2019 Book MicrowaveActiveDevicesAndCircu

Lecture Notes in Electrical Engineering 533
Subhash Chandra Bera
Microwave
Active Devices
and Circuits for
Communication
Lecture Notes in Electrical Engineering
Volume 533
Board of Series editors

Leopoldo Angrisani, Napoli, Italy
Marco Arteaga, Coyoacán, México
Bijaya Ketan Panigrahi, New Delhi, India
Samarjit Chakraborty, München, Germany
Jiming Chen, Hangzhou, P.R. China
Shanben Chen, Shanghai, China
Tan Kay Chen, Singapore, Singapore
Ruediger Dillmann, Karlsruhe, Germany
Haibin Duan, Beijing, China
Gianluigi Ferrari, Parma, Italy
Manuel Ferre, Madrid, Spain
Sandra Hirche, München, Germany
Faryar Jabbari, Irvine, USA
Limin Jia, Beijing, China
Janusz Kacprzyk, Warsaw, Poland
Alaa Khamis, New Cairo City, Egypt
Torsten Kroeger, Stanford, USA
Qilian Liang, Arlington, USA
Tan Cher Ming, Singapore, Singapore
Wolfgang Minker, Ulm, Germany
Pradeep Misra, Dayton, USA
Sebastian Möller, Berlin, Germany
Subhas Mukhopadhyay, Palmerston North, New Zealand
Cun-Zheng Ning, Tempe, USA
Toyoaki Nishida, Kyoto, Japan
Federica Pascucci, Roma, Italy
Yong Qin, Beijing, China
Gan Woon Seng, Singapore, Singapore
Germano Veiga, Porto, Portugal
Haitao Wu, Beijing, China
Junjie James Zhang, Charlotte, USA
Lecture Notes in Electrical Engineering (LNEE) is a book series which reports the
latest research and developments in Electrical Engineering, namely:
• Communication, Networks, and Information Theory

• Computer Engineering
• Signal, Image, Speech and Information Processing
• Circuits and Systems
• Bioengineering
• Engineering
The audience for the books in LNEE consists of advanced level students,
researchers, and industry professionals working at the forefront of their fields. Much
like Springer’s other Lecture Notes series, LNEE will be distributed through
Springer’s print and electronic publishing channels.
More information about this series at http://www.springer.com/series/7818

Microwave Active Devices

and Circuits
for Communication
123
Space Applications Centre
Indian Space Research Organization (ISRO)
Ahmedabad, Gujarat, India
ISSN 1876-1100 ISSN 1876-1119 (electronic)

Lecture Notes in Electrical Engineering
ISBN 978-981-13-3003-2 ISBN 978-981-13-3004-9 (eBook)
https://doi.org/10.1007/978-981-13-3004-9
Library of Congress Control Number: 2018958497
© Springer Nature Singapore Pte Ltd. 2019

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. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
To Him
who gives me inspiration and patience
Foreword
It has been a pleasure for me to see this book titled ‘Microwave Active Devices and
Circuits for Communication’ written by Subhash Chandra Bera based on his long
experience on the design and development of on-board communication circuits and
systems at Space Applications Centre, ISRO, Ahmedabad.
This book provides extensive coverage in the field of microwave engineering for
graduate students, practical circuit designers and researchers. This book begins with
basics of device physics and ends with the design of microwave communication
systems through detailed design, analysis and realization of different circuits and
systems. Apart from classical topics in microwave active devices such as p-i-n
diode, Schottky diode, step recovery diode, BJT, HBT, MESFET, HFET and
various microwave circuits such as switch, phase shifter, attenuator, amplifier,
multiplier, mixer, the coverage extends to some modern topics such as Class-F
power amplifier, direct frequency modulator, linearizer, equalizer. The written text
of each topic is supplemented with suitable diagrams, and each chapter has various
types of solved problems for clear understanding.
It is hoped that this book will be very useful for graduate students, circuit
designers and researchers.
Ahmedabad, India Tapan Misra

May 2018 Distinguished Scientist
Director, Space Applications
Centre (ISRO)
vii
Preface
This book is about active devices and circuits for microwave communications
appropriate for undergraduate and postgraduate students, practical circuit designers
and researchers in the field of electronics and communication engineering. It pre-
sents the design and analysis of various linear and nonlinear circuits for microwave
communication systems after discussing the working principle and behaviour of
microwave diodes and transistors. This book begins with the basics of device
physics and ends with the design of microwave communication systems through
detailed design, analysis and realization of different circuits and systems. Though
this book is written focusing on microwave communications, much of the material
of this book is entirely generic and will be useful for other microwave applications.
This book starts with describing the diodes which are omnipresent in all
solid-state circuits across all the frequency ranges starting from DC to terahertz even
extending to optical frequency range as an isolated diode or as a part of other
devices. Proper understanding about diodes leads to ease of understanding about
other devices such as different types of bipolar and unipolar transistors in their
homo-junction and hetero-junction structures which are discussed subsequently.
Schottky, p-i-n, step recovery and tunnel diodes are popularly used in linear and
nonlinear microwave communication circuits such as variable attenuator, power
limiter, phase shifter, linearizer, frequency multiplier, mixer for their high-frequency
performance and simplicity in operation. With the advancement of material tech-
nologies, there are various modern transistors such as hetero-junction bipolar and
field effect transistors based on GaAs, InP and GaN technologies. These devices
offer very high-frequency performance with low noise figure and high-power-
handling capabilities with different device structures and circuit configurations. DC
biasing circuits for the diodes and transistors are very relevant which determine the
RF performance including temperature behaviour of the circuits. This book also
addresses the temperature behaviour and compensation mechanism of the micro-
wave circuits for practical applications.
ix
x Preface
This book has evolved from the author’s design and development experience on
microwave circuits and systems for Indian national satellite communication systems
at Space Applications Center, Indian Space Research Organization (ISRO), India.
This book is believed to be a useful one for students and microwave circuit
designers. Any suggestions for improvement of the quality of this book would be
highly appreciated.
Ahmedabad, India Subhash Chandra Bera

Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Microwave Communications . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Microwave Active Circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Microwave Active Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4 Microwave Circuit Analysis and Measurements . . . . . . . . . . . . 9
1.5 Book Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2 P-I-N Diode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Basics of P-I-N Diode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3 P-I-N Diode Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.4 Nonlinearity of P-I-N Diode . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.5 Temperature Behaviour of P-I-N Diode . . . . . . . . . . . . . . . . . . 20
2.6 Temperature-Invariant RF Resistance of P-I-N Diode . . . . . . . . 22
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3 Schottky Diode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2 Basics of Schottky Diode . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.3 Schottky Diode Characteristics . . . . . . . . . . . . . . . . . . . . . . . . 36
3.4 Temperature Behaviour of Schottky Diodes . . . . . . . . . . . . . . . 39
3.5 Temperature Invariant RF Resistance . . . . . . . . . . . . . . . . . . . . 41
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4 Special Microwave Diodes . ....... . . . . . . . . . . . . . . . . . . . . . . . . 47
4.1 Introduction . . . . . . . ....... . . . . . . . . . . . . . . . . . . . . . . . . 47
4.2 Step Recovery Diode ....... . . . . . . . . . . . . . . . . . . . . . . . . 47
4.2.1 Characteristic of SRD . . . . . . . . . . . . . . . . . . . . . . . . . 49
xi
xii Contents
4.3 Tunnel Diodes . . . . . .............. . . . . . . . . . . . . . . . . . 51

4.3.1 Characteristic of Tunnel Diode . . . . . . . . . . . . . . . . . . 52
4.4 Backward Diode . . . . .............. . . . . . . . . . . . . . . . . . 54
4.5 Varactor Diode . . . . . .............. . . . . . . . . . . . . . . . . . 56
References . . . . . . . . . . . . . .............. . . . . . . . . . . . . . . . . . 57
5 Microwave Bipolar Transistors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.2 Bipolar Junction Transistor (BJT) . . . . . . . . . . . . . . . . . . . . . . 60
5.2.1 Frequency Limitation of BJT . . . . . . . . . . . . . . . . . . . 65
5.2.2 Temperature Behaviour of BJT . . . . . . . . . . . . . . . . . . 69
5.3 Hetero-junction Bipolar Transistor (HBT) . . . . . . . . . . . . . . . . . 70
5.3.1 SiGe HBT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.3.2 III–V Group Semiconductor HBT . . . . . . . . . . . . . . . . 74
5.3.3 GaN HBT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6 Microwave Field Effect Transistors . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.2 Metal–Semiconductor Field Effect Transistors . . . . . . . . . . . . . . 81
6.3 Hetero-Structure Field Effect Transistors (HFETs) . . . . . . . . . . . 85
6.3.1 High-Electron-Mobility Transistors (HEMTs) . . . . . . . . 87
6.3.2 Pseudo-morphic HEMTs (pHEMTs) . . . . . . . . . . . . . . 90
6.3.3 Meta-morphic HEMTs (mHEMTs) . . . . . . . . . . . . . . . 92
6.4 Microwave GaN HEMTs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6.5 Equivalent Circuit of Microwave FETs . . . . . . . . . . . . . . . . . . 94
6.5.1 Transconductance Gain (gm ) . . . . . . . . . . . . . . . . . . . . 95
6.5.2 Output Conductance (1/rds ) . . . . . . . . . . . . . . . . . . . . . 97
6.5.3 Gate–Source and Gate–Drain Capacitances
(Cgs and Cgd ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
6.5.4 Charging Resistance (Ri ) . . . . . . . . . . . . . . . . . . . . . . . 101
6.6 Maximum Frequency of Operation . . . . . . . . . . . . . . . . . . . . . . 107
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
7 Microwave Circuit Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
7.2 Transmission Line Theory and Analysis . . . . . . . . . . . . . . . . . . 112
7.3 Microwave Transmission Lines . . . . . . . . . . . . . . . . . . . . . . . . 120
7.3.1 Losses in Transmission Lines . . . . . . . . . . . . . . . . . . . 120
7.3.2 Coaxial Transmission Lines . . . . . . . . . . . . . . . . . . . . 123
7.3.3 Waveguides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
7.3.4 Cut-Off Frequency and Guide Wavelength . . . . . . . . . . 133
7.3.5 Planar Transmission Lines . . . . . . . . . . . . . . . . . . . . . 139
7.4 Transmission Line Elements . . . . . . . . . . . . . . . . . . . . . . . . . . 147
Contents xiii
7.5 Smith Chart Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

7.6 Network Theory of Circuits and Transmission Lines . . . . . . . . . 158
7.6.1 S-Parameter Network Representation . . . . . . . . . . . . . . 159
7.6.2 S-Parameters for Matched Reciprocal Lossless
Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
7.6.3 ABCD Parameter for Network Representation . . . . . . . 174
7.6.4 Conversion in Between ABCD and S-Parameters . . . . . 177
7.7 Power Transfer in Microwave Networks . . . . . . . . . . . . . . . . . . 179
7.7.1 Power Transfer from Source to a Load . . . . . . . . . . . . 179
7.7.2 Power Transfer to and from a 2-Port Network . . . . . . . 181
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
8 Microwave Switches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
8.2 Switch Circuits Based on P-I-N Diode . . . . . . . . . . . . . . . . . . . 200
8.3 Series Switch Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
8.4 Shunt Switch Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
8.5 Compound Switch Configuration . . . . . . . . . . . . . . . . . . . . . . . 209
8.5.1 Series–Shunt Configuration . . . . . . . . . . . . . . . . . . . . . 209
8.5.2 TEE Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
8.6 Compound Switch Analysis Using ABCD Parameter . . . . . . . . 217
8.7 Switch Circuits Based on FETs . . . . . . . . . . . . . . . . . . . . . . . . 219
8.8 Applications of RF/Microwave Switches . . . . . . . . . . . . . . . . . 221
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
9 Microwave Attenuators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
9.2 Diode-Based Attenuator Circuits . . . . . . . . . . . . . . . . . . . . . . . 236
9.3 Series- or Shunt-Connected Element . . . . . . . . . . . . . . . . . . . . 237
9.4 Multiple Shunt-Connected Element . . . . . . . . . . . . . . . . . . . . . 240
9.5 Matched Attenuator Circuits . . . . . . . . . . . . . . . . . . . . . . . . . . 244
9.5.1 TEE Attenuator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
9.5.2 p Attenuator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
9.5.3 Quadrature Hybrid Matched Attenuator . . . . . . . . . . . . 251
9.6 Driver Circuit for P-I-N Diode Attenuators . . . . . . . . . . . . . . . . 255
9.7 Effect of Nonideal Components in Driver Circuits . . . . . . . . . . 259
9.8 Experimental Determination of VOPTM . . . . . . . . . . . . . . . . . . . 262
9.9 FET-Based Attenuators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
10 Microwave Phase Shifters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
10.2 Phase Shift and Time Delay . . . . . . . . . . . . . . . . . . . . . . . . . . 279
xiv Contents
10.3 Types of Phase Shifter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281

10.4 Realization of Phase Shifters . . . . . . . . . . . . . . . . . . . . . . . . . . 283
10.4.1 Switched Transmission Line Phase Shifter . . . . . . . . . . 283
10.4.2 Varactor Diode-Based Analog Phase Shifters . . . . . . . . 285
10.4.3 Four-Quadrant Continuously Variable Phase
Shifter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
10.4.4 Four-Quadrant Digital Phase Shifter . . . . . . . . . . . . . . 292
10.5 Applications of RF/Microwave Phase Shifter . . . . . . . . . . . . . . 293
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294
11 Microwave Modulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
11.2 Amplitude Modulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
11.3 Phase Modulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
11.3.1 Bi-Phase Modulators . . . . . . . . . . . . . . . . . . . . . . . . . 298
11.3.2 Bi-Phase-Balanced Modulators . . . . . . . . . . . . . . . . . . 300
11.4 I–Q Vector Modulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
11.5 PSK Modulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
11.6 QAM Modulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
12 Amplitude Tilt Microwave Equalizers . . . . . . . . . . . . . . . . . . . . . . . 307
12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
12.2 Different Configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
12.3 Equalizer with Adjustable Amplitude Slope . . . . . . . . . . . . . . . 311
12.4 Equalizer with Adjustable Parabolic Gain Slope
for Broadband MPM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
12.5 Equalizer with Adjustable Positive Gain Slopes
for Solid-State Circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
12.6 Equalizer with Adjustable Positive as Well as Negative
Gain Slopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
12.7 Versatile Equalizer with Variable Gain Slope
and Insertion Loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
13 Microwave Detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
13.2 Microwave Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334
13.3 Diode Detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
13.4 RMS Power (Average) Detector . . . . . . . . . . . . . . . . . . . . . . . 346
13.5 Envelope and Peak Power Detector . . . . . . . . . . . . . . . . . . . . . 347
13.6 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
Contents xv
14 Microwave Solid-State Amplifiers . . . . . . . . . . . . . . . . . . . . . . . . . . 349

14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
14.2 Types of Microwave Amplifiers . . . . . . . . . . . . . . . . . . . . . . . . 351
14.2.1 Dynamic Range of Amplifier . . . . . . . . . . . . . . . . . . . 353
14.2.2 Spurious-Free Dynamic Range . . . . . . . . . . . . . . . . . . 358
14.3 Stability of Microwave Amplifier . . . . . . . . . . . . . . . . . . . . . . . 363
14.4 Single-Stage Amplifier Design . . . . . . . . . . . . . . . . . . . . . . . . . 376
14.4.1 Amplifiers Using Unconditionally Stable Device . . . . . 377
14.4.2 Amplifier Using Conditionally Stable Device . . . . . . . . 380
14.5 Amplifier with Specific Gain . . . . . . . . . . . . . . . . . . . . . . . . . . 382
14.5.1 Amplifier with Specific Transducer Power Gain . . . . . . 382
14.5.2 Amplifier with Specific Available Power Gain . . . . . . . 386
14.5.3 Amplifier with Specific Operating Power Gain . . . . . . . 394
14.6 Small Signal Amplifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 400
14.6.1 Low Noise Amplifier (LNA) Design . . . . . . . . . . . . . . 400
14.6.2 High-Gain Amplifier Design . . . . . . . . . . . . . . . . . . . . 411
14.7 Large Signal Amplifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 420
14.7.1 Linear Power Amplifier . . . . . . . . . . . . . . . . . . . . . . . 424
14.7.2 Load-Pull Characterization Technique . . . . . . . . . . . . . 434
14.7.3 Amplifiers with Reduced Conduction Angle . . . . . . . . 435
14.7.4 Nonlinear Power Amplifiers . . . . . . . . . . . . . . . . . . . . 445
14.7.5 Class-F Power Amplifier . . . . . . . . . . . . . . . . . . . . . . . 451
14.8 FETs Output Power Capability . . . . . . . . . . . . . . . . . . . . . . . . 459
14.8.1 Microwave Power Combing Techniques . . . . . . . . . . . 463
14.8.2 In-Phase Power Combiners . . . . . . . . . . . . . . . . . . . . . 464
14.8.3 Balanced Power Combining . . . . . . . . . . . . . . . . . . . . 469
14.9 Temperature Compensation of Microwave Amplifiers . . . . . . . . 476
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 476
15 Microwave Limiters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
15.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
15.2 Limiter Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 480
15.3 P-I-N Diode Limiters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 482
15.4 Schottky Diode Limiters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487
15.5 Amplifier-Based Limiter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 492
15.6 Closed-Loop Limiter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494
15.6.1 Temperature Behaviour of OLC System . . . . . . . . . . . 495
15.6.2 Temperature Compensation of OLC System . . . . . . . . . 496
15.7 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
15.7.1 OLC for Out-of-Band Carrier Mitigation
in Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 498
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 499
xvi Contents
16 Microwave Linearizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501

16.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501
16.2 Types of Distortion and Amplifier Nonlinearity . . . . . . . . . . . . 502
16.3 Types of Linearizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 504
16.3.1 Feedback Linearizers . . . . . . . . . . . . . . . . . . . . . . . . . 504
16.3.2 Feedforward Linearizers . . . . . . . . . . . . . . . . . . . . . . . 505
16.3.3 Predistortion Linearizers . . . . . . . . . . . . . . . . . . . . . . . 506
16.4 Implementation of Predistortion Linearizers . . . . . . . . . . . . . . . 507
16.4.1 Schottky Diode Predistortion Linearizers . . . . . . . . . . . 508
16.4.2 Linearizer with Vector Modulator Configuration . . . . . 513
16.4.3 Temperature Behaviour of Diode Linearizer . . . . . . . . . 517
16.4.4 Broadband Linearizers . . . . . . . . . . . . . . . . . . . . . . . . 520
16.5 Digital Predistortion Linearizers . . . . . . . . . . . . . . . . . . . . . . . . 523
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525
17 Microwave Frequency Multipliers . . . . . . . . . . . . . . . . . . . . . . . . . . 527
17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 527
17.2 Principle of Multiplier Operation . . . . . . . . . . . . . . . . . . . . . . . 528
17.3 Diode Multiplier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533
17.3.1 Schottky Diode Frequency Multiplier . . . . . . . . . . . . . 533
17.3.2 Varactor Diode Frequency Multiplier . . . . . . . . . . . . . . 535
17.3.3 SRD Frequency Multiplier . . . . . . . . . . . . . . . . . . . . . 537
17.4 Transistor Multiplier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 540
17.5 Realization of FET Multipliers . . . . . . . . . . . . . . . . . . . . . . . . . 549
17.6 Balanced Frequency Multipliers . . . . . . . . . . . . . . . . . . . . . . . . 550
17.6.1 Balanced Frequency Multipliers Using Diodes . . . . . . . 551
17.6.2 Balanced Frequency Multiplier Using
Transistors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 552
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 553
18 Microwave Frequency Mixers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 555
18.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 555
18.2 Working Principle of Frequency Mixers . . . . . . . . . . . . . . . . . . 555
18.2.1 Image Frequency in Mixer . . . . . . . . . . . . . . . . . . . . . 558
18.2.2 Conversion Loss of Frequency Mixer . . . . . . . . . . . . . 560
18.2.3 Noise Performance of Frequency Mixer . . . . . . . . . . . . 562
18.3 Single-Ended Mixers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565
18.4 Balanced Mixers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 570
18.4.1 Single-Balanced Mixers . . . . . . . . . . . . . . . . . . . . . . . 570
18.4.2 Double-Balanced Mixers . . . . . . . . . . . . . . . . . . . . . . . 574
18.4.3 Image Rejection Mixers . . . . . . . . . . . . . . . . . . . . . . . 575
18.5 Subharmonic Mixers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 576
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 580
Contents xvii
19 Microwave Communication Systems . . . . . . . . . . . . . . . . . . . . . . . . 583

19.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 583
19.2 Mobile Communication Systems . . . . . . . . . . . . . . . . . . . . . . . 584
19.2.1 Receiver Architecture for Mobile
Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 585
19.2.2 Transmitter Architecture for Mobile
Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 588
19.2.3 Transceiver for Mobile Communication . . . . . . . . . . . . 589
19.3 Satellite Communication Systems . . . . . . . . . . . . . . . . . . . . . . 593
19.4 Receiver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 606
19.4.1 Local Oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 610
19.5 Satellite Transmitter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 612
19.5.1 Driver Amplifier (DA) . . . . . . . . . . . . . . . . . . . . . . . . 612
19.5.2 Travelling Wave Tube Amplifier (TWTA) . . . . . . . . . . 616
19.5.3 Solid-State Power Amplifier (SSPA) . . . . . . . . . . . . . . 618
19.6 Linearizer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 620
19.7 Microwave Power Module (MPM) . . . . . . . . . . . . . . . . . . . . . . 622
19.8 Multiport Amplifier (MPA) . . . . . . . . . . . . . . . . . . . . . . . . . . . 623
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 645
20 Multiple Choice Questions with Answers . . . . . . . . . . . . . . . . . . . . 647
20.1 Multiple Choice Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . 647
20.2 Answers of MCQs with Explanations . . . . . . . . . . . . . . . . . . . . 678
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 689
About the Author
Subhash Chandra Bera received his B.Sc. degree

(with honours) in physics from Presidency College,
Calcutta, and B.Tech. and M.Tech. degrees in radio
physics and electronics from the Institute of Radio Physics
and Electronics, University of Calcutta. He also received a
Ph.D. degree in microwave and antenna engineering from
Gujarat University, India. Since 1994, he has been with the
Space Applications Centre, Indian Space Research
Organization (ISRO), Ahmedabad, India, where he has
been involved in the design and development of
microwave circuits and systems for various INSAT and
GSAT series of Communication Payload projects as well
as GAGAN/IRNSS Navigation Payload projects. He has
developed several state-of-the-art microwave subsystems
that are operational in various Indian national satellites.
His research interests include microwave active circuits in
general and solid-state power amplifiers, channel ampli-
fiers, linearizers, limiters, attenuator and equalizers in
particular for spacecraft use. Currently, he is Division
Head of Satcom and Navigation Systems Engineering
Division, Space Applications Centre (ISRO), Ahmedabad,
and Associate Project Director of GSAT-24 and GSAT-31
Communication Payloads. He has authored more than 50
research publications in international journals and presented
numerous papers at national and international conferences
and symposiums. He has also delivered invited talks in the
field of microwave active devices and circuits at various
workshops and symposiums. He has also been granted four
patents.
xix
Chapter 1
Introduction
Microwave frequency region is generally considered from 1–30 GHz in electro-

magnetic spectrum which corresponds to a free-space wavelength (λo ) of 30–1 cm,
respectively. Frequencies within the range of 30–300 GHz (λo 10−1 mm) are
considered as mmwave frequency band and submillimetre wave frequency band is
considered beyond 300 GHz (λo <1 mm). The microwave and higher frequency cir-
cuits and systems show characteristics which are not seen in the lower frequency
range. The differences in characteristics are due to the smaller wavelength of the
microwave and higher frequency range which becomes comparable to the size of the
circuit elements, i.e. size of components (resistors, capacitors, inductors, diodes and
transistors) and even the transmission line elements. Due to the smaller wavelength
compared to the circuit elements, the phase of the signal varies along the dimension
of the elements, consequently, the circuit design and analysis techniques, power gen-
eration, amplification as well as measurement techniques are different from those at
lower frequencies. Thus, materials, structures of the active and passive components,
topology, and realization of circuits at microwave frequency are different from those
at lower frequencies. Microwave frequencies are used in communication, naviga-
tion, remote sensing, radar applications, industrial measurements, and even heating
and drying of food products. This book will focus on active devices and circuits for
microwave communications. However, much of the materials covered by the book
are entirely generic and will be useful for other applications.
1.1 Microwave Communications
Microwave frequencies have distinct advantages over lower frequencies due to its
higher percentage bandwidth, higher directive coverage and higher penetration capa-
bility through the atmosphere. It is easy to design a circuit with 10% bandwidth, for
example, to get 1 GHz bandwidth around 10 GHz centre frequency. Whereas, it is
very difficult to design a circuit of 200% bandwidth to achieve the same bandwidth
© Springer Nature Singapore Pte Ltd. 2019 1
S. C. Bera, Microwave Active Devices and Circuits for Communication, Lecture Notes
in Electrical Engineering 533, https://doi.org/10.1007/978-981-13-3004-9_1
2 1 Introduction
around 500 MHz centre frequency. Another advantage for microwave communi-
cation is the ability to use smaller aperture antenna with high directivity because
antenna beam-width is inversely proportional with the ratio of antenna aperture size
to the wavelength of the signal to be transmitted/received. Due to the availability
of higher percentage bandwidth and highly directive coverage, it is possible for
very high data rate communication of the order of multi-Gbps. With the maturity of
microwave technology, it is used as the major trunk channel for long distance com-
munication. Microwave communication is easier over difficult terrain and removes
the requirement of land acquisition as required for cable systems. Infrastructure
needs for short-term communication systems are possible without the high cost of
installing permanent network cabling that will only be required for a limited time.
These are the reasons for using microwave communication in spite of the difficulty in
implementation of microwave circuits and limitation of communications within the
line-of-sight range. As the lower frequency bands become congested, the communi-
cation systems need to shift towards higher and higher frequency bands to meet the
capacity and data rate targets. Higher microwave and millimetre wave frequency
bands are less congested and can potentially provide multi-gigahertz spectrum
[1, 2]. This is the reason that satellite communication systems already using the
higher microwave and mmwave frequency bands for high throughput applications. It
is also envisaged that the next-generation (5G) mobile communication system needs
to use higher microwave and mmwave frequency bands to meet the requirement of
spectrum [3].
Block diagram of a wireless communication system such as mobile communi-
cation is shown in Fig. 1.1. It consists of a transmitter, a receiver, and a channel.
RF section of the transceiver mainly consists of low-noise amplifier (LNA), power
amplifier (PA), up- and downconverters, IF amplifiers, phase shifters, modulators,
switches, and filters. Signal quantization, coding and decoding are performed in the
digital domain. Practically, at each end of the link, a transmitter and a receiver, i.e.
integrated transceiver is used. In a transmitter, the signal modulates a carrier using a
particular modulation scheme and transmitted through a transmit antenna, whereas
the receiver recovers the information from the received signal through the receiving
antenna. In case of mobile communication system, at least one of the transceivers is
mobile. It may be a handheld unit used by a pedestrian or at on-board a vehicle that
can move at high speed.
The transmitter consists of a modulator, an upconverter, filters and PA. Baseband
signals are modulated by the modulator then filtered, amplified and upconverted by
the upconverter to a higher frequency carrier. The upconverted signal further ampli-
fied, filtered and finally amplified by a PA and transmitted by the transmitting antenna.
The receiver consists of a low-noise amplifier, downconverter, bandpass filters and
demodulator. The receiver shall perform the signal amplification and filtering func-
tionality to meet the system sensitivity, channel selectivity, spurious response and
power control requirements to meet the required signal-to-noise ratio.
Unlike lower frequencies, microwave frequencies are not reflected and practically
not absorbed by the ionosphere on account of their small wavelength, and thus they
penetrate the atmosphere without significant attenuation except at certain absorp-
1.1 Microwave Communications 3
Fig. 1.1 Block diagram of a mobile communication system
tion bands of atmospheric gases & subject to clear weather. This leads to the use
of microwave frequencies for satellite communication. Presently, microwave fre-
quencies are widely used in satellite communication for telephony, broadcasting,
television systems and various data communication systems. Satellite microwave
communication is also very useful for disaster-prone regions due to the possibil-
ity of quick and easy establishment of ground infrastructure considering the prior
existence of space segment for satellite communication. Block diagram of a satel-
lite microwave communication system is shown in Fig. 1.2. The system consists of
ground and space segments. Both the segments consist of transmitter, receiver and
antenna systems like any other communication systems.
A ground transmitter section starts with the processing of information to be sent
which is then suitably modulated. The filtered modulated signal is then upconverted
(frequency translated) to microwave frequency by using a mixer and followed by
several stages of amplification. Finally, the microwave signal is amplified by a high
power amplifier (HPA) and transmitted by a highly directive antenna towards satellite.
Generally, the high power amplifier operates in power back-off condition to minimize
the addition of noise due to nonlinearities of the power amplifier. Sometimes, a
linearizer is used before the HPA to improve the linearity of the transmitter. In general,
the signal is modulated by a low-frequency carrier and then upconverted to microwave
frequency as shown in the block diagram. Thus, the upconverter and amplifier section
operate in microwave frequency band. However, the direct modulation scheme also
can be used to modulate the microwave carrier frequency directly by the baseband
signal. In this case modulator, upconverter and amplifier sections operate in the
microwave frequency band.
4 1 Introduction
Fig. 1.2 Block diagram of satellite microwave communication system
The space segment is a repeater in space with a sizable amount of

gain and frequency translation, sometimes with signal processing also. It
consists of receiving antenna followed by a preselect filter to select only the desired
frequency band transmitted by the ground transmitter and rejecting undesired fre-
quency bands. Since the received signal is very feeble (of the order of pW), the signal
is first amplified by a low-noise amplifier (LNA) and then downconverted using a
mixer circuit. The output of the downconverter is filtered and processed by a signal
processing (S/P) unit. The signal processing unit may be a transparent or regenerative
processor. Transparent processing is only to provide flexibility in frequency, band-
width and signal routing at lower Intermediate Frequency (IF) level for multichannel
multibeam transponders. Regenerative processor extracts the information by using
a demodulator and then processed. The processed signal is then remodulated before
being upconverted to the transmit microwave frequency. The upconverted microwave
signal is subsequently amplified by a driver amplifier (DA) and high power amplifier
(HPA) before being transmitted by a transmit antenna. A predistortion linearizer is
1.1 Microwave Communications 5
used to improve the linearity of a satellite transmitter which is cascaded at the input
of the high power amplifier.
The ground receiver consists of receive antenna followed by a low-noise ampli-
fier. The amplified signal is downconverted to an intermediate frequency (IF) to ease
the signal demodulation and processing for information extraction. Thus, a typical
microwave communication system consists of various diode and transistor-based
microwave circuits and systems such as modulator, demodulator, receiver, IF ampli-
fiers, high power amplifier, linearizer, converters, etc.
1.2 Microwave Active Circuits
In microwave communication systems, various types of amplifiers are used in receive

and transmit sections. A microwave amplifier takes the signal of low or intermedi-
ate level as its input over a range of frequency band and significantly boosts its
power level. In addition to amplification of the desired signal, a practical amplifier
introduces noise generated within it and also distorts the signal. Based on the signal
power level to be amplified, there are three types of microwave amplifiers: low-noise
amplifier (LNA), high gain amplifier and high power amplifier (HPA). Low-noise
amplifiers are used as the front end of a receiver where the strength of the desired
signal is very low. High gain amplifiers are used in the intermediate stages of a trans-
mitter–receiver section to boost the signal strength to the desired level. The high
power amplifiers (HPAs) are used as the final power amplifying unit of a transmitter.
At low frequencies, the design of an amplifier involves mainly designing a proper DC
bias circuit for maximum power output. However, at microwave frequency range, in
addition to the design of appropriate DC bias circuit, it involves the design of input
and output matching networks based on the requirements of noise figure, gain and
output power over the desired frequency band.
The linearity of the amplifier plays a major role in a communication system.
Ideally, throughout the receive–transmit (repeater) chain, linear amplification of mod-
ulated signal is required to accurately decode the information. Amplifiers’ nonlinear-
ity becomes more severe when higher order modulations such as 8APSK, 16APSK,
16QAM, etc. are used to achieve improved spectrum efficiency. Problem arises when
linear amplification is required throughout the chain where power amplifiers are used.
To operate the power amplifiers in their linear region, it is required to operate them
in power back-off condition which reduces the DC to RF efficiency. The solution is
the use of amplifier near its saturation condition and use a linearizer to improve the
overall linearity of the transmitter section. There are various types of linearization
techniques used in baseband as well as at RF and IF level. At the baseband level, dig-
ital linearization technique is used. However, at RF level, diode-based predistortion
linearization technique is mostly used due to its low power consumption, broadband
performance and compact size.
Various sensitive microwave devices are used in receivers, solid-state power
amplifiers, detectors, etc. Limiter circuits are used to protect sensitive devices from
6 1 Introduction
intentionally or unintentionally exposure of high power levels. Basically, two types of

limiter circuits are used, one is open-loop limiter and another is closed-loop limiter.
Limiting device of an open-loop limiter works with self-bias condition. Whereas
closed-loop limiter circuit uses a separate detector circuit to detect the RF power
level. Then, the detected signal is applied to other device such as voltage-variable
amplifier/attenuator to limit the power level in a closed-loop configuration. Schottky
diodes, p-i-n diodes as well as various types of Field Effect Transistors (FETs) are
the most suitable limiting devices for microwave applications.
Microwave signal power level is an important performance parameter of any
communication system. At lower frequency range, signal power is calculated from the
measured voltage and current. Direct measurement of power is more convenient and
accurate starting from the microwave frequency range to optical spectrum. Thus, as
the frequency increases, power level is measured from which voltage and current are
calculated. Microwave power detectors are used to detect/measure the power level.
Schottky barrier and tunnel diodes are also popularly used as microwave power level
detectors. Depending upon the signal structure and applications; average, envelop,
peak or pulse powers are measured.
Mixers are used in up- and downconverters to convert/translate signals in fre-
quency domain. They use a stable local oscillator to translate the spectrum containing
information to other frequency band. It is fundamentally a multiplier having three
ports: a radio frequency (RF) port, a local oscillator (LO) port and an intermediate
frequency (IF) port. In practice, the LO signal is CW, whereas the RF and IF signals
are modulated carriers. Generally, Schottky barrier diodes and various types of FETs
and Bipolar Junction Transistors (BJTs) are commonly used as nonlinear device for
realization of microwave mixers.
Frequency multipliers are used to realize microwave frequency source, e.g. local
oscillator (LO) of higher frequency from a low-frequency standard. Performance
of microwave solid-state oscillator degrades with increasing frequency. To over-
come this problem, highly stable oscillators operating at a lower frequency are used
followed by frequency multipliers with low conversion loss. The basic realization
principle for frequency multiplication is the generation of multiple higher frequen-
cies by distorting (amplitude or phase) a sinusoidal signal and then filtering out the
desired higher frequency component. The distortion is realized using nonlinear char-
acteristic of semiconductor devices. Varactor diodes, Schottky barrier diodes (SBD),
step recovery diodes (SRD) and bipolar junction transistors (BJTs) or field effect
transistors (FETs) are used as nonlinear device.
Microwave and RF amplifiers are used in many broadband applications such as
meeting the increasing demand for Internet and multimedia applications. Modern
microwave receivers use equalizers to compensate the excess passband negative
gain slope due to the reduction in gain of transistors as the frequency increases.
Amplitude tilt equalizer provides predetermined gain/loss slope over an operating
range of frequency. It is mainly used in cascade with other circuits or systems to
improve the overall frequency response. Microwave-variable gain slope equalizers
are used for the compensation of gain roll-off of microwave and RF circuits and
systems using the provision of slope adjustability.
1.2 Microwave Active Circuits 7
Modulators are three port networks. One of the ports is for modulating signal
and other two ports are RF ports. A conventional modulator is designed by using
a complex chain of downconverter, mixers, filters and upconverter. Most frequently
used digital modulation techniques are the discrete phase modulation, i.e. M-ary
phase shift keying (PSK). BPSK and QPSK are the most popularly used digital
phase modulations techniques for wireless and satellite communication systems.
Direct carrier modulation at RF/microwave frequency avoids the use of upconverter
and downconverter assemblies leading to less hardware complexity and cost. In case
of direct modulation techniques, p-i-n diode or FET-based voltage/current-controlled
RF resistance are used to modulate the amplitude and phase of the carriers by using
the modulating signal as control voltage/current of the device.
Phase shifter provides a known and controllable phase shift when a RF/microwave
signal passes through it. Phase shifters are used for the realization of phase array
antenna systems, linearizers for high power amplifiers, various types of phase mod-
ulators, etc. True time delay (TTD) and phase shifters (PS) are the basic functional
blocks for complex signal processing used in variety of applications including arbi-
trary waveform generation, filtering, antenna beam forming and reconfiguration abil-
ity in communication systems.
Attenuators are used to control signal level in the signal path. They provide fixed
attenuation or variable/adjustable attenuation depending upon control voltage or cur-
rent. Fixed attenuators are realized using a combination of RF resistive elements with
fixed values. Voltage/current controlled attenuators are realized using combination
of voltage/current dependent resistors. At RF/microwave frequency range, forward-
biased p-i-n diodes are used as voltage/current dependent RF resistors. Voltage-
variable p-i-n diode attenuators are used to control signal level of communication
systems by controlling gain of the system. In case of satellite transponders, p-i-n diode
based attenuators are used for on-board gain control of a communication transponder
by issuing command from ground or automatically by sensing RF power.
RF/microwave switches are used to reconfigure communication systems by divert-
ing signals from one path to another. The switches are also used to realize stepwise
variable attenuators, phase shifters and other elements. Realization point of view,
there are two types of RF and microwave switches: electromechanical and solid-
state switches. Generally, solid-state switches are very compact in size, faster in
operation and more reliable than their electromechanical counterparts. However,
solid-state switches have higher insertion loss and also exhibit more nonlinearity
at higher signal power. Though CMOS and GaAs FETs are also used as switching
devices, p-i-n diodes are used widely as solid-state switching device due to high RF
power handling capability and low insertion loss at high-frequencies.
Microwave circuits are used for many applications including airborne and space-
based systems where temperature-invariant performance is always demanded over
a very wide operating temperature range. The success of electronic systems for
these applications relies on the ability to design high performance, highly reliable
circuits which function in demanding thermal environments. Performance parameters
of the microwave diodes and transistors change with temperature. Attenuation and
phase shift of p-i-n diode circuits, limiting power level, attenuation and phase shift
8 1 Introduction
of Schottky barrier diode circuits and gain, saturated power, nonlinear parameters
of transistor-based circuits are all function of temperature. Therefore, microwave
active devices need proper bias circuit with temperature compensation mechanism
to achieve temperature-insensitive performance [4].
1.3 Microwave Active Devices
Transistors are three-terminal solid-state devices used for the realization of ampli-
fiers, multipliers, mixers, linearizers and other electronic circuits in communication
systems. Among the various types of transistors, bipolar junction transistor (BJT) was
the first three-terminal device and still being used for many low frequency as well
as microwave applications. With the advancement of hetero-junction technology,
hetero-junction bipolar transistor (HJBT or HBT) uses a high bandgap semiconduc-
tor as emitter material offers better performance in terms of gain at higher microwave
frequency and provides sufficient gain over mmwave frequency regions.
With the advancement of microwave transistors fabrication facilities, solid-state
amplification is dominated by the use of three-terminal device transistors. Among the
microwave transistors, all the low-noise amplifiers for communication systems are
dominated by the psuedomorphic and metamorphic High Electron Mobility Transis-
tors (pHEMTs and mHEMTs). High gain with moderate output power amplifiers are
also dominated by MEtal Semiconductor FET (MESFET) and Heterojunction FETs
(HFETs). With the advancement of high bandgap semiconductor materials such as
GaN and SiC, single-ended high power amplification beyond 100 W is achievable
from solid-state technology. However, at higher microwave frequency bands and
higher output power, travelling wave tube amplifier (TWTA) is the most preferable
option and is the only choice in today’s scenario.
Currently, Field Effect Transistor (FET) is the most utilized three-terminal semi-
conductor device used in microwave and mmwave frequency bands. FETs are used
as linear, nonlinear as well as control devices. These are used for realization of low-
noise amplifier (LNA) and high-gain amplifier as well as high power amplifiers.
As nonlinear devices, these are used to realize multipliers, mixers and oscillators.
FETs are also used in RF control device applications as switches and attenuators
in which they dissipate essentially no power and can easily be designed into broad-
band circuits. There are various types of field effect transistors (FETs) based on the
use of different materials and structures with different doping profiles. FETs can be
fabricated from a wide variety of materials like Si, SiGe, GaAs, InP, InGaAs, GaN,
SiC, etc. These are unipolar devices since only one type of carrier, mostly electrons,
take part for the channel current. Being a unipolar device, FET does not suffer from
minority carrier storage delay time & thus have higher cut-off frequency compared to
bipolar devices like BJTs. Field effect transistors based on III-V group semiconduc-
tors provide superior performance at microwave and mmwave frequency bands for
their higher electron velocity at low electric field. MEtal Semiconductor Field Effect
Transistors (MESFETs) based on GaAs are the better microwave device compared
1.3 Microwave Active Devices 9
to junction field effect transistors (JFETs) and Metal Oxide Semiconductor Field
Effect Transistors (MOSFETs) with respect to gain, noise figure and power handling
capability at microwave and mmwave frequency bands.
Hetero-structure Field Effect Transistor (HFET) are also known as MOdulation-
Doped FET (MODFET) and High Electron Mobility Transistor (HEMT) because of
the doping variations within the structure and higher electron mobility, respectively.
Depending on structure, doping profile and materials, there are different variations of
HFETs suitable for various applications. InP-based HEMT provides the highest fre-
quency of operation with the best noise figure. On the other hand, high bandgap semi-
conductor such as GaN- and SiC-based HFETs provides hundreds of watt microwave
power due to their higher breakdown voltage, higher electron mobility at high electric
field and high thermal conductivity.
Schottky barrier diode is a metal–semiconductor junction diode where only major-
ity carrier takes part for device operation. The non-existence of minority carrier
injection and extraction leads it to a very high speed of operation. High-speed oper-
ation capability of the Schottky diode makes it suitable to use widely for microwave
and mmwave applications. Lower cut-in voltage of the Schottky diode makes it suit-
able for very low-voltage high-current operation. The device is used as RF detector,
multiplier, mixer, limiter, attenuator, etc. This is also used to generate nonlinear char-
acteristic for linearizer and microwave phase shifter for microwave beam forming
network.
A microwave p-i-n diode behaves as current controlled resistance at microwave
frequency. The resistance of the diode may vary from 10,000 to less than 1 by
control of its bias current. Though all the junction diodes show this type of character-
istic, structure and doping profile of p-i-n diodes is designed to achieve wide variable
resistance range and linearity with low bias current level. These characteristics of the
p-i-n diode made it most suitable for variable attenuators, switches, limiters, phase
shifters, modulators, etc.
Step recovery diode (SRD), varactor diode and tunnel diodes are popular devices
used for microwave communications. The step recovery diode has ultra-fast switch-
ing speed, typically in the range of picosecond, and is used as nonlinear device for
multiplier, mixer, etc. Nonlinear behaviour of tunnel diodes is used for low-level RF
power detection and multiplier applications. Nonlinear characteristics of varactor
diodes are also used to realize multiplier, mixer and linearizer.
1.4 Microwave Circuit Analysis and Measurements
At low frequencies, where circuit dimensions are relatively smaller than the oper-
ating wavelength, circuit elements can be considered as lumped components as the
phase change along the circuit elements is negligible. Similarly, at low frequencies,
the interconnection between two elements is treated as an ideal connection, with no
loss, no characteristic impedance and no transmission phase angle. At microwave fre-
quencies, circuit components and interconnecting elements may not conveniently be
10 1 Introduction
scaled down with the reducing wavelength. Thus, the circuit analysis cannot be done
using conventional circuit voltage and current methodology. Microwave circuits may
consist of elements with dimensions such that the voltage amplitude and phase over
the length of the device can vary significantly. In this case, the current that flows in
the interconnecting wires or through the circuit elements is not the same at differ-
ent points along the wire or elements. To accurately calculate the behaviour of an
element, it is required to consider its length, width and thickness of metal and its
proximity to the ground plane, i.e. the element is considered as a distributed element.
Similarly, when interconnects become an appreciable fraction of the signal wave-
length, these interconnections themselves must be treated as distributed elements or
transmission lines.
1.5 Book Outline
The book consists of 20 chapters including this introduction.

