonic 24, 181
NPN 341 Panasonic FC 189
NPN-half 197 paralleled input stage 256
parallel operation 193
passive, fully 5ff
O Pauler acoustics 168
Occam's Razor 254, 313 PCB(s) 11ff, 84, 109, 141, 181
OP27 51, 102, 275 PCB, main 15ff
OPA1632 54 PCI 233
OPA627 6, 51, 255 PCS 52 E, BRAUN 385
OPA827 255 performance (PMMA) 266ff
op-amp driven 4 PFMA 240f, 266, 268, 275ff
op-amp output 252 P-half 133ff
op-amps, quadruple 49 pham, 2- 6
operating collector current (DC) 132 pham, 3- 7, 165
operating current gain 132 phase
operating gain (BJT) 348ff
422 Index TSOS-E

- response(s) 5, 105ff, 165ff record specifications 394

- shift 53, 365f record(s), test 168, 385ff
- splitter, cathodyne 191 recording studio 183
- splitter, concertina 191 red noise 130
piano 168 reduced mutual conductance 137
picture(s) 11ff reduction, noise 307ff
pink noise 168ff, 175 reference level, input 191
Pioneer 168 reference signal 390
phenomenon, transient 127, 141 regulated 5
phono-amps 6 relay(s) 6
physical constants xxv requirements 5
PL-L 1000 168 resistance(s)
plugged-in 12ff - base-emitter 342
plug-in 84 - base-spreading 130ff
PMMA 235, 240f, 247ff, 254ff, 264ff, 275 - collector-emitter 342
PNP 341 - input 7, 103, 137, 249, 251ff
point(s) 5 - coil 277
polypropylene 189 - output 7f, 10, 243, 247, 251ff, 272f
Popa, Ovidiu 197ff - output (BJT) 347ff
post-measurement work 233 resistor, cathode 254
pot, trimming 7 resistor cooling 309
power supply 5, 84f resistor, load 254
pre-amp(s) 29, 61, 181, 187ff response, frequency 5
pre-amp, MC 125 response, phase 5
pre-pre-amp 197 result(s) (draft designs) 201ff
primary 104 results, Engine II 176ff
printed circuit board(s) 12ff results, noise 170f
PSRR 269, 281 Revox A77 385
PSU 5, 11ff, 15, 167, 281 RIAA
- ground lead 189 - encoder 165, 168, 244
- (PMMA) 269 - equalized 5, 62, 140, 312, 317
- soli-state 16 - equalization 104
- triode 17 - network(s) 6, 24, 49
- triode heater 18 - transfer 104
psychology 184 - transfer function 5f, 26f, 54f
Putzey, Bruno 274 - voltage divider 51
Richter, Svatoslav 168
right channel 11
Q
Riley 169
Quad 169 ringing 141
quadruple op-amps 49 RMMA 274
Rohde und Schwarz (R&S) 247
R role, low-pass 252
roughly 195
Rams, Prof. 169, 387 R&S 254
ratio, common mode rejection 6 rule-of-thumb 195, 260
ratio, signal-to-noise 5
Razor, Occam's 254, 313
Razor (personal) 390 S
Razor (Vogel's) 184 S-AES17 248, 274
R-C network 101 sample rate 234, 266
rear 13ff Schaltungstips 49
record, Mono 168 Schenk, C. 393
record(s), calibration 385ff Schüller, Peter 170
Index TSOS-E 423