Chapters 2 and 3 describe microwave p-i-n diodes, and Schottky barrier diodes
respectively. Step recovery, tunnel, backward and varactor diodes are discussed in
Chap. 4. These chapters also describe temperature behaviour of the diodes and effect
on the performance of the diode-based circuits.
Chapters 5 and 6 describe microwave bipolar and field effect transistors respec-
tively. Chapter 5 also describes hetero-junction bipolar transistors (HBTs) based on
GaAs and SiGe semiconductors. Chapter 6 gives a detailed description about metal
semiconductor FETs (MESFETs), various types of hetero-junction FETs (HFETs)
including pseudomorphic HEMT (pHEMT), metamorphic HEMT (mHEMT) and
GaN-based HEMTs.
Chapter 7 describes details about microwave circuit analysis including the concept
of S-parameters, ABCD parameters and their inter-relationships through various
examples.
The working principle of microwave switches is given in Chap. 8, and about atten-
uators and their performance dependency on operating temperature and compensa-
tion techniques are given in Chap. 9. Chapter 10 describes RF and microwave phase
shifters. Chapter 11 describes various RF and microwave modulators. Chapter 12
describes various amplitude tilt active equalizers operating at RF and microwave
frequency range. Chapter 13 described various types of microwave detectors for
measurement of power level.
The design and analysis of different types of microwave amplifiers is given in
Chap. 14. The chapter contains design and analysis of low-noise amplifier (LNA),
high gain amplifier and high power amplifier with examples using transistors. This
chapter also gives detailed design and analysis of Class-F amplifiers. Chapters 15
and 16 describe design and analysis of various types of microwave limiters and
linearizers respectively.
1.5 Book Outline 11
Chapter 17 describes microwave frequency multipliers using various types of

diodes and transistors. Chapter 18 provides details on the design and analysis of
microwave mixers.
Chapter 19 presents an overview of terrestrial mobile and satellite communication
systems with examples of design and development of complete microwave subsys-
tems such as receiver, local oscillator, channel amplifier, and linearized high power
amplifiers. Chapter 20 provides various types of multiple choice questions (MCQs)
and answers with their explanations.
References
1. Huang X, Guo YJ, Zhang J, Dyadyuk V (2012) A multi-gigabit microwave backhaul. IEEE
Commun Mag 50(3):122–129
2. Huang X, Guo YJ, Zhang JA (2014) Multi-gigabit microwave and millimeter-wave communi-
cations research at CSIRO. In: 2014 14th International Symposium on Communications and
Information Technologies (ISCIT)
3. Wang CX, Haider F, Gao X, You XH, Yang Y, Yuan D, Aggoune H, Haas H, Fletcher S, Hepsaydir
E (2014) Cellular architecture and key technologies for 5G wireless communication networks.
IEEE Communications Magazine, February 2014
4. Bera SC (2006) Temperature behavior of active circuits at microwave frequency. Ph.D. disser-
tation, Gujarat University, India
Chapter 2
P-I-N Diode
Abstract This chapter presents the basic working principle of p-i-n diodes including
their characteristics, equivalent circuits at different bias conditions and applications
in communication systems. Nonlinear behaviours of p-i-n diodes under different
bias conditions are presented. Temperature behaviour of p-i-n diodes and concept
of equiresistance curves for achieving temperature-invariant RF resistance is also
presented.
2.1 Introduction
A microwave p-i-n diode behaves as current-controlled resistance at microwave fre-

quency. The resistance of the diode may vary from 10,000 to less than 1 by
controlling its bias current. Though all the junction diodes show this type of charac-
teristic; structure and doping profile of p-i-n diodes are designed to achieve a wide
range of variable resistance and linearity with low-bias current level. These char-
acteristics of the p-i-n diodes made them most suitable for realization of variable
attenuators, switches, limiters, phase shifters, modulators, etc.
2.2 Basics of P-I-N Diode
A p-i-n diode has a high resistivity nearly Intrinsic (I) layer sandwiched in between
a heavily doped p-region and a heavily doped n region as shown in Fig. 2.1 [1]. The
thickness of the intrinsic layer is usually in the range of 10–200 μm. The intrinsic
layer is either lightly doped p-type or n-type having resistivity of about 1000 ohm-
cm. Practically, the lightly doped layer is ν-type or π -type depending upon whether
the conductivity is n-type or p-type, respectively.
At zero-bias condition, in case of diodes with purely intrinsic middle layer, there
will be no mobile charge carriers in the I (intrinsic) layer, and thus the p-i-n diode
exhibits very high resistance. However, in case of ν-type or π -type middle layer,

14 2 P-I-N Diode
Fig. 2.1 Impurity profile of

a p-i-n diode
a very thick depletion region will be formed in the middle layer and a very thin
depletion layer will be formed in p- or n-layer, respectively. In this case, at zero-bias
condition, the middle layer will not be completely depleted of mobile carriers. A
diode exhibits very high resistance when a small reverse-bias voltage, called punch-
through voltage, is applied to sweep free of mobile carriers.
When forward bias is applied across a p-i-n diode, carrier injection takes place into
the middle layer. Electrons are injected in the middle layer from the n-layer, and holes
are injected from the p-layer. Due to the recombination of the carriers in the middle
layer, carrier concentration diminishes with the depth. The carrier concentration in
the middle layer increases with the increase of forward bias, thus increasing the con-
ductivity, i.e. lowering its resistance. The current-controlled variable RF resistance
with high OFF resistance and low ON resistance makes the p-i-n diode suitable for
various microwave applications.
2.3 P-I-N Diode Characteristics
The impedance of p-i-n diode is the most important electrical parameter that deter-
mines the p-i-n diode-based circuit’s performance. In forward-bias condition, the
amount of stored charge in the I-layer is forward-bias current (Id ) times the ambipo-
lar lifetime (τa ). Thus, with the increase of diode forward current, conductivity of
the I-layer increases, i.e. intrinsic layer resistance Ri decreases. However, when AC
signal is applied with the DC forward bias, then impedance of the p-i-n diode not
only depends on the DC forward current but also depends on the AC signal current
depending on its frequency.
The conductance of the p-i-n diode is determined by the stored charge (Q d ), i.e.
current to the diode. The total diode current Id (t) is given by [2, 3]
Q d dQ d (t)
Id (t) + (2.1)
τa dt
2.3 P-I-N Diode Characteristics 15
The first term is due to the excess carrier distribution that is replaced in every
ambipolar carrier life lime τa seconds, which happens when the diode biased with
constant current. Thus, if τa is relatively long and the I-layer is not too thick, the
I-layer becomes flooded with carriers at a reasonable forward-bias level and the diode
exhibits a low RF resistance. The second term is time-dependent excess charge varia-
tion term dQ d /dt. This time-varying storage charge is due to the time-varying signal
to the diode. The effect of controlling the stored charge in the I-layer by controlling
the forward current is known as conductivity modulation. This phenomenon of the
p-i-n diode is used in electronically controlled attenuators and amplitude modula-
tor. When a time-varying signal is applied with the constant-current bias, the stored
charge is modulated by the presence of time-varying signal. The degree of modu-
lation depends on the frequency of the time-varying modulating signal. When the
signal frequency is much higher than the inverse of the ambipolar carrier lifetime
and amplitude of the AC current is much less than the applied DC current, then AC
modulation of the stored charge in the I-region is negligible. Thus, at this higher
frequency, the conductivity modulation due to the AC signal is negligible compared
to the DC current effect. At this higher frequency, the I-region Ohmic resistance Ri
of a p-i-n diode of area A, width W , and carrier density over I-region ρ(x) is given
by [1]
W
1
Ri ρ(x)dx (2.2)
A
0
Under DC forward current, Ido and ambipolar diffusion length L a , the intrinsic
region resistance is given by

4kT W W π
Ri sinh tan−1 exp − (2.3)
q Ido 2L a 2L a 4
Here, q is the electron charge, k is the Boltzmann constant and T is diode temper-
ature. The expression (2.3) is valid for any value of W/L a . For narrow p-i-n diodes,
where W L a , it is assumed that the stored charge is uniformly distributed over
the entire I-region and the value of Ri is given by
W2 W2
Ri (2.4)
2μa Q d 2μa τa Ido
Here, μa is the ambipolar carrier mobility. This relationship shows that the RF
resistance of p-i-n diode depends on device geometry (W ), electronic properties of its
semiconductor material (μa , τa ) and forward-bias current (Ido ). Thus, for small-size
p-i-n diode (W L a ), the diode’s intrinsic region resistance Ri changes inversely
with the diodes forward-bias current (Ido ).
16 2 P-I-N Diode
Example 2.1 Calculate the intrinsic resistances of a narrow p-i-n diode of I-region
width 100 μm at forward current of 10 μA and 10 mA. Consider μa 1800 cm2 /Vs,
and τa 1 μs.
Solution
The intrinsic resistance of a narrow p-i-n diode is given by (2.4)
Ri W 2 /(2μa τa Ido )
2
100 × 10−6 1000
×
2 × 1800 × 10−4 × 1 × 10−6 Ido (in mA)
27.8

Ido (in mA)
Therefore, Ri at 10 μA (27.8 × 100) ≈ 2.8 k

And Ri at 10 mA (27.8/10) ≈ 2.8 .
Practically, minimum resistance of a forward-bias diode is limited by its para-
sitic series resistance which is typically about 1–2 . On the other side, maximum
resistance of the diode is limited by the value of its equivalent shunt capacitance.
Modulation of the stored charge in the I-region takes place under low AC fre-
quency. The time variation of the stored charge is equivalent to reactance across the
I-region. This reactance may be capacitive or inductive depending on the size of the
p-i-n diodes. In general, for small-size p-i-n diodes, the reactance is capacitive. The
value of I-region capacitor (Ci ) diminishes with the increase of frequency of the AC
signal.
Junctions of a p-i-n diode will not contribute to the RF impedance of the p-i-n
diode at high frequency due to the stored charge (Q d ) across the junctions. In equiva-
lent circuit terminology, the stored charge effect is equivalent to a capacitance called
diffusion capacitance Cd that is given by [3]
∂ Qd q
Cd τa Id Vd j , T (2.5)
∂ Vd j ηkT
Here, V dj is voltage across junction of the diode. Thus, the capacitance Cd depends
upon the ambipolar carrier lifetime τa of the I-region, determined by the structure of
the diode, and diode forward current Id . The diffusion capacitance increases linearly
with the increase of Id .
At low frequencies, there will be time variation of stored charge in the entire I-
region. Thus, low-frequency AC signal will modulate the I-region conductivity and
junction resistance will be the major contribution to the impedance of p-i-n diode.
The small-signal junction resistance will be governed by the slope of the following
DC I–V relationship of the diode:

q Vd j
Id Vd j , T Io exp −1 (2.6)
ηkT
Here, I o is the reverse saturation current of the diode. Thus, small-signal low-
frequency AC junction resistance of the forward-biased p-i-n diode is given by
∂ Id ηkT
R j (Id , T ) 1/ (2.7)
∂ Vd j q Id Vd j , T
At low-frequency operation, the p-i-n diode behaves as simple junction diode.

The junction capacitance C j of the diode of area A and width of the transition region
W is given by
εA
Cj (2.8)
W
This junction capacitance will contribute to the low-frequency diode impedance.
The equivalent circuit of a p-i-n diode is shown in Fig. 2.2. Junction resistance R j
and capacitance C j correspond to p–n junction of the diode. Ri represents resistance
of the middle I-layer excluding the swept out region. Ci is I-region capacitance due
to the modulation of the stored charges under AC signal. Cd is diffusion capacitance
across the junction representing the charge storage due to the diode current given by
(2.5). Rs is the ohmic series resistance. L p and C p are parasitic inductance and capac-
itance related to bond wire and packaging of the diode. The value of the equivalent
components depends upon size, bias condition and operating AC frequency.
The diffusion capacitor appears in parallel with the junction resistance. Thus, the
degree of modulation decreases with the increase of the current as well as frequency.
The cut-off frequency f c is given by
1
fc (2.9)
2π Cd + C j R j
In general, under sufficient forward-bias condition, Cd C j . Thus, putting the

values of Cd and R j from (2.5) and (2.7), respectively, the cut-off frequency becomes
1
fc (2.10)
2π τa
Fig. 2.2 Equivalent circuit

of p-i-n diode
18 2 P-I-N Diode
At frequencies well below the cut-off frequency f c , the voltage-variable junction

resistance dominates the impedance of a p-i-n diode and thus it behaves as an ordinary
p–n junction diode. Therefore, RF signal of frequency well below f c incident to
the diode will be rectified and will generate considerable distortion. However, at
frequencies well above f c ( f f c ), effect of junction resistance and the I-region
reactance due to the stored charge modulation in the I-region will be negligible and
the diode impedance is purely resistive.
Example 2.2 Calculate the cut-off frequency of a p-i-n diode of ambipolar carrier
lifetime τa 0.01 μs.
Solution
From (2.10), the cut-off frequency is given by
f c 1/2π τa
1
Hz
2 × π × 0.01 × 10−6
15.9 MHz
Therefore, this diode can behave as voltage-variable RF resistance at its forward-

bias condition without much distortion when the operating frequency f 15.9
MHz. Operating frequency f > 10 f c is considered for using the p-i-n diode as RF
resistance without much distortion. Practically, minimum frequency of operation of
Fig. 2.3 Equivalent circuit of a forward biased, b zero biased, c reversed biased p-i-n diode for
f 1/2π τa
a p-i-n diode as RF resistance not only determines by only f c . It also depends on

operating bias voltage/current of the diode and operating RF power level.
The equivalent circuit of a p-i-n diode is dependent on frequency and bias con-
dition. At frequencies much higher than f c , the equivalent circuit of a p-i-n diode at
different bias conditions is as shown in Fig. 2.3. At forward-bias condition, the effect
of junction capacitance and junction resistance can be neglected due to the presence
of diffusion capacitance at operation frequency much greater than the cut-off fre-
quency. Thus, the equivalent circuit at forward-bias condition will be as shown in
Fig. 2.3a. However, at zero-bias condition, junction resistance R j is very high, and
there will be no presence of diffusion capacitance. Thus, the equivalent circuit of the
p-i-n diode at zero-bias condition is as shown in Fig. 2.3b. At reverse-bias condition,
if there is no charge carrier in the I-region, then there will be no existence of Ri .
Thus, the equivalent circuit at reverse-bias condition is as shown in Fig. 2.3c.
2.4 Nonlinearity of P-I-N Diode
For controlling RF and microwave signals, p-i-n diode-based voltage/current-

controlled attenuators, switches and phase shifters are widely used. In forward
and reverse-bias condition, p-i-n diode may create signal distortion for its nonlinear
high-frequency current–voltage characteristic. The modulation of the I-region stored
charge due to the applied AC signal causes time-varying conductivity modulation.
This generates harmonics and intermodulation distortion of the diode current output.
In forward-bias condition of p-i-n diode, low-frequency ( f f c ) RF signal mod-
ulates the I-region carriers as the DC current does. Stored charge modulation by the
AC signal decreases with the increase of frequency. At high frequencies ( f f c ),
AC signals have less influence on I-region stored charge. However, at microwave
frequencies also, there is still a small amount of charge injection by the AC signal.
I-region resistance Ri is inversely proportional to the stored charge Qd as it is clear
from (2.4). Therefore, Ri will also vary with the high-frequency AC signal due to the
time variation of stored charge causing distortion generation of the AC signal. Stored
charge modulation in the I-region depends upon frequency as well as amplitude of
the applied AC signal for a particular p-i-n diode. However, width (W ) of the I-region
of p-i-n diodes also influences the degree of modulation of the stored charge. Larger
width p-i-n diodes exhibit less variation of stored charge with increase of the AC
signal frequency [4].
Reverse-biased p-i-n diodes are used for RF and microwave switching appli-
cation. Reverse-biased p-i-n diodes also generate signal distortion. The reason for
distortion of reverse-biased p-i-n diode is the finite time-varying capacitance due
to the RF and microwave signal. The capacitance of an ideal p-i-n diode is gen-
erally assumed as constant for all reverse-bias voltage beyond punch-through volt-
age. However, in actual p-i-n diodes, reverse-bias capacitance decreases for voltages
beyond its punch-through due to the advancement of depletion layer into the end
regions with the increase of reverse voltage. Thicker p-i-n diodes exhibit flatter
20 2 P-I-N Diode
capacitance–voltage relationship than thinner diodes. This results in a lower rate

of capacitance change with reverse voltage for thick p-i-n diodes. The magni-
tude of the reverse-bias distortion is inversely proportional to the slope of the
capacitance–voltage characteristic and frequency. This is in contrast to the forward-
bias p-i-n diode distortion case where the distortion improves with increasing AC
signal frequency.
2.5 Temperature Behaviour of P-I-N Diode
RF resistance, Ri , of p-i-n diodes depends on forward-bias current (Ido ) and electronic

properties of its semiconductor material, i.e. ambipolar carrier mobility (μa ) and
ambipolar carrier lifetime (τa ) as given by (2.4). The I-region ambipolar mobility
μa , and the ambipolar carrier lifetime τ a , are both functions of temperature; therefore,
high-frequency resistance of p-i-n diodes also depends on temperature. Forward-bias
current Ido also changes with temperature if the diode is not driven by a constant
current source.
The temperature dependency of mobility is well known and it decreases with
nearly square of increasing temperature in the temperature range of interest
223–473 K (−50–220 °C). This temperature dependency can be approximated by [1]
n
To
μ(T ) μ(To ) (2.11)
T
Here, μ(To ) and μ(T ) are the carrier mobility at temperature To and T , respec-
tively and n is a temperature independent constant.
For pure materials, the experimental value of n is 2.42 and 2.2 for n- and p-type
Si, respectively; 1.0 and 2.1 for n- and p-type GaAs, respectively; 1.66 and 2.33 for
n- and p-type Ge, respectively.
Based on measurements and works on silicon devices, carrier lifetime has been
found to increase with increasing temperature. The temperature dependency of carrier
lifetime may be modelled as
m
T
τ (T ) τ (To ) (2.12)
To
Here, τ (To ) and τ (T ) are the carrier lifetime at temperature To and T , respec-
tively and m is a temperature independent constant. The value of m depends on variety
of factors influencing the carrier lifetime temperature characteristic. The important
factors are the diode geometry, i.e. I-region width, diameter and the type of surface
passivation material. Depending on these factors, in general, the value of m varies
from 0.5 to 1.5 [5, 6]. However, in case of most Si microwave p-i-n diodes, the value
of m is near about 1.0 [7].
2.5 Temperature Behaviour of P-I-N Diode 21
Therefore, from (2.4), (2.11) and (2.12), temperature dependency of high-

frequency resistance of p-i-n diodes can be written as
W 2 T n−m
Ri (Id , T ) (2.13)
2μa (To )τa (To )Ido (T )
Here, μa (To ) and τa (To ) are the ambipolar carrier mobility and ambipolar carrier
lifetime, respectively, at temperature To , whereas putting the temperature-dependent
reverse saturation current I o to (2.6), the temperature-dependent forward-bias static
DC current Id (T ) of p-i-n diodes is given by [1]

−q E b q Vd j
Id Vd j , T AT ν exp exp −1 (2.14)
ηkT ηkT
Here, A is a temperature-independent constant depending on area and electronic

properties of material of the diode, ν is a constant ~3 for p-i-n diodes, E b is the
bandgap potential in volt of diode’s semiconductor material, Vd j is the voltage across
the junction of the diode, η is the ideality factor nearly equal to 2.0 due to the recom-
bination of current in p-i-n diodes, q is the electron charge and k is the Boltzmann
constant.
Typical current–voltage relationship of p-i-n diodes under forward-bias condi-
tion at three different temperatures is shown in Fig. 2.4. Three different types of
bias load lines on the same plot are also shown. Temperature dependency of RF
resistance of p-i-n diodes is also shown in Fig. 2.5 under fixed-current- and fixed-
voltage-bias condition based on Eqs. (2.13) and (2.14) for n 2, m 1, ν 3
Fig. 2.4 Typical temperature dependency of V –I curve and different bias load line
22 2 P-I-N Diode
Fig. 2.5 Typical

temperature dependency of
p-i-n diode RF resistance at
fixed-current- and
fixed-voltage-bias conditions
and η 2 [8]. Operation of diodes under constant-voltage-bias condition never

recommended due to its large variation of current with the change of operating tem-
perature. The RF resistance is also very sensitive with the change of temperature
under constant-voltage-bias condition, and the RF resistance decreases drastically
with the increase of temperature. Diodes are generally known as current-controlled
device. Under constant-current-bias condition, i.e. when driver circuit of p-i-n diode
provides constant current irrespective of the temperature of the diodes, the RF resis-
tance of the diode increases slowly with the increase of temperature. Thus, with the
increase of temperature, the RF resistance of p-i-n diode increases when it is biased by
constant-current-bias, and however decreases when it is biased by constant-voltage-
bias condition. These are the very interesting characteristics of p-i-n diodes. There-
fore, it is possible to achieve temperature-invariant RF resistance of the p-i-n diodes
under optimum bias load-line condition which is in between constant-voltage- and
constant-current-bias condition.
2.6 Temperature-Invariant RF Resistance of P-I-N Diode
Microwave and RF control circuits based on p-i-n diodes are used in airborne and
space-based systems where always temperature-invariant performance demanded
over very wide operating temperature range. The success of electronic system
for these applications relies on the ability to design high-performance, high-
reliable circuits which function in demanding thermal environments. However, high-
frequency resistance of p-i-n diodes that determine the performance of the circuits is
temperature-dependent as discussed in the previous section. The temperature varia-
tion of RF resistance is due to the temperature dependency of carrier mobility and
carrier lifetime. The carrier mobility is the property of the diode’s semiconductor
material. However, temperature characteristic of carrier lifetime influenced by the
diode geometry, especially by the I-region area-to-width ratio and the type of mate-
rial, is used to passivate the diode surface [5, 6]. For larger size silicon p-i-n diodes (n
2.6 Temperature-Invariant RF Resistance of P-I-N Diode 23
~ 2) with glass or silicon dioxide passivated surface, the value of m becomes nearly 2
and leads to temperature-insensitive RF resistance. However, the diodes with larger
I-region area-to-width ratio are having larger capacitance value across it that leads
to limit high-frequency circuit performance.
In case of most of the commercially available microwave p-i-n diodes, the carrier
lifetime linearly increases with the increase of temperature, i.e. m 1. Therefore, for
most of the practically available microwave p-i-n diodes (n − m) ~1, that means,
RF resistance linearly increases with the increase of diode temperature under con-
stant current bias. In other words, the well-known constant current biasing will not
provide acceptable temperature stability of the p-i-n diode-based circuit performance
depending upon the RF resistance of the p-i-n diode such as attenuators and modula-
tors. Therefore, proper temperature compensation mechanism is required to achieve
acceptable temperature stability.
There are numerous conventional compensation mechanisms in the circuit design
level, which uses temperature sensor to produce a temperature-dependent signal. This
signal modifies the current applied to the diode in such a manner that it maintains same
RF resistance over the operating temperature. Some of the compensation mechanism
is based entirely on analog control signal, while others contain digital control signal
acting at discrete temperature steps. However, the successful adjustment of such
control circuits is a time-consuming task, often involves trial and error; and the
circuits themselves are relatively complex.
Figure 2.6 shows a simple biasing circuit for a p-i-n diode with supply voltage
V S and load resistor R L . Forward-biased DC equivalent circuit of p-i-n diode is
also shown in Fig. 2.6. Here, Rs is the parasitic series resistance, Ri is the I-region
resistance and R j is the DC junction resistance of the diode. Therefore, the volt-
age–current relationship of the p-i-n diode can be written in terms of diode terminal
voltage as

ν −q E b q[Vd (T ) − Id Rs − Id Ri ]
Id (Vd , T ) AT exp exp −1 (2.15)
ηkT ηkT
For (n − m) ~1 and neglecting reverse saturation current compared to the forward

current, (2.15) can be written as
Fig. 2.6 Simple bias circuit

and DC equivalent circuit of RL
forward-biased p-i-n diode
Rs
Id
Vd Rj Vdj Vd
VS
Ri
24 2 P-I-N Diode

q[Vd (T ) − Id (Vd , T )Rs − E b ]
Id (Vd , T ) A T ν exp (2.16)
ηkT
Here, A is a temperature-independent constant. The temperature and current

dependency of RF resistance of practical diodes (2.13) can be rewritten in more
generic form as
T n−m
Ri (Id , T ) z (2.17)
Ido (Vd , T ) p
Here, p and z are constants. The value of p is near about 1. To achieve temperature-
invariant RF resistance of the diodes, the bias network must meet the current require-
ment to achieve a constant RF resistance, say Rio over the operating temperature.
Therefore, current–voltage relationship (2.16) is to be satisfied by the bias network
to achieve temperature-invariant RF resistance over the temperature range under the
constraint imposed on Vd (T ) as given by
n−m 1p
T ηkT z
Vd (T ) E b + z Rs + ln p T (n−m−v p) (2.18)
Rio qp A Rio
This parametric equation is the equation of ideal bias-point locus that will satisfy
Ri Rio over the diode temperature range. This locus is called equiresistance curve
for the required RF resistance Rio [7, 9, 10].
Fig. 2.7 Equiresistance curve and optimum bias load line along with I-V characteristic of p-i-n
diode
I–V characteristic and corresponding equiresistance curve are shown in Fig. 2.7,
for m 0.97, p 0.9, η 2 and n 2.3 over the temperature range of −260
to +300 °C. The plot shows that the equiresistance curve is highly linear over the
temperature range of interest. Thus, the load line of a simple bias circuit, as shown
in Fig. 2.6, can be adjusted to coincide with the equiresistance curve within the
temperature range of interest and it will maintain the desired RF resistance. The
voltage of the bias network will be voltage axis intercept of the bias load line, and
resistance will be the reciprocal of slope of the bias load line. Combining (2.16),
(2.17) and (2.18) and eliminating T , another form of equation of the equiresistance
curve can be written as
⎡ ⎤
p n−m
1 n−m
v
ηk Rio Id z 1
Vd (T ) E b + ln⎣ vp
⎦ (2.19)
q z Rio n−m −1
AI d
The equation of tangent to the equiresistance curve of (2.19) at diode bias voltage
Vdo and corresponding current Ido at temperature To is given by

p ηkTo pv Id
Vd (T ) Vdo + (E b − Vdo ) + −1 × 1− (2.20)
n−m q n−m Ido
Taking this tangent to be the optimum bias load line, its voltage axis intercept
VOPT , called optimum voltage, is given by

p ηkTo pv
VOPT Vdo + (E b − Vdo ) + −1 (2.21)
n−m q n−m
This optimum voltage VOPT will be the open-circuit voltage of the bias network.
The voltage axis intercept is seen to be practically independent of the selected bias
point (Vdo , Ido ), implying that the optimum no-load output voltage of the bias circuit
is the same for all resistance values.
For a particular type of diode, the parameter values m, n and p are constants and
independent of temperature. Putting typical values of these constants, i.e. m 1.0,
p 1.0, v 3, η 2 and n 2.0 in (2.21), the optimum open-circuit voltage
becomes
2kTo
VOPT E b + (2.22)
q
Thus, open-circuit voltage is very close to the bandgap potential E b of the semi-
conductor material of the p-i-n diode since the last term is of the order of a few
millivolts. Parameter E b is the property of the diode’s material; it is independent of
process technology and does not vary from one die to another. Moreover, the tem-
perature dependency of bandgap potential (E b ) is also very weak [11] over a prac-
tical operating temperature range. Thus, this optimum bias load-line technique for
26 2 P-I-N Diode
achieving temperature-invariant performance is very stable under different parameter

variations of the diodes.
The optimum bias load-line technique for achieving temperature-invariant per-
formance is also applicable for many other junction diodes [10]. For example, RF
resistance of Schottky barrier diode and light intensity (brightness) of light-emitting
diode can be made temperature-insensitive using the optimum bias load-line tech-
nique [12, 13].
Example 2.3 Calculate open-circuit voltage for temperature-invariant intrinsic layer

resistance of a silicon p-i-n diode over the temperature range of −20 to 80 °C.
Solution From (2.22), the optimum open-circuit voltage is given by

2kTo
VOPT E b +
q
2 × 1.38 × 10−23
1.12 + To V
1.6 × 10−19
1.12 + 0.0001725 × To V
Putting the values of To , the optimum voltage (V OPT ) is calculated and is shown in
Fig. 2.8. It shows that the optimum voltage is practically independent of temperature
and is slightly higher than the value of bandgap potential (E b ) of the semiconductor.
Example 2.4 Current-dependent RF resistance, Ri , and current–voltage (Id − Vd )

relationship of a p-i-n diode are

aT Vd − E b
Ri and Id b × exp
Id cT
Here, a, b, c are constants. Derive optimum open-circuit voltage for zero temper-
ature coefficient of RF resistance of the diode.
Solution
Temperature coefficient of the p-i-n diode resistance is given by
Fig. 2.8 Dependence of

optimum open-circuit
voltage on temperature
∂ Ri a aT ∂ Id
− 2 (2.23a)
∂T Id Id ∂ T
For zero temperature coefficient of the RF resistance, ∂ Ri /∂ T 0. Putting this

condition to (2.23a), we can write
∂ Id Id
(2.23b)
∂T T
Suppose the diode is biased through a bias resistance, Rb and supply voltage VS .
Thus,
VS Id Rb + Vd (2.23d)
Using (2.23d), the I–V relationship of the p-i-n diode can be written as

VS − I d Rb − E b
Id b × exp (2.23e)
cT
Differentiating (2.23e) w.r.t. temperature T ,

∂ Id Id Rb Id (VS − Id Rb − E b )
1+ − (2.23f)
∂T cT cT 2
For zero temperature coefficient, suppose VS VOPT and Id Ido . Putting these
values to (2.23f) and using (2.23b),

Ido Ido Rb Ido (VOPT − Ido Rb − E b )
1+ − (2.23f)
T cT cT 2
This leads to the value of optimum supply voltage for zero temperature coefficient
of RF resistance of the p-i-n diode as
VOPT E b − cT (2.23g)
Example 2.5 A microwave circuit as shown in Fig. 2.9 consists of n numbers of

batch matched p-i-n diodes in shunt configuration (connected in DC parallel path).
Current-dependent RF resistance, Ri , and current–voltage (Id − Vd ) relationship of
the p-i-n diodes are

aT Vd − E b
Ri and Id bT × exp
3
Id cT
Here, a, b, c are constants. Derive the optimum open-circuit voltage for zero
temperature coefficient of the RF resistance.
28 2 P-I-N Diode
Fig. 2.9 RF circuit using

multiple shunt p-i-n diodes
Solution For achieving zero temperature coefficient of RF resistance of the p-i-n

diodes,
∂ Ri a aT ∂ Id
− 2 0 (2.24a)
∂T Id Id ∂T
∂ Id Id
or, (2.24b)
∂T T
As shown in Fig. 2.9, the diodes are biased through a bias resistance Rb and supply
voltage VS . Here, DC current Id flows through each diode and voltage across each
diode is Vd . Thus,
VS n Id Rb + Vd (2.24c)
Using (2.24c), the I–V relationship of the p-i-n diode can be written as

VS − n I d Rb − E b
Id bT 3 × exp (2.24d)
cT
Differentiating (2.24d) w.r.t. temperature T ,

∂ Id n Id Rb 3Id Id (VS − n Id Rb − E b )
1+ − (2.24e)
∂T cT T cT 2
values to (2.24e) and using (2.23b),

Ido n Ido Rb 3Id Ido (VOPT − n Ido Rb − E b )
1+ − (2.24f)
T cT T cT 2
of RF resistance as
VOPT E b + 2cT (2.24g)
This shows that the optimum supply voltage for zero temperature coefficient of
RF resistance of diodes is independent on the number of diodes connected in parallel
DC path.
Example 2.6 A microwave circuit as shown in Fig. 2.10 consists of n numbers of

matched pair p-i-n diodes connected in DC series path. Current-dependent RF resis-
tance, Ri , and current–voltage (Id − Vd ) relationship of the p-i-n diodes are

aT Vd − E b
Ri and Id bT 3 × exp
Id cT
temperature coefficient of RF resistance of the diodes.
Solution For achieving zero temperature coefficient of RF resistance of the p-i-n

diodes,
∂ Ri a aT ∂ Id
− 2 0 (2.25a)
∂T Id Id ∂T
∂ Id Id
or, (2.25b)
∂T T
As shown in Fig. 2.10, the diodes are biased through a bias resistance Rb and
supply voltage VS . Here, DC current Id flows through each diode, and voltage across
each diode is Vd . Thus,
Fig. 2.10 RF circuit

diagram using multiple
series p-i-n diodes
30 2 P-I-N Diode
VS Id Rb + nVd (2.25c)

VS − I d Rb − n E b
Id bT 3 × exp (2.25d)
ncT

∂ Id Id Rb 3Id Id (VS − Id Rb − n E b )
1+ − (2.25e)
∂T ncT T ncT 2

Ido Ido Rb 3Ido Ido (VOPT − Ido Rb − n E b )
1+ − (2.25f)
T ncT T ncT 2
of RF resistance as
VOPT n(E b + 2cT ) (2.25g)
This shows that the optimum supply voltage for n numbers of series DC path
connected diodes is n times the optimum voltage corresponding to a single diode.
References
1. Sze SM (1981) Physics of semiconductor devices, 2nd edn. Wiley Eastern Limited
2. Wilson P (1967) Recombination in silicon p-i-n diodes. Solid State Electron 10:145–154
3. Slatter JAG, Whelan JP (1980) p-i-n diode recovery storage time. Solid State Electron
23:1235–1242
4. Caverly R (2004) Distortion modeling of PIN diode switches and attenuators. In: IEEE MTT-S
digest, pp 957–960
5. Caverly RH, Hiller G (1993) Temperature insensitive PIN diode attenuators. Appl Microw 88,
Summer 1993
6. Caverly RH, Hiller G (1995) Temperature effect on PIN diode forward bias resistance. Solid
State Electron 38(11):1879–1885
7. Bera SC, Bharadhwaj PS (2005) Insight into PIN diode behavior leads to improved control
circuit. IEEE Trans Circuits Syst II 52(1):1–4
8. Bera SC, Singh RV, Garg VK, Sharma SB (2007) Optimum bias load-line compensates
temperature variation of junction diode’s RF resistance. IEEE Trans Microw Theory Tech
55(2):215–221
9. Bera SC, Singh RV, Garg VK (2005) A temperature dependent PIN diode model leads to simple
temperature invariant attenuator circuit. Microw J 48(02):104–116
10. Bera SC, Singh RV, Garg VK (2006) Inherent temperature-compensating behavior of junction
diodes. Proceedings of international conference on computer and devices for communication
References 31
11. Bludau W, Onton A, Heinke W (1974) Temperature dependence of the band gap of silicon. J
Appl Phys 45(4):1846–1848
12. Bera SC, Singh RV, Garg VK (2005) Temperature behavior and compensation of light emitting
diode. IEEE Photonics Tech Lett 17(11):2286–2288
13. Bera SC, Singh RV, Garg VK (2009) A device for compensating the effect of temperature
variation on brightness of light emitting diodes. India Patent, IN235617
Chapter 3
Schottky Diode
Abstract This chapter presents the basic working principle and characteristics of
Schottky barrier diodes including its equivalent circuit and applications in commu-
nication systems. Temperature behaviour of Schottky barrier diodes and its effect on
performance of the diode-based circuits are also presented.
3.1 Introduction
Metal–semiconductor junction diode is known as Schottky barrier diode where only

majority carrier takes part for device operation. Non-existence of minority car-
rier injection and extraction leads it to a very high-speed operating device. High-
speed operation capability of the Schottky diode makes it suitable to use widely for
microwave and mmwave applications. Lower cut-in voltage of the Schottky diode
makes it suitable for very low voltage and high current operation. The device is used
as RF detector, multiplier, mixer, limiter, attenuator, phase shifter, etc. This is also
used to realize phase shifter and distortion generator for microwave beam forming
network and linearizer, respectively.
3.2 Basics of Schottky Diode
Schottky contact is a metal–semiconductor contact that generates potential barrier for

carriers to move from one material to another and generates nonlinear voltage–current
characteristic similar to p–n junction diode. The semiconductor material may be n-
or p-type silicon (Si), Gallium Arsenide (GaAs), Gallium Nitride (GaN), Silicon
Carbide (SiC), etc. and metal may be Aluminium (Al), Gold (Au), Tungsten (W),
Titanium (Ti), etc. [1–4].
Early detector diodes were point contact diodes, consists of a thin metal whisker
pressing on a semiconductor material generally silicon. However, Schottky barrier
diodes are fabricated by deposition of a metal layer on a semiconductor. Practically,

34 3 Schottky Diode
Fig. 3.1 Schematic structure of Schottky barrier diode
n-type semiconductors are used for all microwave Schottky diodes because of supe-
rior mobility of electrons than mobility of holes. The basic schematic structure of
a Schottky barrier diode is shown in Fig. 3.1 [1, 2]. Metal–semiconductor Schottky
junction is formed in between metal and n-type active layer. The n-type active layer is
generally grown epitaxially over an n+ substrate. Very low resistance ohmic contact
is formed in between n+ substrate and metal as electrode.
The energy band diagram of an n-type Schottky barrier diode when metals and
semiconductors are away from each other is shown in Fig. 3.2. Here, E Fm and E FS
are Fermi level energies, qφ m and qφ s are work functions of the metal and semi-
conductor respectively. E C and E V are conduction and valance band-edge ener-
gies respectively and qχ is electron affinity. For the n-type Schottky diode, the work
function of metal (φm ) is higher than that of semiconductors (φs ). Electron flows
from semiconductor to metal to establish thermodynamic equilibrium when metals
and semiconductors come into contact. Due to the charge transfer, Fermi levels align
and barrier potential (φb φm −χ ) is developed between metal and n-type semicon-
ductor junction as shown in Fig. 3.3. The Fermi level and band profile of the metal
will not change at all due to very high-electron density of the metal side. A depletion
region (W ) is formed adjacent to the metal in the semiconductor and the depletion
region width is given by (3.1) for donor concentration of Nd [2], where ε is dielectric
constant of the semiconductor. The depletion layer of width W 1 for Schottky barrier
contact and W 2 for Ohmic contact are shown in Fig. 3.3.
1/2
2εVbi
W (3.1)
q Nd
The positive charge within the depletion region W due to the uncompensated
donor ions matches the negative charge in the metal. It generates the equilibrium
contact potential (Vbi ), which is the difference of work functions of metal and semi-
conductor and is given by Vbi (φm − φs ). The generated contact potential prevents
further net electron diffusion in between them. The potential barrier height φb for
3.2 Basics of Schottky Diode 35
Fig. 3.2 Energy band diagram of Schottky barrier diode when metals & semiconductors are away
from each other
Fig. 3.3 Energy band diagram of Schottky barrier diode when metals and semiconductors are
in contact
electron injection from the metal into the n-type semiconductor conduction band is
(φm − χ ). Here, qχ (the electron affinity) is measured from the vacuum level to
the semiconductor conduction
√ band edge. The width W of the depletion region is
inversely proportional to Nd . Thus, as the doping concentration in the semicon-
ductor increases, the width of the depletion region decreases. This property is used
to make ohmic contact with the heavily doped semiconductor and suitable metal as
shown in Fig. 3.3. The locally heavy doping in the semiconductor decreases the width
of the depletion region (W2 ). Thus, probability of tunnelling the electrons through the
barrier increases and the junction behaves as ohmic contact, i.e. I–V characteristic
of the ohmic junction becomes linear with the same slope under forward and reverse
bias condition.
36 3 Schottky Diode
3.3 Schottky Diode Characteristics
The potential difference Vbi can decrease or increase due to the application of either
forward or reverse bias voltage. The contact potential is reduced from Vbi to (Vbi −V )
with the application of forward voltage V applied across the Schottky barrier junction.
Due to reduction of the contact potential, electrons from the semiconductor can
diffuse across the depletion region to the metal, this rises forward current from metal
to semiconductor through the junction. However, due to the barrier potential φb , the
flow of electrons from metal to semiconductor is negligible. Thus, the forward current
is due to the only injection of majority carriers from the semiconductor into the metal.
Whereas with the application of reverse voltage Vr , the barrier potential increases
from Vbi to (Vbi + Vr ) that reduced to negligible electron flow from semiconductor
to metal. In this case, also due to the barrier potential φb , the flow of electrons from
metal to semiconductor is negligible.
At normal operation of Schottky barrier diodes, there is no minority carrier injec-
tion. This eliminates the associated storage delay time, which is an important feature
of Schottky barrier diodes. Therefore, Schottky barrier diodes are essentially major-
ity carrier devices, although some minority carrier injection occurs at high-current
levels. These properties lead to the superior high-frequency properties and switching
speed compared to typical p–n junctions.
Under thermal equilibrium with no bias condition, the current from the metal to
the semiconductor is equal in magnitude and opposite in direction to the current from
the semiconductor to the metal. At this condition, the current in one direction say
from the semiconductor to the metal (I S→M ) is given by Sze [1] as follows:

∗ −qφb
I S→M A AT exp 2
(3.2)
kT
Here, A∗ is the Richardson constant and A is area of the diode. Under forward
bias condition of junction voltage Vd j , incorporating ideality factor (η), the diode
current is given by

∗ −qφb q Vd j
Id A AT exp
2
exp −1 (3.3)
ηkT ηkT

q Vd j
Io exp −1 (3.4)
ηkT
In case of Schottky barrier diodes, the ideality factor η ∼ 1.0–1.2, because of

negligible recombination in the depletion region. Io is defined as the reverse satura-
tion current. The Schottky diode does not have charge storage delay. Thus, its DC
transfer characteristic (3.3) is also valid for ac transfer characteristic up to a very
high frequency, extended to microwave even up to mmwave frequency range.
3.3 Schottky Diode Characteristics 37
Suppose, RF signal of amplitude Vs and angular frequency ω is applied to a finitely

DC biased (Vo ) Schottky diode. The diode voltage can be represented by (3.5) as
follows:
Vd j Vo + Vs cos(ωt) (3.5)
The diode current can be written as

Vo + Vs cos(ωt)
Id Io exp −1 (3.6)
Vt
Here Vt (ηkT /q). Representing the exponential term containing AC part by

series expansion up to fourth-order term, the diode current can be written as

Vo Vs cos(ωt)
Id Io exp × exp −1
Vt Vt

Vo
Io exp −1
Vt

Vo Vs V2 V3
+ Io exp cos(ωt) + s 2 cos2 (ωt) + s 3 cos3 (ωt)
Vt Vt 2Vt 6Vt
4
V
+ s 4 cos4 (ωt) + · · · (3.7)
24Vt

Vo
Io exp − 1 ⇒ dc current due to the bias voltage Vo
Vt

Vo Vs
+ Io exp × cos(ωt) ⇒ fundamental ac component
Vt Vt
2
Vo Vs Vs2
+ Io exp × + cos(2ωt) ⇒ rectified dc + 2nd harmonic
Vt 4Vt2 4Vt2
3
Vo Vs Vs3
+ Io exp × cos(ωt) + cos(3ωt)
Vt 8Vt3 24Vt3
⇒ fundamental + 3rd harmonic
4
Vo Vs Vs4 Vs4
+ Io exp × + cos(2ωt) + cos(4ωt)
Vt 64Vt4 48Vt4 192Vt4
⇒ dc + 2nd harmonic + 4th harmonic
+ ··· (3.8)
The expression (3.8) shows that all the even-order nonlinearity produces DC term
and even-order harmonics. Whereas, odd-order nonlinearity produces only odd-order
harmonic components. The property of generation of DC term related to RF signal
by Schottky diodes is used as power detector. Whereas, the property of generation
of higher order harmonics of the applied RF signal by Schottky diodes is utilized as
multiplier and mixer.
38 3 Schottky Diode
Fig. 3.4 I–V characteristic of Schottky diode in presence of RF signal
The expression (3.6) is written under the assumption that the exponential I–V
relationship is also valid under RF signal operation. This assumption is true if Vs
Vt . Otherwise, due to the significant rectification of the RF signal exponential, I–V
characteristic of the Schottky diode will not remain valid. A typical I–V characteristic
of Schottky diodes in presence of RF signal is shown in Fig. 3.4. The characteristic
deviates more from its exponential nature with the increase in RF signal level.
RF resistance of Schottky diodes is given by the inverse slope of the I–V curve at
the particular operating condition and is given by
∂ Id
Rd (Id ) 1/ (3.9)
∂ Vd j
Under small signal operating condition (Vs Vt ), the RF resistance of Schottky

diodes is given by
∂ Id ηkT 1
Rd (Id ) 1/ (3.10)
∂ Vd j q (Id + Io )
However, at large signal operating conditions, the exponential I–V relationship is

not valid to determine the RF resistance using it. Therefore, the expression (3.10) for
RF resistance of the diodes is also not valid under large signal operating conditions.
Under large signal RF operation, DC voltage and current across the diode are a
function of RF signal level and also depends on bias load line as shown in Fig. 3.4.
It is clear from the figure that for bias load line L 1 , RF resistance of the diode
decreases with the increase of RF signal level. Whereas, in case of bias load line L 2 ,
RF resistance of the diode increases with the increase of RF signal level. Thus, RF
3.3 Schottky Diode Characteristics 39
Fig. 3.5 RF equivalent

circuit of forward
biased Schottky barrier diode
resistance of the diode may increase or decrease with the increase in RF power level
depending on the bias load line [5, 6].
RF equivalent circuit of a forward-biased Schottky diode is shown in Fig. 3.5.
Here, Rd is the junction resistance of the diode, is a function of DC bias voltage as
well as operating RF power of the diode as discussed in the previous paragraphs.
The capacitance Cd is the depletion layer capacitance and is nearly independent
of operating RF power level. Thus, the capacitor Cd is considered as linear capacitor.
Rs is equivalent series resistance of the diode caused by the part of N layer, N+
layer and Ohmic contact as shown in Fig. 3.1. The resistor Rs is also considered as
linear resistor. L p and C p are the parasitic inductance and capacitance related to the
packaging of the diode.
3.4 Temperature Behaviour of Schottky Diodes
DC and RF characteristics of the Schottky barrier diodes are very much sensitive to
its operating temperature. Temperature dependency of DC and RF parameters of the
diodes in a circuit is also determined by the way it is biased. Temperature-dependent
I–V relationship of the Schottky diodes is given by

∗ −qφb q Vd (T ) − Id (Vd (T ), T )Rs
Id (Vd (T ), T ) A AT exp
2
exp −1
ηkT ηkT
(3.11)
where Vd (T ) is the temperature-dependent voltage across the diode and is related to

the junction voltage Vd j (T ) by
Vd (T ) Vd j (T ) − Id (Vd (T ), T )Rs (3.12)
A typical I–V characteristic of a Schottky diode is shown in Fig. 3.6 for three
different temperatures. Three different bias load lines are also shown in the figure to
show the current and voltage variations along the load lines under varying temperature
conditions. It shows that the diode current is highly temperature sensitive under
40 3 Schottky Diode
Fig. 3.6 I–V characteristic dependency on temperature and load line
Fig. 3.7 Temperature

dependency of Schottky
diode RF resistance under
different bias conditions
constant-voltage-bias condition. Thus, bias load line not only determines the RF
performance of the diode-based circuits it also determines the temperature sensitivity
of the circuit parameters.
Under sufficiently forward bias condition, where Id Io the temperature-
dependent small signal RF resistance Rd can be written from (3.10) as [3]
ηk T
Rd (Id ) (3.13)
q Id (Vd (T ), T )
The variation of the RF resistance of a Schottky diode with temperature, given by

(3.13), under fixed-current and fixed-voltage-bias conditions are shown in Fig. 3.7.
It shows that, under fixed-current bias, the RF resistance increases linearly with the
increase of temperature. Whereas, it decreases exponentially under fixed-voltage-
bias-condition.
The practical value of equivalent series resistance Rs is very small compared to
the RF resistance Rd of diodes and is defined by (3.13). Thus, in spite of small
temperature coefficient of Rs , its temperature effect on overall RF performance is
negligible. Similarly, in most of the practical applications, at the operating frequency,
the capacitive reactance of the capacitor Cd is very high compared to the diode
3.4 Temperature Behaviour of Schottky Diodes 41
resistance Rd . Thus, in spite of small temperature coefficient of the capacitance

Cd , its temperature effect on overall RF performance can be neglected compared to
the temperature effect of Rd . Thus, making temperature invariant RF resistance Rd ,
determined by (3.13), temperature-independent RF performance of the circuit can
be achieved.
3.5 Temperature Invariant RF Resistance
Schottky diode-based microwave circuits such as multiplier, mixer, linearizer, etc. are
used in various airborne and space-based systems. In all these applications, always
temperature invariant performance is demanded. However, the RF performance of
the circuits determined by RF resistance of the diodes is very sensitive to the circuit
temperature as discussed in the previous section.
To achieve temperature invariant RF resistance of the diode say Rd Rdo , and
neglecting reverse saturation current compared to the forward current, the I–V rela-
tionship will be given by Eq. (3.12) under the constraint imposed on Vd (T ) as [5, 6]

∗ q[Vd (T ) − Id (Vd (T ), T )]Rs − φb
Id (Vd (T ), T ) A AT exp
2
(3.14)
ηkT

z Rs ηkT z
Vd (T ) φb + T+ ln ∗
(3.15)
Rdo q A A Rdo T
Here, z ηk/q. Taking the typical values of the diode parameters as φb = 0.69 V
and η 1.2, the plot of Eq. (3.14) over the temperature range of −273 to 300 °C
with the constraint given by Eq. (3.15) is shown in Fig. 3.8. This curve in I–V plane
over the diode temperature is called ‘equiresistance curve’. The equiresistance curve
intersects the voltage axis at the diode’s barrier height voltage (φb ). An important
characteristic of the equiresistance curve is that it is a straight line over a very wide
range of temperature. Therefore, a bias circuit consists of a constant voltage source
with a series connected resistor of predefined value can provide temperature invariant
RF resistance of the diode, as shown in Fig. 3.8. The value of the constant voltage
and the series connected resistor are such that the bias load line passes along the
equiresistance curves over the operating temperature range.
Combining (3.14) and (3.15) and eliminating T , the equation of the equiresistance
curve can be written as

ηk Id Rdo z 2 1
Vd φb + Id Rs + ln (3.16)
q z Rdo A∗ AId
Differentiating the Eq. (3.16), the slope (Mo ) of the equiresistance curve at bias
point (Vdo , Ido ) corresponding to To can be written as
42 3 Schottky Diode
Fig. 3.8 Equiresistance curve and optimum bias load line

1 ∂ Vd ηk To q(Vdo − Ido Rs − φb )
Rs + −1 (3.17)
Mo ∂ Id q Ido ηkTo
The optimum bias load line is nothing but the tangent of the equiresistance curve
at temperature To . Thus, optimum load line at diode bias point Vdo and Ido can be
written as
1
Vd Vdo + (Id − Ido ) (3.18)
Mo
Combining (3.17) and (3.18), the optimum load-line equation can be written as

ηkTo Id
Vd (Id ) Vdo + (φb − Vdo + Ido Rs ) + − Ido Rs 1 − (3.19)
q Ido
The optimum open-circuit voltage (VOPT ) of the bias circuit is the voltage axis
intercept of the load line that can be obtained by putting Id 0 in the load line
Eq. (3.19). Thus, the optimum open-circuit voltage (VOPT ) is given by
ηkTo
VOPT φb + (3.20)
q
The expression (3.20) of the optimum open-circuit voltage shows that VOPT is
independent of the diode current or voltage. Therefore, the required open-circuit
voltage of the bias circuit for achieving temperature invariant RF resistance is the
same for all values of the RF resistance. The second term of the expression (3.20) is
about 0.03 V at ambient temperature, thus it can be considered as negligible com-
pared to the metal-to-semiconductor barrier height (φb ) of the diode. Thus, VOPT is
determined by the physical parameter (φb ) of the diode and is practically independent
on temperature.
Figure 3.9 shows the small signal RF resistance variation with temperature under
optimum load-line bias and fixed-current bias for φb 0.69 V and η 1.2. This
3.5 Temperature Invariant RF Resistance 43
Fig. 3.9 Temperature

dependency of small
signal RF resistance under
constant current and
optimum load-line bias
conditions
shows that the RF resistance variation decreases from 24% to about 0% over the
temperature range of −10 to +60 °C when the diode is biased from conventional
constant current bias to optimum open-circuit voltage bias.
Thus, RF resistance of Schottky diodes is (a) highly temperature sensitive under
fixed-voltage-biased condition, (b) less sensitive under constant-current-biased con-
dition and (c) minimal sensitivity of about 0% when it is biased by the optimum load
line technique.
The optimum open-circuit voltage VOPT as given by (3.20) corresponds to small
signal RF resistance. With the increase of RF power level, the I–V characteristic of
Schottky diodes deviates from exponential characteristic due to self-biasing effect.
Thus, the expression for VOPT also differs from (3.18). Practically, it is observed that
VOPT increases with the increase of RF power level under large signal operation.
Example 3.1 Calculate open-circuit voltage for temperature invariant small signal
RF resistance of a Silicon Schottky diode. Consider barrier height 0.69 V and
ideality factor 1.2.
Solution
From (3.20), the optimum open-circuit voltage is given by
ηkTo
VOPT φb +
q
1.2 × 1.38 × 10−23
0.69 + To V
1.6 × 10−19
0.69 + 0.0001035 × To V
The second term is temperature dependent but it is very small. Considering tem-
perature To 300 K:
VOPT 0.721 Volt
Example 3.2 Small signal RF resistance, Rd and current–voltage (Id − Vd ) relation-

ship of a Schottky barrier diode under forward bias condition are
44 3 Schottky Diode

aT Vd − φb
Rd and Id bT 2 × exp
Id cT
Here, a, b, c are temperature-independent constants. Derive optimum open-circuit

voltage for zero temperature coefficient of RF resistance of the diode.
Solution
Temperature coefficient of the small signal RF resistance of the Schottky barrier
diode is given by
∂ Rd a aT ∂ Id
− 2 (3.21a)
∂T Id Id ∂ T
For zero temperature coefficient of the RF resistance, ∂ Rd /∂ T 0. Putting this

condition to (3.21a), we can write
∂ Id Id
(3.21b)
∂T T
Suppose the diode is biased through a bias resistance, Rb and supply voltage VS .
Thus,
VS Id Rb + Vd (3.21d)
Using (3.21d), the I–V relationship of the diode can be written as

V S − I d R b − φb
Id bT × exp
2
(3.21e)
cT
Differentiating (3.21e) with respect to temperature T :

∂ Id Id Rb 2Id Id (VS − Id Rb − φb )
1+ − (3.21f)
∂T cT T cT 2
For zero temperature coefficient, suppose, VS VOPT and Id Ido . Putting these
values to (3.21f) and using (3.21b), we can write

Ido Ido Rb 2Ido Ido (VOPT − Ido Rb − φb )
1+ − (3.21f)
T cT T cT 2
of RF resistance of the Schottky diode as
VOPT φb + cT (3.21g)
References 45
References
2. Streetman BG (1986) Solid state electronic devices, 2nd edn. Prentice Hall of India Pvt. Ltd.,
New Delhi
3. Erikson J, Rorsman N, Zirath H (2001) Microwave silicon carbide Schottky diodes. Electron
Lett 37(4):250–252
4. Erikson J, Rorsman N, Zirath H (2003) 4H-silicon carbide Schottky barrier diodes for microwave
applications. IEEE Trans Microw Theory Tech 51(3):796–804
5. Bera SC, Singh RV, Garg VK, Sharma SB (2007) Optimum bias load-line compensates tempera-
ture variation of Junction diode’s RF resistance. IEEE Trans Microw Theory Tech 55(2):215–221
6. Bera SC, Singh RV, Garg VK (2008) Temperature behavior and compensation of Schottky barrier
diode. Int J Electron 95(5):457–465
Chapter 4
Special Microwave Diodes
Abstract This chapter presents construction, working principle and characteristic

of Step Recovery Diodes (SRDs), tunnel diodes, backward diodes, and varactor
diodes. The applications of these diodes in microwave communication systems are
also presented.
4.1 Introduction
Step recovery diodes (SRD), tunnel diodes and varactor diodes are popular devices
used for microwave communications. The step recovery diodes have very fast switch-
ing speed, in the range of picosecond, and used as a nonlinear device for multiplier,
mixer, etc. Nonlinear behaviour of tunnel diodes are used for low-level RF power
detection and frequency multiplier applications. Nonlinear characteristics of varac-
tor diodes are used to realize multiplier and mixer for receivers and linearizer for
high-power amplifiers.
4.2 Step Recovery Diode
When the current through a p–n junction diode is reverse biased from its forward
bias, the voltage across the diode does not change from its positive voltage to a
negative voltage instantaneously. Thus, the current through the diode also does not
stop instantaneously. This happens due to the charge storage effect in the diodes
[1–4]. Typical current and voltage transient under a step input for the circuit using
p–n junction diode is shown in Fig. 4.1a. The time t S is storage delay time required
for the stored charge in the diode to become zero. The time t R is the transition
time required to build the reverse bias across the diode junction by depletion of the
charge carriers.

48 4 Special Microwave Diodes
(a)
v(t)
+V
0 t v(t)
-V
i(t)
id(t)
If
(b)
0 t id(t)
If
Ir
tS tR t
0
vd(t)
tR
t Ir
0
tS
-Vr
Fig. 4.1 Pulse response of a conventional p–n junction diode
Example 4.1 A p–n junction diode in series with a 100 resistor is forward biased
so that a current of 50 mA flows. What will be the current through the circuit just
after the voltage across the combination is suddenly reversed to 10 V?
Solution
Initially, the p–n junction diode is in forward biased condition with forward current
of 50 mA. Just after providing the reverse voltage of 10 V to the combination, the
diode will remain in its forward bias condition due to its charge storage effect with
small voltage drop across the diode. Thus, just after reversing the supply voltage, the
reverse current (opposite direction of current) will be 10 V/100 100 mA.
Step recovery diodes, also known as ‘snap-off-diode’ and ‘charge storage diode’,
are used as frequency multipliers and harmonic generators. These diodes are also
used for generating pulses having very short rise and fall time for its ultra-fast switch-
ing time. For a reverse recovery diode, the fall time t R should be very small as shown
in Fig. 4.1b. Due to the rapid reverse recovery, it generates various higher harmonic
frequency components when RF signal is applied to the diode. To generate the very
sharp current pulse, a special type of impurity profile is required for step recovery
diode [5]. The impurity profile is made such that the generated build-in potential
pushes the minority carriers away from the semiconductor junction. The impurity
doping profile for a step recovery diode is shown in Fig. 4.2. Here, impurity con-
centrations within the p-and n-type semiconductors are further away from its p–i
and n–i junctions. This impurity profile creates very short transition time (t R ) going
4.2 Step Recovery Diode 49
Fig. 4.2 Typical doping

profile of SRD p i n
Anode Cathode
Doping
x
0
from its conduction to non-conduction phase within very short time in its reverse
bias condition.
Under forward bias condition, SRD acts as a normal p–n junction diode. When
the SRD is biased from forward to reverse, initially it has a very high conductivity
of current like normal p–n junction diode. However, this high-conductivity situation
only maintains for a certain period of time. Then, it suddenly stops conducting current.
This property of SRD makes it function as generator of higher order harmonics with
sufficient magnitudes, when driven by a sinusoidal signal. This property of the diode
makes it suitable to use as higher order multiplier and mixer.
Epitaxial p-i-n structure with quite narrow I-layer is used to construct SRDs. Thus,
the charges injected under forward bias condition are stored very close to the p–i and
n–i junctions. Under reverse bias, the narrow intrinsic layer width ensures very low-
dynamic punch-through voltage compared to operating voltage. A significant electric
field is generated throughout the entire I-layer. Thus remaining mobile carriers are
rapidly swept out, causing very sharp current transition from high- to low-final value
of the diode.
4.2.1 Characteristic of SRD
The step recovery diode behaves as an ideal nonlinear capacitor [5]. Ideally, it works
as a two-state capacitor with zero switching time between the two states. In one state,
it acts as a large (ideally infinite) capacitance under forward charge storage condition.
The diffusion capacitance of the I-layer is the forward storage capacitance. In other
state, it acts as a small capacitance (the depletion layer capacitance) under its reverse
charge storage condition. The capacitor charge consists of electrons and holes stored
in the I-layer and are related to the effective minority carrier lifetime τa .
I-V characteristic of an SRD under forward bias condition follows the same
equation as of simple p–n junction diode, and can be written as
Fig. 4.3 SRD equivalent

circuit
Lp
Rc
Cp
rd(vdj) cd(vdj) vdj

q Vd j
Id Vd j , T Io exp −1 (4.1)
ηkT
Thus, under forward bias condition, the SRD is modelled as a p–n junction diode
and a parallel diffusion capacitance to account the stored charge as given by (4.2)
neglecting the depletion charge that is negligible under sufficient forward bias current.
∂ Qd q
cd Vd j τa Id Vd j , T (4.2)
∂ Vd j ηkT
Under reverse bias condition, the junction capacitance will be the depletion layer
capacitance of the diode of area A and depletion layer width W and is given by (4.3)
as follows:
εA
cd Vd j (4.3)
W
The most contribution on W is the width of the intrinsic region thus it is nearly
independent of reverse bias voltage and is very low.
Under forward bias condition, the diffusion capacitance (cd ) is determined by
large amount of free carriers within a volume of depletion width W and area A.
Therefore, it is not possible for instantaneous transition from forward capacitance to
reverse capacitance. An important parameter of SRD is time constant tt for switching
from its forward capacitance to reverse capacitance. The effect of finite switching time
should be taken into account for operating frequencies higher than 1/tt . Equivalent
circuit of SRD is shown in Fig. 4.3 for the operating frequency range of greater than
1/τa and less than 1/tt . At frequencies lower than 1/τa , it is required to consider
the effect of carrier recombination and at frequencies higher than 1/tt ; the effect of
finite switching time should be taken into account.
In a semiconductor, the minority carrier lifetime is strongly temperature depen-
dent. The major temperature effect on SRDs is an increase of minority carrier lifetime
of about 0.5%/°C. This may not be a problem when diodes are used for generating
harmonics for microwave signal generation, as the temperature-dependent optimum
bias voltage is used. In general, this temperature dependency is compensated using
4.2 Step Recovery Diode 51
temperature-dependent resistor for its biasing. However, the problem is severe when
an SRD is used for generation of pulses and the position in time of the pulses is
important. In this case, a change of minority carrier lifetime causes a change in
snap time, this leads to a change in the pulse position in time relative to a driving
waveform.
4.3 Tunnel Diodes
Tunnel diode is based on the electron tunnelling effect discovered by Leo Esaki, thus
also known as Esaki diode. Tunnel diodes have lower 1/ f noise, higher rectification
current sensitivity, low-temperature variation and broadband performance capability.
Tunnel diode is a highly doped p–n junction diode. A part of its I-V characteristic
shows a decrease of current with the increase of voltage. That means its I-V
characteristic exhibits negative differential resistance [6]. The depletion region
becomes very thin ~100 Å, since the p- and n-regions of the tunnel diodes are highly
doped (about 1024 –1025 m−3 ). Due to the very thin depletion layer, electrons can
tunnel from conduction band of n-region of the diode to the valence band of p-region
without change of particle energy. Hence, the flow of current is due to the tunnelling
of particle, thus called tunnelling current and the diode is called tunnel diode.
Tunnel diodes are made using highly doped semiconductor, called degenerate
semiconductor, where for the n-type semiconductor Fermi level goes above the con-
duction band and for p-type semiconductor, it goes below valence band. At zero bias,
there is no current flow as shown in Fig. 4.4a. Under the application of small for-
ward bias, there will be no conventional forward current due to very high-potential
barrier. However, under this small forward bias condition, the current will be due
to the tunnelling of electrons from conduction band of the n-region to the valence
band in the p-region, as shown in Fig. 4.4b. With the application of certain higher
forward bias as shown in Fig. 4.4c, the majority electron’s energy in the n-region
becomes equal to energy of the p-region’s empty states in the valence band. This
results in very high-tunnelling current as shown in Fig. 4.4c. With further increase of
forward bias, the number of n-region’s electrons decrease that are directly opposite
to the p-region’s valence band empty states. Therefore, decrease in the tunnelling
current will start as shown in Fig. 4.4d. This region of the I-V characteristic shows
negative differential resistance of the diode. With further increase of applied forward
bias voltage, tunnelling current decrease and ultimately drops to zero. However, due
to the lowering of potential barrier as shown in Fig. 4.4e, the conventional forward
current of the diode starts to increase due to the injection of electron hole through the
junction. Whereas under reverse bias condition, current flows in the reverse direction
due to the tunnelling of valence band electrons of the p-region directly to the n-region
empty states present in the conduction band as shown in Fig. 4.4f with the increase
of reverse bias tunnelling current increases which leads to increase of the reverse
current.
no bias forward bias reverse bias
(a) (b) (c) (d) (e) (f)

I I I I I I
V V V V V V
Fig. 4.4 Typical I-V characteristic of tunnel diode at various bias conditions
4.3.1 Characteristic of Tunnel Diode
Figure 4.5a shows I-V characteristic of a tunnel. There are two very distinct features
of tunnel diode’s I-V characteristics. First, the I-V characteristic is strongly nonlinear
and second the characteristic consists of a negative differential resistance region.
The negative differential resistance between V p and Vv is important, and when the
diode is used in this region it acts as voltage-controlled negative resistance. The plot
of differential resistance versus diode voltage is shown in Fig. 4.5b. It shows that
negative differential resistance occurs over the diode voltage of V p to Vv with a peak
near about V p . The current ratio I p /Iv is used as figure of merit of a tunnel diode.
The total diode current I d of a tunnel diode is given by [6]

Vd j Vd j m Vd j Vd j − VV q Vd j
Id exp − + exp + Io exp − 1 (4.4)
Ro Vo RV Vex ηkT
The first term is the tunnel current, where m is a constant of value ranging from
1 to 3, and Ro is the tunnel diode resistance in the ohmic region, i.e. near Vd j 0 V.
Second term of the total current expression is the excess current of the tunnel
diode which is due to the additional parasitic tunnelling current. The last term is
conventional p–n junction current. The parameters Vo , V v , RV and Vex are the curve
fitting empirical parameters. The maximum negative differential resistance of the
tunnel diode is given by

exp 1+m
Rnd Ro m
(4.5)
m
The negative resistance of the tunnel diode is used for realization of microwave
oscillators and amplifiers. High-frequency equivalent circuit of a tunnel diode is
4.3 Tunnel Diodes 53
(a) I
La
IP
Q3
Q1
(c)
Q2
Lp
IV Lb
V
0 VP VV Rs
Cp
R C
(b) dI/dV
0 V
VP VV
Fig. 4.5 Typical a current–voltage characteristic and b conductance, c equivalent circuit of the
tunnel diode
shown in Fig. 4.5c. Here, R is differential resistance and Rs is parasitic resistance

due to the lead resistance, ohmic resistance and spreading resistance of the semicon-
ductor. Two typical bias load lines L a and L b are shown in the Fig. 4.5a. Quiescent
operating point Q1 corresponding to the bias load line L a corresponds to a circuit of
resistance less than the negative resistance of the diode. It will provide stable or unsta-
ble amplification/oscillation depending upon the reactance of the circuit. Whereas,
quiescent point Q1 corresponding to the bias load line L b is not stable as it may go to
the point Q2 or Q3 depending upon its previous condition. As the tunnelling current
does not depend strongly on operating temperature, the performance of the tunnel
diode-based circuits is very stable with change in operating temperature.
As there is no time delay in the tunnelling current as present in the drift and
diffusion process, tunnel diodes are capable to operate up to hundreds of GHz range.
However, due to the limitations in high-current handling capacity of the tunnel diode
and also due to the presence of many other high-speed devices based on advanced
material and processes, tunnel diodes are not being used frequently.
(a) I
(b) I
Fig. 4.6 Energy band diagrams of backward diodes with three different doping configurations
a no bias, b reverse bias
4.4 Backward Diode
Backward diode is also a highly doped p–n junction device but it has slightly lower
impurity concentration than the tunnel diode. In case of backward diodes, tunnelling
current flows only in its reverse bias condition. There is no tunnelling current when
a backward diode is forward biased. I-V characteristic of a backward diode has very
sharp knee near 0 V. This characteristic enables rectification of very small magnitudes
of RF signal and is used as sensitive RF detector.
Energy band diagram of backward diodes for three different combinations of
p- and n-region doping concentrations at zero bias and reverse bias conditions are
shown in Fig. 4.6. The three different doping combinations in p- and n-region of the
backward diodes are such that in thermal equilibrium, (i) the Fermi levels of both
p-and n-regions are very close to the band edges, (ii) the Fermi level of p-region is
very close to the band edge, however, Fermi level in n-region is above the conduction
band edge and (iii) the Fermi level of n-region is very close to the band edge, however,
Fermi level in p-region is below the valence band edge.
Energy band diagram and associated current-voltage characteristics of a backward
diode are shown in Fig. 4.7. Under forward bias condition of the backward diode,
there will be negligible (ideally zero) tunnel current thus it behaves like a conventional
p–n junction diode. However, when a small reverse voltage is applied to the backward
diode, electron can easily tunnel from the valence band of p-region to the conduction
band of n-region and give rise to a tunnelling current. The reverse bias tunnel current
(I R ) can be expressed as [6]:

|VR | m
I R A1 exp + (4.6)
A2
4.4 Backward Diode 55
I I I
V V V
(a) no bias (b) forward bias (c) reverse bias
Fig. 4.7 Typical I-V characteristic of backward diode
Fig. 4.8 Typical I

current–voltage
characteristic of backward
diode
V
0
Here, A1 , A2 and m are all positive constants and are very weak functions of the
applied reverse voltage VR . The expression shows that the reverse current of the
backward diode increases exponentially with the applied voltage.
A typical forward- and reverse-biased I-V characteristic of the backward diode is
shown in Fig. 4.8. It shows that for small reverse bias, the current is larger than the
current under the same amount of forward bias voltage. This enables backward diode
to rectify small signals of amplitude smaller than the forward bias cut-in voltage.
The reverse bias current is due to the tunnelling of the carriers. Thus, it is fast due to
absence of minority carriers charge storage effect.
Tunnelling current is insensitive to the change of temperature. Therefore, the
performance of the backward diode is not sensitive to its operating temperature.
4.5 Varactor Diode
Varactor diode is basically a p–n junction or Schottky contact diode usually operated
in reverse bias condition to act as voltage variable capacitance [7]. Varactor diode-
based electronically tunable elements like capacitive switch banks and continuously
variable capacitances are used for voltage variable phase shifters, frequency tunable,
i.e. multiband amplifiers, antennas and oscillators. These elements are used for real-
ization of truly generic transceivers that feature adjustable frequency, impedance, and
system transfer functions for flexible communication systems by dynamic reconfig-
uration. The varactor diodes are also used as a nonlinear device for realization of
multiplier, mixer, linearizer, etc.
Under reverse biased condition, absolute capacitance and the change of capaci-
tance with the variation of reverse bias voltage are two important parameters for a
varactor diode. Absolute capacitance with the other reactance of a circuit determines
the frequency of operation whereas the change of capacitance with the change of
reverse bias voltage determines the range of frequency or phase adjustability of RF
circuit.
Doping profile of a varactor diode determines the absolute capacitance and
change of capacitance over the reverse bias voltage. Doping distribution (N(x))
along the length (x) of the diode can be expressed as
N ∝ xm (4.7)
In case of uniform doping distribution, m 0 and for linearly graded doping

distribution, m 1. Whereas for hyper-abrupt doping distribution, m < 0. The junction
capacitance C j dependency on the applied bias voltage (V ) can be expressed as
C j ∝ (V + Vbi )−1/(m+2) (4.8)
Here, Vbi is the built-in voltage of the diode. Thus, with the increase of reverse
bias voltage, the junction capacitance decreases. Larger the value of 1/(m + 2), the
capacitance variation with the biasing voltage will be larger. This parameter is called
sensitivity of a varactor diode. For the hyper-abrupt junction, sensitivity of the varac-
tor diode increases with the increase of |m|. A typical hyper-abrupt doping profile is
shown in Fig. 4.9. GaAs-based hyper-abrupt Schottky contact layer used to realize a
high-frequency performance varactor diode to operate in the higher microwave and
mmwave frequency range.
Equivalent circuit of a reverse biased varactor diode is shown in Fig. 4.10. Here,
C j is the bias-dependent junction capacitance. R p is the parallel equivalent deple-
tion layer resistance corresponding to reverse bias leakage current. Generally, for
high-quality diodes, R p increases with the increase of reverse bias voltage. Rs is the
equivalent series resistance of the diode consisting of contact resistance, and resis-
tance of the un-depleted regions of the diode. L p and C p are the package inductance
and capacitance respectively.
4.5 Varactor Diode 57
Fig. 4.9 Typical

hyper-abrupt doping profile p+ hyper- n+
of a varactor diode Anode abrupt n Cathode
Doping
p+
x
0
n+
Fig. 4.10 Equivalent circuit

of reverse-biased varactor
diode Lp
Rs
Cp
Rp Cj
The capacitance of a varactor diode also changes with temperature. With the
increase of diode temperature, capacitance increases. Temperature sensitivity of the
diode capacitance increases with increase of capacitance sensitivity with the applied
voltage.
References
1. Streetman BG (1986) Solid state electronic devices, 2nd edn. Prentice Hall of India Pvt. Ltd.,
New Delhi
2. Ko Wen-Hsiung (1961) The forward transient behaviour of semiconductor junction diodes.
Solid-State Electron 3:59
3. Ko WH (1961) The reverse transient behaviour of semiconductor junction diodes. IRE Trans
Elec Dev ED-8:123
4. Kuno HJ (1964) Analysis and characterization of P-N junction diode switching. IEEE Trans
Elec Dev 11:8
5. Moll JL, Hamilton SA (1969) Physical modeling of the pulse and harmonic step recovery diode
for generation circuits. Proc IEEE 57(7):1250
6. Karlovsky J (1967) The curvature coefficient of germanium tunnel and backward diodes. Solid-
State Electron 10:1109
7. Ulriksson B (1979) Continuous varactor-diode phase shifter with optimized frequency response.
IEEE Trans Microw Theor Tech 27(7):650–654
Chapter 5
Microwave Bipolar Transistors
Abstract This chapter describes various types of microwave bipolar junction tran-
sistors (BJTs). Frequency limitation of BJT and structural modification for suit-
able operation at microwave frequency band are also discussed. The chapter also
describes GaAs- and SiGe-based hetero-junction bipolar transistors (HBTs) suitable
for microwave and mmwave applications. Temperature dependency of BJTs and its
effects on circuit performance are also discussed.
5.1 Introduction
Transistors are three-terminal solid-state devices used for realization of amplifiers,

multipliers, mixers, linearizers and various other active circuits for microwave com-
munication systems. A transistor should have sufficient gain at microwave frequency
range for its use as an amplifying device. Thus, the device should have ability of large
linear change of voltage/current at its output due to a small change of voltage/current
at its input. Various types of amplifiers are realized using transistors for communi-
cation systems like low-noise amplifier (LNA), high-gain amplifier and high power
amplifiers (HPA). Transistors required for LNA should amplify the signal with the
addition of minimum noise, i.e. the device should have low-noise figure at microwave
frequency. Power transistors should have higher output current handling capability
with high output breakdown voltage, i.e. it should have high output power handling
capability, whereas transistor’s nonlinear characteristics are used for realization of
multiplier, mixer, linearizer, etc. for communication systems.
Depending on material, doping profile, and device structures, there is a wide
range of transistors suitable for different applications. Among the various types of
transistors, bipolar junction transistor (BJT) was the first three-terminal device and is
still being used for many low-frequency as well as microwave applications. With the
advancement of hetero-junction technology, hetero-junction bipolar transistor (HJBT
or HBT) uses a high bandgap semiconductor as emitter material and hence offers
better performance in terms of gain at higher microwave frequency, and provides
sufficient gain over mmwave frequency regions.

60 5 Microwave Bipolar Transistors
5.2 Bipolar Junction Transistor (BJT)
In spite of the existence of various advanced unipolar junction transistors, the first
developed three-terminal solid-state device, the bipolar junction transistor (BJT), is
still being used an account of its microwave applications for its lower realization
cost, lower 1/f noise and high packing density using well-established foundry. BJTs
are not only used for amplification, but also used as nonlinear devices for realization
of multipliers, mixers and linearizers. Basically, the bipolar junction transistor can be
considered as two back-to-back p–n junction diodes with appropriate doping profile
and structure. A schematic structure of a BJT suitable for microwave applications is
shown in Fig. 5.1, and its cross-sectional view is shown in Fig. 5.2. All microwave
bipolar transistors are n-p-n type due to the higher mobility of electron than holes. The
emitter is heavily n-type doped, shown as n+ , compared to the lightly p-type doping
of base. The collector is also lightly n-type doped. Suppose a BJT is working in the
forward-active mode, i.e. base–emitter junction is forward-biased and base–collector
junction is in reverse-biased condition by applying the bias voltages V BE and V CB ,
respectively, as shown in Fig. 5.2. Energy band diagrams of the BJT under unbiased
and forward-active-bias condition are shown in Fig. 5.3. The basic I–V relationships
across the junctions are same as p–n junction diode. Working principle of BJT is
based on injection of minority carrier electrons into the base from the emitter. The
amount of carrier injection is determined by the applied base-to-emitter voltage
V BE . The injected minority carriers in the base region diffuse across the base region
and are swept to the collector as the base–collector junction is in reversed bias
condition. Actually, the base width W B is much smaller than the diffusion length
of the electrons in the base region. Thus, most of the injected electrons cross the
base region without recombination. The relationship between the emitter current and
base–emitter voltage is same as diode I–V relationship and can be written for the
common-base-bias condition as [1]
n+ n+ n+
E B p C n+
n+
n+
p
wb
Fig. 5.1 A schematic structure and doping profiles of a BJT

5.2 Bipolar Junction Transistor (BJT) 61
Vbe Vcb
B
E C
n+
n+
wb p
Fig. 5.2 Cross-sectional view of the BJT
Fig. 5.3 Band diagram of (a) Emitter Base Collector

a unbiased,
b forward-active-mode-bias WB
BJT
Ec
Ev
(b)
qVBE p
+
n
qVCB
Ec
n
Ev

q(VB E − I B R B − I E R E )
I E Io exp −1 (5.1)
ηkT
Here, Io is the reverse saturation current, and R B and R E are the base and emit-
ter terminal parasitic resistances, respectively. The well-known static common-base
current gain αo is defined as
∂ IC ∂ In E ∂ InC ∂ IC
αo (5.2a)
∂ IE ∂ I E ∂ In E ∂ InC
Here, the first term of the common-base current gain ∂ In E /∂ I E is the emitter
efficiency γ ; the second term, ∂ InC /∂ In E , is the base transport factor αt ; and the last
term is called the collector multiplication factor M. Thus, the common-base current
gain can be expressed as
αo γ αt M (5.2b)
The currents of the three terminals are related by
I E IC + I B (5.3)
Thus, the static common-emitter current gain βo , defined as ∂ IC /∂ I B , can be

expressed as
∂ IC αo
βo (5.4)
∂ IB 1 + αo
The collector current can be written as

q(VB E − I B R B − I E R E )
IC αo I E αo Io exp −1 (5.5)
ηkT
Therefore, an input signal voltage across the base–emitter junction will produce
an output current at the collector terminal. In the presence of bias, consider a small
change of input voltage, i.e. base-to-emitter voltage, then the change of output current
can be determined by
∂ IC q IC
gm (5.6)
∂ VB E ηkT
This is called transconductance gain gm of the BJT. Transconductance of BJT

is very high due to the exponential relationship of collector current with the
base–emitter voltage. The expression suggests that transconductance gain of a BJT
is linearly proportional to the collector bias current, IC .
A typical output characteristic of BJTs in common-emitter configuration is shown
in Fig. 5.4. It shows that over the active region, the collector current (IC ) increases
with the increase of collector to emitter voltage (VC E ). With the increase of VC E , the
base width W B decreases, and this causes an increase in current gain βo . The increase
of current gain with the increase of VC E results in the increase in collector current.
The lack of saturation of the collector current in the active region is known as early
effect. The voltage axis intercept point with the extrapolated output curves is called
the early voltage as denoted by V A as shown in Fig. 5.4.
Equivalent circuit model of a forward-active-bias BJT in common-emitter con-
figuration, including the terminal parasitic components, is shown in Fig. 5.5 [2].
Here, R B , R E , and RC are the terminal parasitic ohmic resistances in series with
base, emitter and collector terminals and C E B and CC B are parasitic capacitances
associated with both the p–n junctions. Output of the intrinsic transistor consists of
IC (mA)
ΔIC
ΔVCE
VCE (Volt)
VA
Fig. 5.4 A typical I–V characteristic (output characteristic) of common-emitter BJT for different
base currents
RB RC
B C
IC /α
RE
Fig. 5.5 Equivalent circuit of forward-active mode-biased BJT
a current source IC given by (5.5) with parallel combination of output resistance ro

as given by
∂ VC E
ro (5.7)
∂ IC
The input of the intrinsic transistor is equivalent to a p–n junction diode to model
the dynamic resistance of the base-to-emitter junction, and Cd is the diffusion capac-
itance to model the carrier storage effect.
Example 5.1 Common–base current gain of a bipolar transistor is 0.99. Calculate

the common-emitter current gain of the transistor.
Solution
Common–base current gain of the transistor
α 0.99
Common-emitter current gain

α 0.99
β 99
1−α 1 − 0.99
Example 5.2 A BJT is having common-emitter current gain 100. Considering 10 V

supply, V BE 0.7 V and RE 100 , what is the value of RB , and RC to set the
quiescent point at I C 10 mA and V CE 8 V.
10 V
RB RC
RE = 100
Solution
Here, collector current IC 10 mA
Common-emitter current gain β 100
Thus, base current
I B IC /β 10/100 mA 0.1 mA
Therefore, emitter current
I E IC + I B (10 + 0.1) mA 10.1 mA
Thus, voltage drop across the emitter resistance
VE I E × R E 10.1 mA × 100 1.01 V

Thus, voltage drop across the base resistor R B
VB (10 − 0.7 − 1.01) V 8.29 V
Thus, base resistor
R B VB /I B 8.29 V/0.1 mA 82.9 k
Collector resistor
(10 − 1.1 − 8)V
RC 90
10 mA
5.2.1 Frequency Limitation of BJT
At high frequencies, BJTs are used to realize amplifier, multiplier, mixer, linearizer
and other circuits. BJT should provide sufficient gain over microwave frequency
range to realize high-gain/low-loss circuits. High-frequency performance of a BJT
is limited mainly by the minority carrier storage effect in the base region, time delay
due to various parasitic capacitances, and other transit time delays along the carrier
flow path through the device [2, 3].
Carrier injected from the emitter pass through the base region by diffusion process
due to the gradient in distribution of the stored minority carriers in the base and go
to the collector. The diffusion process is relatively slower than the carrier transport
in the FET channel which is governed by drift of carrier by the applied electric field.
The carrier storage in the base region can be modelled by a diffusion capacitance Cd
and is given by
∂ QB ∂(IcB )
Cd gm τ B (5.8)
∂ VB E ∂ VB E
Here, Q B is the stored charge in the base region and τ B is the forward transit
time of the carriers. The transit time τ B can be considered as the average time a
carrier spends in the base region before going into the collector. The transit time is
determined by the structure and material properties of the BJT and is given by [1]
W B2
τB (5.9)
γ Dn
Here, W B is the un-depleted width of the base region as shown in Fig. 5.6, Dn is
the electron diffusion coefficient and γ is a constant depending on the build-in field.
In case of uniformly doped base region, γ 2. The transit time τ B determines the
alpha cut-off frequency f α and can be expressed in terms of the charging time τ B as
Wb
Wb WC
E C
Fig. 5.6 Diagram showing depletion layers of forward-active-biased BJT
1
fα (5.10)
2π τ B
From this it is clear that for microwave BJT, the base region width W B should be
as small as possible. Though minority carrier storage time limit the high-frequency
response, BJTs can work very well around the lower microwave frequency range
due to the possibility of achieving very small base width (less than 100 nm) in Si
technology.
The microwave BJTs, in which τ B is reduced by minimizing the base width,
are limited by the device fabrication technology, and delays due to the various para-
sitic capacitors which have become the dominant part of its high-frequency perfor-
mance limitation.
Example 5.3 Calculate alpha cut-off frequency of Si transistor with uniformly
doped base region of width 0.25 μm. Consider electron diffusion coefficient for
Si, Dn 20 cm2 /s.
Solution
The alpha cut-off frequency is given by
1 1 2Dn 1 Dn
f∝ 2

2π τ B 2π W B π W B2
Putting W B 0.25 μm, Dn 20 cm2 /s.

1 20
f∝ Hz
π 0.25 × 10−4 2
10.2 GHz
Various parasitic capacitors and resistors will contribute to the total delay time.
One of these delay times which determines the higher cut-off frequency of BJT is
Fig. 5.7 Common-emitter β

current gain of BJT with its
collector current
IC
the emitter charging time, τ E , which is related to the capacitance of depletion layer
and can be expressed as
C E B + CC B
τE (5.11)
gm
Thus, to minimize the charging time τ E , high gm is required. To achieve this,

emitter region should be heavily doped compared to base doping and as per (5.6) the
transistor should also be operated at sufficiently high current level. With the increase
of collector current, current gain initially increases and attains a maximum value.
For further increase of collector current, the injected minority carrier density from
emitter to the base approaches the majority carrier density and causes a decrease of
emitter efficiency. The decrease in current gain with the increase of collector current
is known as Webster effect as shown in Fig. 5.7 [4].
Another factor of delay time which also contributes to the higher cut-off frequency
of BJT is the base–collector junction depletion layer delay and can be expressed as
WC
τC (5.12)
vs
where WC is the width of the collector–base junction depletion layer and vs is the
carrier saturation velocity.
Therefore, the high-frequency cut-off ( f C ) of the BJTs can be expressed as
1
fC (5.13)
2π (τ B + τ E + τC )
or
2 −1
WB C E B + CC B WC
f C 2π + + (5.14)
γ Dn gm vs
Fig. 5.8 Typical

dependency of cut-off
frequency on output current
From this expression, it is clear that for achieving higher cut-off frequency ( f C ),
the BJT should have (a) a very narrow base width W B , (b) higher transconductance
gain gm and (c) narrow collector region WC .
Base width W B is limited by the fabrication technology, and the present Si technol-
ogy is capable to fabricate submicron base width. Higher value of transconductance
can be achieved by operating the transistor at higher current level. However, a max-
imum current level exists up to which the current gain increase; above this level
the current gain decreases and the high-frequency performance drops as shown in
Fig. 5.8. Actually, under high current conditions, effective base width increases and
extended towards the collector region which degrades the high-frequency perfor-
mance. This effective increase of base width at very high-level injection of current
is known as Kirk effect [5]. To maximize the higher cut-off frequency, collector
width should be minimized. However, with decrease of collector width, there will
be decrease in collector breakdown voltage, which limits the high-voltage operation
for high-power handling.
Another way to increase common-emitter current gain is by increasing the doping
density of emitter region compared to the doping density of base. This requirement
may be met by lowering the base region doping. However, lower base doping will
increase the base resistance that will adversely affect the BJT performance. The
other way to meet this requirement is to dope the emitter heavily maintaining base
doping lower than emitter to ensure lower base resistance. However, heavy doping
of emitter region will lead to bandgap shrinkage called the bandgap narrowing effect
[6]. Therefore, for a fixed-base doping, with the increase of emitter doping, though
initially the current gain increases, the current gain starts to decrease beyond certain
level of doping.
From the above discussion, it is clear that the requirements of heavy emitter
doping, lower base doping, smaller base width, smaller collector width, etc. are
conflicting requirements and cannot be met by simple structure of BJT. To overcome
these conflicting requirements, new materials with advanced fabrication technology
are used to develop advanced bipolar devices such as SiGe and GaAs HBTs.
5.2.2 Temperature Behaviour of BJT
Common-emitter configuration of a BJT is the most extensively used configuration

for its largest current and power amplification. Almost all the expressions of the BJT
parameters as derived in earlier sections contain temperature T [7]. This is due to the
fact that the current is governed by the minority carrier density, which is determined
by the temperature in addition to other factors. The collector current (IC ) of a BJT is
strongly related to the temperature through the exponential temperature dependency
of reverse saturation current Io as expressed in (5.5). Temperature dependency of
static characteristic of BJT is shown in Fig. 5.9. The collector current increases with
the increase of temperature for a fixed-base-bias current. In addition to these com-
ponents, the other parasitic components also have temperature dependency. Though
common-emitter current gain varies differently with temperature in different types
of bipolar transistors, however, in most cases it increases with increasing tempera-
ture. Therefore, the common-emitter current gain as well as transconductance gain
increases with the increase of temperature which leads to positive temperature coef-
ficient of power gain in BJT-based microwave amplifiers.
In addition to the temperature dependency of gain of BJT-based amplifiers, the
higher cut-off frequency is also sensitive to the change of operating temperature.
The diffusion coefficient Dn , transconductance gain gm and saturation velocity vs of
the carrier determine the higher cut-off frequency as expressed in (5.14). The most
dominant factor which determines the higher cut-off frequency is the stored charge
effect in the base region which is governed by diffusion coefficient. The diffusion
coefficient is determined by the Einstein relationship Dn μkT /q. As μ ∝ T −2
(approximately) in the range of operating temperature for Si, the diffusion coefficient
decreases with the increase of temperature and thus decreases the cut-off frequency.
At high-power applications, due to the power dissipation (IC × VB E ) within the
BJT, there is thermal heating and rise of the device temperature. With this increase
of temperature, there will be further increase of current. With the further increase of
Fig. 5.9 Temperature IC (mA)

dependency of static output
characteristic of BJT IB2> IB1
T2
T2> T1
IB2
T1
T2 IB1
T1
VCE (Volt)
current, there will be increase of heat dissipation. With this process, the device can
burn out, known as thermal runaway, if proper protection is not taken in the circuit
design. Thus, the temperature sensitivity of the BJT affects the performance of the
circuit. In circuit level, various compensation mechanisms are adopted to minimize
the temperature-related performance variations and protection from thermal runaway.
5.3 Hetero-junction Bipolar Transistor (HBT)
Bipolar junction transistors are made with the same material, in general Si, for its base,
emitter and collector regions. In case of hetero-junction bipolar transistors (HBTs),
different (at least two) materials, i.e. materials having different bandgap energies, are
used for different regions [8–19]. Thus, base–emitter and/or base–collector junctions
are formed with different bandgap materials; hence, these devices are called hetero-
junction device.
In case of homojunction bipolar transistor, it is seen that to achieve higher cut-off
frequency, the emitter doping should be much higher than the base region doping and
the base width should be as small as possible. This requirement can be met by having
a thin base with low doping density. However, a thin base with low doping creates
a large base resistance which degrades the device performance. Another way to
meet the requirement is doping the emitter region heavily with moderate base region
doping. However, heavy doping in the emitter region causes bandgap shrinkage
problem. Therefore, though at the lower doping level of base region, current gain
increases with the increase of emitter doping, it starts to decrease beyond certain
doping level.
From the above discussion, it is clear that these conflicting requirements cannot
be met by BJT using a single-material structure. Hetero-junction bipolar transis-
tor (HBT) meets these requirements due to its hetero-junction construction [8].
Schematic structural diagram of an HBT is shown in Fig. 5.10. Emitter of an
HBT is made with a wider bandgap material such as GaAlAs, whereas base is
made with narrower bandgap material such as GaAs. Aluminium gallium arsenide,
Alx Ga1−x As, is a ternary semiconductor alloy, where x is mole fraction of aluminium
in the alloy and the value is in between 0 and 1. The Alx Ga1−x As has very nearly
the same lattice constant as GaAs, and at 300 K its energy bandgap is given by [25]
E g (x) 1.424 + 1.155x + 0.37x 2 for x < 0.43 (5.15a)

1.424 + 0.125x + 0.143x 2
for x > 0.43 (5.15b)
Therefore, the energy bandgap of Alx Ga1−x As varies in between 1.424 eV corre-
sponding to GaAs (x 0) and 2.168 eV corresponding to AlAs (x 1). As shown
in the diagram, for x 0.28, the bandgap energy of the Al0.28 Ga0.72 As material is
1.776 eV, about 0.35 eV larger than the bandgap energy of GaAs material. Thus,
5.3 Hetero-junction Bipolar Transistor (HBT) 71
Emitter
Base Emitter is with wider

n type (Eg1)
bandgap material
Collector p type (Eg2)
n type (Eg2) Eg1 > Eg2
n+ sub-collector (Eg1)
Semi insulating substrate
Fig. 5.10 Schematic basic structural diagram of HBT
emitter–base junction becomes hetero-junction whereas collector–base junction is

homojunction.
Energy band diagrams of an HBT under no-bias and forward–active-bias condi-
tions are shown in Fig. 5.11. Dotted lines represent abrupt p–n hetero-junction and
solid lines represent graded junction. As marked in the figure,
E c and
E v are the
conduction band and valence band discontinuities due to the abrupt hetero-junction.
In case of graded hetero-junction, the notch in the conduction band discontinuity
is very much reduced. The figure shows that the barrier to hole injection from the
p-type base to the n-type emitter region is qφ Bh , whereas barrier to electron injection
from n-type emitter to p-type base is qφ Be . It is to be noted that the barrier to the
holes in both abrupt and graded hetero-junctions are same, whereas it is reduced for
electron barrier, which provides improved performance. The dependence of electron
current In and the hole current I p on the barrier potential can be expressed as
In ∝ exp(−qφ Be /kT ) (5.16a)

I p ∝ exp(−qφ Bh /kT ) (5.16b)
In case of emitter–base homojunction, the potential barrier seen by electrons and

holes is the same and only means for controlling the currents are relative doping den-
sities. For hetero-junction, barrier seen by holes is more than the barrier seen by the
electrons. As seen in the Fig. 5.11, in graded hetero-junction, the bandgap difference
E g between emitter material and base material appears across the valence band
barrier which leads to highly diminishing of hole current contribution compared to
the electron current contribution in the total emitter current. The reduction of hole
current contribution in total emitter current for HBT with respect to the case for
homojunction BJT can be expressed as