SC-02 233 standard Model, MM 307f

secondary 104 stereo (use) 11
section, center 6 stereoplay 247, 387
section, input 6 strip connectors, female 84
Seelmann, Prof. 170 studio, recording 183
Self, Douglas 125, 187f, 239, 248, 309, 393 sub-net 7
sengpielaudio 394 subtractor(s) 53, 56, 59f
sequence 7 subtraction stage 262
sequence, amp 51 subwoofer 169
sequence connected 9 succ-apps 137
servo, DC 108, 195ff, 269, 274, 372 successive approximation 137
shielding 266, 271 summing stage 262
shot noise 130 summary results, Engine II 176ff
Shure viii switchable 6
Shure V15 IV 385 switches 20
Shure V15 V 194 switching possibilities 13
sideband artefacts 238 symbols xvii
Siemens 169, 175 synthesis, load 309
signal chain 5 SYS 2722 248, 274
signal conditioner 233
signal level, nominal 6
T
signal path(s) 7, 9
signal-to-noise ratio(s) 5, 234ff, 346 tables (listing) xxxvii
single ended 101, 181 Tchaikovsky 168
single-ended intermediate vii TCR(s) 385ff
sink, current (constant) 21, 25 temperature (T) 130, 309
situation, overload 125 temperature, absolute 342
size(s), mathematical xxv test-board (case) 244f
slew rate 5, 181 test(s), listening 167ff
SLOA054D (TI) 371 test noise 168
slope 130ff test point(s) 125
slope figure 130ff test record(s) 168, 385ff
small signal 341f Texas Instruments 371
SME II 385 THAT Corp. 58
SMPTE RP120 238 THAT1646 275
SN 5 THD 6, 23, 89, 169, 171ff, 181, 236ff
SN (BJT) 346 THD + N 23
SN calculation(s) 60f, 86f, 137ff, 260ff, THD performance (UBC) 241ff
279f, 317 Third octave 168
SN calculation, DIFCF 27f thread, overload 126
SN improvement figure 23 TI Application 371
solid-state 7 Tietze, U. 393
solid-state PSU 16 time constant(s) 7f, 51
sound 6, 167ff tonearms 3
source(s) 6 tools, measurement 233ff
spike(s) 165f, 241, 266 Toshiba 124, 132
spike level, distortion 23 TQ2 181
SSM2142 275 tracking force (measurement instrument) 385ff
SSM2210 128 Trafo 23
SST Brüggemann 390 transconductance 13
stabilizing circuit 20 transconductance amplifier 199
stacked 20 transfer function, RIAA 5f, 26f, 55
standard cartridge 310ff, 317 transformer(s) 7, 182
424 Index TSOS-E

- driven 4, 187 valve driven 4

- gain 103 Vatter, Martin 168
- house 167 VC-A 199
- input 101, 181 VEB-Deutsche Schallplatten 389
- input voltage 101 velocity 5, 390
- output 6, 83, 88, 165 Vienna Symphony Orchestra 168
- output voltage 101 Vierfach Op-Amps 49
- output load 101 vinyl vii
transient phenomenon 127, 141 visible effects 165ff
transit frequency (BJT) 128 voltage divider 243, 252
treble control 181 - effect(s) 105, 193, 251
Trimming 233ff, 244ff - input 51
trimming possibilities 5 - RIAA 51
trimming procedure 191 voltage swing, output 52
trimming pot 7 volume knob 180
triode 254
- driven 187
W
- heater PSU 18
- output resistance 7 Walton, John 15
- PSU 17 white noise 23, 128, 247f, 341
- system(s) 21, 23 Whitlock, Bill 268
TSOS extension vii Whitt, S. 239
TSOS-1 3, 394 wild oscillation 109, 141
TSOS-2 3, 394 WIMA 189
tubecad 395 WIN 2k 233
tubedata 394 WIN 7 233
turns ratio(s) 102ff, 190 WIN XP 233
turntable(s) 3, 168 winding(s) 277
Tyler, Les 58 Wily-Interscience 59
Wireless World 239, 394
wiring, main board 19
U wobbling 168ff
un-balanced vii Worsening Figure 201, 234ff, 250f, 254, 272ff
- connectors viii worsening goal 254
- inputs 5
- output 83, 88f
X
- to balanced converter 105, 168
UBC 105, 165, 236ff, 239ff x (slope figure) 130ff
un-correlated, noise voltage 28, 59f XLR 203, 266, 277
Y
V y (slope figure) 132ff
value pair 134 Yaniger, Stuart 193
valve 7