I pE for HBT
E g
exp − (5.17)
I pE for BJT kT
EMITTER BASE COLLECTOR

E C
n AlGaAs p GaAs n GaAs
ΔEc
(a)
EG2 Ec
EG2
ΔEv
EG1
Ev
EG1 > EG2

qφBe
(b)
p
qVBE
n+
qφBh
qVCB Ec
n
Ev
Fig. 5.11 Energy band diagram of a unbiased, b forward-active-mode-bias HBT
Due to the exponential relationship of the hole current reduction with the dif-
ference of bandgap, emitter efficiency ∂ In E /∂ I E in HBT increases tremendously.
Therefore, the advantages for the hetero-junction bipolar transistors (HBTs) com-
pared to the homojunction bipolar transistors (BJTs) are (a) higher emitter efficiency,
due to the higher barrier potential in the valence band for the hole injection from
base to emitter, and (b) reduced base resistance, because base can be heavily doped
without sacrificing the emitter efficiency. Both the advantages ultimately lead to the
improved frequency response due to the achievement of higher current gain.
In the high-power wireless communication, HBT technologies are becoming
important for their distinct advantages compared to FET technology. HBTs typi-
cally exhibit much higher device transconductance, as a result of the exponential
relationship between the input voltage and the output current, as opposed to the
quadratic relationship for FETs. In an HBT, the turn-on voltage is determined by
the intrinsic material properties of the epitaxial layers, whereas for FET, its turn-on
voltage is dependent on the actual dimensions of the device, and therefore fluctuates
significantly between devices. Furthermore, electron transport occurs in the vertical
direction in case of HBTs, whereas laterally in a FET that leads to the overall smaller
area for the HBT. A greater power density is thus achieved, and integrated circuits can
be designed using less die area for HBT-based circuits. HBTs also provide excellent
linearity and significantly lower 1/f noise. There are varieties of HBTs based on the
different material technologies which are discussed in the following sections.
5.3.1 SiGe HBT
The silicon–germanium hetero-junction bipolar transistors (SiGe HBT) are realized

using silicon processing compatible technology [9–12]. The base layer of the SiGe
HBT developed using strained and composition-graded SiGe. The SiGe HBTs pro-
vide RF performance comparable to GaAs technologies with fabrication cost and
reliability similar to the silicon process. This is an Si-based device, and thus it has
intrinsically higher thermal conductivity than GaAs HBTs. In comparison to the Si
BJTs, it also offers the flexibility of bandgap engineering as well as base and emitter
doping adjustment capabilities to achieve better RF performance.
Because of the much matured Si technology with high reliability and low-cost,
Si-based devices are more competitive with their III–V group semiconductor-based
counterparts, even though the latter show superior high-frequency performance. The
smaller bandgap of base layer of Si1−x Gex causes an increase in current gain in
n-Si/p-Si1−x Gex /n-Si HBT. Schematic basic structure of an SiGe/Si HBT is shown
in Fig. 5.12. Practically graded-base SiGe HBT is realized in which the Ge is graded
across the base with the higher Ge content at the collector side. In case of silicon
homojunction transistors, the valence and conduction bands are flat, implying that
there is no electrostatic field in the base. However, alloy grading of the Ge in the
base region of a graded SiGe HBT changes the band structure and introduces a
quasi-electric field that drives the electrons across the base. The graded bandgap in
the base region of SiGe HBT results in rapid transport of minority carriers through
the base region. With the increase of Ge mole fraction x, the cutoff frequency of the
HBT increases. Hetero-junction bipolar transistor based on the SiGe/Si technology
material also shows its higher cutoff frequency beyond 350 GHz.
The processing compatibility of SiGe and Si-CMOS technology leads to a power-
ful combination technology called BiCMOS technology for high-frequency mixed-
signal circuits [13]. The BiCMOS circuit consists of both bipolar hetero-junction
transistors (HBTs) and MOS transistors on a single substrate. This enables com-
bination of high-speed switching and high output driving capabilities of the bipo-
lar junction transistors, and the low-power and high-density characteristics of the
CMOS circuit. This aids realization of low-cost very compact integrate transceiver
for microwave wireless communication, particularly mobile handsets. This tech-
nology enhances output and gate drive capabilities for high-speed logic and power
switching circuits, which are required for specialized analog applications.
Emitter
Base Emitter is with wider

n Si
bandgap material
Collector p Si1-xGex
n Si
n+ Si subcollector
Semi insulating substrate
Bandgap of Si > band gap of SiGe
Fig. 5.12 Schematic structural diagram of Si/SiGe-based HBT
5.3.2 III–V Group Semiconductor HBT
III–V group semiconductor-based HBTs are attractive for high-speed applications

due to their excellent electron transport properties which lead to improved high-
frequency performance. Semi-insulating property of GaAs material due to higher
bandgap energy makes it suitable for monolithic microwave integrated circuits.
Due to its semi-insulating (i.e. high resistivity) property, the parasitic capacitance
in between substrate and the other terminals is very less and can have improved
high-frequency performance. Schematic structural diagram of three different basic
HBTs, i.e. n-AlGaAs/p-GaAs, n-InP/p-InGaAs and n-InAlAs/p-InP are shown in
Fig. 5.13 [20–22]. The Inx Ga1−x As systems with high indium content have several
unique advantages. Their high electron mobility and large peak saturation velocity
extend the performance of electronic devices to higher frequency at lower power
dissipation. HBT structure based on the InP/InGaAs is shown in Fig. 5.13b, and the
schematic HBT structure based on InAlAs/InGaAs is shown in Fig. 5.13c. These
are very simple structures; actual devices consist of many other layers with different
doping densities for improving the performance of HBTs. The complicated struc-
tures are made by growing epitaxial layers using metal beam epitaxy (MBE) and
metal–organic chemical vapour deposition (MOCVD) system.
5.3.3 GaN HBT
The GaN material system has enormous potential for high-power applications across
a range of frequencies. High breakdown field, high electron saturation velocity and
high thermal conductivity of GaN make it especially useful for delivering high output
power at high frequencies for wireless base stations and satellite communications.
(a) E (b) E (c) E
B n AlGaAs B n InP B n InAlAs
C p GaAs C p InGaAs C p InGaAs
n GaAs n InGaAs n InAlAs
n+ GaAs subcollector n+ InGaAs subcollector n+ InAlAs subcollector
Semi insulating GaAs Semi insulating InP Semi insulating InP
Al0.28Ga0.72As : Eg ~ 1.774 eV InP : Eg ~ 1.34 eV In0.52Al0.48As: Eg ~ 1.4 eV

GaAs : Eg ~ 1.424 eV In0.53Ga0.47As: Eg ~ 0.75 eV In0.53Ga0.47As: Eg ~ 0.75 eV
Fig. 5.13 Schematic structural diagram of GaAs/AlGaAs-based HBT
Fig. 5.14 Simplified E

structural diagram of
InGaN/GaN-based HBT B n GaN
p InGaN
C n GaN C
GaN Sub-collector
In the high-power wireless communication, GaN-based HBT is becoming one of the

most important semiconductor device technologies for its distinct advantages com-
pared to FET technology [23, 24]. A simplified structural diagram of an InGaN/GaN-
based HBT is shown in Fig. 5.14. Here, narrower bandgap material InGaN is used
as base for HBT action. The energy bandgap of the Inx Ga1−x N is given by
E g (x) 3.427 − 3.86x (x < 0.25) (5.18)
Practically graded-base InGaN HBT is realized in which the In is graded across

the base with the higher In content at the collector side. The graded bandgap in the
base region of InGaN HBT results in rapid transport of minority carriers through the
base region leads to improved high-frequency performance.
References
1. Sze SM (1981) Physics of semiconductor devices, 2nd edn, Wiley Eastern Limited
2. Schwierz F, Schippel C (2005) Performance trends of Si-based RF transistors. In: 2005 IEEE
conference on electron devices and solid-state circuits, Dec 2005, pp 299–304
3. Johnson E (1965) Physical limitations on frequency and power parameters of transistors. IRE
Int Conv Rec 13:27–34
4. Webster WM (1954) On the variation of junction-transistor current amplification factor with
emitter current. IRE Proc 42(6):914–920
5. Kirk CT (1962) A theory of transistor cutoff frequency fT falloff at high current densities. IRE
Trans Electron Devices 9(2):164–174
6. Lanyon HPD, Tuft RA (1978) Bandgap narrowing in heavily doped silicon. In: IEEE technical
digest on international electron device meet, p 316
7. Arora N, Hauser J, Roulston D (1982) Electron and hole mobilities in silicon as a function of
concentration and temperature. IEEE Trans Electron Devices ED-29:292
8. Kroemer H (1957) Theory of a wide-gap emitter for transistors. Proc IRE 45(11):1535–1537
9. Bardin JC (2009) Silicon-germanium heterojunction bipolar transistors for extremely low-noise
applications. Ph.D. thesis, California Institute of Technology, Pasadena, California
10. Cressler J, Crabbe E, Comfort J, Stork J, Sun J-C (1993) On the profile design and optimization
of epitaxial Si- and SiGe-base bipolar technology for 77 K applications. IEEE Trans Electron
Devices 40(3):542–556
11. Krithivasan R, Lu Y, Cressler J, Rieh J-S, Khater M, Ahlgren D, Freeman G (2006) Half
terahertz operation of SiGe HBTs. IEEE Electron Device Lett 27(7):567–569
12. Yuan J, Krithivasan R, Cressler J, Khater M, Ahlgren D, Joseph A (2007) On the frequency
limits of SiGe HBTs for terahertz applications. In: Bipolar/BiCMOS circuits and technology
meeting, 2007, BCTM’07. IEEE, 30 2007–2 Oct 2007, pp 22–25
13. Pruvost S, Delcourt S, Telliez I, Laurens M, Bourzgui N, Danneville F, Monroy A, Dambrine
G (2005) Microwave and noise performance of SiGe BiCMOS HBT under cryogenic temper-
atures. IEEE Electron Device Lett 26:105–108
14. Woods B, Mantooth H, Cressler J (2007) SiGe HBT compact modeling for extreme tempera-
tures. In: 2007 international semiconductor device research symposium, Dec 2007, pp 1–2
15. Tasker P, Fernandez-Barciela M (2002) HBT small signal T and Pi model extraction using a
simple, robust and fully analytical procedure. In: IEEEMTT-S international microwave sym-
posium digest, 2002, vol 3, pp 2129–2132
16. Gobert Y, Tasker P, Bachem K (1997) A physical, yet simple, small-signal equivalent circuit
for the heterojunction bipolar transistor. IEEE Trans Microw Theory Tech 45(1):149–153
17. Jia S, Yang R, Guo H (2011) SiGe heterojunction bipolar transistor and its applications in
microwave communication systems. In: 7th international conference on wireless communica-
tions, networking and mobile computing (WiCOM), Sept 2011
18. Escotte L, Phillippe Roux J, Plana R, Graffeuil J, Gruhle A (1995) Noise modeling of microwave
heterojunction bipolar transistors. IEEE Trans Electron Devices 42(5):883–889
19. Schamacher H, Erben U, Gruhle A (1994) Low noise performance of SiGe heterojunction
bipolar transistors. In: Proceedings of IEEE MTT-S microwave symposium, pp 1167–1170
20. Monier C, Cavus A, Sandhu RS, Lange MD, Chang PC, Sawdai DJ, Gambin VF, Gutierrez
Aitken AL (2004) High indium content InAlAs/InGaAs HBT technology for low-power, high-
speed applications. In: 205th meeting, The Electrochemical Society Inc.
21. Griffith Z, Lind E, Rodwell M, Fang X-M, Loubychev D, Wu Y, Fastenau J, Liu A (2007)
Sub-300 nm InGaAs/InP type-i DHBTs with a 150 nm collector, 30 nm base demonstrating
755 GHz fmax and 416 GHz ft. In: IEEE 19th international conference on indium phosphide
and related materials, 2007, IPRM’07, May 2007, pp 403–406
References 77
22. Radisic V, Sawdai D, Scott D, Deal WR, Dang L, Li D, Cavus A, Fung A, Samoska L, To R,
Gaier T, Lai R (2008) Demonstration of 184 and 255-GHz amplifiers using InP HBT technology.
IEEE Microw Wirel Compon Lett 18(4):281283
23. McCarthy LS (1999) AlGaN/GaN heterojunction bipolar transistor. IEEE Electron Device Lett
20(6):277–279
24. McCarthy LS, Smorchkova IP, Xing H, Fini P, Limbo J, Pulfery DL, Speck JS, Rodwel MJW,
DenBaars SP, Mishra UK (2001) GaN HBT: toward an RF device. IEEE Trans Electron Devices
48(3):543–551
25. Allford Craig (2016) Resonant Tunnelling in GaAs/AlGaAs Triple Barrier Heterostructures.
Ph.D. dissertation, Cardiff University
Chapter 6
Microwave Field Effect Transistors
Abstract This chapter presents structures, working principles and various character-
istics of metal–semiconductor FET (MESFET) and different types of heterojunction
FETs (HFETs) including pseudo-morphic HEMT (pHEMT), meta-morphic HEMT
(mHEMT) and GaN-based HEMTs. Equivalent circuits of the FETs and details of
the associated elements are also discussed.
6.1 Introduction
Field effect transistors (FETs) are the most utilized three terminal semiconductor
devices in microwave and mmwave frequency bands. They consist of a semicon-
ductor active channel connected to two ohmic contact terminals named source and
drain. Another terminal called gate to which voltage is applied to controls current
flow through the channel by changing the electric field, and hence the name field
effect transistor. There are various types of field effect transistors based on the use
of different materials and structures with different doping profiles [1–23]. FETs can
be made from a wide variety of materials like Si, SiGe, GaAs, InP, InGaAs, GaN,
SiC, etc. These are unipolar devices, hence only one type of carrier, mostly elec-
trons, takes part for the channel current. Being a unipolar device, FET does not
suffer from minority carrier storage delay time, and thus enhances the higher cut-off
frequency compared to a bipolar device like BJT. FETs are negative temperature coef-
ficient devices, thus with the increase in temperature, current of the device decreases,
this prevent damage to FETs from thermal runaway. Field effect transistors based
on III–V group semiconductors provide superior performances at microwave and
mmwave frequency bands. The superior performances are for their higher electron
velocity compared to Si semiconductor as shown in Fig. 6.1 [1]. Electron velocity of
GaAs and InGaAs are higher at low electric field, thus these materials are suitable
for low power high-frequency performance. Whereas, electron velocities of GaN and
SiC materials are higher at higher electric field, thus these are suitable for high power
at higher frequency applications.

80 6 Microwave Field Effect Transistors
Electron Velocity
Electron Velocity
(107 cm/s)
(107 cm/s)
3 InGaAs
GaN 3
InP
2 2
GaAs
SiC
1 Si
1
GaAs Si
0 0
0 100 200 300 400 0 5 10 15 20
Electric Field (kV/cm) Electric Field (kV/cm)
Fig. 6.1 Electron velocity with electric field of different materials
Metal–semiconductor field effect transistors (MESFETs) based on GaAs is the

better microwave device compared to junction field effect transistors (JFETs) and
metal oxide semiconductor field effect transistors (MOSFETs) with respect to gain,
noise figure and power handling capability at microwave and mmwave frequency
bands. Schottky barrier gate-channel junction is used in MESFET due to its superior
high-frequency performance compared to p–n junction and MOS junction used in
JFETs and MOSFETs, respectively.
Hetero-structure field effect transistor (HFET) uses dissimilar bandgap materi-
als in their structures that offer further enhanced performance compared to single
material structure MESFET at microwave and mmwave frequency range. HFETs
are also known as modulation doped FET (MODFET) and high-electron-mobility
Transistor (HEMT) because of the doping variations within the structure and higher
electron mobility. Depending on the structure, doping profile and materials, there are
different variations of HFETs suitable for various applications. InP-based HEMT
provides the highest frequency of operation with the best noise figure. On the other
hand, high bandgap semiconductor such as GaN- and SiC-based HFETs provides
hundreds of watt microwave power due to their high breakdown voltage and higher
thermal conductivity.
FETs are used as linear, and nonlinear devices for various types of amplifiers
and control devices. These are used for the realization of low noise amplifier (LNA),
high gain amplifier and high power amplifier. GaAs material is more preferable
than the Si material for MESFET due to the former’s better low field mobility,
8500 cm2 /V-s compared to 1300 cm2 /V-s of Si in addition to many more advantages.
Power handling capability of GaAs FETs is about 1 W/mm gate width whereas it is
about 10 W/mm for GaN-based power FETs. As nonlinear device, these are used to
6.1 Introduction 81
realize multiplier, mixer and oscillator. FETs are also used as RF control devices to
realize switches and attenuators in which they dissipate essentially no extra power
from the control signal and can easily be designed into broadband circuits.
6.2 Metal–Semiconductor Field Effect Transistors
Metal–semiconductor field effect transistors (MESFETs) consist of metal–semicon-

ductor Schottky barrier gate contact to control the current flow through the active
channel [1–4]. MESFET structure consists of a conducting channel positioned in
between a source and a drain contact region. Schematic structural diagram of a MES-
FET is shown in Fig. 6.2. In between source and drain contact region Schottky contact
is placed which controls the carrier flow through the channel from source to drain. The
channel current between source to drain is controlled by varying the depletion layer
width which modulates the thickness of the conducting channel. The dimensions of
the structural diagram are not drawn in proper scale. The length of the gate is very
critical for determining the maximum frequency limit. For microwave/millimetre
wave frequencies, the gate length L g is of the order of 0.1 μm. Whereas, the channel
depth, a is only about one-tenth of the gate length, i.e., a ≈ 0.1 × L g . Shortening
of gate length, maintaining the aspect ratio of gate length to channel depth is one of
the most critical technological challenges. High-frequency FETs as well as the fab-
rication technology are called in terms of the gate length and material. For example,
GaAs FETs of 0.25 μm gate length is called 0.25 μ GaAs FET and the technology
are known as 0.25 μ GaAs technology. The width (Z ) of a FET is responsible for
maximum current, i.e., power handling capability. For low noise and low power appli-
cations, relatively smaller gate width devices are used. Larger gate width devices are
used for higher power applications. To increase the power handling capability, FETs
are designed with larger periphery configuring the gate structure with multiple gate
fingers.
Invariably n-channel MESFETs are preferable for higher electron mobility than
hole mobility. Under the source and drain terminals, n+ doping is used to make
both the contacts ohmic. At zero gate bias, as shown in Fig. 6.2, the active channel
is partially depleted. When a negative voltage on the gate terminal is applied, it
reverse-biases the gate Schottky diode which will increase the depletion region thus
constricting the channel width. The control of the channel width is done by varying
the depletion layer width under the gate contact that modulates the thickness of the
conducting channel for controlling the current between source and drain. In this way,
the source to drain carrier flow is controlled by the Schottky metal gate as shown in
Fig. 6.3. Thus, by controlling the gate-bias voltage, MESFETs are used as voltage-
controlled RF resistor. Figure 6.4 shows the change in drain current (I d ) with the
variation of drain-to-source voltage (V ds ) at different fixed gate control voltage (V g ).
The current–voltage relationship over this lower range of drain-to-source voltage is
linear. Over this region, MESFETs are used as linear RF resistor. As the gate reverse
bias voltage is increased the channel current decreases until the channel is pinched
Fig. 6.2 Schematic structural diagram of MESFET
(a) (b) (c)

S G D S G D S G D
 
 
n+ n+ n+ d n+ n+ n+
   
  
  
  
  
  
  
 
  
a
  
   


  
  
    
n-type channel

     
  
   

    
   

    
   

 
 


   
   

  
   
   
   
   
   
   
   
   
   
   
   
   
   
    
  
S. I. substrate S. I. substrate S. I. substrate
Fig. 6.3 Schematic structural diagram of MESFET with very small drain bias with a zero gate
bias, b moderate reverse gate bias and c reverse gate bias up to pinch off
Id
Vg1
Id |Vg1| < |Vg2| < |Vg3|
Vg Vds Vg2
Vg3
VDS
Fig. 6.4 Channel current is linearly related with the drain voltage. With the increase of gate reverse
voltage, channel opening decreases thus increases the channel resistance
off and in this condition, there will be no free carrier within the channel under the
gate. This situation is shown in Fig. 6.3c.
MESFETs at zero drain bias are used as variable resistor to realize voltage vari-
able amplifier or attenuator (VVA). Gate-bias voltage is used to control the channel
resistance, i.e. resistance in between drain and gate terminals which are used as linear
6.2 Metal–Semiconductor Field Effect Transistors 83
Fig. 6.5 MESFETs are used RF RF RSISTOR RF

as voltage variable RF IN OUT
resistor for realization of
VVA and switches
Control
voltage
RF resistor where the drain and source terminals are used as RF ports as shown in
Fig. 6.5. Ideally, the gate is isolated from the channel, thus there will be no current
through the gate terminal, thus it dissipates essentially no power of the control signal
and can be easily designed into broadband circuits. The RF signal flows from source
to drain, and the RF isolated gate is used for controlling the resistance value. The
channel resistance (Rc ) can be expressed as [1]
1 channel length
Rc × (6.1a)
channel conductivity channel area
1 Lg
Rc × (6.1b)
q Nd μn Wg a − d Vg
Here, Nd is the channel electron density which is equal to the free carrier den-
sity within the channel, μn is the low-field electron mobility, L g is the gate length
approximately equal to the channel length, Wg is the gate periphery, a is channel
depth and d(Vg ) is the depletion layer depth which is a function of the applied gate
control voltage. Linearity of the MESFETs as VVA maintained as long as the oper-
ating RF power level is very small compared to the applied gate control voltage.
Otherwise, depletion layer width will be modulated by the electric field due to the
signal amplitude in between source–drain ports.
MESFETs are also used as microwave/mmwave switch. By applying a control
voltage to the gate terminal which stops current through the channel, high impedance
OFF state is attained. In this condition, the source–drain channel is pinched off, thus
offers very high resistance path in between source and drain. The OFF condition of
the MESFET equivalently can be considered as a capacitance of value about 0.25
pF per mm of gate width. The ON state occurs when zero control voltage is applied
to the gate. At this condition, the channel from source to drain is full of free carriers
and represents a very low RF resistance about 2–3 per mm of gate periphery.
At a fixed gate-bias voltage, channel current increases with the increase of drain
voltage towards positive value. The increase of channel current is due to the increase
of electron velocity with the increase of electric field in between source and drain.
The drain current (Id ) is given by
Id channel area × channel conductivity × electric field (6.2a)

dV
Wg × (a − d) × q Nd μn (6.2b)
dx
(a) (b)
VDS VDS
S VGS G D S VGS G D
n+ n+ n+ n+
n-type channel d′
Semi-insulating substrate Semi-insulating substrate
Fig. 6.6 Schematic diagram of MESFET showing depletion layer with non-zero drain bias a when
Vds < Vds (sat), b when Vds > Vds (sat)
Wg × (a − d) × q Nd vn (6.2c)
Here, dV /dx is the electric field experienced by the carrier in the direction of
carrier flow from the source towards the drain and vn is the drift velocity of the
carriers. With the increase of drain bias voltage, potential difference in between
drain and gate will increase, and thus there will be an increase of the depletion
layer width towards the drain as shown in Fig. 6.6a. The channel then starts to
pinch-off beyond certain value of drain voltage and current starts to saturate. For
silicon-like materials, carrier velocity increases monotonically with the increase of
applied electric field. However, for materials like GaAs, initially, the carrier velocity
increases and attains a peak of about 2 × 107 cm/s at about 3 kV/cm electric field
then decreases and finally saturates as shown in Fig. 6.1. Depending upon the gate
length, channel pinch-off may coincide with the saturation velocity of the carriers.
For very small gate length, carrier saturation may occur before channel pinch-off.
Considering the onset of channel pinch-off coincides with the attaining of carrier
saturation velocity, the saturated drain current can be written as
Id (sat) Wg × d × q Nd vn (sat) (6.3)
Here, d is the channel opening near drain at pinch-off condition as shown in

Fig. 6.6b and vn (sat) is the carrier saturation velocity. Beyond onset of pinch-off
voltage, with the further increase of drain voltage, there will be lateral extension of
depletion region from drain towards source as shown in Fig. 6.6b, which leads to
unchanged drain current, i.e. saturation of drain current as shown in Fig. 6.7. There
will be sharp increase of drain current with further increase of the drain voltage
when it reaches to drain-to-gate breakdown voltage corresponds to drain-to-gate
breakdown electric field. The drain-to-gate breakdown voltage, Vdg B , mainly depends
on the material property, doping concentration and structural configuration. With the
increase of gate reverse bias voltage, the drain breakdown voltage will be lesser by
the amount of applied gate reverse voltage as shown in Fig. 6.7.
6.3 Hetero-Structure Field Effect Transistors (HFETs) 85
Fig. 6.7 Typical I–V Id

characteristics of a n-channel Linear Saturation Breakdown
MESFET region region region
Vgs > 0
Vgs = 0
Vgs < 0
Vds
Breakdown voltage
6.3 Hetero-Structure Field Effect Transistors (HFETs)
Hetero-structure field effect transistors (HFETs) utilize dissimilar bandgap materi-

als in their structures enabling current carrying electrons to flow within an undoped
semiconductor material to offer higher electron mobility compared to single mate-
rial structure MESFET [8–23]. Thus, HFETs offer further improved performance
compared to MESFETs at microwave and mmwave frequency range.
The carriers in the MESFET channel are provided by dopants. Electron mobil-
ity in a semiconductor is highly dependent on the dopant concentration as shown
in Fig. 6.8. The dopant in the channel causes ionization impurity scattering, which
reduces the mobility of the carriers. Thus, electron velocity also decreases with the
increase of dopant density as shown in Fig. 6.9 [1]. It shows that the low field elec-
trons drift velocity is more in case of undoped and lightly doped materials compared
to heavily doped materials. To increase electron mobility in MESFET, doping con-
centration in its channel cannot be reduced beyond a certain limit; otherwise carriers
responsible for drain current will be reduced. Reduction of doping concentration in
MESFET channel ultimately will reduce the drain current and gain of the device.
Hetero-structure FET utilizes two dissimilar bandgap materials in their structure
with modulation doping to provide carriers in the undoped region, hence the device
is called modulation doped field effect transistor (MODFET). Here, channel elec-
trons are provided from the highly doped layer of higher bandgap materials which
is spatially separated from the undoped channel of lower bandgap material. Thus,
channel carriers do not suffer from the ionized impurity scattering hence achieving
high-electron mobility. The HFETs are also called high-electron-mobility transistor
(HEMT) because the structure takes the advantage of the higher mobility of the car-
riers in the undoped layer. In the HFET structure, the carriers flow under the gate
like two-dimensional electron gas, thus it is also called two-dimensional-gate FET
(TEGFET). Hetero-structure field effect transistors (HFETs) have various advan-
tages compared to the simple MESFETs. HEMTs offer higher cut-off frequency
9000
Electron mobility (cm /V.Sec)

2
7000
5000
3000
1000
1013 1014 1015 1016 1017 1018 1019 1020
Doping density (cm-3)
Fig. 6.8 Dependence of electron mobility on doping concentration of GaAs
Fig. 6.9 Dependence of

electron velocity on doping 3
concentration of GaAs
Electron Velocity (107 cm/S)
Un-doped GaAs
2 Lightly doped GaAs
1
Heavily doped GaAs
0
0 5 10 15 20
Electric Field (kV/cm)
(higher gain) as well as better noise performance due to the reduction of source
resistance. Some HFET structures also offer comparatively higher output power
at higher microwave/mmwave frequency range. High–electron-mobility transistors
have different varieties depending on the doping profiles and use of materials for
different layers.
One of the advantage in modulation doped field effect transistor (MODFET)
is the use of any superior material as channel since only about 200 Å layer is
required in which the carrier transport takes place. Therefore, it enables to use a
very high mobility material for the channel. Electron mobility of some semiconduc-
tor materials is shown in Table 6.1 with their bandgap energy and lattice constant.
InSb and InAs are having very high-electron mobility but these materials cannot be
Table 6.1 Electron mobility and bandgap

Semiconductor Electron mobility @ Bandgap @ 300 K Lattice constant (Å)
material 300 K (cm2 /V s) (eV)
Si 1500 1.12 5.431
GaAs 8500 1.424 5.653
AlAs 200 2.163 5.660
InSb 80,000 2.26 6.479
InAs 33,000 0.36 6.058
InP 4600 1.35 5.8687
Alx Ga1−x As 8500–200 1.424 + 1.247x ≈same as GaAs
Inx Ga1−x As 8500–33,000 1.424 − 1.064x 5.653–6.058
In0.53 Ga0.47 As 11,000 0.8 ≈same as InP
used in MESFETs since processing of these materials is difficult. However, in case

of MODFETs, a very narrow layer of these materials can be used as channel.
Another aspect of making junction in between two dissimilar bandgap materials
is that ideally lattice constants of both the semiconductors should be same. How-
ever, an epitaxial layer of a semiconductor whose lattice constant is close but not
equal to the lattice constant of the substrate can also be grown with a very thin
layer though resulting in a coherent strain. Structures of lattice-matched HEMTs
consists of semiconductors with the same lattice constant but different bandgap
energy. The hetero-junction structures grown with slightly different lattice constants
are known as pseudo-morphic HEMTs or pHEMTs. Heterojunction also formed
using two unequal lattice constant materials with graded lattice constants. These
devices are called meta-morphic-HEMT (mHEMT). For military, space and com-
mercial applications, at mmwave frequencies, where low noise figure and high gain
are required, conventional MESFET technology has been replaced by differnt high-
electron-mobility transistors (HEMTs, pHEMTs, mHEMTs). For high-efficiency and
high-power amplifications, at microwave and mmwave frequency range, pHEMTs
with multiple channels are also being used.
6.3.1 High-Electron-Mobility Transistors (HEMTs)
The high-electron-mobility transistor (HEMT) is a hetero-structure FET where

the structure takes the advantage of superior electron mobility in the near-
intrinsic/undoped semiconductor materials to achieve better performance primar-
ily in terms of higher gain and better noise figure particularly at microwave and
millimetre-wave frequency range.
HEMTs are basically FETs, its structure and working principles are similar to
MESFET as discussed in the previous section. The epitaxial layer structures of
HEMTs are different from that of MESFETs [5–23]. In the HEMT structure, compo-
sitionally different layers are grown to improve high-frequency performance. Such
types of structures are fabricated by epitaxial techniques such as metal–organic chem-
ical vapour deposition (MOCVD) and molecular beam epitaxy (MBE) with the use
of recessed gate technology of gate length down to about 0.05 μm. Common mate-
rials Alx Ga1−x As and GaAs having different bandgap energies are used as different
layers with GaAS as substrate to construct hetero-structure of HEMTs. Alx Ga1−x As
has larger bandgap energy than GaAs. The bandgap energy of Alx Ga1−x As increases
with the increase of AlAs mole fraction in the Alx Ga1−x As material. Here, lattice
constants of both the materials are nearly the same, thus this is also simply referred
to as lattice-matched HEMT [5–7]. The epitaxial structure of a basic HEMT of
AlGaAs/GaAs is shown in Fig. 6.10 and its energy bandgap diagram is shown in
Fig. 6.11. The HEMT structure is fabricated on a semi-insulating GaAs substrate.
The HEMT structure consists of an Alx Ga1−x As spacer layer, a n+ donor Alx Ga1−x As
layer, and an n Alx Ga1−x As Schottky contact layer. The highly doped n+ GaAs layer
is used for low resistance ohmic contacts. The formation of two-dimensional elec-
tron gas (2DEG) is shown in Fig. 6.11. The higher bandgap energy of Alx Ga1−x As
allows free electrons to diffuse from the Alx Ga1−x As to the lower bandgap GaAs
near the interface. This forms the two-dimensional electron gas (2DEG). The elec-
trons from the donor atoms in the high barrier material AlGaAs spill over into the
low bandgap material GaAs conduction band creating the 2DEG. A potential barrier
then confines the electrons to a thin sheet of charge known as the 2DEG. Thus, the
2DEG electron is free to move in the 2D plane of the device but are confined in the
vertical directions. The 2DEG structures have very less Coulomb scattering, leads to
high-electron-mobility structure compared to doped MESFET channel having lots
of ionized donors. The spacer layer serves to separate the 2DEG from any ionized
donors in the n+ donor AlGaAs layer. Here, the source of electrons is the n+ donor
layer and n Schottky layer. Thickness of the Schottky layer of depletion mode devices
is chosen to overlap the depletion region at the Alx Ga1−x As/2DEG interface. This
is to avoid the possibility of conventional MESFET effects that is conduction of
electrons through the doped Alx Ga1−x As layer in HEMTs. Higher Schottky barrier
height of the metal–semiconductor junction on the AlGaAs is advantageous. How-
ever, increase of donor concentration in donor layer of HEMTs decreases breakdown
voltage that leads to lowering output power capability. However, structures of power
HEMTs with higher breakdown voltages designed and developed using either single
or double recess gate technology (as shown in Fig. 6.10) or by reducing the doping
in the layer just below the Schottky gate.
The AlGaAs/GaAs HEMT was the first implemented lattice-matched HFET. It
shows significant performance improvement, compared to the GaAs MESFETs, in
terms of noise figure and gain at microwave frequency due to the 2DEG at undoped
GaAs layer. However, amount of the 2D charge density in the GaAs channel is limited
due to the limited band discontinuity at the AlGaAs/GaAs interface. Moreover, with
availability of semiconductors such as InSb, InAs of higher mobility and advance-
ment of material processing technology, other HEMT structures have been developed
and being used. On GaAs and InP substrates Inx Ga1−x As layer as active channels
Source Drain
Gate
+
n GaAs n+ GaAs
n+ n AlGaAs n+
+
n AlGaAs donor layer
Very narrow undoped AlGaAs spacer
2DEG
Undoped GaAs
S. I. substrate
Fig. 6.10 Schematic diagram of AlGaAs/GaAs HEMT structure
Schottky n & n+ Undoped Undoped

Metal contact AlGaAs AlGaAs GaAs
spacer
2DEG
EC
--
---
- EFm
--
EV
Fig. 6.11 Energy band diagram of HEMT
can be made to achieve better performance for superior mobility of Inx Ga1−x As.
Indium gallium arsenide (InGaAs) is an alloy composed of two III–V semiconduct-
ing materials, GaAs and InAs. The lattice constant of InGaAs can be varied by
changing the ratio of InAs and GaAs in the alloy. Though In0.53 Ga 0.47 As is lattice
constant matched with the InP, it does not matches with the lattice constant of GaAs.
Thus, in order to match the lattice constant of InP and to avoid mechanical strain,
In0.53 Ga0.47 As is used and structure as shown in Fig. 6.12 is known as InP HEMT.
GaInAs device is normally grown on an indium phosphide (InP) substrate. InP-based
HEMTs are advantageous for power HEMTs relative to GaAs-based HEMTs due
Source Drain
Gate
+
n InGaAs n+ InGaAs
n+ n+
n InAlAs donor layer
2DEG
Undoped In0.53Ga 0.47As channel
InAlAs buffer layer
InP Substrate
Fig. 6.12 Schematic structural diagram of InP HEMT
to superior thermal conductivity of InP (0.68 W/o C cm) compared to that of GaAs
(0.46 W/o C cm). InP HEMTs also have superior microwave performance compared
to GaAs HEMTs and its current gain cutoff frequency is in excess of 500 GHz.
Designing circuits using HEMTs are similar to MESFETs including biasing tech-
niques. The Schottky layer of HEMTs depleted with application of a small negative
gate bias as in the case of MESFET. With the further increase of negative gate bias,
number of free carriers in the 2DEG layer decreases. This is the modulation of the
channel carriers (2DEG) due to the applied negative bias. The device provides ampli-
fication by the modulation of the carriers until the channel is pinched off. At low drain
current, the transconductance of HEMTs is very high because of the conduction of
carriers takes place within the undoped channel which is well confined. In addition
to this, HEMT provides lower noise figure and higher gain for the higher mobility
compared to MESFETs resulting from lower parasitic drain and source resistances.
6.3.2 Pseudo-morphic HEMTs (pHEMTs)
The limitation of the 2D charge density in the channel due to the limited band
discontinuity at the AlGaAs/GaAs interface can be solved by reducing the bandgap
of the channel material. Further improvement can also be made by using the channel
material with higher electron mobility thus higher electron velocity in addition to
the lower bandgap material. To take the advantage of matured GaAs processing
technology, on GaAs substrate AlGaAs/InGaAs interface is used where InGaAs
layer used as channel for its higher electron mobility and lesser bandgap energy
compared to GaAs and AlGaAs (Table 6.1).
Source Drain
Gate
+
n GaAs n+ GaAs
n+ n AlGaAs n+
+
n AlGaAs donor layer
Very narrow undoped AlGaAs spacer
2DEG
Undoped InxGa1-xAs channel layer
S. I. GaAs substrate
Fig. 6.13 Schematic structural diagram of pHEMT
On GaAs substrates, Inx Ga1−x As layer as active channels is made to achieve better
performance for superior mobility of Inx Ga1−x As. Indium gallium arsenide is an alloy
composed of GaAs and InAs, the two III–V semiconducting materials. The lattice
constant of InGaAs can be changed by changing the ratio of InAs and GaAs in the
alloy. Though interatomic spacing, i.e. lattice constant of In0.53 Ga 0.47 As matches with
the InP as discussed in case of InP HEMT, it does not matches with the lattice constant
of GaAs. Thus, the device formed having interface of two materials of different
lattice constant is called pseudo-morphic HEMT (pHEMT) [8, 9]. In this case, a
special constraints is adopted to keep the mechanical strain from differences in lattice
constants within allowable limit. For the GaAs-based pHEMT using Inx Ga1−x As, the
InAs mole fraction is constrained for x < 0.3 and the layer thickness kept very thin,
and typically 100–200 Å. A schematic diagram of AlGaAs/InGaAs-based pHEMT
structure on GaAs substrate is shown in Fig. 6.13.
Advantages of the AlGaAs/InGaAs pHEMT structure include the enhanced elec-
tron mobility (≈13000 cm2 /V-s) and higher saturation velocity of InGaAs. The large
conduction band discontinuity in between AlGaAs and InGaAs enables higher 2DEG
sheet charge density and improved confinement of carriers in the channel by the
InGaAs quantum well. In addition to all these advantages, this pHEMT structure
also enjoys the use of more matured GaAs process technology for realization in
MMIC.
6.3.3 Meta-morphic HEMTs (mHEMTs)
The AlGaAs/InGaAs-based pHEMT structure on a GaAs substrate enjoys the use

of matured and low price fabrication technology (compared to InP processing) in
addition to the superior electron transport properties in InGaAs. However, excessive
lattice mismatch in between GaAs and Inx Ga1−x As does not permit to increase the
Indium (In) content in the InGaAs channel to increase the electron mobility in the
InGaAs channel. The lattice mismatch limits the In content in Inx Ga1−x As below
30% (x < 0.3) which limits the electron mobility, thus limits the high-frequency
performance in GaAs pHEMT. On the other hand, InAlAs/InGaAs HEMT devices
on indium phosphide (InP) substrates have achieved cutoff frequencies of more
than 500 GHz due to the higher In content of 53%. Unfortunately, InP substrates
remain expensive. Hence, it is attractive to explore other ways to produce such high-
performance devices, preferably using MOCVD and GaAs substrates. Another way
to use semiconductor materials of different lattice constants is to place a buffer layer
of varying lattice constant in between them. This is done in the low strain, i.e. meta-
morphic HEMT (mHEMT) [10–13]. This is a structural advancement compared to
that of the pHEMT structure. Schematic diagram of an InAlAs/InGaAs mHEMT
structure is shown in Fig. 6.14. The buffer layer of InAlAs, with graded concentra-
tion of the indium content is used to match its one side lattice constant with GaAs
substrate and another side with the Inx Ga1−x As channel. The graded indium concen-
tration in the buffer layer gives another advantage that practically any concentration
of Indium in the channel can be realized to optimize different parameters of the
device for different applications. Practically, for the realization of device of low
noise figure, low indium concentration is used, whereas high indium concentration
is used to achieve high transconductance gain.
6.4 Microwave GaN HEMTs
The wide bandgap GaN based HEMT is an emerging class of field effect transis-
tor for microwave high-power amplifiers which are going to replace the presently
used travelling wave tube amplifiers (TWTA) for high power requirements in broad-
band microwave communications [14–23]. Properties of GaN semiconductor with
other contemporary materials are shown in Table 6.2. Wide bandgap energy of
GaN (3.4 eV), high channel charge density (≈1013 /cm2 ) of the AlGaN/GaN hetero-
structure interface and high thermal conductivity of the GaN (1.7 W/cm-K) channel
on the higher thermal conducting SiC (3.5 W/cm-K) substrate makes the structure
suitable for handling very high microwave power of the order of 10 W/mm gate
width [17–19].
Schematic diagram of basic HEMT of AlGaN/GaN hetero-structure is shown in
Fig. 6.15. Due to the generation of tensile strain at the heterointerface of AlGaN/GaN,
there will be spontaneous and piezoelectric polarizations. It generates free charge
6.4 Microwave GaN HEMTs 93
Source Drain
Gate
+
n InGaAs n+ InGaAs
n+ n InAlAs n+
+
n InAlAs donor layer
Very narrow undoped InAlAs buffer
2DEG
Undoped InxGa1-xAs channel layer
Undoped InxAl1-xAs buffer layer (x = 0-0.53)
S. I. GaAs substrate
Fig. 6.14 Schematic structural diagram of mHEMT
Table 6.2 Properties of GaN compared to other materials

Material Electron Bandgap @ Breakdown Electron Thermal
mobility @ 300 K (eV) field (v/cm) saturation conductivity
300 K velocity (W/cm-K)
(cm2 /V s) (cm/s)
Si 1350 1.12 6 × 105 1 × 107 1.5
GaAs 8500 1.424 6.5 × 105 1.3 × 107 0.46
SiC 800 2.163 3.5 × 106 2 × 107 3.5
GaN 1600 3.4 3.5 × 106 2.5 × 107 1.7
density at the interface of AlGaN/GaN epitaxial layer. Due to the difference of

bandgap energy of AlGaAs and GaN materials, a quantum well exists in the conduc-
tion band discontinuity. The quantum well is of the order of 25 Å, thus the induced
charge density behaves as 2DEG. The 2DEG sheet charge density at the interface
depends on the Al content in the AlGaN. With the increase of Al content, there will
be an increase in the sheet charge density. The achievable 2DEG charge density is
of the order of 1013 /cm2 . The origin of the electrons in the 2DEG is different in
the polar hetero-structures compared to the conventional HFETs. The channel elec-
trons in conventional HFETs are provided by a doped donor layer, whereas they
come from ionized surface donor states in GaN-based HFETs. Therefore, there may
not be any intentional doping of the AlGaN layer in the GaN HEMTs. In some
AlGaN/GaN hetero-structure, Si impurity is added in the AlGaN which donate elec-
tron in the crystal, which then tends to accumulate just beneath the AlGaN/GaN
interface which contributes to the sheet 2DEG charge density. Similar to the other
Fig. 6.15 Schematic

diagram of GaN HEMT
structure VDS
S VGS G D
n+ AlxGa1-xN n+
High purity GaN
SiC substrate
HFETs, in case of GaN HEMTs also, a Schottky barrier gate terminal is used which
modulates the sheet charge in the 2DEG.
There is a tremendous increase of the power densities, about 10 W/mm gate
width, in the wide bandgap GaN HFETs for its higher breakdown voltage (high volt-
age operation) and high sheet charge density (high current operation). Furthermore,
high voltage (drain supply ≈ 50 V) operation makes the higher input and output
impedances lead to the possibility of a very broad frequency band matching for
maximum output power.
High thermal conductivity of GaN channel and higher thermal conductivity of
the SiC substrate makes lower thermal resistance path to dissipate out the heat from
the device. However, there is a challenge on packaging techniques to dissipate out
the large amount of generated heat in the device for keeping the device’s channel
temperature within the specified maximum allowable temperature limit due to the
continuous device miniaturization. In spite of various efforts, there exists a thermal
issue for high power GaN devices for its very high power density and parasitic effects
due to the associated packaging that affects the circuit performances reliability and
lifetime.
The GaN devices are mainly targeted for high power application at microwave
frequency range and single device has the capability of CW output power about 250 W
at S-band. There is also development of GaN amplifiers at mmwave frequency range
and reported operating frequency range goes to beyond 100 GHz.
6.5 Equivalent Circuit of Microwave FETs
To design microwave circuits using FETs, it is important to know its equivalent

circuit. Design and analysis of linear circuits such as low noise amplifiers and low
power high gain amplifiers need small-signal equivalent circuit, whereas, non-linear
6.5 Equivalent Circuit of Microwave FETs 95
Intrinsic FET
Rg cgd Rd
cgs cdc
gmvgs rds
ri
Rs
Fig. 6.16 Equivalent circuit of a FET
circuits such as multiplier, mixer, or power amplifiers operating in nonlinear region

need large signal equivalent circuit. An equivalent circuit of FET is shown in Fig. 6.16
and the origin of the equivalent circuit elements are shown in Fig. 6.17 [24–40].
This is a generalized equivalent circuit which is applicable for MESFET as well as
for different types of HEMTs. However, the element values depend not only on the
types of FETs it also changes with the structural dimensions and doping profile of the
device. The equivalent circuit elements can be divided into two distinct categories, i.e.
extrinsic and intrinsic elements. The first category presents the different terminals
of access like the source, drain and gate resistances Rs , Rd and Rg . The intrinsic
elements represent the intrinsic characteristics of the FETs. The most important
intrinsic elements are the transconductance gain gm and drain-to-source resistance
rds , its inverse is also known as drain conductance gd . Both the parameters determine
the DC I–V characteristics of the FETs. Performance of a FET based circuit can
be analyzed using the equivalent circuit. The important elements of the equivalent
circuits are defined and details are discussed in the following sections.
6.5.1 Transconductance Gain (gm )
The transconductance (gm ) of a device is defined as the slope of its Ids −Vgs charac-
teristics with the drain–source voltage (Vds ) kept constant. Mathematically it can be
expressed as:
∂ Ids
gm V (6.4)
∂ Vgs dsconstant
Ls Lgs Lg Lgd Ld
Wg a
Rg
Cgs Cgd
Ri
Rs Cdc Rd
Rds
gmVgs
Cds
Fig. 6.17 Schematic structure of a FET indicating the origin of its equivalent circuit elements
Typical Ids −Vgs characteristics and transconductance dependence on gate volt-

age of FETs are shown in Fig. 6.18. The transconductance is zero, for the gate-bias
voltage below the pinch-off voltage and it increases with the reduction of reverse
bias voltage towards zero. Sensitivity of drain current on the change of gate–source
voltage determines the transconductance of a FET as shown in the plots. Transcon-
ductance variation with the gate-bias voltage for different drain bias voltages is shown
in Fig. 6.19. Transconductance is nearly invariant with drain bias voltage when it is
above the drain saturation currents. However, over very low drain bias voltage, cor-
responds to below the drain saturation current, the transconductance becomes very
low. With all other characteristics same, a FET with higher transconductance will
provide greater gain and superior high frequency performance. The device transcon-
ductance is greatly affected by the device dimensions, properties of channel material
and doping profile. With the decrease of gate length, transconductance increases.
In general, the transconductance is inversely proportional to the gate length of the
FETs. The HEMT device always provides better transconductance than GaAs FETs.
As with the channel current, FET transconductance is directly proportional to the
gate width of the device. For this reason, device comparisons are made by comparing
transconductance per unit gate width.
To achieve highest gain from a FET, it is desirable to bias the device at its maximum
transconductance point when it is operating under small-signal conditions. However,
dependency of transconductance on gate–source and drain–source bias levels intro-
duces non-linearity when the FETs are operated under large signal conditions. For
large signal operation, such as in case of high power amplifiers, the transconductance
needs to remain constant in the entire operating range of the gate voltage to achieve
low distortion in the amplified output signal. Such a transconductance characteristic
is shown in Fig. 6.20. This type of characteristics is achieved in case of power FETs
by designing it with modified device structure and doping profiles [3].
(a) (mA) (b) (mS)

80 100
60 75
FET1 FET1
40 50
FET2 20 FET2 25
0 0
-2.0 -1.5 -1.0 -0.5 0 -2.0 -1.5 -1.0 -0.5 0
(V) (V)
Fig. 6.18 Typical a drain current and b transconductance variation with gate voltage
= 4.0 V (mS)
=3.0 V
= 1.0 V 100
= 0.5V
= 0.3V 75
50
25
0
-2.0 -1.5 -1.0 -0.5 0
(V)
Fig. 6.19 Typical transconductance variation with gate bias voltage for various drain bias voltages
6.5.2 Output Conductance (1/rds )
Output conductance (gds ) is the measure of drain current (Ids ) variation with the
change of drain–source voltage (Vds ) while keeping the gate–source voltage (Vgs )
constant. Output characteristics of FETs are often more conveniently expressed in
terms of output resistance (rds ) which is inverse of the output conductance (gds ).
Mathematically it can be written as:
1 ∂ Ids
gds V (6.5)
rds ∂ Vds gsconstant
(mA/mm) 400
(mS/mm) 300
200
100
0
-2.0 -1.5 -1.0 -0.5 0
(V)
Fig. 6.20 Typical drain current and transconductance variation with gate voltage for a power FET
with V ds 5 V
Thus, the output conductance of FETs is determined by the slope of its Ids − Vds
characteristics. The output conductance of a FET is important for any circuit-level
applications. It plays a significant role in determining the output matching network
for a FET based circuit. In general, a FET with low value of output conductance, i.e.
higher value of output resistance is desirable. This implies that the FETs Ids − Vds
characteristic should have ideally zero slopes. A typical output resistance character-
istic with the change of drain supply voltage for different values of gate-bias voltages
is shown in Fig. 6.21. At low drain–source bias levels, the output resistance is low
and then increases with the increase of Vds towards drain current saturation. This
behaviour is valid for all the gate–source bias levels except when it reaches near
the pinch-off levels as shown in Fig. 6.21. At the pinch-off bias levels, the active
channel shows high value of resistance which is nearly independent of drain–source
bias voltage. From this characteristic, it is also clear that the FET at its lower level of
drain–source bias level acts as a voltage-controlled microwave resistor controllable
by the gate–source voltage. This operating region is also used as non-linear resistance
for the realization of microwave multiplier and mixer circuits.
The output resistance of a FET is affected by the device dimensions, material
properties and doping profile. It decreases with the increase of device periphery as
the channel current increases. Power FETs are having very large gate periphery,
thus its output resistance is very small. For a typical case of 0.5 μm gate length
GaAs MESFET, the output resistance is about 150 /mm gate width. Thus, for a
power FET of output power of about 7.5 W, the gate periphery will be about 10 mm
(considering output power capability 0.75 W/mm), leads to the output resistance
of about 15 . Thus, designing an amplifier using this power FET it is required
to transfer this 15 output resistance to the load resistance of 50 . Larger the
(Ω)
= - 1.5 V
3 = - 1.0 V
10
= - 0.5V
= - 0.3V
103
102
101 (V)
0 3.0 6.0
Fig. 6.21 Typical output resistance variation with drain voltage
difference in between the output resistance and the load resistance, there will be
difficulty in designing the matching network over a broad frequency range. For further
higher power MESFET chip, there will be further reduction of output resistance and
narrower bandwidth of the power amplifier. GAN material based HEMTs are having
higher output resistance per gate width. This is one of the reasons that GaN power
amplifiers provide broadband output power performance.
6.5.3 Gate–Source and Gate–Drain Capacitances

(Cgs and Cgd )
Amount of charge stored under the gate terminal is governed by the applied terminal
voltages. In normal operation of FETs, reverse bias of the gate–drain region is more
than the reverse bias of gate–source region. Thus at the drain end, the gate depletion
region is deeper compared to that at the source end. Also, the depletion region is
closer to the drain end compared to the source end. Due to the redistribution of
stored charge under the gate terminal by the change of gate–drain and gate–source
reverse voltages, there are two capacitances known as gate–source capacitance (Cgs )
and gate–drain capacitance (Cgd ).
The gate–source capacitance is defined as when the gate–drain voltage remains
constant, the rate of change of free charge under the gate terminal with respect to the
change of gate–source voltage. Mathematically, it can be written as
∂ Q g
Cgs V (6.6a)
∂ Vgs gdconstant
Here, Q g is the free charge under the gate terminal. In general, FETs are operated at
common source configurations where gate and drain voltages are applied with respect
to the source terminal. The controllable voltages are gate–source and drain-to-source
voltages. Therefore, the gate–source capacitance is also defined with respect to the
constant drain–source voltage as below
∂ Q g
Cgs V (6.6b)
∂ Vgs dsconstant
Capacitance values, as per the two different definitions, differ slightly. For
microwave circuit applications, the gate–source capacitance has an important role.
Input impedance as well as the microwave frequency response depends on this capac-
itance. Neglecting the contribution of other capacitances, the input impedance of a
FET is mainly governed by the gate–source capacitive reactance in series with a
few ohms of resistance. At a higher frequency of operation the input capacitance
acts as a short circuit. The lower the Cgs value, higher the frequency at which the
input impedance becomes short. At such frequencies, the FETs will not be able to
provide sufficient gain. Thus, it is desirable to have lower gate–source capacitance
value to achieve high-frequency performance. The gate–source capacitance is also
depends on the device structure as well as channel doping. Higher the channel doping,
higher is the value of Cgs . With the increase of gate periphery, the capacitance value
increases. The capacitance value decreases with the decrease of gate–strip length.
Thus, the value of Cgs is proportional to both the gate width and gate length. Actually,
the gate–source capacitance is not only due to the charge under the gate strip, it is also
due to the fringing charge in the gate–drain and gate–source spaces. The gate–source
capacitance is also a nonlinear component. Its value depends on the gate and drain
voltage levels. The dependence of gate–source capacitance on the terminal voltages
is shown in Fig. 6.22. For a fixed drain–source voltage, the gate–source capacitance
increases monotonically with the increase of gate–source voltage towards zero value
as it is expected due to the decrease of depletion layer thickness with the increase of
gate–source voltage. The Cgs value also increases with the increase of drain–source
voltage this is due to the extension of the depletion layer towards drain and source
regions, thus effectively it increases the area of the gate–source capacitor.
The gate–drain capacitance (Cgd ) is also due to the charge stored under the gate
terminal. The gate–drain capacitance is defined as the rate of change of free charge
under the gate terminal with respect to the gate–drain voltage when the gate–source
voltage remains constant. Mathematically, it can be written as
∂ Q g
Cgd V (6.7)
∂ Vgd gsconstant
In circuit applications, the capacitance Cgd is responsible for the reverse isolation.
It determines the stability of a FET based amplifier. The smaller the gate–drain
capacitance value higher will be the output to input isolation and greater will be
the stability of the circuit. In case of oscillator circuit, the Cgd will provide the
(pF)
= 5.0 V
= 3.0 V 0.5
= 1.5 V
= 0.5 V
0.4
0.3
0.2
(V) 0.1
-2.0 -1.5 -1.0 -0.5 0
Fig. 6.22 Typical gate–source capacitance variation with gate voltage
(pF)
= -1.5 V
= -1.0 V 0.15
= -0.5 V
0.01
0.05
(V) 0.00
-6.0 -4.5 -3.0 -1.5 0
Fig. 6.23 Typical gate–drain capacitance variation with the gate–drain voltage
positive feedback for oscillation. As in the case of Cgs , the gate–drain capacitance
also increases with the increase of gate length and width. It also increases with the
increase of channel doping density. In general, for a particular FET, the value of
Cgd is always smaller than the value of Cgs . The dependence of Cgd on the terminal
voltages is shown in Fig. 6.23.
6.5.4 Charging Resistance (Ri )
The charging resistance Ri represents the time required to charge the gate depletion
region capacitance Cgs . The value of Ri is about few ohms with a strong dependency
on gate–source as well as drain–source voltages. In general, it increases with the
increase of gate as well as drain reverse bias voltages. Practically, the contribution
Fig. 6.24 Equivalent circuit of a FET for Example 6.1
of Ri to signal distortion is negligible compared to the contributions due to the other

strongly nonlinear elements such as gm , rds and Cgs . However, bias dependency of
Ri is important to consider its contribution in input resistance of the FETs.
Example 6.1 Compute short circuit current gain of a FET whose equivalent circuit is
as shown in Fig. 6.24. Also, compute the frequency at which its short circuit current
gain becomes unity.
Solution
From Fig. 6.24, the voltage across the gate to source capacitor is given by
i in ( f )
vgs ( f )
j2π f cgs
Here, i in is the input current at frequency f . The short circuit output current i out
is given by
i out ( f ) −gm vgs ( f )
Therefore, the short circuit current gain β is given by

i out ( f ) −gm vgs ( f ) jgm
β( f )
i in ( f ) j2π f cgs vgs( f ) 2π f cgs
gm
or, |β( f )|
2π f cgs
Suppose, the frequency for which the short-circuit current gain becomes unity is
f T . Therefore, we can write
gm
|β( f T )| 1
2π f T cgs
gm
Therefore, the short circuit current gain, fT (6.8)
2π cgs
Example 6.2 Compute the frequency for unity short circuit current gain of a FET
whose equivalent circuit is as shown in Fig. 6.25. Calculate its value for gm
0.1 mho, cgs 0.25 pF, cgd 0.01 pF, rds 400 , rin 2.5 .
Solution
From Fig. 6.25, the voltage across the gate to source capacitor is given by

i in ( f )/ j2π f cgd 1
vgs ( f ) ×
ri + 1/ j2π f cgs + 1/ j2π f cgd j2π f cgs
i in ( f )

j2π f cgs + cgd − 4π 2 f 2 rin cgs cgd
Here, i in is the input current at frequency f . The short circuit output current i out ( f )
is given by

i in ( f ) × rin + 1/ j2π f cgs
i out ( f ) −gm vgs ( f ) +
rin + 1/ j2π f cgs + 1/ j2π f cgd

−gm i in ( f ) i in ( f ) × j2π f rin cgs + 1 cgd
+
j2π f cgs + cgd − 4π 2 f 2 rin cgs cgd j2π f rin cgs cgd + cgs + cgd
Therefore, the short circuit current gain β( f ) is given by

i out ( f ) − gm + 4π 2 f 2 rin cgs cgd + j2π f cgd
β( f )
i in ( f ) −4π 2 f 2 rin cgs cgd + j2π f cgs + cgd
2 2 1/2
gm + 4π 2 f 2 rin cgs cgd + 2π f cgd
or, |β( f )| 2 2 (6.9)
4π 2 f 2 rin cgs cgd + 2π f cgs + cgd
In practice, rin 2π f cgs , and gm 2π f cgd . Therefore, the output short-circuit

current gain can be written as
gm
|β( f )| (6.10)
2π f cgs + cgd
Suppose, the frequency for which the short-circuit current gain becomes unity is
f T . Therefore, we can write
gm
|β( f T )| 1
2π f T cgs + cgd
gm
Therefore, the short circuit current gain, fT
2π cgs + cgd
Putting the values for gm 0.1 mho, cgs 0.25 pF, cgd 0.01 pF, the fre-
quency where the short current gain becomes unity is
0.1 × 1000
fT GHz 61.2 GHz
2π (0.25 + 0.01)
Putting |β( f T )| 1 in (6.9), and using gm 0.1 mho, cgs 0.25 pF, cgd
0.01 pF, rds 400 , rin 2.5 , one can calculate the frequency for unity short
circuit current gain without any approximation, it is about 61.83 GHz.
Example 6.3 Derive the maximum frequency of oscillation ( f max ) which is the
frequency for which the maximum available power gain of the device becomes unity
of a FET. The equivalent circuit of the FET is as shown in Fig. 6.26 inside the dotted
area. Also, derive relationship in between the maximum frequency of operation (also
called maximum frequency of oscillation), f max and frequency for unity short circuit
current gain, f T .
Solution
The input signal vs of impedance Z s is applied at the input of the FET and the output
is terminated with impedance Z L as shown in Fig. 6.26, the voltage across the gate
to source capacitor is given by
vs ( f ) 1
vgs ( f ) × (6.11a)
rin + 1/ j2π f cgs + Z s j2π f cgs
For complex conjugate matching of input port, the source impedance will be

Z s rin − 1/ j2π f cgs (6.11b)
Putting this value of Z s , the voltage across cgs is given by

vs ( f )
vgs ( f ) (6.11c)
j4π f cgs rin
The current through the output load impedance is given by

−vs ( f )gm rds
i out ( f ) × (6.12a)
j4π f cgs rin rds + Z L
Thus, the output signal voltage is given by

−vs ( f )gm rds Z L
vout ( f ) × (6.12b)
j4π f cgs rin rds + Z L
For complex conjugate matching at the output port:
Z L rds (6.12c)
Putting Z L rds , the output signal voltage under complex conjugate matched
condition is given by
−vs ( f )gm rds
vout ( f ) (6.12d)
j8π f cgs rin
Under complex conjugate matching at the input port, the input signal voltage is
given by
vs ( f )
vin ( f ) (6.13)
2
Thus, the maximum available power gain of the FET is given by
2
(vout ( f ))2 /rds vout ( f ) rin
Pavs ( f ) × (6.14)
(vin ( f ))2 /rin vin ( f ) rds
Putting vout ( f ) and vin ( f ) from (6.12d) and (6.13), respectively, in (6.14), the
maximum available power gain is given by
2
−gm rds 2 ri gm rds

Pavs ( f ) × × (6.15)
j4π f cgs rin rds 4π f cgs rin
As per the definition of the maximum frequency of operation: Pavs ( f ) 1, for

f f max . Thus, from (6.15), we can write
2
gm rds
× 1 (6.15)
4π f max cgs rin

gm rds
or, f max × (6.16)
4π cgs rin
From, the previous example we have seen

gm
fT (6.18)
2π cgs
Thus, the maximum frequency of operation of the FET is given by

fT rds
or, f max × (6.19)
2 rin
Example 6.4 Calculate the frequency, f T for unity short circuit current gain and
maximum frequency of operation, f max of a FET whose equivalent circuit is as
shown in Fig. 6.27, for gm 0.15 mho, cgs 0.65 pF, rds 500 , rin 3.5 .
Solution
The frequency for unity short circuit current gain is given by
gm
fT
2π cgs
Putting, gm 0.15 mho and cgs 0.30 pF:

0.15 × 1000
fT GHz 36.73 GHz
2π × 0.65
The maximum frequency of operation is given by

fT rds
f max ×
2 rin
Putting, f T 36.73 GHz, rds 500 and rin 3.5 :

36.73 500
f max × 219.5 GHz
2 3.5
6.6 Maximum Frequency of Operation
Maximum frequency of operation for a FET is mainly determined by transit time of

carriers through the channel [40]. The transit time is the time required for carriers
to travel from source to drain. At low-field condition, when the mobility is constant,
the transit time is given by
Lg
τ (6.20a)
μE x
Here, L g is the gate length and E x is the electric field along the channel. Suppose
the drain supply voltage is VD with respect to the source terminal. Therefore,
Vd
Ex (6.20b)
Lg
Putting E x from (6.20b) to (6.20a):
L 2g
τ (6.20c)
μVd
Whereas, at saturated velocity condition, the transit time is given by

Lg
τ (6.21)
vs
Here, vs is the saturated velocity of the carriers. It can also be shown that the
ratio of transconductance gm and gate to source capacitance cgs is equal to vs /L g .
Therefore, the frequency for unity short circuit current gain can also be written as
vs
fT (6.22)
2π L g
Example 6.5 Calculate the frequency, f T for a GaAs MESFET of gate length 1 μm.
Consider the electrons are moving with their saturated velocity of 1.3 × 107 cm/s.
Solution
Under the condition of saturation velocity of carriers, f T is given by (6.22)
vs
fT
2π L g
1.3 × 107
Hz
2π × 10−4
20.7 GHz
References
2. Statz H, Newman P, Smith I, Pucel R, Haus H (1987) GaAs FET device and circuit simulation
in spice. IEEE Trans Electron Devices ED-34(2):160–169
3. Tkachenko AY, Bartle D, Wei C-J (1999) Correlation between ungated recess width and linearity
of GaAs MESFETs. Microw J 1–8
4. Pal D, Noll A (2014) Improvement of transconductance flatness of GaAs MESFETs. In: CS
mantech conference, pp 219–222, Denver, Colorado, USA, 19th –22nd May 2014
5. Lai R, Mei X, Deal W, Yoshida W, Kim Y, Liu P, Lee J, Uyeda J, Radisic V, Lange M, Gaier T,
Samoska L, Fung A (2007) Sub 50 nm InP HEMT device with fmax greater than 1-THz. In:
IEEE international electron devices meeting, IEDM 2007, pp 609–611, Dec 2007
6. Grundbacher R, Lai R, Barsky M, Tsai R, Gaier T, Weinreb S, Dawson D, Bautista J, Davis
J, Erickson N, Block T, Oki A (2002) 0.1 m InP HEMT c from X-band to W-band. In: 14th
Indium phosphide and related materials conference, IPRM, pp 455–458
7. Greenberg D, Del Alamo L, Bhat R (1995) Impact ionization and transport in InAlAs/n+-InP
HFET. IEEE Trans Electron Devices 42(9):1574–1582
8. Lin YC, Chang EY, Yamaguchi H, Wu WC, Chang CY (2007) A δ-doped InGaP/InGaAs
pHEMT with different doping profiles for device linearity improvement. IEEE Trans Electron
Devices 54(7):1617–1625
9. Yang MT, Chan YJ (1996) Device linearity comparisons between doped channel and modula-
tion doped designs in Pseudomorphic Al0.3 Ga0.7 As/In0.2 Ga0.8 As heterostructures. IEEE Trans
Electron Devices 43(8):1174–1180
10. Leuther A, Tessmann A, Massler H, Losch R, Schlechtweg M, Mikulla M, Ambacher O (2008)
35 nm metamorphic HEMT MMIC technology. In: 20th international conference on indium
phosphide and related materials, MoA3.3, May 2008
11. Tessmann A, Leuther A, Massler H, Kuri M, Loesch R (2008) A metamorphic 220–320 GHz
HEMT amplifier MMIC. In: IEEE compound semiconductor integrated circuit symposium,
CSICS 2008, IEEE, Monterey, California, Piscataway, NJ, 12–15 Oct 2008, pp 31–34
12. Hülsmann A, Leuther A, Kallfass I, Weber R, Tessmann A, Schlechtweg M, Ambacher O
(2009) Advanced mHEMT technologies for space applications. 20th international symposium
on space terahertz technology, Charlottesville, 20–22 April 2009
13. Leuther A, Tessmann A, Kallfass I, Losch R, Seelmann-Eggebert M, Wadefalk N, Schafer F,
Puyol JDG, Schlechtweg M, Mikulla M, Ambacher O (2009) Metamorphic HEMT technology
for low-noise applications. In: 21st IEEE international conference on indium phosphide &
related materials, IPRM, pp 188–191, 411
References 109
14. Pengelly RS, Wood SM, Milligan JW, Sheppard ST, Pribble WL (2012) A review of GaN on
SiC high electron-mobility power transistors and MMICs. IEEE Trans Microw Theory Tech
60(6)
15. Kazutaka I, Norihiko U, Seigo S (2011) High power and high efficiency GaN-HEMT for
microwave communication applications. In: IEEE IMWS-IWPT proceedings
16. Ambacher O, Smart J, Shealy JR, Weimann NG, Chu K, Murphy M, Schaff WJ, Eastman
LF, Dimitrov R, Wittmer L, Stutzman M, Rieger W, Hilsenbeck J (1999) Two-dimensional
electron gases induced by spontaneous and piezoelectric polarization charges in N- and Ga-
face AlGaN/GaN heterostructures. J Appl Phys 85(6):3222–3333
17. Wu YF et al (1996) Very high breakdown voltage and large transconductance realized on GaN
heterojunction field-effect transistors. Appl Phys Lett 69:1438–1440
18. Wu YF, Kapolnek D, Ibbetson JP, Parikh P, Keller BP, Mishra UK (2001) Very-high power
density AlGaN/GaNHEMTs. IEEE Trans Electron Devices 48:586–590
19. Mitani E, Aojima M, Maekawa A, Sano S (2007) An 800-W AlGaN/GaN HEMT for S-band
high-power application. In: CS mantech on-line Digest
20. Chung Jinwook W, Hoke William E, Chumbes Eduardo M, Palacios Tomas (2010) AlGaN/GaN
HEMT With 300-GHz fmax . IEEE Electron Device Lett 31(3):195–197
21. Palacios T, Chakraborty A, Rajan S, Poblenz C, Keller S, DenBaars SP, Speck JS, Mishra UK
(2005) High-power AlGaN/GaN HEMTs for Ka-band applications. IEEE Electron Device Lett
26(11):781–783
22. Quay R, Musser M, van Raay F, Maier T, Mikulla M (2009) Managing power density of high-
power GaN devices. In: IEEE MTT-S international microwave symposium workshop Digest,
Boston, MA, pp 71–86
23. Wu Y-F, Kapolnek D, Ibbetson JP, Parikh P, Keller BP, Mishra UK (2001) Very-high power
density AlGaN/GaN HEMTs. IEEE Trans Electron Devices 48:586–590
24. Curtice W, Ettenberg M (1985) A nonlinear GaAs FET model for use in the design of output
circuits for power amplifiers. IEEE Trans Microw Theory Tech 33(12):1383–1393
25. Minasian RA (1977) Simplified GaAs MESFET model to 10 GHz. Electron Lett
13(18):549–551
26. Kondoh H (1986) An accurate fet modelling from measured S-parameters. In: IEEE MTT-S
international microwave symposium digest, pp 377–380, June 1986
27. Fukui H (1979) Determination of the basic device parameters of a GaAs MESFET. Bell Syst
Tech J 58(3):771–795
28. Chen T, Kumar M (1988) Novel GaAs FET modeling technique for MMICs. In: IEEE GaAs
IC Symposium Digest, pp 49–52
29. Dambrine G et al (1988) A new method for determining the FET small-signal equivalent circuit.
IEEE Trans Microw Theory Tech 36(7):1151–1159
30. Berroth M, Bosch R (1991) High-frequency equivalent circuit of GaAs FET’s for large-signal
applications. IEEE Trans Microw Theory Tech 39(2):224–229
31. Ooi B et al (1997) A novel approach for determining the GaAs MESFET small-signal
equivalent-circuit elements. IEEE Trans Microw Theory Tech 45(12):2084–2088
32. Angelov I, Zirath H, Rorsman N (1992) A new empirical nonlinear model for HEMT and
MESFET devices. IEEE Trans Microw Theory Tech 40(12):2258–2266
33. Jeon K, Kwon Y, Hong S (1997) A frequency dispersion model of GaAs MESFET for large-
signal applications. IEEE Microw Guided Wave Lett 7(3):78–80
34. Lazaro A, Pradell L, O’Callaghan J (1999) FET noise-parameter determination using a
novel technique based on 50- noise-figure measurements. IEEE Trans Microw Theory Tech
47(3):315–324
35. Rorsman N, Garcia M, Karlsson C, Zirath H (1996) Accurate small-signal modeling of HFET’s
for millimeter-wave applications. IEEE Trans Microw Theory Tech 44:432–437
36. Dambrine G, Cappy A, Heliodore F, Playez E (1988) A new method for determining the FET
small-signal equivalent circuit. IEEE Trans Microw Theory Tech 36:1151–1159
37. Curtice WR, Ettenberg M (1985) A nonlinear GaAs FET model for use in the design of output
circuits for power amplifiers. IEEE Trans Microw Theory Tech MTT-33(12):1383–1393
38. Kompa G, Lin F (1990) FET modelling using an analytical extraction method based on
broadband S-parameter measurement. In: 20th European microwave conference proceedings,
Budapist, Hungary, pp 778–783, September 1990
39. Kompa G, Novotny M (1992) Highly consistent FET model parameter extraction based on
broadband S-parameter measurements. IEEE MTT-S international microwave symposium
digest, pp. 293–296
40. Higashiwaki M, Mimura T, Matsui T (2008) AlGaN/GaN heterostructure field-effect transistors
on 4H-SiC substrates with current-gain cutoff frequency of 190 GHz. Appl Phys Express
1:021103
Chapter 7
Microwave Circuit Analysis
Abstract This chapter describes detailed transmission line theory and analysis
including concept of S-parameters, ABCD parameters and their interrelationships
through various examples. It gives Smith chart analysis and describes various trans-
mission lines like waveguide, coaxial, stripline, microstrip and coplanar transmis-
sion lines. This chapter also presents detailed analysis about power transfer theory
describing available power gain, transducer power gain and operating power gain for
microwave networks.
7.1 Introduction
Objective of the microwave circuit analysis is to provide a complete description of

terminal behaviour of the circuit as of low-frequency circuit analysis. Microwave
circuit analysis needs the knowledge about transmission line theory, microwave net-
works, network analysis and impedance matching.
At low frequencies where circuit dimensions are relatively smaller than the oper-
ating wavelength, the circuit elements can be considered as lumped components as
the phase change along the circuit elements is negligible. Similarly, at low frequen-
cies, the interconnection between two elements is treated as an ideal connection with
no loss, no characteristic impedance and no transmission phase angle. At microwave
frequencies, dimension of circuit components and interconnecting elements may not
be possible to conveniently made smaller compared with the operating wavelength.
Thus, the circuit analysis technique cannot be done using conventional circuit voltage
and current. Microwave circuits may consist of elements with dimensions such that
the voltage amplitude and phase over the length of the device can vary significantly.
In this case, the current that flows in the interconnecting wires or through the circuit
elements is not the same at different points along the wire or elements. To accurately
calculate the behaviour of an element, it is required to consider its length, width
and thickness of metal and its proximity to the ground plane, i.e. the element is
considered as a distributed element. Similarly, when interconnections become an

112 7 Microwave Circuit Analysis
appreciable fraction of the signal wavelength, these interconnections themselves

must be treated as distributed elements or transmission lines.
Though all electronic circuits are nonlinear in character; however, some circuits
can be treated as linear circuits depending upon their operating conditions. Amount
of nonlinearity not only depends on particular device but it also depends on bias
condition, signal level and operating frequency. A circuit that behaves as nonlinear
circuit at some operating signal level may behave as a linear circuit under compara-
tively lower signal power level. In general, microwave switches, attenuators, phase
shifters, modulators, low noise and IF amplifiers operate at their linear operating
regions, whereas high power amplifiers, linearizers and limiters shall be treated as
nonlinear circuits.
7.2 Transmission Line Theory and Analysis
Transmission lines are used to guide and carry electromagnetic waves from one
point of a system to another. At the higher operating frequency, where length of the
transmission lines used as interconnecting elements is comparable or more than the
wavelength, the approximations useful for calculations at lower frequencies are no
longer accurate. In this case, the voltage and current along the transmission line at a
given time may not be the same at all points. Thus, the length of the interconnecting
line is important when the signal includes frequency components with corresponding
wavelengths comparable to or less than the length of the line.
For analysis, a transmission line can be represented by a network with circuit
components distributed throughout the line as shown in Fig. 7.1a. Here, R in /m is
resistance, L in H/m is inductance in both conductors, G in S/m is conductance of
the dielectric media and C in F/m is capacitance between the conductors, all are per
unit length. The capacitance C and inductance L represent electromagnetic field that
passes along the transmission line, whereas the resistance R and the conductance G
represent ohmic loss along the line and dielectric loss across the transmission lines,
respectively.
Kirchhoff’s voltage and current laws can be applied to this line section to derive the
propagation constant γ and the characteristic impedance Z o of the line [1]. Applying
Kirchhoff’s voltage law to the circuit shown in Fig. 7.1a,
∂i(l, t)
v(l, t) − Rl i(l, t) − Ll − v(l + l, t) 0 (7.1a)
∂t
or
v(l + l, t) − v(l, t) ∂i(l, t)
− R i(l, t) + L (7.1b)
l ∂t
7.2 Transmission Line Theory and Analysis 113
(a)
(b)
Fig. 7.1 a Equivalent circuit of a transmission line of very small length l, b EM wave travelled
in a transmission line of characteristic impedance Z o terminated by a load impedance Z L
For l → 0
∂v(l, t) ∂i(l, t)
− Ri(l, t) + L (7.1c)
∂l ∂t
Similarly, applying Kirchhoff’s current law to the circuit shown in Fig. 7.1a,
∂v(l + Δl, t)
i(l, t) − Gl v(l + Δl, t) − CΔl − i(l + Δl, t) 0 (7.2a)
∂t
For l → 0
∂i(l, t) ∂v(l, t)
− Gv(l, t) + C (7.2b)
∂l ∂t
The first-order differential equations (7.1c) and (7.2b) are the generalized
transmission line equations valid for any voltage and current waveforms along the
transmission line [1]. Suppose, the voltage and current are time-varying sinusoidal
functions represented by

v(l, t) Re V (l)e jωt (7.3a)

i(l, t) Re I (l)e jωt (7.3b)
Using (7.3a) and (7.3b), the generalized transmission line equations can be written
as
dV (l)
− (R + jωL)I (l) (7.4a)
dl
dI (l)
− (G + jωC)V (l) (7.4b)
dl
Differentiating (7.4a) and using (7.4b), we get,
d2 V (l)
(R + jωL)(G + jωC)V (l) (7.5a)
dl 2
Similarly, differentiating (7.4b) and using (7.4a), we get,
d2 I (l)
− (R + jωL)(G + jωC)I (l) (7.5b)
dl 2
Defining, the propagation constant,

γ α + jβ (R + jωL)(G + jωC) (7.6a)
For lossless transmission lines, R 0 and G 0. Putting these values, for

lossless transmission lines, the propagation constant can be expressed as
√
γ jβ jω LC (7.6b)
Thus, the phase velocity v p is given by

√
v p ω/β 1/ LC (7.6c)
(7.5a) and (7.5b) can be written as
d2 V (l)
γ 2 V (l) (7.7a)
dl 2
and
d2 I (l)
− γ 2 I (l) (7.7b)
dl 2
(7.7a) and (7.7b) are the ordinary differential equations. The solutions can be
written as
V (l) V + e−γ l + V − e+γ l (7.8a)

+ −γ l − +γ l
I (l) I e −I e (7.8b)
The two terms in each solution represent travelling waves along positive l and
negative l directions respectively. Here, V + (I + ) and V - (I - ) are the amplitudes of
voltage (current) waves travelling in positive and negative l drection respectively.
The characteristic impedance Z o is the ratio of voltage and current for an infinitely
long transmission line. In case of an infinitely long transmission line, there will be
no reflecting waves. Thus, using (7.4a) and (7.8a), the characteristic impedance can
be written as

V+ (R + jωL) R + jωL
Zo + (7.9)
I γ G + jωC
For lossless transmission lines, R 0 and G 0. Putting these values, for

lossless transmission lines, the characteristic impedance can be expressed as

Zo L/C (7.10)
When a signal is fed to a transmission line of infinite length, there will be no

reflection of the signal. The voltage and current ratios at any location and time on the
line give the same impedance as if the signal generator is terminated by an impedance
of Z o . This impedance is called characteristic impedance of the transmission line.
When a transmission line of finite length is terminated by its characteristic impedance
(Z o ), then there also will be no reflection of the signal and the line is called matched
transmission line. In case of lossless transmission line, the characteristic impedance
is purely resistive and is denoted as Ro .
In practice, a transmission line may be terminated by any load, i.e. other than its
characteristic impedance. In this case, when a wave travelling from the generator
reaches the load impedance, the wave (V −)is incident on it and some of the power
+
gets absorbed in the load. Some wave V is reflected from the load as backward
wave, as shown in Fig. 7.1b. The load Z L is connected in the line and l is along the
transmission line. Origin of the axis is chosen at the load point, i.e. at l 0. As the
generator impedance is matched to the characteristic impedance of the line (Z o ), the
reflected wave that reaches by the generator is absorbed by the generator impedance.
At steady-state condition, under sinusoidal excitation of angular frequency ω, the
voltage and current waves along the lossless transmission line at l is given by

v(l, t) Re V (l)eiωt (7.11)

i(l, t) Re I (l)eiωt (7.12)
where V (l) and I (l) can be expressed as
V (l) V + e− jβl + V − e+ jβl (7.13)

+ −
V − jβl V + jβl
I (l) e − e (7.14)
Zo Zo
Here, β is the phase constant and is given by β ω/c 2π/λ. Ratio of the
reflected wave amplitude and incident wave amplitude is called reflection coefficient
(Γ ) and is given by
V−
Γ (7.15)
V+
Thus, in terms of the reflection coefficient, the voltage and current waves along
the lossless transmission line at l is given by

V (l) V + e− jβl 1 + Γ e+ j2βl (7.16)
V + − jβl
I (l) e 1 − Γ e+ j2βl (7.17)
Zo
At the load end, V (0) Z L I (0). Therefore,

1+Γ
Z L Zo (7.18)
1−Γ
Thus, the reflection coefficient for a load impedance of Z L of the transmission

line of characteristic impedance Z o is given by
Z L − Zo
Γ (7.19)
Z L + Zo
In microwave frequency, it is very convenient to express load impedance by reflec-

tion coefficient involving incident and reflected waves instead of expressing it by the
ratio of voltage and current as it is done for low-frequency circuits.
The reflection coefficient including its phase term can be written as
V − (z) V − e+ jβl
Γ (l) + − jβl Γ e+ j2βl (7.20)
V (z)
+ V e
Therefore, under sinusoidal excitation of angular frequency ω, the steady-state

voltage and current waveforms can be written as

v(l, t) Re V + e− jβl (1 + Γ (l))e jωt (7.21)
+
V − jβl
i(l, t) Re e (1 − Γ (l))e jωt
(7.22)
Zo
These expressions give the amplitude variation of voltage and current waveforms
along the transmission line. These variations are due to the generation of standing
wave for the load impedance other than its characteristic impedance.
Figures 7.2 and 7.3 show the normalized amplitude of voltage and current waves
along the transmission line for different values of load impedances. It is clear from
the figures that amplitude of the voltage waveform changes along the transmission
line due to the generation of standing wave. Comparing (7.12) with (7.11), it is also
clear that the position of the voltage maximum will be the position of the current
minimum along the transmission line. For example, in case of open load condition,
there will be maximum voltage amplitude and minimum current amplitude at the
load end, whereas there will be minimum voltage amplitude and maximum current
amplitude at the shorted load end.
The ratio maximum voltage amplitude (Vmax ) to the minimum voltage
of the
amplitude Vmin is defined as voltage standing wave ratio (VSWR), i.e.
Vmax 1 + |Γ |
VSWR (7.23)
Vmin 1 − |Γ |
Therefore, when transmission lines are not terminated by its characteristic

impedance, variation of voltage and current amplitudes leads to the variation of
impedance along the transmission line. The input impedance Z in (l) of a transmis-
sion line is the impedance looking from one end of the transmission line and is given
by
V (l) 1 + Γ (l)
Z in (l) Zo (7.24)
I (l) 1 − Γ (l)
Putting the value of Γ (l) from (7.20), the input impedance becomes
(a) Voltage λ/2 Current
0
Along the Tr. line
(b)
TRANSMISSION LINE
ZO ZL
Point of voltage Point of voltage
minima maxima
Vg
Time Time
Fig. 7.2 Normalized voltage and current wave along the transmission line with load impedance
Z L Z o a amplitude along the transmission line, b time domain voltage waveform at minima and
maxima on the transmission line
1 + Γ e+ j2βl Z L − j Z o tan(βl)
Z in (l) Z o Zo (7.25)
1−Γe + j2βl Z o − j Z L tan(βl)
Therefore, the input impedance of a transmission line at a distance l from the load
end towards the generator (i.e. at the source end) terminated with Z L can be derived
by putting l −l in (7.25) as given by
Z L + j Z o tan(βl)
Z in Z o (7.26)
Z o + j Z L tan(βl)
This is the most widely used formula for microwave circuit design and analysis
using transmission lines.
From (7.25), the impedance of a transmission line at a distance l towards the
generator from the load impedance of reflection coefficient Γ L can be written as
1 + Γ L e− j2βl 1 + |Γ L |e j(φ−2βl)
Z in Z in (−l) Z o Zo (7.27)
1 − ΓL e + j2βl 1 − |Γ L |e j(φ−2βl)
Here, φ is the phase of the reflection coefficient at the load end (i.e. load reflection
coefficient). This (7.27) gives the variation of impedance along a transmission line
when we move from the load end towards generator. Suppose lmax is the distance
TRANSMISSION LINE
Zo ZL
Vg
(a) 2
ZL = ∞
(b) 2
ZL = 0
(c) 1
ZL = 50Ω
0
Voltage amplitude
Current amplitude
Fig. 7.3 Normalized voltage wave along the transmission line with load impedance a Z L open,
b Z L short and c Z L Z o
from the load end towards generator where the impedance is maximum, i.e. voltage
maximum and current minimum point. Then,
2βlmax φ (7.28)
Therefore, the voltage minima point lmin from the load end will be
2βlmin φ + π (7.29)
From (7.27), the impedance at the voltage maxima point is given by:
1 + |Γ L |
Z max Z o (7.30a)
1 − |Γ L |
Z o × VSWR (7.30b)
This is a pure resistance. Similarly, the impedance at the voltage minimum point
can be written as
1 − |Γ L |
Z min Z o (7.31a)
1 + |Γ L |
Z o /VSWR (7.31b)
7.3 Microwave Transmission Lines
Low insertion loss, negligible signal distortion, high transmitted power capability,
wide operation frequency band, low electromagnetic interference (EMI) and proper
electromagnetic compatibility (EMC) are the requirements for any transmission line.
In addition to these, easy and low-cost production, and easy integration of passive
and active devices & components are also demanded from a good transmission line.
Waveguides and coaxial lines transmit electromagnetic waves totally confined
by metallic walls, so both satisfy all the above-mentioned requirements except easy
integration of the components, whereas open transmission lines such as microstrip
lines, striplines, coplanar waveguides, etc., where waves propagate along the line
and its close vicinity have less power transmitting capability, higher insertion loss,
poorer EMI/EMC and higher coupling to neighbouring circuitry.
7.3.1 Losses in Transmission Lines
The propagation constant for a transmission line is given by (7.6a)

γ α + jβ (R + jωL)(G + jωC)

√ R 1/2 G 1/2
jω LC 1 + 1+
jωL jωC

√ R R2 G G2
jω LC 1 + − 2 2 2 + ··· 1 + − 2 2 2 + ···
j2ωL j 8ω L j2ωC j 8ω C
For a low-loss transmission line, neglecting higher order terms,

√ 1 R G R2 G2 RG
γ α + jβ jω LC 1 + + + + −
j2ω L C 8ω2 L 2 8ω2 C 2 4ω2 LC
Equating the real terms from both the sides,

√ 1 R G
α LC + (7.32a)
2 L C
7.3 Microwave Transmission Lines 121
√
Putting, Z o 1/Yo L/C in (7.32a),
R G
α + αc + αd (7.32b)
2Z o 2Yo
The first term (αc ) of (7.32b) is related to the series equivalent resistance of
the transmission line, this corresponds to the conductor or resistive loss of the line.
Resistive losses within a transmission line arise from the resistance of the conductors.
When an electromagnetic wave flows through a conductor, amplitude of the field
decreases with the depth of the conductor. Thus, most of the currents flow through
skin of the conductor. The skin depth (δ) is defined as the distance at which the field
falls 1/e of its value at its surface where the wave enters. The skin depth (δ) is given
by

2
δ , σ is conductivity of the conductor (7.33)
μo ωσ
Thus, the actual area through which the current flows through the conductor is
limited by the skin effect. As frequency ω 2π f increases, area through which
current flows decreases and thus the resistive loss increases.
Example 7.1 Conductivity of silver is 3 × 107 S/m. Calculate its skin depth at
10 GHz.
Solution
μo 4π × 10−7 H/m
σ 3 × 107 S/m
ω 2π × 1010 Hz
From (7.33), the skin depth is given by

2 2
δ m
μo ωσ 4π × 10 −7 2π × 1010 3 × 107
0.92 × 10−6 m
0.92 µm
This example shows that at microwave frequency, the skin depth for a good con-
ductor is very small. Thus, the difference microwave performance (loss) between a
component using pure good conductor and a component using bad conductor plated
with good conductor would be expected to be negligible. This enables the use of
silver- or gold-plated low-cost microwave components such as silver- or gold-plated
waveguides and planar transmission lines instead of using fully gold- or silver-based
costly component, without compromising their RF performance.
The second term (αd ) of (7.32b) is related to the shunt conductance (G) of the
transmission line, and this corresponds to the loss within the dielectric support mate-
rial of the transmission line. Within the dielectric material, displacement current
cannot leads the field strength by π/2 due to the polarization delay. This leads to the
appearance of an active displacement current component which is in phase with the
field strength. Under this condition, the permittivity can be expressed as
εr εr − jεr (7.34)
This leads to the power dissipation within the dielectric material under alternate
electromagnetic field. The power dissipation per unit volume within the dielectric
material is given by
Pd ωεo εr E 2 tan(δ) (7.35)
where δ εr /εr and E is the electric field. This (7.35) shows that the dielectric
loss increases linearly with frequency, whereas due to the decrease of skin depth,
the resistive loss increases with square root of the frequency. Therefore, at higher
operating frequency, the dielectric loss dominates over resistive loss.
Radiation is other source of transmission line losses of electromagnetic energy
when it propagates through a transmission line. Though it is negligible at lower
range of microwave frequency, it becomes significant at higher microwave frequency
range. Transmission line losses due to radiation are generally of not much concern
for waveguide, coaxial line, stripline and grounded coplanar waveguide due to their
closed configurations. In all these configurations, the signal path is surrounded by
some form of grounded conductors, whereas in microstrip line configurations, the
signal line is open to environment. Thus, microstrip line configurations are more
prone to radiation loss and there are several EMI-related issues related to the radiation.
Mismatch loss is another source of transmission losses of electromagnetic energy.
When a transmission line is terminated to a load of impedance different from its
characteristic impedance, then some of the incident power is reflected back from the
load. The amount of power that is not delivered to the load due to the impedance
mismatch is the mismatch loss. The mismatch loss is defined as the power delivered
to the load to the power available from the transmission line network. With reference
to Fig. 7.4, the mismatch loss is given by
Zo
TRANSMISSION LINE ZL
Fig. 7.4 Mismatch loss in transmission line

PI − PR
L MISMATCH (7.36a)
PI
PR
1− (7.36b)
PI
Here, PI is the power incident to the load and PR is the power reflected from the
load. The (7.36b) can be written as
L MISMATCH 1 − Γ L2 (7.36c)
Here, Γ L is the load reflection coefficient and is given by Γ L PR /PI . Thus, in

decibel, the mismatch loss is given by

(L MISMATCH )dB 10 log 1 − Γ L2 (7.36d)
7.3.2 Coaxial Transmission Lines
A coaxial transmission line consists of a centre conductor with another conductor

around it and the space in between the conductors is filled with dielectric material
as shown in Fig. 7.5. The principal mode of transmission that propagates along
coaxial transmission lines is TEM since it is a two-wire transmission line. The useful
frequency range of a coaxial transmission line is restricted to the principal mode of
operation which is below the cut-off frequency of first higher order mode. Maximum
power handling of a coaxial line is set by the dielectric breakdown and heat dissipation
within the line due to attenuation losses contributed by resistive and dielectric losses.
The inductance and capacitance per unit length of a coaxial transmission line can be
calculated considering the entire current flow through the surface of the conductor
as follows.
Suppose 2a and 2b are the diameters of inner and outer conductors of a coaxial
transmission line as shown in Fig. 7.5. The electric and magnetic field distributions
of TEM wave propagating through a coaxial line is as shown in Fig. 7.6a. The
electric field is in the radial direction and varies as 1/r and has no variation in the
φ direction, whereas the magnetic field is in the φ direction and varies as 1/r in the
radial direction. Suppose current I flows through the line. Therefore, the magnetic
flux B everywhere is tangent to a circle centred on the centre conductor as shown in
Fig. 7.6b. Therefore, at a distance x from the centre of the centre conductor, we can
write

B.dl 2π x B (7.37a)
Using Ampere’s circuital law, we can write

Center conductor
Dielectric fill
Outer conductor
Fig. 7.5 Cross section of coaxial transmission line
(a) (b) (c)
Fig. 7.6 a Field distribution for TEM wave of a coaxial line, b diagram to determine L and c
diagram to determine C

B.dl 2π x B μo μr I (7.37b)
Here, μo is permiability in vacuum and μr is relative permeability of the medium.

Thus,
μo μr I
B (7.37c)
2π x
Thus, magnetic flux dΦ at a distance x from the centre of the conductors through
a strip of thickness dx and length l is given by
μo μr I
dΦ Bldx ldx (7.37d)
2π x
Suppose inductance per unit length is L; therefore, inductance of the line of length
l is Ll. Thus, total magnetic flux is given by
b

μo μr I μo μr I l b
Φ ldx ln Ll I (7.37e)
2π x 2π a
a
Thus, inductance per unit length is given by

μo μr b
L ln (7.37f)
2π a
The capacitance per unit length for a coaxial line can be obtained by evaluating
the voltage difference between the conductors for a given charge on each. Suppose
ξ is the charge density per unit length of the conductor and E is the electric field at a
distance x as shown in Fig. 7.6b. The flux through the Gaussian surface, as shown in
Fig. 7.6b, is E × 2π xl. The total charge along the length l is lξ . By applying Gauss’
law,
E × 2π xl lξ/εr εo (7.38a)
Thus, electric field at a distance x is given by

ξ 1
E (7.38b)
2π εr εo x
The voltage between the cylinders can be obtained by integrating the electric field
along a radial line as shown below:
b

ξ 1 ξ b
V dx ln (7.38c)
2π εr εo x 2π εr εo a
a
Suppose equivalent capacitance per unit length is C. Therefore, the total capaci-
tance Cl is given by

ξl ξ b
Cl ξl/ ln (7.38d)
V 2π εr εo a
or capacitance per unit length is given by

2π εr εo
C (7.38e)
ln ab
Thus, for a lossless coaxial transmission line, the characteristic impedance can be
written from (7.10), (7.37f) and (7.38e) as

L 1 μo μr b
Zo ln (7.39a)
C 2π εo εr a
Putting μo 4π × 10−7 H/m and εo 8.854 × 10−12 F/m

377 μr b
Zo ln (7.39b)
2π εr a

μr b
Z o 138 log10 (7.39c)
εr a
Example 7.2 Inner and outer diameters of an air–dielectric coaxial cable are 0.05
and 0.15 in., respectively. Calculate its characteristic impedance.
Solution
Characteristic impedance is given by (7.39c)

μr b
Z o 138 log10
εr a
Here, a 0.05 in., b 0.15 in., μr εr 1. Thus,

0.15
Z o 138 × log10
0.05
138 × 0.477
65.8
In a coaxial transmission line, the space between the two conductors can sup-
port higher order modes, which is similar to waveguide mode of propagation. The
approximate cut-off wavelength (λc ) for the first higher order waveguide mode is
given by
√
λc π (a + b) εr (7.40a)
Thus, the first higher order cut-off frequency ( f c ) is given by

c
fc √ , c is speed of light in vacuum (7.40b)
π (a + b) εr
c/a
√ (7.40c)
π (1 + b/a) εr
A coaxial line is useful only below its cut-off frequency of the first higher order
mode. The cut-off frequency is determined by the size (a) of the coaxial transmission
line for a given dielectric constant (εr ) and characteristic impedance (b/a) of the line.
High
conducting
metal plating
Circular Rectangular
Fig. 7.7 Waveguide structures with circular and rectangular cross section
7.3.3 Waveguides
Literally, a transmission line which guides electromagnetic (EM) waves permitting

its propagation along it, confining its energy within it or in vicinity of it can be con-
sidered as waveguide. Under this definition, coaxial line, twin-wire line, microstrip
line, fibre optic cables, hollow metal pipe, all are waveguides. However, in microwave
engineering, a hollow metal (conductor) pipe of rectangular, circular or elliptic cross
section used to guide and carry EM waves from one place to another with minimum
loss of energy is known as waveguide. Structure of such waveguides is shown in
Fig. 7.7. The waveguides are superior transmission lines in terms of lower atten-
uation and higher power handling capability compared to coaxial, microstrip and
stripline transmission lines. For these superior performances, waveguides are pop-
ularly used for high-power and low-loss applications in microwave and mmwave
frequency range.
Waveguides can be analysed using the following Maxwell’s electromagnetic equa-
tions in terms of electric (E) and magnetic (H ) fields inside the waveguides [1, 2].
∂D
∇×H J+ (7.41a)
∂t
∂B
∇×E − (7.41b)
∂t
∇·Dρ (7.41c)
∇·B0 (7.41d)
Here, D εE and B μH are electric and magnetic flux densities, respectively,

constants ε and μ are permittivity and permeability, respectively, of the medium
through which EM wave propagates, J and ρ are the electric current density and elec-
tric charge density, respectively. In case of waveguide, considering perfect dielectric
medium containing no charge and no conduction currents, we can write, J 0 and
ρ 0 in (7.41).
Taking curl (∇×) of (7.41b) and substituting (7.41a), under the condition of J 0,
we can write,

∂B ∂2 E
∇ × ∇ × E −∇ × −με 2 (7.42)
∂t ∂t

Using the vector identity, ∇ × ∇ × E ∇∇ · E − ∇ 2 E, (7.42) can be written
as
∂2 E
∇∇ · E − ∇ 2 E −με (7.43)
∂t 2
Here, ρ 0. Thus, from (7.41c), ∇ · D 0 implies ∇ · E 0. Thus, (7.43) can

be written as
∂2 E
∇ 2 E με (7.44a)
∂t 2
This is the wave equation for the electric field E. Similar wave equation for the
magnetic field can be derived and written as
∂2 H
∇ 2 H με (7.44b)
∂t 2
The time-varying electric and magnetic fields can be expressed as E(r, t)

E(r )e jωt and H (r, t) H (r )e jωt . Thus, wave equations can be expressed as
∇ 2 E −ω2 μεE (7.45a)

∇ H −ω με H
2 2
(7.45b)
Analysis of a waveguide can be done using these wave equations putting appropri-
ate boundary conditions corresponding to the waveguide. The boundary conditions
for the waveguide walls are the tangential component of electric field E should be
zero and the normal component of the magnetic field H should be zero. Consider
a rectangular waveguide in a coordinate system as shown in Fig. 7.8, and in gen-
eral always a > b. Suppose the waveguide is invariant in the z direction and the
wave is propagating in the positive z direction as e−γ z (in −z direction can also be
considered), here, γ α + jβ. Equation (7.41b) can be written as
Fig. 7.8 Cross section of a Y

rectangular waveguide of
infinite length Metal
Wall
Metal
Wall
X
a
Z
⎡ ⎤
x̂ ŷ ẑ
⎢ ⎥
⎢ ∂ ∂ ∂ ⎥ − jωμ x̂ Hx + ŷ Hy + ẑ Hz (7.46)
⎣ ∂x ∂y ∂z ⎦
Ex E y Ez
Equating the x, y, z components from left and right sides,

∂ Ez ∂ Ey
− − jωμHx (7.47a)
∂y ∂z
∂ Ez ∂ Ex
− − jωμHy (7.47b)
∂x ∂z
∂ Ey ∂ Ex
− − jωμHz (7.47c)
∂x ∂y
Considering spatial variation of the wave in z direction as e− jωz :

∂ e−γ z
−γ e−γ z
∂z
Equations (7.47) can be simplified to

∂ Ez
+ γ E y − jωμHx (7.48a)
∂y
∂ Ez
+ γ E x jωμHy (7.48b)
∂x
∂ Ey ∂ Ex
− − jωμHz (7.48c)
∂x ∂y
Similarly, following expressions can be derived from Eq. (7.41a)

∂ Hz
+ γ Hy jωεE x (7.49a)
∂y
∂ Hz
+ γ Hx − jωεE y (7.49b)
∂x
∂ Hy ∂ Hx
− jωεE z (7.49c)
∂x ∂y
Combining (7.48a) and (7.49b) and putting
kc2 γ 2 + ω2 με (7.50)
The x component of magnetic field (Hx ) can be expressed as

γ ∂ Hz ωε ∂ E z
Hx − +j 2 (7.51)
kc ∂ x
2 kc ∂ y
Similarly, combining (7.48b) and (7.49a) y component of magnetic field (Hy ) can
be expressed as
γ ∂ Hz ωε ∂ E z
Hy − −j 2 (7.52a)
kc ∂ y
2 kc ∂ x
Combining (7.49a) and (7.48b) x component of electric field (E x ) can be expressed

as
γ ∂ Ez ωμ ∂ Hz
Ex − −j 2 (7.52b)
kc ∂ x
2 kc ∂ y
Similarly, combining (7.49b) and (7.48a) y component of electric field (E y ) can

be expressed as
γ ∂ Ez ωμ ∂ Hz
Ey − +j 2 (7.52c)
kc ∂ y
2 kc ∂ x
Equations (7.52a) to (7.52c) show very interesting and important characteristics

of waves transmitting through waveguide. These equations show that the x and y
component of magnetic fields (Hx , Hy ) and electric fields (E x , E y ) can be expressed
in terms of only z component of electric field (E z ) and magnetic field (Hz ). In short, all
the transverse components can be determined from only the axial components. This
finding implies that, if E z and Hz both are zero, all the fields within the waveguide
vanish. To support wave propagation through a waveguide, either an electric field
along the direction of propagation (E z ) or a magnetic field along the direction of
propagation (Hz ) must exist.
Depending on the existence of either electric field or magnetic field along the
direction of propagation, there are two modes of propagation in a waveguide. One
is transverse magnetic (TM) mode of wave propagation, for which magnetic field
along the wave propagation is zero, i.e. Hz 0. Another is transverse electric (TE)
mode of wave propagation, for which electric field along the wave propagation is
zero, i.e. E z 0. In general, the propagation of EM wave through a waveguide may
be a combination of both these types of waves.
7.3.3.1 Transverse Magnetic (TM) Wave
Characteristics of the transverse magnetic (TM) waves propagating along a rectan-

gular waveguide can be derived by putting appropriate boundary conditions in the
wave equations.
For TM wave, magnetic field component along the direction of propagation is zero.
Thus, Hz 0. The wave equation for the electric field (7.44a) along the direction of
propagation (z) can be written as
∂ 2 Ez ∂ 2 Ez
+ + γ 2 E z −ω2 μεE z (7.53a)
∂x2 ∂ y2
This equation is to be solved under the following boundary conditions, for the
rectangular waveguide as shown in Fig. 7.8:
E z 0, at x 0, x a, y 0 and y b (7.53b)
The solution of the partial differential equation (7.53a) can be written as
E z (C1 cos Ax + C2 sin Ax)(C3 cos By + C4 sin By)e−γ z (7.54)
Here, kc2 A2 + B 2 , the constants C1 , C2 , C3 , C4 , A and B can be selected to

satisfy the boundary conditions (7.53b) as follows:
mπ nπ
C1 0, C3 0, A and B (7.55)
a b
Putting the values of constants from (7.55) to Eq. (7.54), the electric vector of
TM wave along the propagation can be written as replacing C2 C4 C,
mπ nπ
E z C sin x sin y e−γ z (7.56a)
a b
Other electric and magnetic field components can be written from (7.52a) to
(7.52c) as follows:
γ mπ mπ nπ
E x −C 2 cos x sin y e−γ z (7.56b)
kc a a b
γ nπ mπ nπ
E y −C 2 sin x cos y e−γ z (7.56c)
kc b a b
ωε nπ mπ nπ
Hx jC sin x cos y e−γ z (7.56d)
kc2 b a b
ωε mπ mπ nπ
Hy − jC 2 cos x sin y e−γ z (7.56e)
kc a a b
The set of equations from (7.56a) to (7.56e) determines the electric and mag-
netic field variations of transverse magnetic (TM) wave in a rectangular waveguide
propagating in z direction.
The constants m and n can have any integer value. Thus, multiple solutions for
the electric and magnetic fields of the TM wave propagation in a waveguide are
possible. These are called mode of propagation. These transverse magnetic modes
of propagation are designated as T M mn modes. From (7.56a), for either m 0
or n 0, the electric field vector of the wave along the direction of propagation
becomes zero, i.e. E z 0. This implies that there is no existence of TM waves for
either m 0 or n 0. Minimum possible values for m and n are 1 and the mode is
designated as T M 11 .
7.3.3.2 Transverse Electric (TE) Wave
Characteristics of the transverse electric (TE) waves propagating along a rectangular

waveguide can be derived using the wave equations by putting appropriate boundary
conditions.
For TE waves, the electric field component along the direction of propagation
is zero. Thus, E z 0. The wave equation for the magnetic field (7.44b) along the
direction of propagation (z) can be written as
∂ 2 Hz ∂ 2 Hz
+ + γ 2 Hz −ω2 με Hz (7.57a)
∂x2 ∂ y2
This equation is to be solved under the following boundary conditions for the
rectangular waveguide as shown in Fig. 7.8:
∂ Hz
0, at x 0, x a and
∂x
∂ Hz
0 at y 0, y b (7.57b)
∂y
The solution of the partial differential equation (7.57a) can be written as
E z (C1 cos Ax + C2 sin Ax)(C3 cos By + C4 sin By)e−γ z (7.58)
Here, kc2 A2 + B 2 , the constants C1 , C2 , C3 , C4 , A and B can be selected to

satisfy the boundary conditions (7.57b) as follows:
mπ nπ
C2 0, C4 0, A and B (7.59)
a b
Putting the values of constants from (7.59) to Eq. (7.58), the magnetic field vector
of TE wave along the propagation can be written as, replacing C1 C3 C:
mπ nπ
Hz C cos x cos y e−γ z (7.60a)
a b
Other electric and magnetic field components can be written from (7.52a) to
(7.52d) as follows:
ωμ nπ mπ nπ
E x jC 2 cos x sin y e−γ z (7.60b)
kc b a b
ωμ mπ mπ nπ
E y − jC 2 sin x cos y e−γ z (7.60c)
kc a a b
γ mπ mπ nπ
Hx C 2 sin x cos y e−γ z (7.60d)
kc a a b
γ nπ mπ nπ
Hy C 2 cos x sin y e−γ z (7.60e)
kc b a b
The set of equations from (7.60a) to (7.60e) determines the electric and magnetic
field variations of transverse electric (TE) wave in a rectangular waveguide propagat-
ing in z direction. The constants m and n can have any integer value. Thus, multiple
solutions for the electric and magnetic fields of the TE wave propagation in a waveg-
uide are possible. From (7.60a), it is clear that for T E mn waves, it is possible to make
either m or n equal to zero (however, both m and n cannot be zero simultaneously)
without resulting all fields to vanish. Therefore, possible modes for T E mn are T E 10 ,
T E 01 , T E 11 , T E 12 , T E 21 , etc.
7.3.4 Cut-Off Frequency and Guide Wavelength
Putting kc2 A2 + B 2 in (7.50),
γ 2 A2 + B 2 − ω2 με (7.61)
Values of A and B are the same for T Mmn mode (7.55) and T E mn mode (7.59).
Putting these values of A and B, and γ α + jβ in (7.61),

mπ 2 nπ 2
γ α + jβ + − ω2 με (7.62)
a b
Equation (7.62) determines the propagation constant for T Mmn and T E mn modes
in a rectangular waveguide. At low frequencies, where
mπ 2 nπ 2
+ > ω2 με
a b
γ is real, implies β, the propagation phase constant is zero, i.e. γ α, where α
is the attenuation constant. In this condition, there is no transmission of EM waves
at these frequencies. At the frequencies, where
mπ 2 nπ 2
+ < ω2 με
a b
γ is imaginary. It implies α, the attenuation constant is zero. The propagation
constant γ jβ. At these frequencies, the EM wave transmission will be without
any attenuation (in case of no conductor loss and no dielectric loss in the waveguide).
The frequency below which the EM wave propagation does not take place is called
cut-off frequency. The cut-off frequency ωc is given by

1 mπ 2 nπ 2
ωc √ + (7.63)
με a b
The frequencies above the cut-off, the propagation constant is imaginary and is
given by
mπ 2 nπ 2
jβ j ω2 με − + (7.64a)
a b
Putting cut-off frequency from (7.63), the phase constant can be written as

β ω2 με − ωc2 με (7.64b)
Therefore, the phase velocity v p (ω/β), can be written as

1 1 c
v p (ω/β) √ √ (7.65)
με 1 − ωc2 /ω2 μr εr 1 − ωc2 /ω2
√
where μ μo μr , ε εo εr and c 1/ μo εo is the wave velocity of EM waves in
vacuum (unguided).
The expression (7.65) of phase velocity in waveguide indicates that the propa-
gation velocity is greater than the velocity (c) of EM wave in free space. With the
increase of frequency (ω) above the cut-off frequency (ωc ), the phase velocity v p
decreases from very large value towards the velocity (c) in free space.
Wavelength of the wave propagating within the waveguide is called guide wave-
length λg and can be written from (7.64) as
2π 2π
λg (7.66a)
β ω με − ωc2 με
2
2π
(7.66b)
ω2 με − ωc2 με
c 1
√ (7.66c)
μr εr 1 − ( f c / f )2
f
λo
√ (7.66d)
μr εr 1 − (λo /λc )2
where λo c/ f is wavelength in vacuum for the frequency f .

Example 7.3 Derive expression of cut-off frequency, guide wavelength and phase
velocity of T M11 mode in a rectangular waveguide of dimensions a and b.
Solution
Putting m n 1 in (7.63), the cut-off frequency of T M11 mode for a rectangular
waveguide of dimension (a, b) can be written as follows:

1 π 2 π 2
f c(T M11 ) √ + (7.67a)
2π με a b
The guide wavelength of T M11 mode for a rectangular waveguide of dimension

(a, b) can be written from (7.66d) as
c 1
λg √ 2 (7.67b)
f μr εr
1 − f c(T M11 ) / f
Here, f is the operating frequency. The phase velocity can be written from (7.65)
as
1 c
vp √ 2 (7.67c)
μr εr
1 − f c(T M11 ) / f
Example 7.4 Derive expression of cut-off frequency, guide wavelength and phase
velocity of T E 10 mode in a rectangular waveguide of dimensions a and b.
Solution
Putting m 1 and n 0 in (7.63), the cut-off frequency of T E 10 mode for a
rectangular waveguide of dimension (a, b) can be written as follows (Fig. 7.9):
1 π 1 c
f c(T E10 ) √ √ (7.68a)
2π με a μr εr 2a
Comparing (7.68a) with (7.67a),
f c(T E10 ) < f c(T M11 )

Fig. 7.9 E- and H-field lines of TE 10 in rectangular waveguide cross section
This implies that the cut-off frequency of T E 10 is lower than the cut-off frequency
of T M11 for a rectangular waveguide. Considering a > b, it can be concluded
that cut-off frequency of T E 10 is the lowest among all T E mn and T Mmn modes
in a rectangular waveguide. Thus, T E 10 is the lowest order mode in a rectangular
waveguide and is called dominant mode.
The guide wavelength of T E 10 mode for a rectangular waveguide of dimension
(a, b) can be written as
c 1
λg √ 2 (7.68b)
f μr εr
1 − f c(T E10 ) / f
Here, f is the operating frequency. The phase velocity can be written from (7.65)
as
1 c
vp √ 2 (7.68c)
μr εr
1 − f c(T E10 ) / f
Example 7.5 Derive expressions of electric and magnetic fields for T E 10 waves prop-
agating in a rectangular waveguide of dimensions a and b, and also diagrammatically
show the electric and magnetic field lines.
Solution
Putting m 1 and n 0 in (7.60a) to (7.60e), and replacing kc2 A2 + B 2
(π/a)2 , the electric and magnetic field components can be written for T E 10 mode in
a rectangular waveguide as
Ex 0 (7.69a)
E-lines
b
H-lines
Fig. 7.10 E- and H-field lines of TM 11 in rectangular waveguide cross section
ωμa π
E y − jC sin x e−γ z (7.69b)
π a
Ez 0 (7.69c)
γa π
Hx C sin x e−γ z (7.69d)
π a
Hy 0 (7.69e)
π
Hz C cos x e−γ z (7.69f)
a
Based on these expressions, the electric and magnetic field lines are shown in
different cross sections of a rectangular waveguide in Fig. 7.9.
Example 7.6 Derive expressions of electric and magnetic fields for T M11 waves
propagating in a rectangular waveguide of dimensions a and b. Also diagrammati-
cally show the electric and magnetic field lines.
Solution
Putting m 1 and n 1 in (7.56a) to (7.56e), the electric and magnetic field
components for T M110 mode can be written in a rectangular waveguide as
γ π π π
λE x −C 2 cos x sin y e−γ z (7.70a)
kc a a b
γ π π π
E y −C 2 sin x cos y e−γ z (7.70b)
kc b a b
π π
E z C sin x sin y e−γ z (7.70c)
a b
ωε π π π
Hx jC 2 sin x cos y e−γ z (7.70d)
kc b a b
ωε π π π
Hy − jC 2 cos x sin y e−γ z (7.70e)
kc a a b
Hz 0 (7.70f)
Here, kc2 A2 + B 2 (π/a)2 + (π/b)2 .

Based on these expressions, the electric and magnetic field lines are shown in
different cross sections of a rectangular waveguide in Fig. 7.10.
Example 7.7 Calculate the cut-off frequencies for different

T E 10 , T E 11 , T E 20 , T E 21 and T M11 , T M21 , T M12 modes in a rectangular
waveguide of dimension a 10 mm b 5 mm, and filled with dielectric of
μr 1, εr 4.
Solution
The cut-off frequencies for a rectangular waveguide is given by (7.63)

1 mπ 2 nπ 2
ωc(mn) √ +
με a b

cπ m 2 n 2
√ +
μr εr a b
Therefore,

c m 2 n 2
f c(mn) √ + (7.71)
2 μr εr a b
Putting μr 1, εr 4, a 10 mm 1 cm, b 5 mm 0.5 cm and

c 3 × 1010 cm/s in (7.71),

3 × 1010 m 2 n 2
f c(mn) √ + Hz
2 1×4 1 0.5
30 √ 2
m + 4n 2 GHz
4 √
7.5 × m 2 + 4n 2 GHz
Putting the values of m, n cut-off frequencies of different modes are shown in the
following Table:
Mode Cut-off frequency (GHz)

T E 10 7.5
T E 01 , T E 20 15.0
T E 11 , T M11 16.77
T E 21 , T M21 21.21
T E 12 , T M12 30.92
Example 7.8 Determine all the possible propagating modes in a rectangular waveg-
uide of dimension a 10 mm b 5 mm and filled with dielectric of μr 1, εr
6.0 operating at 10 GHz.
Solution
The cut-off frequencies for a rectangular waveguide are given by (7.71)

c m 2 n 2
f c(mn) √ +
2 μr εr a b
Putting μr 1, εr 9.9, a 10 mm 1 cm, b 5 mm 0.5 cm and

c 3 × 1010 cm/s,

3 × 1010 m 2 n 2
f c(mn) √ + Hz
2 1 × 9.9 1 0.5
√
4.767 × m 2 + 4n 2 GHz
Putting the values of m, n cut-off frequencies of different modes are shown in the
following Table:
Mode Cut-off frequency (GHz)

T E 10 6.123
T E 01 , T E 20 12.247
T E 11 , T M11 13.693
T E 21 , T M21 17.321
The table shows that the only mode which has cut-off frequency below the operat-
ing frequency (10 GHz) is T E 10 . Therefore, only T E 10 mode can propagate through
the waveguide at operating frequency 10 GHz.
7.3.5 Planar Transmission Lines
Planar transmission lines have structures that allow controlling its characteristic
impedance by controlling its dimension in a single plane. Planar transmission lines
consist of multiconductor lines comprising a solid dielectric substrate having single or
multilayers of metallization. In the structure, signal and ground currents flow through
separate conductors. These structures are very compact and suitable for realization
of microwave integrated circuits (MICs) using simpler fabrication process. There are
various types of planar transmission line structures. Among various configurations
stripline, microstrip line and coplanar waveguide are commonly used for realization
of microwave integrated circuits.
Fig. 7.11 Stripline configuration
7.3.5.1 Stripline
Stripline configuration consists of three flat conductor strips; the centre strip is the
signal line that runs in between two ground strips filled with suitable dielectric
material. Sketch of stripline configuration is shown in Fig. 7.11. Width (W ), thickness
(t) of the centre strip, height of the stripline (b) and dielectric constant of the filled
dielectric material determine the characteristic impedance of the line.
Stripline supports propagation of TEM mode because the region between
the two ground conductor strips contains a single medium. Thus, the phase
velocity and the characteristic impedance of a stripline do not vary with fre-
quency. Electric and magnetic field distributions in stripline are shown in
Fig. 7.12. Here, electromagnetic field is strongly confined near the centre strip
and within the top and bottom ground planes. In a stripline, the return current
path for a high-frequency signal will remain directly above and below the sig-
nal strip on the ground planes. The high-frequency signal thus remains entirely
inside the substrate, minimizing emissions and providing natural shielding against
any incoming radiation. The minimum emission leads to negligible radiation and thus
the stripline transmission lines have two types of losses, conductor loss and dielectric
loss. The dielectric loss is proportional to frequency and dominates at higher oper-
ating frequencies. The ohmic skin losses in the strip conductor and ground planes
depend on the conductivity of the metal conductors and surface roughness of the
conductors.
The characteristic impedance Z o of the stripline depends on the dielectric constant
and cross-sectional geometry of the structure. The characteristic impedance is very
sensitive to the ratio of centre strip width W to the dielectric thickness b. However, it
is relatively insensitive to the thickness of the centre conductor strip. The approximate
characteristic impedance Z o of a stripline can be expressed in the following simple
form:
E field
H field
Fig. 7.12 E- and H-fields distribution in stripline configuration
30π b
Zo √ (in ) (7.72)
εr We + 0.441b
Here, We is the effective width of the centre strip conductor and is given by
We W W
for > 0.35
b b b

W W 2 W
− 0.35 − for < 0.35
b b b
It shows that the characteristic impedance Z o of the stripline decreases with the
increase of width W of the strip for fixed substrate height b.
7.3.5.2 Microstrip
The microstrip line is the most widely used planar transmission line for RF and
microwave circuits. It consists of a single dielectric substrate with ground plane on
one side and a strip on the other side as shown in Fig. 7.13. The widespread use of
microstrip line is due to its planar nature, ease of fabrication using various simple
processes, easy integration of surface-mount devices (SMD) due to easy access of the
top surface, good heat sinking and good mechanical support. Width (W ), thickness
(t) of the upper strip, height (h) and dielectric constant (εr ) of the dielectric material
determine the characteristic impedance of the line. Microstrip configuration is a strip
over a ground plane and thus it tends to radiate due to openness of the upper strip as
the spacing between the ground plane and the strip increases. However, a substrate
thickness of a few percent of a wavelength (or less) minimizes radiation without
restricting width of the strip.
Electric and magnetic field distributions in a microstrip configuration are shown
in Fig. 7.14. Here, electromagnetic fields extend in the space above the dielectric.
Thus, this configuration becomes a mixed dielectric transmission structure (air and
dielectric). To ensure that EM fields are confined near the strip and mostly within
the dielectric, it is necessary to use high dielectric constant material as substrate.
Another advantage of using high dielectric constant substrate is the reduced circuit
dimension due to the reduction of phase velocity.
Fig. 7.13 Microstrip line configuration
E field
H field
Fig. 7.14 E- and H-fields distribution in microstrip line configuration
In contrast to stripline, the two-media nature (substrate discontinuity due to the

air above upper strip) of microstrip causes its dominant mode to be hybrid (not TEM)
called quasi TEM. This fact can be ascertained from Maxwell’s equations as follows.
The tangential component of the electric field is continuous at the air–dielectric
interface. Thus,
(E dielectric )x (E air )x (7.73a)
Here, x is tangential to the dielectric–air interface and normal to the conductor

strip. Since εdielectric εair , the tangential component of displacement density D
becomes discontinuous, i.e
(Ddielectric )x (Dair )x (7.73b)
Using Maxwell’s (∇ × H ) equation and combining with (7.73b), we can write,

(∇ × H )x dielectric (∇ × H )x air (7.74)
If we assume that the mode of EM wave propagation is pure TEM, then Hz 0.

Putting this in (7.74), we can write,

∂ Hy ∂ Hy
(7.75)
∂z dielectric ∂z air
This leads to,

Hy dielectric
Hy air (7.76)
This inequality violates the required field continuity condition for the normal
components of magnetic field. This implies that the assumption Hz 0 is not
true. This leads to the conclusion that a pure TEM mode cannot be supported by a
microstrip line configuration. However, in most of the practical configurations, major
portion of the electromagnetic energy is confined within the dielectric material below
the strip and the electric flux crossing the air–dielectric boundary is small. Due to the
quasi-TEM mode of propagation, the phase velocity, characteristic impedance and
field variation in the guide cross section all become mildly frequency dependent.
Taking the effect of quasi-TEM mode of propagation, concept of effective
dielectric constant εeff is used for simple analysis in design of microstrip line.
Practically, part of the electric field is in air (εr 1) and other is in the dielectric
(εr ). Considering the effective dielectric constant εeff , it is assumed that most of the
electric fields are constrained within the substrate, but a fraction of the total energy
exists within the air above the strip. The value of εeff is somewhere in between 1
and εr , which is used to calculate capacitance per unit length of the line. With the
increase of strip width or decrease of substrate thickness more and more electric
lines of forces will be within the dielectric. Thus, εeff increases with the increase in
the strip width and decrease in substrate thickness.
With increasing frequency, the effective dielectric constant gradually increases
towards that of the substrate, so that the phase velocity gradually decreases. This is
true even with a non-dispersive substrate material. The εeff changes with the free-
space wavelength, λo . The dispersion becomes more pronounced with the decreasing
ratio of strip width to substrate thickness, (W/ h). Dispersion is less pronounced as
the strip width becomes relatively wider, and in this condition, the effective dielectric
constant, εeff εr .
The effective dielectric constant is given by
⎡ ⎤

εr + 1 εr − 1 ⎣ 1 W 2⎦ W
εeff + + 0.04 1 − for <1 (7.77)
2 2 1 + 12h h h
W
⎡ ⎤
εr + 1 εr − 1 ⎣ 1 ⎦
Else, εeff +
2 2 1+ 12h
W
whereas characteristic impedance of microstrip lines can be expressed as

Ground conductor Ground conductor

W strip
strip
dielectric ( )
t
s s
h h
strip (signal line)
Fig. 7.15 Coplanar waveguide configuration

60 8h W W
Zo √ ln + when <1
εeff W 4h h
120π 1 W
√ W W when ≥1 (7.78)
εeff h + 1.393 + 3 ln h + 1.444
2 h
Practically, the microstrip impedance (Z o ) with finite width ground plane is equal
to the impedance value with infinite width ground plane, if the ground width is at
least greater than three times of strip width (W ).
The sources of attenuation of microstrip lines are due to finite conductivity of the
strip conductor, finite resistivity of the substrate and its dumping phenomena, i.e.
polarization effect, radiation effects and magnetic loss in case of magnetic substrates
such as ferrites.
Microstrip lines with wider strip width have lower current density and thus
lower ohmic loss. One way to lower the conductor loss of microstrip line is to
suspend the substrate over the air, which increases the width of the strip due to the
lower effective dielectric constant. In addition, the insertion loss of the microstrip
is reduced because air essentially has no/less dielectric loss compared to standard
substrates. However, in this case, the separation between the signal and ground paths
increase, which leads to an increase of radiation loss. The radiation loss depends on
the dielectric constant, substrate thickness and the circuit geometry. The higher the
dielectric constant, the higher the concentration of energy in the substrate region that
leads to lesser radiation loss. Another benefit of having a higher dielectric constant
is the reduction of circuit size. To minimize radiation loss in a circuit, the number of
discontinuities, such as bends and T-junctions, should be made as small as possible.
Dielectric loss can be reduced using substrates with a low dielectric loss tangent.
E field
H field
Fig. 7.16 E- and H-fields distribution in CPW configuration
Power handling capacity of a microstrip line is limited by heating caused by ohmic

and dielectric losses and by dielectric breakdown. An increase in temperature due to
conductor and dielectric losses limits the average power handling of the microstrip
line, whereas the substrate breakdown in between the strip conductor and the ground
plane limits the peak-power-handling capability.
Metallic enclosure is used for microwave active as well as passive microstrip
circuits such as amplifiers, mixers, filters, etc. to protect them from environment and
to provide EM shielding. The presence of conducting (metallic) top and side walls
affect both the characteristic impedance, Z o and the effective dielectric constant,
εeff . To reduce the effect of metallic enclosure for practical circuits, the height of the
cover from the circuits should be more than ten times the substrate thickness, and
the distance to walls should be more than six times the substrate thickness.
7.3.5.3 Coplanar Waveguide (CPW)
Coplanar waveguide (CPW) consists of a centre strip separated by a narrow gap from
two ground planes on either side as shown in Fig. 7.15. Characteristic impedance of
a CPW configuration is determined by the dimensions of the centre strip, the gap, the
thickness and permittivity of the dielectric substrate. The gap in the CPW configu-
ration is usually very small, about half of the dielectric substrate thickness, supports
electric fields primarily concentrating the fields in the substrate area to minimize
radiation. CPW can support propagation of EM waves down to zero frequency. The
mode of EM wave propagation is quasi TEM for the presence of mixed dielectric
medium. Typical electric and magnetic field distribution in a coplanar waveguide is
shown in Fig. 7.16.
Two ground planes of the CPW ideally to be maintained at the same potential to
prevent propagation of unwanted modes. In case of different potentials, the CPW
mode will become uneven, with a higher field in one gap than the other resulting
in combined propagation of desired coplanar mode and undesired parasitic slotline
mode. To suppress the propagation of undesired slotline mode, air bridges between
ground planes have to be applied as shown in Fig. 7.17. Practically, the spacing
between the bond wires must be one-quarter wavelength apart or less.
Coplanar waveguides are widely used for high-frequency MMIC applications,
for good access to grounding points and lower dispersion characteristic. It provides
Fig. 7.17 Coplanar Ground Strips

waveguide with bond wires
Bond
Wires
broadband performance as it does not need via holes for shunt and series-connected
elements.
7.3.5.4 Substrate Integrated Waveguide (SIW)
Metallic waveguides are having low-loss and high-power handling capability suitable
for high-Q and high-power component realization and no EM radiation. However,
they are bulky, voluminous, expensive and non-planar, so their integration with pla-
nar components and packaged circuits is impractical. Whereas planar transmission
lines such as microstrip and coplanar waveguide are convenient for integration of
surface-mount devices and realization of compact components and packaged cir-
cuits but suffers from higher insertion loss and EM radiation. Alternative option
is the substrate integrated waveguide (SIW) which shows almost all advantages of
metallic waveguides and planar transmission lines [3–7].
Schematic diagram of surface integrated waveguide (SIW) is shown in Fig. 7.18.
Series of filled via holes are used for vertical walls. The vertical walls have a mesh
structure and are able to reflect all components of electric field. Upper and lower
conductive layers are the main conductive layers. The two periodic rows of metallic
vias are connecting the top and bottom ground planes of a dielectric substrate. The
SIW structure can be manufactured by standard lamination techniques. SIW structure
is a promising alternative to conventional all-metal waveguides for the design of
microwave/mmwave components and systems. Using SIW structure, relatively low-
loss and high-Q waveguide structure can be realized as highly integrated planar
microwave/mmwave components with compact size, low cost with very low EM
radiation.
7.4 Transmission Line Elements 147
Port-2
Via Holes
d
b Substrate
a
Metal Planes
Port-1
Fig. 7.18 Schematic of substrate integrated waveguide (SIW) structure
7.4 Transmission Line Elements
Transmission line elements with particular termination and particular length are
having interesting features and thus frequently used as circuit elements. Some of the
examples are given below:
(1) Short-circuited transmission line (Fig. 7.19): Putting Z L 0 in (7.26), the input
impedance of a short-circuited transmission line of length l will be
Z in (short) j Z o tan(βl) (7.79)
It is inductive for βl < π/2, i.e. l < λ/4 and capacitive for π/2 < βl <
π, i.e. λ/4 < l < λ/2. Therefore, a short-circuited transmission line can be
used as a capacitive or inductive element in a circuit.
(2) Open-circuited transmission line (Fig. 7.20): Putting Z L ∝ in (7.26), the
input impedance of an open-circuited transmission line of length l will be
Z in (open) − j Z o cot(βl) (7.80)
Fig. 7.19 Short-circuited

transmission line
Short
is capacitive or inductive
It is capacitive for βl < π/2, i.e. l < λ/4 and inductive for π/2 < βl <
π, i.e. λ/4 < l < λ/2. Therefore, an open-circuited transmission line can also
be used as a capacitive or an inductive element in a circuit.
(3) Half-wavelength transmission line section (Fig. 7.21): Putting l λ/2 in (7.26),
the input impedance of the transmission line,
Z in (λ/2) Z L (7.81)
It implies that the input impedance of a transmission line of half wavelength is

the load impedance. Due to the one-to-one impedance transformer property of
the half-wavelength transmission line, it is used for physical spacing between
two parts of a circuit without changing its impedance at a particular frequency.
(4) Quarter-wavelength transmission line section (Fig. 7.22): Putting l λ/4 in
(7.26), the input impedance of the transmission line,
Z o2
Z in (λ/4) (7.82)
ZL
It implies that the input impedance is resistive if the load impedance is resis-
tive but with different value. This property of the quarter-wavelength transmis-
sion line is used as impedance transformer to transform resistive impedance to
another resistive impedance of different value. For Z L 0, the input impedance,
Z in ∝. This implies that the input impedance of a quarter-wavelength short-
circuited transmission line tends to infinity. This property is used to realize an
RF choke for providing DC bias to a microwave circuit. Similarly, for Z L ∝,
the input impedance, Z in 0. This implies that the input impedance of a
quarter-wavelength open-circuited transmission line tends to zero. This prop-
erty is used to provide short-circuit termination for a particular frequency. Thus,
open-circuit λ/4 stub transmission line can act as band-stop filter.
(5) One-eighth wavelength transmission line section (Fig. 7.23): Putting l λ/8
in (7.26), the input impedance of the transmission line,

λ Z o (Z L + j Z o )
Z in (7.83)
8 Zo + j Z L
It implies that if Z L is real, then |Z in | Z o , i.e. independent of the value of

load impedance Z L at a particular frequency. However, other than this partic-
ular frequency, the input impedance is dependent on the load impedance. This
Fig. 7.20 Open circuitted

transmission line
Open
is capacitive or inductive
7.4 Transmission Line Elements 149
Fig. 7.21 Half-wavelength λ

transmission line
Zin = ZL
Fig. 7.22 Quarter- λ

wavelength transmission
line
Fig. 7.23 One-eighth λ

wavelength transmission line
property is used to realize amplitude tilt equalizer. In a particular case, when

Z L is complex and Z o |Z L |, then the input impedance Z in becomes real.
This property of the one-eighth wavelength transmission line section is used to
transform complex impedance to real impedance satisfying Z o |Z L |.
7.5 Smith Chart Analysis
Smith chart is one of the most useful graphical tools for high-frequency circuit
design and analysis. The Smith chart is a graphical representation of complex
impedance/admittance on a complex reflection coefficient plane. It is a graph show-
ing both the normalized impedance and corresponding reflection coefficient in a
single plane. Smith chart is a simple and very convenient tool for transmission line
and circuit analysis. It is also a practical tool to design impedance matching circuit
at microwave frequency.
Complex impedance Z of a transmission line can be written as
Z (l) R + j X (7.84)
Suppose the characteristic impedance of the line is Z o . The normalized line

impedance z(l) with respect to the reference impedance Z o can be written as
z(l) (R + j X )/Z o r + j x (7.85)

The reflection coefficient Γ (l) of the line for a system of reference impedance Z o
can be written as
1 + Γ (l)
Z (l) Z o (7.86)
1 − Γ (l)
or
Z (l) 1 + Γ (l)
z(l) (7.87)
Zo 1 − Γ (l)
Suppose
Γ (l) Γr + jΓim (7.88)
Combining (7.85), (7.87) and (7.88),

1 + Γr + jΓim
r + jx (7.89)
1 − Γr − jΓim
1 − Γr2 − Γim
2
+ j2Γim
(7.90)
(1 − Γr ) + Γim
2 2
Equating real and imaginary parts separately,
1 − Γr2 − Γim2
r (7.91a)
(1 − Γr )2 + Γim2
2Γim
x (7.91b)
(1 − Γr )2 + Γim2
(7.91a) can be written as

2
r 2 1
Γr − + Γim
2
(7.92)
1+r 1+r
The result (7.92) for the real part indicates that on the complex plane with coor-
dinates (Γr , Γim ), all the possible impedances with a given normalized resistance r
are found on a circle with
r
Center: ,0 (7.93a)
1+r
and

1
Radius: (7.93b)
1+r
7.5 Smith Chart Analysis 151
Fig. 7.24 Constant resistance (r ) circles on complex Γ plane
Figure 7.24 shows the plots of (7.92) for different constant values of normalized
resistance r . These circles are the constant resistance circles. There are many other
circles which can be drawn for different values of the constant resistance r . Boundary
circle corresponds to r 0. With the increase of resistance values, circles move
towards right and the large value circles collapse towards r ∝. As the normalized
resistance r varies from 0 to ∝, a family of circles completely contained inside the
domain of the reflection coefficient |Γ | ≤ 1.
Similarly, (7.91b) can be written as

2
1 2 1
(Γr − 1) + Γim −
2
(7.94)
x x
The result (7.94) for the imaginary part indicates that on the complex plane with
coordinates (Γr , Γim ), all the possible impedances with a given normalized reactance
x are found on a circle with

1
Center: 1, (7.95a)
x
and

1
Radius: (7.95b)
x
Figure 7.25 shows the plots of (7.95) for different constant values of normalized
reactance x. These are constant reactance arcs. The positive imaginary components
Fig. 7.25 Constant reactance (x) circles on complex Γ plane
indicate that the impedance is inductive and negative imaginary values indicate that
the impedance is capacitive. There are many other such arcs which can be drawn
for different values of reactances. The horizontal axis itself is a constant reactance
arc of infinite radius for x 0. As in case of constant resistance circles, the smaller
reactance values have arcs of larger circles. As the reactance increases, the radius of
the arcs becomes smaller and the arcs collapse towards x ∝. These circles are the
constant reactance circles. As the normalized reactance x varies from −∝ to +∝, a
family of arcs is completely contained inside the domain of the reflection coefficient
|Γ | ≤ 1. Plot of families of constant resistance circles and reactance arcs on a single
reflection coefficient plane gives the impedance Smith chart.
Smith chart containing plot of admittance on Γ plane is also very useful particu-
larly for matching circuit design using shunt elements. Normalized admittance y(l)
is given by
Y (l)
y(l) (7.96a)
Yo
Zo
(7.96b)
Z (l)
using (7.86),
Zo 1 − Γ (l)
y(l) (7.96c)
Z (l) 1 + Γ (l)
From (7.20), the reflection coefficient Γ (l) of a lossless transmission line can be
written as
Γ (l) |Γ |e+ j2βl (7.97a)
Therefore,
Γ (l + λ/4) |Γ |e+ j2β(l+λ/4)

|Γ |e+ j2βl × e+ jπ
−Γ (l) (7.97b)
Thus, (7.96c) can be written as

1 − Γ (l) 1 + Γ (l + λ/4)
y(l) (7.98)
1 + Γ (l) 1 − Γ (l + λ/4)
Comparing (7.98) with (7.87):
y(l) z(l + λ/4) (7.99)
Thus, rotating all the resistance circles and reactance arcs of the Smith chart by
180° around the complex Γ plane provide a mapping of complex admittance onto
the complex Γ plane. Figure 7.26 shows the impedance Z and admittance Y Smith
charts. The distance from origin to any other point on the Smith chart gives magnitude
of the reflection coefficient |Γ |. The angle between x 0 line and the line connecting
the point to the origin gives the phase angle of the reflection coefficient Γ . Smith
chart containing impedance and admittance chart on a single reflection coefficient
plane is shown in Fig. 7.27. Here, the horizontal axis is a constant reactance circle
of value r 0. Thus, all impedances on the horizontal axis are pure resistances. The
upper half of the chart corresponds to inductive impedances of all positive constant
reactances and negative susceptance arcs. The lower half of the chart corresponds to
capacitive impedances of all negative constant reactances and positive susceptance
arcs.
Reflection coefficient Γ (l) of a lossless transmission line at a distance l from a load
towards generator is given by Γ (−l) |Γ |e− j2βl . The magnitude of Γ (−l) remains
constant with the change of transmission line length though its phase changes. There-
fore, moving through a lossless transmission line, magnitude of reflection coefficient
remains constant. This result gives very important set of constant reflection coeffi-
cient circles on Γ plane. The voltage standing wave ratio VSWR is given by (7.23)
1 + |Γ |
VSWR (7.100)
1 − |Γ |
From (7.100), it is clear that the VSWR remains constant with the change of
transmission line length. Figure 7.28 shows constant VSWR circles (dark dashed)
1.0 j -1.0 j
0.5 j 2.0 j -2.0 j -0.5 j
0.0 0.5 1.0 2.0 2.0 1.0 0.5 0.0
-0.5 j -2.0 j 2.0 j 0.5 j

-1.0 j 1.0 j
Z Chart Y Chart
Fig. 7.26 Smith chart: impedance and admittance plots on complex reflection coefficient plane
1
0.5 2
Pure 0.5 1 2
0 ∝
resistance
Z Chart
0.5 2 Y Chart
1
Fig. 7.27 Smith chart: impedance and admittance plots on a single complex reflection coefficient
plane
on complex reflection coefficient plane. Here, it is shown that a load impedance Z L

changes to Z I N through a series transmission line of length l along a constant VSWR
circle while moving towards generator.
Example 7.9 Compute transferred input impedance of a load (25 + 50 j) trans-
ferred through a series-connected transmission line of characteristic impedance 50
and length (a) λ/8 and (b) λ/4.
Fig. 7.28 Constant VSWR circles
Solution
The problem is shown schematically in Fig. 7.29. Here, we have to determine Z I N
for the two cases.
The normalized load impedance is given by
(25 + 50 j)
zL 0.5 + 1 j
50
The normalized load z L is plotted marked as A on the smith chart in Fig. 7.29. The
load reflection coefficient for this load is shown as vector Γ L . A constant VSWR circle
is drawn (dashed circle) corresponding to this reflection coefficient. Input impedance
will change along the constant VSWR circle.
(a) The length λ/8 corresponds to 90° on the Smith chart. Moving the point A by
90° towards generator (clockwise direction), it reaches at the point B which
corresponds to normalized impedance of (4 − 1 j), i.e. (200 − 50 j).
(b) The length λ/4 corresponds to 180° on the Smith chart. Moving the point A by
180° towards generator (clockwise direction), it reaches at the point C which
corresponds to normalized impedance of (0.4 − 0.8 j), i.e. (20 − 40 j). From
the plot, it is clear that this point corresponds to normalized admittance of
(0.5 + 1 j).
Example 7.10 Compute transferred input impedance of a pure resistive load of 100
, when it is shunted by an (a) open-circuited transmission line of length λ/8 and
Fig. 7.29 Impedance transfer for series transmission line
(b) short-circuited transmission line of length λ/8. The characteristic impedances of

the shunt lines are 50 .
Solution
The normalized load impedance is given by
(100 + 0 j)
zL 2 + 0j
50
The normalized load z L is plotted marked as A on the Smith chart in Fig. 7.30.
Connecting an open- or short-circuit stub parallel to a load will change its reactive
part keeping conductance constant. Thus, the input impedance will change along the
constant conductance circle with the change of stub length.
(a) As shown in Fig. 7.30, λ/8 open-circuit stub corresponds to normalized

impedance of −1 j, and thus the normalized susceptance of the λ/8 open-circuit
line is +1 j. Therefore, moving along the constant conductance (1/2 0.5) cir-
cle, the point B is reached. The admittance corresponding to the point B is
Fig. 7.30 Impedance transfer for series transmission line
(0.5 + 1 j). This corresponds to normalized impedance of (0.4 − 0.8 j). Thus,
the transformed impedance is (20 − 40 j) .
(b) As shown in Fig. 7.30, λ/8 short-circuit stub corresponds to normalized
impedance of +1 j, and thus normalized susceptance of the λ/8 open-circuit
line is −1 j. Therefore, moving along the constant conductance (1/2 0.5) cir-
cle, the point C is reached. The admittance corresponding to the point C is
(0.5 − 1 j). This is corresponding to normalized impedance of (0.4 + 0.8 j).
Thus, the transformed impedance is (20 + 40 j).
Lumped elements in different configurations are also used as impedance trans-
former. Figure 7.31 shows the transfer of impedance z L on a Smith chart by the
use of resistance, capacitance and inductance as lumped elements in different
configurations.
0.5 2
H
G
E
F
A
B
0.5 1 2
0 ∝
C D
Z Chart
0.5 2 Y Chart
A : Positive series resistance C : Series capacitance

B : Positive shunt resistance D : Shunt capacitance
G : Negative series resistance E : Series inductance
H : Negative shunt resistance F : Shunt inductance
Fig. 7.31 Impedance transfer using lumped elements
7.6 Network Theory of Circuits and Transmission Lines
Several microwave devices and elements are connected together by transmission

lines to form a microwave network. A large complex network may be divided
into subnetworks usually 1- or 2-port networks for the purpose of analysis and
design before being interconnected to form the complex network. Network anal-
ysis approach enables to know the port behaviours in terms of voltages, currents
and impedances of a network without knowing inside details about it. Therefore, the
network analysis is very useful for the analysis of microwave systems, instead of
working with Maxwell equations. Knowing the network parameters related to the
ports enable us to treat the network as a black box when embedded within a large
network.
7.6 Network Theory of Circuits and Transmission Lines 159
Fig. 7.32 Schematic of an

n-port network
7.6.1 S-Parameter Network Representation
In case of TEM waves, like waves in coaxial lines, unique current and voltage
can be defined; however, it is not possible to uniquely define for non-TEM waves,
like waves in rectangular waveguide. Moreover, there is no suitable instrument to
directly measure voltages and currents in complex value at microwave frequencies.
Therefore, microwave networks are not conveniently represented by impedance or
admittance matrices. For microwave networks, a more appropriate representation is
scattering parameter (S-parameter) matrix representation where directly measurable
reflected, transmitted and incident waves are used. Scattering matrix relates the inci-
dent, transmitted and reflected waves at the ports which can be measured directly
with a network analyzer. Knowing the S-parameters these can be converted to any
other matrix parameters which are suitable for particular analysis of the microwave
network depending upon applications [8–12].
Consider an n-port network as shown in Fig. 7.32. Here, V1+ , V2+ , …, Vn+ are
the amplitudes of the voltage wave and I1+ , I2+ , …, In+ are the amplitudes of the
current wave incident on port-1, port-2, …, port-n, respectively, whereas V1− , V2− ,
…, Vn− are the amplitudes of the voltage wave and I1− , I2− , …, In− are the amplitudes
of the current wave reflected from the port-1, port-2, …, port-n, respectively. The
scattering
matrix [S]
which relates the reflected and incident voltage waves is written
as V − [S] V +
or
⎡ ⎤
⎡ ⎤ ⎡ ⎤
− ⎢ S11 S12 · · · S1n ⎥
⎢ V1 ⎥ ⎢ ⎥ V1+
⎢ −⎥ ⎢ ⎥⎢ ⎥
⎢ V2 ⎥ ⎢ S21 S22 · · · S2n ⎥⎢ V2+ ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥
⎢ ⎥⎢
⎢
⎥⎢
⎥⎢
⎥ (7.101)
⎢ . ⎥ ⎢
⎢ .. ⎥ ⎢ .. ⎥⎢ .. ⎥
⎥
⎣ ⎦ ⎢ . ⎥⎣ . ⎦
⎥
Vn− ⎣ ⎦ Vn+
Sn1 Sn2 · · · Snn
Fig. 7.33 Schematic of a

2-port network
Fig. 7.34 2-port network

with matched output port
Thus, each element of the scattering matrix is defined as
Vi−
Si j , with the condition: Vk+ 0 for k j (7.102)
V j+
This means, there is no incident wave other than the port-j. Therefore, the trans-
mission line on these ports must be terminated by a matched load.
When a 2-port network is driven by a voltage source Vg at its port-1 and port-2
is terminated by load of impedance Z L as shown in Fig. 7.33, then the incident and
reflected voltage waves are related by,
⎡ ⎤
V1− S S V1+
⎣ ⎦ 11 12
(7.103)
V2− S21 S22 V2+
or
V1− S11 V1+ + S12 V2+ (7.104)

V2− S21 V1+ + S22 V2+ (7.105)
Thus, from Fig. 7.34, the S-parameters will be determined by,
V1−
S11 , when V2+ 0 (7.106)
V1+

with matched input port
Fig. 7.36 A 2-port network

with Z o at input and output
Comparing with (7.15), it is clear that S11 is nothing but the reflection coefficient
of the input port when the output port is terminated by its characteristic impedance.
Thus, from (7.19) S11 can be written as
Z in | Z L Z o −Z o
S11 Γin | Z L Z O (7.107)
Z in | Z L Z o +Z o
In similar way from Fig. 7.35, S22 can be determined as
V2−
S22 , when V1+ 0 (7.108)
V2+
Thus, S22 is nothing but the reflection coefficient of the output port when the
input port of the network is terminated by its characteristic impedance Z o as shown
in Fig. 7.36. Thus, from (7.19), S22 can be written as
Z out | Z S Z O −Z o
S22 Γout | Z S Z O (7.109)
Z out | Z S Z O +Z o
From (7.105), S21 is given by
V2−
S21 , when V2+ 0 (7.110)
V1+
Therefore, S21 is the ratio of voltage wave reached to the output matched load to
the incident voltage wave at the input port. S21 is called forward transmission when
the output port is terminated by its characteristic impedance Z o . The incident voltage
wave V1+ at the input port is given by
Vg Z o Vg
V1+ (7.111)
Zo + Zo 2
Since the output port is terminated to its characteristic impedance Z o , the voltage
wave V2− is the output voltage Vout across the output termination Z o . Thus, the
forward transmission coefficient is given by
2Vout
S21 (7.112)
Vg
In a similar way, the reverse transmission coefficient S12 is given by
V1−
S12 , when V1+ 0 (7.113)
V2+
Thus, the reverse transmission S12 is the ratio of voltage wave reached at the input
terminated by Z o to the incident voltage wave at the output port from a source of
impedance Z o . Thus, S-parameters of a network do not depend on the source or load
impedance; however, it depends upon the characteristic impedance of the ports in
which it is connected.
In a general way, the voltage and current at port-n can be written by
Vn Vn+ + Vn− (7.114)

In (Vn+ − Vn− )/Z on (7.115)
Here, Z on is the characteristic impedance of the nth port. Considering Z on as real,

the power delivered to the nth port is given by
1
Pn Re Vn In∗ (7.116)
2
1 2
Re Vn+ − Vn−
2
(7.117)
2Z on
Presence of Z on term implies that the power delivered to the port is dependent
on the characteristic impedance of the line connected to the port. To make power
delivered to a port independent of the characteristic impedance of that port, a new
set of incident and reflected wave is defined. Suppose the new set of incident wave
(an ) and reflected wave (bn ) for port-n is defined as
V+ V−
an √ n and bn √ n (7.118)
Z on Z on
Thus, the voltage and current at the nth port is given by

Vn Vn+ + Vn− Z on (an + bn ) (7.119)
1 + 1
In Vn − Vn− √ (an − bn ) (7.120)
Z on Z on
Now, the power delivered to the nth port is given by

1
Pn Re Vn In∗ (7.121)
2
1
Re (an + bn ) an∗ − bn∗ (7.122)
2
1
Re |an |2 − |bn |2 + an∗ bn − an bn∗ (7.123)
2
or
1
Pn Re |an |2 − |bn |2 (7.124)
2
This shows that the power delivered to the port is independent of the characteristic
impedance (Z on ) of the line connected to that port. The generalized scattering matrix
[S] is the relations in between the normalized incident wave (an ) and the normalized
reflected wave (bn ) and is given by
[b] [S][a] (7.125)
or
⎡ ⎤
⎡ ⎤ ⎢ S11 S12 · · · S1n ⎥⎡ ⎤
b1 ⎢ ⎥ a1
⎢ ⎥ ⎢ ⎥⎢ ⎥
⎢ b2 ⎥ ⎢ S21 S22 · · · S2n ⎥⎢ a2 ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥
⎢ ⎥⎢ ⎥⎢ ⎥ (7.126)
⎢ .. ⎥ ⎢ ⎥⎢ . ⎥
⎣ . ⎦ ⎢
⎢
..
.
⎥⎣ ..
⎥ ⎦
⎢ ⎥ a
bn ⎣ ⎦ n
Sn1 Sn2 · · · Snn
where each element of the S-parameter is defined as

bi
Si j , with the condition: ak 0 for k j (7.127)
aj
The elements of the generalized scattering matrix can be determined by putting

the values of a j and bi from (7.118).
√
Vi− Z oi
Si j √ . + , with the condition: Vk+ 0 for k j (7.128)
Z oi V j
Fig. 7.37 A generalized

2-port S-parameter
representation
and
Vi−
Sii , with the condition: Vk+ 0 for k i (7.129)
Vi+
In case of 2-port network as shown in Fig. 7.37, the generalized S-parameter can
be written as
b1 S11 a1 + S12 a2 (7.130)

b2 S21 a1 + S22 a2 (7.131)
In matrix form,

b1 S11 S12 a1
(7.132)
b2 S21 S22 a2
In the similar way, the S-parameters can be determined as follows:

b1
S11 , when a2 0 (7.133)
a1
Putting the values of a1 and b1 from (7.118)

√
V1− Z o1 V−
S11 √ . + 1+ , when V2+ 0 (7.134)
Z o1 V1 V1
This is the same as (7.106) and is the input port reflection coefficient when the
output port is matched terminated.
Similarly, other parameters can be determined as
√
b1 V− Z o2
S12 √1 , when V1+ 0 (7.135)
a2 Z o1 V2+
√
b2 V2− Z o1
S21 √ , when V2+ 0 (7.136)
a1 Z o2 V1+
√
b2 V2− Z o2 V−
S22 √ . + 2+ , when V1+ 0 (7.137)
a2 Z o2 V2 V2
7.6.2 S-Parameters for Matched Reciprocal Lossless

Networks
The scattering matrices for the matched, reciprocal and lossless networks have spe-
cial properties. The input impedance at each port of a matched network is equal to
the reference impedance Z o when all other ports are terminated by matched loads.
Therefore, the reflection coefficient of each port is zero, and thus there will be no
reflected signal from the port if a signal is incident only on that port. Therefore, for
a matched device/network,
Vi−
Sii 0, for all i (7.138)
Vi+
This means that a matched device has a scattering matrix where all diagonal
elements, Sii are zero.
A reciprocal network does not contain any active device or non-reciprocal medium
such as plasmas or ferrites. For a reciprocal network, the scattering matrix is sym-
metric. Therefore, the condition for a reciprocal network is
Si j S ji (7.139)
For example, a reciprocal network will have S21 S12 or, S23 S32 , etc.
Therefore, for a reciprocal network, the S-parameter matrix satisfies
[S]T [S] (7.140)
For a lossless network, the net power delivered to the network must come out
from it. Thus, for a lossless network there will be no absorption of power by it, i.e.
there will be no power dissipation and generation of heat within the network. With
reference to an n-port network as shown in Fig. 7.32, the power incident to the port-n
is given by
1 ∗
Pn+ an an (7.141a)
2
The total power incident to the network through all the n ports is given by
n
1 ∗
P+ an an (7.141b)
n1
2
Fig. 7.38 Schematic

diagram of two-way resistive
power divider
1
[an ] H [an ] (7.141c)
2
Here, operator H indicates the conjugate transpose.
Similarly, the exiting power, i.e. power coming out from the port-n is given by
1 ∗
Pn− bn bn (7.142a)
2
The total reflected power from the network through all the n ports is given by
n
1 ∗
P− bn bn (7.142b)
n1
2
1
[bn ] H [bn ] (7.142c)
2
Therefore, the resultant power delivered through the port-n to the network is given
by
1
P P + − P − [an ] H [an ] − [bn ] H [bn ] (7.143a)
2
Putting [bn ] [S][an ] in (7.143a)
1
P [an ] H [an ] − [S] H [an ] H [S][an ]
2
1
[an ] H [I ] − [S] H [S] [an ] (7.143b)
2
Here, [I ] is the identity matrix. For a lossless network, the total power delivered
to the network, i.e. the power absorbed by the device must be zero, i.e. P 0.
Thus, for lossless network from (7.143b)
[I ] − [S] H [S] 0
or
[S] H [S] [I ] (7.144)
A matrix that satisfies [S] H [S] [I ] is a special matrix known as unitary matrix.
Thus, S-parameter matrix of a lossless network is a unitary matrix. The columns of
a unitary matrix form an orthogonal set. That means each column of the scattering
matrix will have a magnitude equal to one while the inner product of dissimilar
columns must be zero. Thus, for a lossless network, the S-parameters satisfies

n
Smn 2 1 for all n (7.145a)
m1
and

n
Sni Sn∗j 0 for all i j (7.145b)
n1
For example, the relationship of the S-parameters for a lossless 2-port network
will be
S11 2 + S21 2 1
S12 2 + S22 2 1
and
∗ ∗
S11 S12 + S21 S22 0
∗ ∗
S12 S11 + S22 S21 0
Example 7.11 A 2-port network has the following scattering matrix:

0.467 − 144.01◦ 0.884 − 142.53◦
[S]
0.884 − 142.53◦ 0.467 38.95◦
Determine whether the network is reciprocal, matched and lossless.
Solution
Here, S12 0.884 − 142.53◦ S21 0.884 − 142.53◦ . Thus, S-parameter
matrix is symmetric. So, the network is reciprocal. S11 0.467 − 144.01◦
0 also S22 0.467 38.95◦ 0. Thus, the ports of the network are not matched.
To check for lossless, we can check unitary property of the matrix,
|S11 |2 + |S21 |2 0.4672 + 0.8842 1

|S12 |2 + |S22 |2 0.8842 + 0.4672 1
∗ ∗
S11 S12 + S21 S22 0.467 − 144.01◦ × 0.884 142.53◦
Fig. 7.39 Equivalent circuit for signal Vs at port-1 to calculate Vo
+ 0.884 − 142.53◦ × 0.467 − 38.95◦ 0
∗ ∗
Similarly, S11 S12 + S21 S22 0
Thus, the matrix elements satisfy all the conditions for a unitary matrix. Thus, the
network is lossless.
Example 7.12 Compute S-parameter of a 3-port two-way resistive power divider as

shown in Fig. 7.38.
Solution
Suppose a signal Vs is applied to the port-1 through a source resistance Ro and the
other two ports are terminated by resistance Ro as shown in Fig. 7.38. The impedance
at the junction seeing towards the port-2 and port-3 is the parallel combination of
two resistors of values (Ro /3 + Ro ). Therefore, the input impedance, Z I N at port-1
is given by

Ro Ro Ro
ZIN + + Ro || + Ro
3 3 3
Ro (7.146)
Therefore, the port-1 is a matched port. Since the structure is symmetric, other
two ports are also matched. Thus, S11 S22 S33 0.
To calculate output voltage wave under the application of signal Vs at port-1, the
equivalent circuit is shown in Fig. 7.39. The signal voltage Vo2 at the port-2 is given
by

Vs /2
Vo2 Ro
2Ro /3 + Ro /3 + Ro
Vs
(7.147)
4
Thus, the transmission coefficient S21 is given by
Vo2 Vs /4 1
S21 (7.148)
Vs /2 Vs /2 2
Since the network is symmetric in structure, we can write

Fig. 7.40 Block schematic of two-way lossless equal power divider
S21 S12 S31 S32 S23 S32 1/2 (7.149)
Therefore, the S-parameter matrix of the resistive power splitter is given by

⎡ ⎤ ⎡ ⎤
S11 S12 S13 011
⎢ ⎥ 1
[S] ⎣ S21 S22 S23 ⎦ ⎣ 1 0 1 ⎦ (7.150)
2
S31 S32 S33 110
From the S-parameter matrix, it is clear that the network is matched and reciprocal.
Example 7.13 Compute S-parameter matrix of a lossless two-way equal power

divider as shown in Fig. 7.40.
Solution
The two-way power divider does not contain any active device and non-reciprocal
components. Thus, the network is reciprocal, we can write
S12 S21 , S13 S31 and S23 S32 (7.151)
Suppose the port-1 is the input port and it is matched to the reference impedance.
Thus, S11 0. For equal power division to port-2 and port-3, we can write (without
considering the phase difference if any)
√
S21 S31 1/ 2 (7.152)
Thus, the S-parameter matrix can be written as

⎡ √ √ ⎤
0 1/ 2 1/ 2
⎢ √ ⎥
[S] ⎢
⎣ 1/ 2 S22 S23 ⎦
⎥ (7.153)
√
1/ 2 S23 S33
Since the network is lossless, the matrix is unitary. Thus, applying (7.145a) we
can write
1/2 + |S22 |2 + |S23 |2 1 (7.154a)

Fig. 7.41 Schematic diagram of two-way resistive power divider
and
1/2 + |S23 |2 + |S33 |2 1 (7.154b)
From (7.154a) and (7.154b), we can write
S22 S33 (7.155)
Now, applying (7.145b), we can write

√ √
0 + 1/ 2S22 + 1/ 2S23 0 (7.156a)
and
∗ ∗
1/2 + S22 S23 + S23 S33 0 (7.156b)
Using (7.155), (7.156a) and (7.156b), we can write
S22 S33 −S23 1/2
Therefore, the S-parameter matrix of the lossless 2-way power splitter is

⎡ √ √ ⎤
0 1/ 2 1/ 2
⎢ √ ⎥
[S] ⎢ ⎥
⎣ 1/ 2 1/2 −1/2 ⎦ (7.157)
√
1/ 2 −1/2 1/2
Here, S22 and S33 are non-zero. Thus, for the 3-port lossless network, the other
ports are not matched.
Example 7.14 Is it possible to have a 3-port matched, reciprocal and simultaneously

lossless network?
Solution
Suppose all the 3-ports of the network shown in Fig. 7.41 are matched and the network
is reciprocal. Then, we can write
S11 S22 S33 0 (7.158)
and
S12 S21 , S13 S31 , S23 S32
Thus, the S-parameter matrix of a matched reciprocal 3-port network can be

written as
⎡ ⎤
0 S12 S13
⎢ ⎥
[S] ⎣ S12 0 S23 ⎦ (7.159)
S13 S23 0
Now, if the network is simultaneously lossless. The following conditions for

unitary matrix also should be satisfied.
Since the network is lossless, the matrix is unitary. Thus, applying (7.145a), we
can write
|S12 |2 + |S13 |2 1 (7.160a)

|S12 | + |S23 | 1
2 2
(7.160b)
|S13 | + |S23 | 1
2 2
(7.160c)
and applying (7.145b),

∗
S12 S13 0 (7.161a)
∗
S12 S23 0 (7.161b)
∗
S13 S23 0 (7.161c)
To satisfy (7.161), at least two parameters among S12 , S13 and S23 must be zero
but this is contradictory to Eq. (7.160). Therefore, it is not possible to have a 3-port
network with all ports matched, reciprocal and lossless.
Example 7.15 Compute S-parameter of a series network of impedance Z a as shown

in Fig. 7.42.
Solution
The S-parameters of the series network can be determined as follows:
b1
S11 , when a2 0
a1
V−
1+ , when V2+ 0
V1
Z a /(Z a + 2Z o )
Fig. 7.42 S-parameter

determination of a 2-port
series network
Fig. 7.43 S-parameter determination of a 2-port shunt network
It is a symmetric network, and therefore,
S11 S22 (Z a /(Z a + 2Z o ). (7.162a)
Similarly, from (7.136) and using (7.112), S 21 can be determined as

√
b2 V2− Z o V2−
S21 √ , when V2+ 0
a1 Z o V1+ V1+
2Z o /(Z a + 2Z o ) (7.162b)
Thus, in matrix form, S-parameter of the series network can be written as

⎡ ⎤
Za 2Z o
S11 S12
⎣ ⎦
2Z o +Z a 2Z o +Z a
(7.162c)
S21 S22 2Z o Za
2Z o +Z a 2Z o +Z a
Example 7.16 Compute S-parameter of a shunt network which consists of series

combination of a transmission line (Z o ) of length l in between two resistors Ra and
Rb as shown in Fig. 7.43.
Solution
The shunt network consists of a transmission line of length l and two resistors Ra
and Rb connected in series as shown in Fig. 7.43. Using (7.26), the input impedance
Z a of the shunt network can be written as
Rb + j Z o tan(βl)
Z a Ra + Z o (7.163a)
Z o + j Rb tan(βl)
Here, Z o is the characteristic impedance of the shunt transmission line and port
terminations. The S-parameters of the shunt network can be determined as follows:
b1
S11 , when a2 0
a1
V−
1+ , when V2+ 0
V1
−Z o /(Z o + 2Z a )
It is a symmetric network, and therefore,
S11 S22 −Z o /(Z o + 2Z a ). (7.163b)
Similarly, from (7.136) and using (7.112), S21 can be determined as

√
b2 V − Zo V2−
S21 √2 , when V2+ 0
a1 Z o V1+ V1+
2Z a /(Z o + 2Z a ) S12 (7.163c)
Thus, in matrix form, S-parameter of the shunt network can be written as

⎡ ⎤
−Z o 2Z a
S11 S12
⎣ ⎦
Z o +2Z a Z o +2Z a
(7.163d)
S21 S22 2Z a −Z o
Z o +2Z a Z o +2Z a
In most of the practical cases, ports of a microwave network are terminated by

the same characteristic impedance (mostly Z o 50 ). However, generalized S-
parameters can be used wherever the ports of a network are terminated with different
characteristic impedances (say Z o 25 , 70 , etc.). From the above discussion,
it is clear that in case of S-parameter representations, parameters to be measured are
incident, reflected and transmitted waves.
At this point, the advantages of using S-parameter matrix in microwave circuits
can be summarized as
• At microwave frequency; incident, reflected and transmitted waves, all are directly
measurable quantities, required to determine S-parameters.
Fig. 7.44 ABCD parameter

representation of a 2-port
network
• No suitable instrument is available to measure voltage and current in their complex

form at microwave frequency.
• Matched load required for S-parameters determination are available in broadband
microwave applications, whereas it is difficult to make open and short circuits over
broadband. Practical microwave short often behaves as inductive element and open
behaves as capacitive element.
• Termination of active devices with Z o (mostly 50 ) causes no oscillation to
active devices, whereas short and open at the active device and circuit ports may
result in oscillation, i.e. may not remain stable.
7.6.3 ABCD Parameter for Network Representation
The most practical way to represent a microwave network is by S-parameters since

it is determined from measurable quantities. However, Z, Y and ABCD parame-
ter representation of microwave networks sometimes leads to convenient and easy
way for analysis. Most of the microwave circuits and systems consist of cascaded
connection of several 2-port networks. To analyse these types of circuits, it is conve-
nient to define 2-port parameter representation as transmission parameters or ABCD
parameters. The transmission parameter matrix of circuit or system formed by several
cascaded 2-port networks is the product of the transmission matrices of the individual
2-port networks.
The transmission matrix relates input voltage and current to the output voltage
and current of a given 2-port network as shown in Fig. 7.44 as follows:
V1 AV2 + B I2
I1 C V2 + D I2 (7.164)
or in matrix form,

V1 A B V2
(7.165a)
I1 C D I2
where the matrix elements can be determined as follows:

Fig. 7.45 ABCD parameters of cascaded two 2-port networks
V1
A , open-circuit voltage ratio, when I2 0 (7.165b)
V2
V1
B , short-circuit transfer impedance, when V2 0, (7.165c)
I2
I1
C , open-circuit transfer admittance, when I2 0 (7.165d)
V2
I1
D , short-circuit current ratio, when V2 0 (7.165e)
I2
Suppose two 2-port networks of parameters [A1 B1 C 1 D1 ] and [A2 B2 C 2 D2 ] are

connected in cascade as shown in Fig. 7.45. The two networks can be represented by

V1 A1 B1 V2
(7.166)
I1 C 1 D1 I2

V2 A2 B2 V3
(7.167)
I2 C 2 D2 I3
Combining two matrixes (7.166) and (7.167), the relationship in between port-1
and port-3 can be written as

V1 A1 B1 A2 B2 V3
(7.168)
I1 C 1 D1 C 2 D2 I3
From (7.168), it is clear that the ABCD parameter of cascaded network is just
matrix multiplication of the individual networks.
Example 7.17 A π network is shown in Fig. 7.46. It consists of a series transmission

line of length l and characteristic impedance Z o in between two shunt elements Z a
and Z b , respectively. Determine ABCD parameter of this π network.
Solution
The complete network can be considered as combination of three simple networks
(1, 2 and 3) connected in cascade configuration as shown in Fig. 7.46.
The network-1 can be represented as
Fig. 7.46 A 2-port π network

V1 1 0 Va
(7.169a)
I1 Ya 1 Ia
Here, Ya 1/Z a . The network-2 can be represented as
Va Vb e jβl Vb cos(βl) + j Vb sin(βl) Vb cos(βl) + j Ib Z o sin(βl)

Ia Ib e jβl Ib cos(βl) + j Ib sin(βl) Ib cos(βl) + j Vb Yo sin(βl)
Here, Yo 1/Z o . Thus,

Va cos(βl) j Z o sin(βl) Vb
(7.169b)
Ia jYo sin(βl) cos(βl) Ib
Similarly, network-3 can be represented as

Vb 1 0 V2
(7.169c)
Ib Yb 1 I2
Here, Yb 1/Z b . Therefore, the complete network can be represented by the

matrix multiplication of these three ABCD matrices as

V1 1 0 cos(βl) j Z o sin(βl) 1 0
(7.169d)
I1 Ya 1 jYo sin(βl) cos(βl) Yb 1
Thus, measurable S-parameters can be converted to ABCD or other parameters

for convenience of analysis depending upon the way the 2-port subnetworks are
interconnected to form a larger network.
Fig. 7.47 A two-port

network with matched output
termination
7.6.4 Conversion in Between ABCD and S-Parameters
The conversion relationship in between 2-port S-parameters and ABCD parameters

can be derived in the following way:

b1 S11 S12 a1 V1 A B V2
and
b2 S21 S22 a 2 I1 C D I2
From (7.114) and (7.115),

V1 V1+ + V1− and I1 V1+ − V1− /Z o
or
V1+ (V1 + Z o I1 )/2 and V1− (V1 − Z o I1 )/2

b1 V1− AV2 + B I2 − Z o C V2 − Z o D I2
S11 +
a1 a2 0 V1 V + 0 AV2 + B I2 + Z o C V2 + Z o D I2 V2+ 0
2
From Fig. 7.47, for V2+ 0,
V2 I2 Z o and I2 V2 /Z o V2 Yo
A + BYo − C Z o − D
S11 (7.170a)
A + BYo + C Z o + D
Similarly,

b2 V2− V2 2V2

S21 +
a1 a2 0 V1 (V1 + Z o I1 )/2 AV2 + B I2 + Z o C V2 + Z o D I2 V + 0
V2+ 0 2
or
2
S21 (7.170b)
A + BYo + C Z o + D
Relationship of S 12 and S 22 with ABCD parameters can be determined as follows:

Fig. 7.48 A two-port

network with matched input
termination

b1 V1−
S12 + where,
a2 a1 0 V2 V + 0
1

V2 V2+ + V2− and − I2 V2+ − V2− /Z o
Thus,
V1
S12
(V2 − Z o I2 )/2
The ABCD matrix can also be written as

V2 1 D −B V1
, where AD − BC
I2 −C A I1
Putting V2 and I2 :
2
S12 (7.170c)
A + BYo + C Z o + D
Similarly,

b2 V2−
S22 +
a2 a1 0 V2 V + 0
1
or
V2 + Z o I2
S22
V2 − Z o I2
From Fig. 7.48, the condition V1+ 0, gives V1 −I1 Z o , and I1 −V1 Yo .
Thus,
−A + BYo − C Z o + D
S22 (7.170d)
A + BYo + C Z o + D
In a similar way, ABCD parameters can be expressed by S-parameters as follows:

(1 + S11 )(1 − S22 ) + S12 S21
A (7.170e)
2S21
(1 + S11 )(1 + S22 ) − S12 S21

B Zo (7.170f)
2S21
(1 − S11 )(1 − S22 ) − S12 S21
C Yo (7.170g)
2S21
(1 − S11 )(1 + S22 ) + S12 S21
D (7.170h)
2S21
Scattering matrix relates the port incident, transmitted and reflected waves which
are directly measurable with a network analyzer. Once, the S-parameters of a network
are known, these can be converted to ABCD parameters using the derived relationship
or to any other suitable parameters such as Z or Y parameters for the analysis of the
microwave network.
7.7 Power Transfer in Microwave Networks
At microwave frequency range, for a mismatched load (load other than the char-
acteristic impedance Z o of the line), impedance looking down a transmission line
varies continuously along the line. Thus, it is convenient to express microwave power
transfer in terms of reflection coefficient of the load instead of impedance [13, 14].
7.7.1 Power Transfer from Source to a Load
Suppose a microwave generator of voltage VS with impedance Z S is directly con-

nected to a load of impedance Z L as shown in Fig. 7.49, the load can be represented by
its reflection coefficient Γ L (amplitude including phase) with respect to the reference
impedance Z o as
VL− Z L − Zo
ΓL + (7.171a)
VL Z L + Zo
Fig. 7.49 Power transfer

from source of impedance
Z S to a load Z L
Similarly, source reflection coefficient with respect to the reference impedance

Z o is
Z S − Zo
ΓS (7.171b)
Z S + Zo
The voltage VL across the load is given by

ZL
V L VS VL+ + VL− VL+ (1 + Γ L ) (7.171c)
ZS + ZL
or
VS ZL VS 1 − Γ S
VL+ (7.171d)
1 + ΓL Z S + Z L 2 1 − ΓS ΓL
Similarly,
VS ΓL VS 1 − Γ S
VL− ΓL (7.171e)
1 + ΓL ΓS + ΓL 2 1 − ΓS ΓL
The incident power to the load is given by

2
1 VL+
PincL (7.172a)
2 Zo
or
1 |VS |2 |1 − Γ S |2
PincL (7.172b)
8 Z o |1 − Γ S Γ L |2
And the reflected power from the load is given by

2
1 VL−
Pr e f L (7.173a)
2 Zo
or
1 |VS |2 |1 − Γ S |2
Pr e f L |Γ L |2 (7.173b)
8 Z o |1 − Γ S Γ L |2
Therefore, the power reached to the impedance Z L is given by
PinL PincL − Pr e f L (7.174a)
or
7.7 Power Transfer in Microwave Networks 181
Fig. 7.50 Power transfer to a load Z L

Pr e f L
PinL PincL 1 − (7.174b)
PincL
or

PinL PincL 1 − |Γ L |2 (7.174c)
or
1 |VS |2 |1 − Γ S |2
PinL 1 − |Γ L |2 (7.174d)
8 Z O |1 − Γ S Γ L |2
Equations (7.172b), (7.173b) and (7.174d) are the incident, reflected and delivered
power to the load of impedance Z L from a source of impedance Z S in terms of their
reflection coefficients.
In case of complex-conjugate-matched load for the given source impedance, i.e.
for Z L Z S∗ , the power delivered to the load becomes
1 |VS |2 |1 − Γ S |2
or, PinL | Z L Z S∗ (7.174e)
8 Z O 1 − |Γ S |2
1 |VS |2
or, PinL | Z L Z S∗ (7.174f)
8 Re(Z S )
This is nothing but the maximum power available from the generator. Thus, under
matched load condition, generator delivered maximum power to the load.
7.7.2 Power Transfer to and from a 2-Port Network
A 2-port network is connected in between a source of impedance Z S and a load of

impedance Z L as shown in Fig. 7.50. The network is represented by its S-parameters
with respect to the reference impedance Z o .
As defined in the previous section, the refection coefficients of the load and source
with respect to Z o can be written as
V2+ Z L − Zo
ΓL − (7.175a)
V2 Z L + Zo
V1+ Z S − Zo
ΓS − (7.175b)
V1 Z S + Zo
Thus, S-parameters of the network can be written as
V1− S11 V1+ + S12 Γ L V2− (7.176a)
and
V2− S21 V1+ + S22 Γ L V2− (7.176b)
Input reflection coefficient ΓIN and output reflection coefficient ΓOUT can be writ-
ten as
V1− Γ L S12 S21

ΓIN S11 + (7.177a)
V1+ 1 − Γ L S22
V2− Γ S S12 S21
ΓOUT + S22 + (7.177b)
V2 1 − Γ S S11
The incident voltage V1+ can be derived in the similar way as in previous section
and can be written as
VS 1 − Γ S
V1+ (7.177c)
2 1 − Γ S ΓIN
Power input to the network is given by

2
1 V1+
PIN 1 − |ΓIN |2 (7.177d)
2 Zo
or
1 |VS |2 |1 − Γ S |2
PIN 1 − |ΓIN |2 (7.177e)
8 Z o |1 − Γ S ΓIN | 2
Power delivered to the load is given by

2
1 V2−
PL 1 − |Γ L |2 (7.177f)
2 Zo
From (7.176b), V2− can be expressed as
S21 V1+
V2− (7.177g)
1 − S22 Γ L
Putting this value of V2− in (7.177f) and using (7.177c),

1 |VS |2 1 − |Γ L |2 |1 − Γ S |2
PL |S21 |2
(7.177h)
8 Zo |1 − S22 Γ L |2 |1 − Γ S Γ I N |2
Maximum power available from the source is given by

1 |VS |2 |1 − Γ S |2 1 − |Γ S |2
PAVS PIN |ΓIN ΓS∗ (7.178a)
8 Zo 1 − |Γ S |2 2
or
1 |VS |2 |1 − Γ S |2
PAVS (7.178b)
8 Z o 1 − |Γ S |2
or
1 |VS |2
PAVS (7.179)
8 Re(Z S )
Now, maximum power available from the network PAVN can be determined by
putting the conditions,
∗ ∗
Z L Z OUT , i.e. Γ L ΓOUT (7.180a)
Thus,
1 |VS |2 |1 − Γ S |2
PAVN |S21 |2 (7.180b)
8 Zo 1 − |ΓOUT |2 |1 − S11 Γ S |2
Transducer gain G T is defined as the ratio of power delivered to the load to power
available from the source. This can be represented by
PL 1 − |Γ S |2 2 1 − |Γ L |
2
GT |S21 | (7.181a)
PAV S |1 − Γ S ΓIN |2 |1 − S22 Γ L |2
Thus, the transducer gain G T depends upon both the source impedance Γ S and
the load impedance Γ L .
In case of load and source impedance of Z o as shown in Fig. 7.51, Γ S Γ L 0.
Putting these values in (7.181a), the transducer gain can be represented by
Fig. 7.51 The 2-port network with source and load impedance Z o
PL
GT |S21 |2 (7.181b)
PAVS
Under this condition, the transducer gain in dB is given by

PL
G T (in dB) 10 log 10 log |S21 |2 20 log(|S21 |) (7.181c)
PAV S
Operating power gain G P of a network is defined as the ratio of power delivered

to the load PL to power input PIN to the network. Thus, for a 2-port network, from
(7.177h) and (7.177f), G P can be written as
PL 1 1 − |Γ L |2
GP |S21 |2 (7.182)
PIN 1 − |ΓIN | 2
|1 − S22 Γ L |2
The operating power gain depends on the load impedance Γ L ; however, it is

independent of the source impedance Γ S and output impedance ΓOUT .
Available power gain G A of a network is defined as the ratio of power available
from the network PAVN to power available from the source PAVS . Thus, for a 2-port
network, from (7.180b) and (7.178b), G A can be written as
PAVN 1 − |Γ S |2 1
GA |S21 |2 (7.183)
PAVS |1 − S11 Γ S |2 1 − |ΓOUT |2
The available power gain depends upon the input matching condition, Γ S , because
actual power transferred to the network (device) is not necessarily the same as avail-
able power from the source which happens only for conjugate match at input. How-
ever, available power gain G A is independent of the input impedance ΓIN and load
impedance Γ L .
Example 7.18 A 2-port network has the following scattering matrix:

0 0◦ 1/3 0◦
[S]
1/3 0◦ 0 0◦
Compute the input return loss when the output of the network is terminated by a
(a) matched load, (b) electrical short and (c) electrical open.
Solution
Input reflection coefficient of a 2-port network is given by (7.177a)
Γ L S12 S21
ΓIN S11 +
1 − Γ L S22
(a) In case of matched load, the load reflection coefficient Γ L 0. Thus, input
reflection coefficient ΓIN 0. Thus, input return loss is infinite in dB.
(b) In case of shorted load, Z L 0, the load reflection coefficient,
Z L − Zo 0 − Zo
ΓL −1
Z L + Zo 0 + Zo
Thus, input reflection coefficient,

−1 × 1/3 0◦ × 1/3 0◦
ΓIN 0 0◦ + −1/9
1+0
Thus, input return loss (RL) is given by
RL −20 log(|ΓIN |) 20 log(1/9) 19.08 dB
(b) In case of open load, Z L ∝. The load reflection coefficient,
Z L − Zo
ΓL 1
Z L + Zo
Thus, input reflection coefficient,

1 × 1/3 0◦ × 1/3 0◦
ΓIN 0 0◦ + 1/9
1−0
Thus, input return loss is infinite in dB,
RL −20 log(|ΓIN |) 20 log(1/9) 19.08 dB
Example 7.19 Compute S-parameter matrix of a 2-port network which consists of a

50 series resistor in a 50 system. Also, derive the port return losses and insertion
loss of the network.

series network
Solution
The network is shown in Fig. 7.52. S11 is the input reflection coefficient when output
of the network is terminated by the reference impedance 50 and is given by
Z IN − Z o
S11
Z IN + Z o
Here, Z IN 50 + Z o 100 . Thus,

100 − 50 1
S11
100 + 50 3
The network is symmetric. Thus, S11 S22 1/3.
To compute S21 , input port of the network is driven by a voltage source vg with
source impedance Z o 50 and the output is terminated by load impedance
Z o 50 .
The load voltage vo is given by
vg × Z o vg
vo
Z o + 50 + Z o 3
Thus,
vo 2
S21
vg /2 3

Therefore, the S-parameter matrix of the network is given by

S11 S12 1/3 0◦ 2/3 0◦

S21 S22 2/3 0◦ 1/3 0◦
The same result can also be obtained using (7.162c), putting Z a 50 and
Z o 50 as:

series network
⎡ ⎤
Za 2Z o
S11 S12 1/3 0◦ 2/3 0◦
⎣ ⎦
2Z o +Z a 2Z o +Z a
S21 S22 2Z o Za 2/3 0◦ 1/3 0◦
2Z o +Z a 2Z o +Z a
The input and output return losses of the network (50 series resistor) in a 50
system are
RL −20 log|S11 | −20 log|S22 | −20 log|1/3| 9.54 dB
The insertion loss of the network is
IL −20 log|S21 | −20 log|2/3| 3.52 dB
Example 7.20 Compute S-parameter matrix of a 2-port network which consists of a

50 shunt resistor in a 50 system. Also, derive the port return losses and insertion
loss of the network.
Solution
The network is shown in Fig. 7.53. S11 is the input reflection coefficient when output
of the network is terminated by the reference impedance 50 and is given by
Z IN − Z o
S11
Z IN + Z o
Here, Z IN 50||Z o 25 . Thus,

25 − 50 1
S11 −
25 + 50 3
The network is symmetric. Thus, S11 S22 −1/3.
To compute S21 , input port of the network is driven by a voltage source vg with
source impedance Z o 50 and the output is terminated by load impedance
Z o 50 .
The load voltage vo is given by
vg × 50||Z o vg
vo
Z o + 50||Z o 3
Thus,
vo 2
S21
vg /2 3

Therefore, the S-parameter matrix of the network is given by

S11 S12 1/3 180◦ 2/3 0◦

S21 S22 2/3 0◦ 1/3 180◦
The same result can also be obtained using (7.163d), putting Z a 50 and
Z o 50 as
⎡ ⎤
−Z o 2Z a
S11 S12 1/3 180◦ 2/3 0◦
⎣ ⎦
Z o +2Z a Z o +2Z a
S21 S22 2Z a −Z o 2/3 0◦ 1/3 180◦
Z o +2Z a Z o +2Z a
The input and output return losses of the network (50 series resistor) in a 50
system are
RL −20 log|S11 | −20 log|S22 | −20 log|1/3| 9.54 dB
The insertion loss of the network is
IL −20 log|S21 | −20 log|2/3| 3.52 dB
Example 7.21 The S-parameter matrix of a 3-port network is as follows:

⎡ ⎤
0 0 1 0◦
[S] ⎣ 1 0◦ 0 0 ⎦ (7.184)
◦
0 1 0 0
The 3-port network is converted to a 2-port network by terminating its port-3

with a matched load as shown in Fig. 7.55. Compute the S-parameter matrix of the
converted 2-port network. Identify both the networks.
Solution
The S-parameters of the 3-port network are related to the different signal levels as
b1 0 × a1 + 0 × a2 + 1 × a3 (7.185a)
b2 1 × a1 + 0 × a2 + 0 × a3 (7.185b)
Fig. 7.54 3-port network with third port matched termination

with third port shorted
termination
b3 0 × a1 + 1 × a2 + 0 × a3 (7.185c)
The third port is matched terminated, and thus,
a3 0 (7.186)
Thus, the 3-port network becomes a 2-port network. Combining (7.185) and
(7.186), we can write,
b1 0 × a1 + 0 × a2 (7.187a)
b2 1 × a1 + 0 × a2 (7.187b)
Thus, the S-parameter matrix of the converted 2-port network,

0 0
[S] (7.188)
1 0◦ 0
The 3-port S-parameter matrix given in (7.184) is for a non-reciprocal device

called circulator as shown in Fig. 7.54. S-parameter matrix of the network with
third port matched termination is given by (7.188). This is the S-parameter matrix
of isolators. Thus, a circulator with its one port matched termination works as an
isolator (Fig. 7.54).
Example 7.22 The S-parameter matrix of a 3-port network is as follows:

⎡ ⎤
0 0 1 0◦
[S] ⎣ 1 0◦ 0 0 ⎦ (7.189)
◦
0 1 0 0
The 3-port network is converted to a 2-port network by terminating its port-3

with a shorted load as shown in Fig. 7.55. Compute the S-parameter matrix of the
converted 2-port network.
Solution
The S-parameters of the 3-port network are related to the different signal levels as
b1 0 × a1 + 0 × a2 + 1 × a3 (7.190a)
b2 1 × a1 + 0 × a2 + 0 × a3 (7.190b)
b3 0 × a1 + 1 × a2 + 0 × a3 (7.190c)
The third port is terminated by an electrical short, and thus,
a3 −b3 (7.191a)
Using (7.190c) and (7.191a),
a3 −a2 (7.191b)
Putting a3 −a2 into (7.190a) and (7.190b), we can write
b1 0 × a1 + 0 × a2 + 1 × (−a2 ) (7.192a)
b2 1 × a1 + 0 × a2 + 0 × (−a2 ) (7.192b)
or
b1 0 × a1 + (−1)1 × a2 (7.192c)
b2 1 × a1 + 0 × a2 (7.192d)
Thus, the S-parameter matrix of the converted 2-port network is

0 1 180◦
[S] (7.193)
1 0◦ 0
Example 7.23 A 50 transmission line is terminated by a load impedance of (100 +

j100) . Calculate the magnitude of load reflection coefficient.
Solution
Magnitude of reflection coefficient of load impedance Z L is given by
Fig. 7.56 Circuit diagram of

Example 7.24

Z L − Zo

|S11 |
Z L + Zo
Putting Z L (100 + j100) and Z o 50 ,

100 + j100 − 50

|S11 |
100 + j100 + 50

1 + j2

3 + j2

5

13
Example 7.24 A unilateral transistor has output impedance Z out (10 − j10) .
Calculate the value of the series and shunt components of the matching network as
shown in Fig. 7.56 for complex conjugate match at the output of the device to 50
load.
Solution
The impedance of the network is
Z nework j X 2 + j X 1 ||50
For complex conjugate matching at the output of the device,

∗
Z nework Z out 10 + j10
Therefore,
j X 2 + j X 1 ||50 10 + j10
or
−X 1 X 2 + j50(X 1 + X 2 ) j10(X 1 + 50) + (500 − 10X 1 )
On equating real and imaginary parts equal for both the sides separately
X 1 +25
X 2 −10 for X 1 +25

X 2 +30 for X 1 −25
Thus, the two sets of series and shunt components are
j X 1 + j25, j X 2 − j10
and
j X 1 − j25, j X 2 + j30
Example 7.25 Input resistance of a short-circuited lossless transmission line is 25 .

The input impedance becomes 100 when its other end is kept open. What will be
the characteristic impedance of the transmission line?
Solution
Input impedance of a lossless transmission line of characteristic impedance Z o when
terminated with load impedance Z L is given by (7.26)
Z L + j Z o tan(βl)
Z in Z o
Z o + j Z L tan(βl)
Therefore, input impedance of a shorted (Z L ∞) (Z L = ) line is given by
Z inShort j Z o tan(βl) (7.194)
And the input impedance of an open-circuited (Z L ) line is given by
Z inOpen − j Z o cot(βl) (7.195)
Multiplying (7.194) and (7.195), it can be written as

ZO Z inShort × Z inOpen (7.196)
Putting Z inShort 25 and Z inOpen 100 in (7.196), the characteristic resistance

of the line is
√
Z O 25 × 100
50
Example 7.26 The S-parameters of a transistor at 14 GHz are as follows:
S11 0.498 121◦ , S12 0.1137 − 49◦

S21 3.080 − 30◦ , S22 0.225 − 150◦
Fig. 7.57 Microwave amplifier with its associated matching networks
Calculate (a) transducer power gain, (b) available power gain and (c) operating
power gain for the amplifier with input and output matching networks of reflection
coefficients (Fig. 7.57)
Γ S 0.455 − 160◦ , Γ L 0.58 175◦
Solution
With reference to the Fig. 7.57, The input reflection coefficient (7.177a)
Γ L S12 S21
ΓIN S11 +
1 − Γ L S22
0.58 175◦ × 0.1137 − 49◦ × 3.080 − 30◦
0.498 121◦ +
1 − 0.58 175◦ × 0.225 − 150◦
−0.0212 + j0.2020
−0.256 + j0.427 +
0.8817 − j0.0552
(−0.256 + j0.427) + (−0.0383 + j0.2267)
−0.2947 + j6536 0.717 − 65.726◦
The output reflection coefficient (7.177b)

Γ S S12 S21
ΓOUT S22 +
1 − Γ S S11
0.455 − 160◦ × 0.1137 − 49◦ × 3.080 − 30◦
0.225 − 150◦ +
1 − 0.455 − 160◦ × 0.498 121◦
−0.082 + j0.137
−0.195 − j0.113 +
0.824 + j0.143
(−0.195 − j0.113) + (−0.069 + j0.178)
−0.264 + j0.065 0.272 − 13.885◦
available from the source and is given by (7.181a)
PL 1 − |Γ S |2 1 − |Γ L |2
GT |S21 |2
PAVS |1 − Γ S ΓIN |2
|1 − S22 Γ L |2
1 − |0.455 − 160◦ |2

|1 − 0.455 − 160◦ × 0.717 − 65.726◦ |2
× |3.080 − 30◦ |2
1 − |0.58 175◦ |2
×
|1 − 0.225 − 150◦ × 0.58 175◦ |2
0.798 0.664
× 9.486 ×
0.651 0.780
1.218 × 9.486 × 0.850
9.83 9.92 dB

to the load PL to power input PIN to the network and is given by (7.182)
2 1 − |Γ L |
2
PL 1
GP |S21 |
PIN 1 − |ΓIN |2 |1 − S22 Γ L |2
1

1 − |0.717 − 65.726◦ |2

× |3.080 − 30◦ |2
1 − |0.58 175◦ |2
×
|1 − 0.225 − 150◦ × 0.58 175◦ |2
1 0.664
× 9.486 ×
0.486 0.780
2.058 × 9.486 × 0.850
16.60 12.2 dB

from the network PAVN to power available from the source PAVS and is given by
(7.183)
PAVN 1 − |Γ S |2 1
GA |S21 |2
PAVS |1 − S11 Γ S |2 1 − |ΓOUT |2
1 − |0.455 − 160◦ |2

|1 − 0.498 121◦ × 0.455 − 160◦ |2
× |3.080 − 30◦ |2
1
×
1 − |0.272 − 13.885◦ |2
0.793 1
× 9.486 ×
0.699 0.926
1.134 × 9.486 × 1.080
11.62 10.65 dB
This example shows different gains of the amplifier for the same matching condi-
tion, i.e. for Γ S 0.455 − 160◦ and Γ L 0.58 175◦ . Practically, gain of an ampli-
fier is measured by the ratio of power delivered to the load to the power available
from the source. In this case, the measured power gain is nothing but the transducer
power gain, G T which is 9.92 dB in this example.
Example 7.27 The S-parameters of a transistor at 14 GHz are as follows:
S11 0.653 − 160◦ , S12 0.000

S21 2.565 30◦ , S22 0.452 − 90◦
Calculate (a) transducer power gain, (b) available power gain and (c) operating
power gain for the amplifier with input and output matching networks of reflection
coefficients
Γ S 0.653 160◦ , Γ L 0.452 90◦
Solution
The input reflection coefficient (7.177a)
Γ L S12 S21
ΓIN S11 +
1 − Γ L S22
Here, S12 0.000, therefore:
ΓIN S11 0.653 − 160◦
The output reflection coefficient (7.177b)

Γ S S12 S21
ΓOUT S22 +
1 − Γ S S11
Here, S12 0.000, therefore:
ΓOUT S22 0.452 − 90◦
available from the source and is given by (7.181a)
PL 1 − |Γ S |2 2 1 − |Γ L |
2
GT |S21 |
PAVS |1 − Γ S ΓIN |2 |1 − S22 Γ L |2
◦
1 − |0.653 160 | 2

|1 − 0.653 160◦ × 0.653 − 160◦ |2
× |2.565 30◦ |2
1 − |0.452 90◦ |2
×
|1 − 0.452 − 90◦ × 0.452 90◦ |2
0.574 0.796
× 6.579 ×
0.329 0.633
1.743 × 6.579 × 1.257
14.42 11.59 dB

to the load PL to power input PIN to the network and is given by (7.182)
2 1 − |Γ L |
2
PL 1
GP |S21 |
PIN 1 − |ΓIN |2 |1 − S22 Γ L |2
1

1 − |0.653 − 160◦ |2
× |2.565 30◦ |2
1 − |0.452 90◦ |2
×
|1 − 0.452 − 90◦ × 0.452 90◦ |2
1 0.796
× 6.579 ×
0.574 0.633
1.743 × 6.579 × 1.257
14.42 11.59 dB G T ! !

from the network PAVN to power available from the source PAVS and is given by
(7.183)
PAVN 1 − |Γ S |2 1
GA |S |2
2 21

PAVS |1 − S11 Γ S | 1 − |Γ OU T |2
1 − |0.653 160◦ |2

|1 − 0.653 160◦ × 0.653 − 160◦ |2
× |2.565 30◦ |2
1
×
1 − |0.452 − 90◦ |2
0.574 1
×6.579 ×
0.329 0.796
1.743 × 6.579 × 1.257
14.42 11.59 dB G T ! ! G P ! ! !
This example shows that the same gain value of 11.59 dB for transducer, oper-
ating and available power gain. In this case, it can be noticed that the device
is unilateral, i.e. S12 0.000 and the source and load reflection coefficients
∗ ∗
Γ S 0.653 160◦ S11 and Γ L 0.452 90◦ S22 , respectively, make the
amplifier perfectly (complex conjugate) matched at its both ports. Actually, any
amplifier with input and output ports perfectly matched provides same gain values
corresponding to transducer, operating and available power gain.
References
1. Gupta KC (1983) Microwaves. Wiley Eastern Limited

2. Marks RB, Williams DF (1992) A general waveguide circuit theory. J Res Nat Inst Stand
Technol 97:533–562
3. Uchimura H, Takenoshita T, Fujii M (1998) Development of a laminated waveguide. IEEE
Trans Micro Theory Tech 46(12):2438–2443
4. Deslandes D, Wu K (2003) Single-substrate integration technique of planar circuits and waveg-
uide filters. IEEE Trans Micro Theory Techs 53(9):2968–2977
5. Wu K, Zhang YL, Hong W (2005) Novel substrate integrated waveguide cavity filter with
defected ground structure. IEEE Trans Micro Theory Tech 53(4)
6. Bozzi M, Georgiadis A, Wu K (2011) Review of substrate-integrated waveguide circuits and
antennas. IET Microw Antennas Propag 5:909–920
7. Belenguer A, Esteban H, Boria V (2014) Novel empty substrate integrated waveguide for high-
performance microwave integrated circuits. IEEE Trans Microw Theory Tech 62:832–839
8. Kurokawa K (1965) Power waves and the scattering matrix. IEEE Trans Microw Theory Tech
13(3):194–202
9. Tippet JC, Speciale RA (1982) A rigorous technique for measuring the scattering matrix of a
multiport device with a 2-port network analyzer. IEEE Trans Microw Theory Tech 30:661–666
10. S parameters… circuit analysis and design. Hewlett Packard Application Note, 95A, Feb 1973
11. Frickey DA (1994) Conversions between S, Z, Y, H, ABCD, and T parameters which are valid
for complex source and load impedances. IEEE Trans Microwave Theory Tech 42:205–211
12. Marks RB, Williams DF (1995) Comments on conversions between S, Z, Y, H, A BCD, and T
parameters which are valid for complex source and load impedances. IEEE Trans Microwave
Theory Tech 43:914–915
13. Stanculescu M, Iordache L, Iordache M, Niculae D, Bucata V (2017) Using S parameters in
wireless power transfer analysis. In: The 10th international symposium on advanced topic in
electrical engineering, pp 107–112, March 23–25, 2017
14. Sah AK, Pant DP (2014) Analysis and optimization of wireless power transfer link. In: Proceed-
ings of international conference on technology and innovation management and IOE graduate
conference, pp 82–93, Oct 10–11, 2014
Chapter 8
Microwave Switches
Abstract This chapter presents design and analysis of various types of microwave
switches using p-i-n diodes and transistors. It describes series-, shunt-, TEE- and
π -type switches with their merits and demerits. Various applications of active
switches in microwave communications are also presented.
8.1 Introduction
Microwave switches are used in most modern systems including satellite communi-
cations, cell phone communications, software-defined radios, radars, and many other
applications. The key features of RF switches are the combination of low insertion
loss at ON state, high isolation at OFF state, high power handling capability, high
linearity, low DC power consumption and compact in size enabling easy integration
with other circuitry. The functions of RF/microwave switches are to reconfigure sys-
tems by diverting signals from one direction to another. The switches are also used to
realize step wise variable attenuators, phase shifters and other elements. From real-
ization point of view, there are two types of microwave switches, electromechanical
and solid-state switches. Generally, solid-state switches are very compact in size,
faster in operation and more reliable than electromechanical. However, solid-state
switches have higher insertion loss and also exhibit relatively more nonlinearity at
higher level of signal power. The p-i-n diodes, various FETs and HEMTs are used
as solid-state switching device. Among them, p-i-n diodes are widely used for high-
frequency applications as solid-state switching device for its high RF power handling
capability and comparatively lower insertion loss. MESFETs, and various HEMTs
are used for low-power microwave/RF switch for easy realization in MMIC tech-
nology integrated with other circuitry. Extremely high two-dimensional electron gas
(2-DEG) of GaN HEMTs results in very low ON resistance, and due to its high break-
down fields it is capable to operate at higher operating temperatures which makes it
suitable for realization of high power switches.

200 8 Microwave Switches
8.2 Switch Circuits Based on P-I-N Diode
Low ON resistance at forward bias and high OFF resistance at reverse bias of a p-i-n
diode enables it to be used as a switching device [1–5]. Si- and GaAs-based p-i-n
diodes are mostly used for switching applications. However, SiC based p-i-n diode-
based switches are also developed for high power applications for their higher power
handling capability [4, 5]. The function of a switch is either to allow a signal to a
specific signal path or to block the signal from following this path. In case of ideal
switch, circuit paths are either closed by providing zero impedance or opened by an
ideal open circuit with infinite impedance. RF/microwave switches based on p-i-n
diodes utilize DC voltage/current dependent variable impedance of the p-i-n diode to
direct the signal flow. Practically, RF/microwave p-i-n diodes cannot provide ideally
short or open impedance. Thus, at ON state, always there is some amount of incident
signal absorbed by the switch and some amount also will be reflected by the switch.
This reduction of signal power leads to non-zero value of insertion loss (IL) which
is the ratio of the output signal power to the input signal power. Similarly, at OFF
state of a switch, there is always some small amount of signal passing through the
switch. This leads to finite value of isolation. At OFF state, isolation of a switch is
also defined as the ratio of output power to the input power.
Thus, in case of source and load impedance of Zo , the insertion loss and isolation
of a switch in dB is given by
20 log10 (|S21 |) (8.1)
Here, S 21 is the forward transmission coefficient of the switch at ON and OFF

state for insertion loss and isolation, respectively. Sometimes, isolation of a switch
is defined as the ratio of insertion loss at OFF to ON state, i.e. loss at OFF state with
respect to loss at ON state.
RF/microwave switches are implemented in two ways: reflective-type and
absorptive-type as shown in Figs. 8.1 and 8.2 respectively. In ON state, both types
of switches connect the signal path to the load through a very low-impedance path.
However, in OFF state they work differently. In case of reflective switch, at OFF state,
two extreme impedances, i.e. either short or open impedances are created across the
signal path to reflect back most of the signals, ideally, full signal to its source. Due to
the maximum mismatch in both the extreme impedance cases, signals are reflected
back to provide desired isolation. In case of absorptive-type switch at OFF state,
switch connects the signal to a matched dummy load of impedance Zo . Therefore,
at OFF state, the signals will be absorbed by the dummy load and there will be no
reflected power to the source.
As discussed in Chap. 2 that the p-i-n diode can be used as voltage-/current-
controlled RF/microwave impedance. Figure 8.3 shows the schematic circuit
diagram of switches using p-i-n diode in series or shunt configuration as variable
impedance. Under positive control voltage V C , the p-i-n diode operates in forward-
bias condition and acts as low impedance, whereas under negative voltage V C , the
8.2 Switch Circuits Based on P-I-N Diode 201
(a) (b)
ZS ZS
RF short
RF open ZL ZL
switch switch
Fig. 8.1 Reflective switch is in OFF state a open in the RF path, b short across the RF path
ZS
ZL
ZO
switch
Fig. 8.2 Absorptive RF/microwave switch
(a) VC (b) VC
Cbypass
RF
Cb choke Cb
Cb Cb
ZS RF ZS
choke ZL ZL
switch
Fig. 8.3 Reflective switches configuration a series diode b shunt diode
p-i-n diode operates in reverse-bias condition and acts as high impedance. Thus, in
case of low-impedance state, p-i-n diode operates in forward-bias condition and in
high-impedance state the p-i-n diode operates in reverse-bias condition. The bias
voltage/current is provided to the diode through RF choke and bypass capacitor,
the details of which will be discussed in the driver circuit section. DC blocking
capacitors (Cb ) are used to block DC voltage towards generator and load without
affecting RF performance.
Under forward-bias condition, the equivalent RF circuit of a p-i-n diode is shown in
Fig. 8.4. At low-impedance state, the intrinsic layer resistance Ri is very low compared
to the capacitive reactance (i.e. Ri 1/2π fCi ). The RF/microwave frequency range
Lp
Lp
Cp
Rs Lp
Rs
RT
Ri Ci Ri
Fig. 8.4 Equivalent circuit of p-i-n diode at low-impedance state (forward-bias condition)
Lp
Cp Cj CP CT
Rs
Cj
Fig. 8.5 Equivalent circuit of p-i-n diode at high-impedance state (reverse-bias condition)
where the effect of package capacitance is negligible compared to the highly forward-
biased resistance Ri + Rs , the equivalent low impedance (ZTL ) of the diode can be
considered as series combination of (Ri + Rs ) RT and Lp as shown in Fig. 8.4.
Under reverse-bias condition, the equivalent RF circuit of a p-i-n diode is shown
in Fig. 8.5. At microwave frequency range where the capacitive reactance, i.e.
1/2π fCj Rs and also 1/2π fCj 2π fLP , the equivalent RF circuit of the p-i-
n diode can be considered as a capacitance CT Cj + Cp and the high impedance of
the diode at OFF condition can be written as ZTH 1/j2π fCT .
8.3 Series Switch Configuration
The series-configured switch circuit with equivalent circuit element of p-i-n diodes
can be represented as shown in Fig. 8.6a, and b at ON and OFF state, respectively.
8.3 Series Switch Configuration 203
(a) (b)
RT LP CT
ZS ZS
ZL ZL
Fig. 8.6 Series switch configuration a ON state b OFF state
Fig. 8.7 Series switch configuration
Fig. 8.8 Series switch configuration for S-parameter determination
Therefore, insertion loss, isolation and return loss of the switching circuit when
it is connected in between a source of impedance ZS and a load of impedance ZL as
shown in Fig. 8.7 can be derived as follows:
Here,
ZT ZTL RT + j2π fLP at low-impedance state (8.2a)

1
ZTH at high-impedance state (8.2b)
j2π fCT
For the network of series impedance Z T , the [S] parameter of the series switch
with respect to Zo as shown in Fig. 8.8 is given by (using (7.162c))
⎡ ⎤
ZT 2Zo
S11 S12
⎣ ⎦
ZT +2Zo ZT +2Zo
(8.3)
S21 S22 2Zo ZT
ZT +2Zo ZT +2Zo
The source and load reflection coefficients with respect to the reference impedance
Zo are given by
ZS − Zo ZL − Zo
ΓS , ΓL (8.4a)
ZS + Zo ZL + Zo
The input and output reflection coefficients of the switch can be written
from (7.1774a) and (7.1774b) for the load and source impedances ZL and ZS , respec-
tively, as
ΓL S12 S21 ZT + ZL − Zo
ΓIN S11 + (8.4b)
1 − ΓL S22 ZT + ZL + Zo
ΓS S12 S21 ZT + ZS − Zo
ΓOUT S22 + (8.4c)
1 − ΓS S11 ZT + ZS + Zo
For the series-configured switch, the ratio of power available from the source
(Pavs ) to the power delivered to the load (PL ) can be written as (using (7.181a))

Pavs |1 − ΓS ΓIN |2 1 |1 − S22 ΓL |2
(8.5)
PL 1 − |ΓS |2 |S21 |2 1 − |ΓL |2
Thus, putting the values of reflection coefficients to this equation, insertion loss
and isolation of the series-connected switch can be determined. The first and last
terms are losses due to the input and output mismatches, respectively. The middle
term is the loss of the switch when both of its ports are terminated with reference
impedance Zo and is given from (8.3) as

1
2Zo + ZT
2

(8.6)
|S21 |2 2Zo
Suppose, the load and source impedances are equal and resistive, i.e. Ro then at
ON state, putting ZT RT + j2π fLP RT + jXL , the insertion loss of the series
switch will be

1
2Ro + RT + jXL
2
IL

(8.7a)
|S21 |2 2Ro
or
2 2
RT XL
IL(in dB) 10 log10 1+ + (8.7b)
2Ro 2Ro
8.3 Series Switch Configuration 205
Similarly, at OFF state, putting ZT 1/(j2π fCT ) −jXC , the loss of the series
switch, i.e. isolation is

1
2Ro − jXC
2
Isolation

(8.8a)
|S21 |2 2Ro
or
2
1
Isolation (in dB) 10 log10 1 + (8.8b)
4π fCT Ro
Example 8.1 Determine the insertion loss and isolation of a single p-i-n diode-based
series switch operating at 4 GHz with the following parameters: Rs 2 , Ri 2 ,
Lp 0.1 nH, Cj 0.02 pF, Cp 0.01 pF.
Solution
Here, RT (Ri + Rs ) 4 , CT Cj + Cp 0.03 pF and Lp 0.1 nH.
To calculate insertion loss using (8.7a)
RT 4 ,
XL 2π fLp 2π × 4 × 0.1 2.513
From (8.7a),

1
2Ro + RT + jXL
2
2 × 50 + 4 + j2.513
2
IL

1.0822
|S21 |2 2Ro

2 × 50
or
IL (in dB) 10 log10 [1.082] 0.343 dB
To calculate isolation using (8.8a),
XC 1/2π fCT 1000/2π × 4 × 0.03 1326.29
From (8.8a),

1
2Ro − jXC
2
j1326.29
2
Isolation
|S21 |2 2Ro

1 − 2 × 50
176.905
or
Isolation (in dB) 10 log10 [176.905] 22.48 dB
Thus, for a 50 system, the ON-state insertion loss is 0.343 dB and OFF-state
isolation is 22.48 dB.
8.4 Shunt Switch Configuration
The shunt-configured switch with p-i-n diode as control impedance can be repre-
sented by the circuit diagrams as shown in Fig. 8.9a, b at its ON and OFF states,
respectively. Suppose the shunt impedance of the diode is Z T as shown in Fig. 8.10.
In this case, the shunt impedance of the switch can be written as
1
ZT ZTH −jXC at ON state (8.9a)
j2π fCT
ZTL RT + j2π fLP RT + jXL at OFF state (8.9b)
From (7.163d), using Fig. 8.11, the [S] parameter matrix of the shunt switch with
respect to Zo can be written as
⎡ ⎤
ZT ||Zo −Zo 2ZT
S11 S12 ⎢ ZT ||Zo +Zo 2ZT +Zo ⎥
⎣ ⎦ (8.10)
S21 S22 2ZT ZT ||Zo −Zo
2ZT +Zo ZT ||Zo +Zo
The source and load reflection coefficients with respect to the reference impedance
Zo are given by
ZS − Zo ZL − Zo
ΓS , ΓL (8.11)
ZS + Zo ZL + Zo
(a) (b)
RT
ZS ZS
CT ZL ZL
LP
Fig. 8.9 Shunt switches configuration a ON state b OFF state
Fig. 8.10 Shunt switch configuration for reflection coefficient determination

8.4 Shunt Switch Configuration 207
Fig. 8.11 Shunt switch configuration for S-parameter determination
The input and output reflection coefficients of the switch for the load and source
impedances ZL and ZS , respectively, are
ΓL S12 S21 ZT ||ZL − Zo
ΓIN S11 + (8.12a)
1 − ΓL S22 ZT ||ZL + Zo
ΓS S12 S21 ZT ||ZS − Zo
ΓOUT S22 + (8.12b)
1 − ΓS S11 ZT ||ZS + Zo
For the shunt-configured switch, the ratio of power available from the source
(Pavs ) to the power delivered to the load (PL ) can be written as (using (7.181a))

Pavs |1 − ΓS ΓIN |2 1 |1 − S22 ΓL |2
(8.13)
PL 1 − |ΓS |2 |S21 |2 1 − |ΓL |2
Thus, putting the values of reflection coefficients from (8.12a), insertion loss and
isolation of the shunt-connected switch can be determined. The first and last terms are
losses due to the input and output impedance mismatches, respectively. The middle
term is the loss of the switch when both the ports are terminated with the reference
impedance Zo and is given by

2
1
2ZT + Zo
2

1 + Zo
(8.14a)
|S21 |2
2ZT

2ZT
Suppose, the source and load impedances are resistive, i.e. Zo Ro then putting
ON-state ZT value, the insertion loss of the shunt switch can be written as

1
2ZT + Zo
2
Zo
Zo
2
IL

1 +
1 − (8.14b)
|S21 |2 2ZT
2ZT
j2XC
1
1 + (π fCT Ro )2 (8.14c)
|S21 |2
or

IL (in dB) 10 log10 1 + (π fCT Ro )2 (8.15)
Similarly, putting OFF-state ZT value, loss of the shunt switch, i.e. isolation will be

2
1
Ro
Isolation
1 +
(8.16a)
|S21 |2
2(RT + jXL )
(2RT + Ro )2 + (2XL )2
(8.16b)
(2RT )2 + (2XL )2
or

(2RT + Ro )2 + (2XL )2
Isolation (in dB) 10 log10 (8.16c)
(2RT )2 + (2XL )2
Example 8.2 Determine the insertion loss and isolation of a single shunt p-i-n diode-
based switch operating at 4 GHz for the following parameters: Rs 2 , Ri 2 ,
Lp 0.1 nH, Cj 0.02 pF, Cp 0.01 pF.
Solution
To calculate insertion loss of the shunt diode switch using (8.14b),
1 1000
XC 1326.29
2π fCT 2 × π × 4 × 0.03
From (8.14b),

2
1
Zo
50
IL

1 −
1 −
1.0000355
|S21 |2 j2XC
j2 × 1326.29
or
I L (in dB) 10 log10 [1.000355] 0.002 dB
To calculate isolation using (8.16a),
XL 2π fLp 2π × 4 × 0.1 2.513
From (8.8a),

2

2
1
Ro

50
Isolation

1 +

1 +
37.97
|S21 |2 2(RT + jXL )
2(4 + j2.513)
or

8.4 Shunt Switch Configuration 209
Thus, for a 50 system, the ON-state insertion loss is 0.0002 dB and OFF-state
isolation is 15.79 dB.
Examples 8.1 and 8.2 show that switch with series p-i-n diode configuration
provides better isolation, whereas switch with shunt diode configuration provides
better insertion loss.
8.5 Compound Switch Configuration
Examples of series and shunt switches with same diode parameters show that shunt
switch provides better insertion loss compared to series switch configuration. How-
ever, series switch configuration provides better isolation compared to shunt switch
configuration. Shunt switch provides lower insertion loss due to the absence of any
switch elements in series with the signal path.
At RF and microwave frequencies, practically, it is very difficult to achieve very
high isolation using a single p-i-n diode either in shunt or series configuration. The
causes of this limitation are not only the parasitic reactance of the diode but also it
is due to the radiation effects in the transmission medium and inadequate shielding.
To overcome this limitation, multiple p-i-n diodes can be used in a single switch
to increase the isolation with the penalty of more insertion loss. All p-i-n diodes
may be connected only in series or in parallel. Also, there are switch designs called
compound switch configuration that uses combinations of series and shunt diodes to
achieve improved isolation performance.
Simple series–shunt and TEE configurations are two basic configurations for
compound switch. In the insertion loss (ON) state for a compound switch, the series
diode is operated at forward bias and the shunt diodes are operated at zero- or reverse-
bias conditions, whereas in isolation (OFF) state opposite bias is provided to series
and shunt diodes.
8.5.1 Series–Shunt Configuration
The schematic circuit diagram of a series–shunt configuration switch is shown in

Fig. 8.12. The equivalent RF circuit of the series–shunt-configured switch is shown
in Fig. 8.13. Suppose, the impedance of the series- and shunt-connected p-i-n diodes
are ZT 1 and ZT 2 , respectively, and the switch is connected in between a source and a
load of impedance Zo . With reference to the equivalent circuit as shown in Fig. 8.13,
[S] parameter matrix of the switch can be determined following the similar approach
as it is done in case of series and shunt switch configuration.
With reference to Fig. 8.13, the input and output reflection coefficient of the switch
can be written as
VC
Cb Cb
ZS
ZL
Fig. 8.12 Schematic circuit diagram of series–shunt compound switch configuration
ZT1
ZO
ZT2 Vout ZO
Fig. 8.13 Equivalent RF circuit of the series–shunt switch configuration
ZT 1 + ZT 2 ||Zo − Zo
S11 (8.17a)
ZT 1 + ZT 2 ||Zo + Zo
ZT 2 ||(ZT 1 + Zo ) − Zo
S22 (8.17b)
ZT 2 ||(ZT 1 + Zo ) + Zo
and
2Vout 2ZT 2 ||Zo
S21 (8.17c)
VS Zo + ZT 1 + ZT 2 ||Zo
Similarly, connecting the source VS with impedance at the port-2 of the switch
and a load at the port-1, the reverse transmission coefficient S12 can be written as
2Zo ZT 2
S12 (8.17d)
[Zo + ZT 2 ||(ZT 1 + Zo )][ZT 1 + ZT 2 + Zo ]
It can be shown that S12 S21 , because the network is reciprocal. From (8.17c) the
insertion loss and isolation of the series–shunt switch can be calculated as follows:
8.5 Compound Switch Configuration 211
1 Zo + ZT 1 + ZT 2 ||Zo
(8.18a)
S21 2ZT 2 ||Zo
or
1 Zo + ZT 1 ZT 1
1+ + (8.18b)
S21 2ZT 2 2Zo
Insertion loss can be calculated by putting the values of ZT 1 and ZT 2 corresponding

to the low impedance and high impedance of the p-i-n diodes, respectively, as
ZT 1 ZTL RT + j2π fLP RT + jXL p-i-n diode ON

1
ZT 2 ZTH −jXC p-i-n diode OFF
j2π fCT
Thus,

1 RT XL XL Zo + RT
1+ − +j + (8.19a)
S21 2Zo XC 2Zo 2XC
Therefore,

1 RT XL 2 XL Zo + RT 2
IL 1 + − + + (8.19b)
|S21 |2 2Zo XC 2Zo 2XC
2
RT XL XL Zo + RT 2
IL (in dB) 10log10 1 + − + + (8.19c)
2Zo XC 2Zo 2XC
Isolation of the series–shunt switch can be calculated by putting the values of ZT 1

and ZT 2 corresponding to the high and low impedance of the p-i-n diodes, respectively,
in (8.18b) as

1
j2π fCT
Thus,

1 Zo RT − XC XL XC Zo XL + RT XC
1+ 2 +j + 2 (8.20a)
S21 2 RT + ZL2 2Zo 2 RT + XL2
Therefore,
2 2
Isolation 1+ 2 + + 2 (8.20b)
|S21 | 2
2 RT + XL2 2Zo 2 RT + XL2
⎡ 2 2 ⎤
Zo RT − XC XL XC Zo XL + RT XC
Isolation(in dB) ⇒ 10 log10 ⎣ 1 + 2 + + 2 ⎦
2 RT + XL2 2Zo 2 RT + XL2
(8.20c)
Example 8.3 Determine the insertion loss and isolation of the switch as shown in
Fig. 8.12 using series and shunt p-i-n diodes operating at 4 GHz for the following
parameters: Rs 2 , Ri 2 , Lp 0.1 nH Cj 0.02 pF, Cp 0.01 pF.
Solution
Here, RT (Ri + Rs ) 4 , CT Cj + Cp 0.03pF andLp 0.1nH.
To calculate insertion loss using (8.19b),
RT 4 ,
XL 2π fLp 2π × 4 × 0.1 2.513
XC 1/2π fCT 1000/2π × 4 × 0.03 1326.29
From (8.19b),

1 RT XL 2 XL Zo + RT 2
IL 1 + − + +
|S21 |2 2Zo XC 2Zo 2XC
2 2
4 2.513 2.513 50 + 4
1+ − + + 1.0797
2 × 50 1326.29 2 × 50 2 × 1326.29
or
IL (in dB) 10 log10 [1.0797] 0.333 dB
To calculate isolation using (8.20b),

2 2
Isolation 1+ + + 2
|S21 |2 2 R2T + XL2 2Zo 2 RT + XL2
2 2
200 − 1326.29 × 2.513 1326.29 50 × 2.513 + 4 × 1326.29
1+ + +
2 42 + 2.5132 2 × 50 2 42 + 2.5132
4788.92 + 18208.820.0021 22997.75
or
For a 50 system, the ON-state insertion loss is 0.333 dB and OFF-state isolation
is 43.617 dB.
VC
Cb Cb
ZS
ZL
Fig. 8.14 TEE-type switch configuration
I1 I2
ZT1 ZT1
ZO
ZT2
Vout ZO
Fig. 8.15 TEE-type switch configuration for reflection coefficient derivation
8.5.2 TEE Configuration
The schematic circuit diagram of a TEE configuration switch is shown in Fig. 8.14.
The equivalent RF circuit of the TEE-configured switch is shown in Fig. 8.15. Sup-
pose, the impedances of the series- and shunt-connected p-i-n diodes are ZT 1 and
ZT 2 , respectively, and the switch is connected in between a source and a load of
impedance of Zo . With reference to the equivalent circuit as shown in Fig. 8.15, [S]
parameter matrix of the switch can be determined following the previous approach.
can be written as
ZT 1 + ZT 2 ||(ZT 1 + Zo ) − Zo
S11 S22 (8.21)
ZT 1 + ZT 2 ||(ZT 1 + Zo ) + Zo
and
2Vout 2Zo I2
S21 S12
VS VS
2Zo ZT 2 I1

VS (ZT 1 + ZT 2 + Zo )
2Zo ZT 2
(8.22a)
(Zo + ZT 1 + ZT 2 )(Zo + ZT 1 + ZT 2 ||(ZT 1 + Zo ))
2Zo ZT 2
(8.22b)
2Zo ZT 1 + 2Zo ZT 2 + 2ZT 1 ZT 2 + ZT2 1 + Zo2
Thus, insertion loss and isolation can be determined as follows:
1 2Zo ZT 1 + 2Zo ZT 2 + 2ZT 1 ZT 2 + ZT2 1 + Zo2

(8.22c)
S21 2Zo ZT 2
or
1 ZT 1 ZT 1 ZT2 1 Zo
1+ + + + (8.22d)
S21 ZT 2 Zo 2Zo ZT 2 2ZT 2
Insertion loss of the TEE switch can be calculated by putting the values of ZT 1
and ZT 2 corresponding to the low impedance and high impedance of the p-i-n diodes,
respectively, as

1
j2π fCT
Thus,

1 RT XL (RT + Zo ) XL Zo RT + R2T − XL2 + Zo2
1+ − +j + (8.23a)
S21 Zo Zo XC Zo 2Zo XC
Therefore,
2
RT XL (RT + Zo ) 2 XL Zo RT + R2T − XL2 + Zo2
IL 1 + − + + (8.23b)
Zo Zo XC Zo 2Zo XC
or

RT XL (RT + Zo ) 2
IL(in dB) 10 log10 1 + −
Zo Zo XC
2
XL Zo RT + R2T − XL2 + Zo2
+ + (8.23c)
Zo 2Zo XC
Isolation of the TEE switch can be calculated by putting the values of ZT 1 and
ZT 2 corresponding to the high and low impedance of the p-i-n diodes, respectively,
in (8.22b) as
ZT 1 ZTH 1
j2πfCT
−jXC p-i-n diode OFF
Thus,

1 Zo2 RT − RT XC2 − 2Zo XL XC
1+
S21 2Zo R2T + XL2

XC 2Zo RT XC − XC2 XL − Zo2 XL
−j + (8.24a)
Zo 2Zo R2T + XL2
Therefore,
2
1 Zo2 RT − RT XC2 − 2Zo XL XC
Isolation 1+
|S21 |2 2Zo R2T + XL2
2
+ + (8.24b)
Zo 2Zo R2T + XL2
⎡ 2
Z 2
R − R X 2
− 2Z X X
Isolation (in dB) ⇒ 10log10 ⎣ 1 +
T T o L C
o
C

2Zo R2T + XL2
2 ⎤
XC 2Zo RT XC − XC XL − Zo XL ⎦
2 2
+ + (8.24c)
Zo 2Zo R2T + XL2
Similarly, connecting the source VS with impedance at the port-2 of the switch
and a load at the port-1, the reverse transmission coefficient S12 can be determined. In
this case, the reverse transmission coefficient S12 is the same as forward transmission
coefficient S21 as the network is reciprocal (does not consist of any active element
and non-reciprocal element).
Example 8.4 Determine the insertion loss and isolation of a TEE-type switch as
shown in Fig. 8.14 using series and shunt p-i-n diodes operating at 4 GHz for the
following parameters of the diodes (Figs. 8.4 and 8.5): Rs 2 , Ri 2 , Lp
0.1 nH, Cj 0.02 pF, Cp 0.01 pF.
Solution
Therefore,
RT 4 ,
XL 2π fLp 2π × 4 × 0.1 2.513
XC 1/2π fCT 1000/2π × 4 × 0.03 1326.29
From (8.23b),
2
RT XL (RT + Zo ) 2 XL Zo RT + R2T − XL2 + Zo2
IL 1 + − + +
Zo Zo XC Zo 2Zo XC
2 2
4 2.513(4 + 50) 2.513 50 × 42 − 2.5132 + 502
1+ − + +
50 50 × 1326.29 50 2 × 50 × 1326.29
[1.07795]2 + [0.07509]2 1.1676
or
I L (in dB) 10log10 [1.1676] 0.673 dB
To calculate isolation using (8.24b),

2
Zo2 RT − RT XC2 − 2Zo XL XC
Isolation 1 +
2Zo R2T + XL2
2
+ +
Zo 2Zo R2T + XL2
2
502 × 4 − 4 × 1326.292 − 2 × 50 × 2.513 × 1326.29
1+
2 × 50 42 + 2.5132
2
1326.29 2 × 50 × 4 × 1326.29 − 1326.292 × 2.513 − 502 × 2.513
+ +
50 2 × 50 42 + 2.5132
2957004.87 + 10868818.84 13825823.7
Thus,
Isolation (in dB) 10log10 [13825823.7] 71.4 dB
For a 50 system, the ON-state insertion loss is 0.673 dB and OFF-state isolation
is 71.4 dB.
8.6 Compound Switch Analysis Using ABCD Parameter 217
8.6 Compound Switch Analysis Using ABCD Parameter
It is more convenient to use ABCD parameters for the analysis of compound switch
circuits. Any compound switch can be considered as multiple simple networks con-
nected in cascade. Thus, it is convenient to determine the ABCD matrix of the
individual networks and then multiplying them to determine ABCD parameters of
the compound switch from which S-parameters of the complete switch can be deter-
mined.
can be written as follows:
The TEE switch is divided into three simple networks. Input voltage V1 and input
current I1 of the network-1, as shown in Fig. 8.16, is related by
V1 A1 V2 + B1 I2
I1 C1 V2 + D1 I2
The network-1 consists of only one element ZT 1 connected in the series line. Thus,
the parameters can be determined as

V1
V1
A1 1, B1 ZT 1
V2
I2 0 I2
V2 0

I1
I1
C1 0, D1
1
V2
I2 0 I2 V2 0
Input voltage V2 and input current I2 of the network-2, as shown in Fig. 8.16, is
related by
V2 A2 V3 + B2 I3
I1 I2 I3 I4
ZT1 ZT1
ZO
V1 V2 ZT2 V3 V4 ZO
1 2 3
Fig. 8.16 TEE-type switch configuration for reflection coefficient derivation

I2 C2 V3 + D2 I3
Whereas, the network-2 consists of only one element ZT 2 connected in shunt with
the line. Thus, the parameters can be determined as:

V2
V2
A2 1, B2 0
V3
I3 0 I3
V3 0

I2
I2
C2 1/ZT 2 , D2
1
V3
I3 0 I3 V3 0
Therefore, the ABCD matrix of the TEE configuration switch can be expressed
as

A B 1 ZT 1 1 0 1 ZT 1
(8.25a)
CD 0 1 1/ZT 2 1 0 1

A B 1 + ZT 1 /ZT 2 ZT 1 1 ZT 1
(8.25b)
CD 1/ZT 2 1 0 1
⎡ ⎤
ZT 1 ZT2 1
A B ⎢ 1 + ZT 2
2ZT1 + ZT 2 ⎥
⎣ ⎦ (8.25c)
CD 1 ZT 1
1 + ZT 2
ZT 2
The [S] parameter matrix of the switch can be determined using the relationship
between S-parameters and ABCD parameters as determined in Chap. 7. The forward
transmission coefficient S21 of the switch can be determined using the following
equation:
2
S21
A + BYo + CZo + D
Putting the values of ABCD parameters from (6.25c):

2
S21 ZT2 1
ZT 1 2ZT 1 Zo ZT 1
1+ ZT 2
+ Zo
+ Zo ZT 2
+ ZT 2
+1+ ZT 2
2Zo ZT 2
(8.26)
2Zo ZT 1 + 2Zo ZT 2 + 2ZT 1 ZT 2 + ZT2 1 + Zo2
This is the same result as derived in (8.22c). Other elements of the S-parameters
matrix can be derived in the similar way.
8.7 Switch Circuits Based on FETs 219
8.7 Switch Circuits Based on FETs
Switches based on p-i-n diodes require higher DC current at ON condition and also
behave as nonlinear element at lower microwave frequency. Therefore, p-i-n diode-
based switches are not suitable over lower microwave frequency of operation and
applications where sufficient DC current is not available. This is the reason FETs are
used in place of p-i-n diodes as switching device for mobile applications.
All types of field effect transistors (MESFETs, HEMTs and pHEMTs) at zero
drain bias are used as variable resistors to construct microwave switches [14–16].
Among varieties of FETs, mostly n-channel depletion mode FETs are used as vari-
able RF resistor, which has very low and very high source-to-drain resistance values
in absence and presence of gate bias voltage, respectively [6–10]. As discussed in
Chap. 6, with the application of negative voltage at gate terminal, the produced elec-
tric field below the gate causes the conduction channel to narrow, which increases the
resistance in between source to drain. Ideally, the gate is isolated from the channel,
thus there will be no current through the gate terminal, and it dissipates essentially
no DC power and can easily be designed as broadband switches. Thus, MESFETs at
zero drain bias are used as variable RF resistance to realize microwave switches by
applying control voltage to its gate terminal. Realization of MMIC-based microwave
systems using MESFETs and various HEMTs switches integrated with other circuits
are more convenient than using p-i-n diode-based circuits. Therefore, for low-power
applications, FET-based switches are more convenient.
MESFETs and various HEMTs are used for low-power microwave/RF switch for
easy realization in MMIC technology integrated with other circuitry. Extremely high
two-dimensional electron gas (2-DEG) of GaN HEMTs results in very low ON resis-
tance and due to its high breakdown fields it is capable to operate at higher operating
temperatures which makes it suitable for realization of high power switches.
Schematic diagram of a MESFET is shown in Fig. 8.17. The channel resistance
can be written as (6.1b)
1 Lg
Rc × (8.27)
qNd μn W a − d Vg
S G D
D RF RSISTOR S
RF RF
IN OUT
n + d n +
a
Lg G
substrate Control
voltage
Fig. 8.17 MESFET used as voltage-variable linear RF resistor for microwave switch
CDS
LS RS distributed channel RC RD LD
S D
CGS CGD
RG
LG
S D
Fig. 8.18 Equivalent circuit of a zero-gate-bias FET
sity within the channel, μn is the low field electron mobility, Lg is the gate length
approximately equal to the channel length, W is the gate periphery, a is channel
depth and d (Vg ) is the depletion layer depth which is function of the applied gate
control voltage. Equivalent circuit model of a zero-gate-biased MESFET is shown in
Fig. 8.18. The voltage-variable channel resistance can be considered as distributed
RC network, where the resistances and capacitances are voltage variable. Lumped
elements are included in each terminal to take into account the parasitics of the
extrinsic MESFET. This FET model is used to analyse switches using CAD tool.
However, for simplicity the equivalent circuit of a FET, at zero drain-bias condition,
can be considered as combination of resistors and capacitors which are represented
as a parallel RC circuit with an effective variable resistance and an effective capaci-
tance as shown in Fig. 8.18. However, these parasitic capacitances have a significant
influence on the switch performance at higher microwave frequencies.
For switch circuit realization, FET’s are arranged in both series and shunt con-
figurations. For compound switch configuration, series FETs provide a through-path
for the ON state, while the shunt FETs provide isolation for the OFF state. For the
operation of series FETs and shunt FETs associated with each switch state requires
opposite or complementary conduction states, and therefore opposite or complemen-
tary gate biases.
Circuit diagram for MESFET-based switches are the same as p-i-n diode-based
circuits, only p-i-n diodes are replaced by FETs. Schematic circuit diagrams of TEE-
and π -type switches are shown in Fig. 8.19 and 8.20, respectively. At ON-state con-
dition, the series FETs should operate in low-impedance state and parallel transistors
should operate in high-impedance state. Thus, in case of TEE- and π -type switch
configurations as shown in Figs. 8.19 and 8.20, the control voltage V G1 will be high
negative voltage and V G2 will be zero-bias voltage at ON-state condition and it will
be reversed in OFF-state condition.
8.8 Applications of RF/Microwave Switches 221
ZT1 ZT1 RF ZT1 ZT1 RF

IN OUT
ZT2
ZT2 RG RG
RG
VG1
VG2
Fig. 8.19 FET-based TEE-type switch circuit diagram
RF ZT1 RF
IN OUT
ZT1
ZT2 ZT2
ZT2 ZT2
RG
RG RG
VG2
VG1
Fig. 8.20 FET-based π -type switch circuit diagram
8.8 Applications of RF/Microwave Switches
High-performance communication systems need in-build reconfiguration abil-

ity to achieve optimum performance with change of applications. Routing of
RF/microwave signals is required to add more communication functionality. In
case of satellite communication payloads, more complex redundancy measures are
required to make the system more reliable to ensure long life. Space-based com-
munication systems also need reconfigurable antennas that can support multiple
frequency bands or reconfigurable service area. To achieve all these flexibilities, it is
required to change the signal path, signal level or signal phase. Microwave switches
are the essential building blocks to realize all these flexibilities. Adaptive power
amplifiers with tuneable frequency bands by switching matching networks can also
be used to maximize efficiency of a communication system over different bands of
frequencies as per requirement. Stepwise variable attenuators and phase shifters are
developed using microwave switches. RF/microwave switches are also used in auto-
mated measurement systems to control signal paths. With the help of switch matrix,
an instrument can measure multiple DUTs. Multiple tests with different instruments
Antenna
VC1 VC2
From To
SW1 λ/4 line λ/4 line SW2
Tx Rx
only in case of shunt switch
Fig. 8.21 Single antenna used to transmit as well as receive the signals
can also be run on the same DUT or multiple instruments can test multiple DUTs
very efficiently.
Other than to provide reconfiguration ability to communication systems, the
switches are also used to protect a subsystem just by disconnecting it, to prevent
overstress. For example, in case of radar systems microwave switch protects the
receiver from high-level transmitted power where a single antenna is used as transmit
and receive antenna. Switches used in the transceiver application are known as
transmit–receive (TR) switch. This TR switch connects antenna to transmitter in the
transmit state and to receiver in the receive state. Schematic diagram of the system
using single antenna for transmit and receive is shown in Fig. 8.21. In transmit state,
switch-1 (SW1) is ON and switch-2 (SW2) is OFF, thus transmitter gets connected
to the antenna and receiver is disconnected, whereas in receive state, SW2 is
ON and SW1 is in OFF condition; therefore, the antenna gets connected to the
receiver and transmitter is disconnected. Switches in their OFF state should provide
isolation by providing high impedance to the antenna; otherwise, it will change
the impedance matching condition of the antenna. To satisfy this condition, at least
one series diode should be in the switch configuration. However, shunt-configured
switches can also be used by connecting one-half wavelength transmission lines
in between antenna and switches. Instead of using two separate switches, single
switch with different configurations can also be used as shown in Fig. 8.22. This is a
single-pole-double-throw (SPDT) switch that acts as TR switch. Under sufficiently
positive-bias voltage VC1 and negative or zero voltage VC2 , the transmitter will be
connected to the antenna and the receiver will be isolated. In the other state, VC2
will be sufficiently positive-bias voltage and VC1 will be zero or negative voltage,
thus antenna will be connected to the receiver and transmitter will be isolated.
FET-based SPDT switch configuration is shown in Fig. 8.23. Under sufficiently
negative control voltage to V G1 and zero control voltage to V G2 , low-loss signal
path will be established in between RF1 and RF2 and there will be high isolation in
between RF1 and RF3 path. To establish low-loss signal path in between RF1 and
RF3, high negative control voltage is to be applied to V G2 and zero control voltage
to V G1 . This will provide high isolation in between RF1 and RF2 path.
Antenna
VC1 VC2
From To
Transmitter Receiver
Fig. 8.22 TR switch with series–shunt configuration
RF1
RF2 RF3
RG RG RG RG
VG1
VG2
Fig. 8.23 FET-based reflective-type SPDT Switch
The SPDT switch diagrams as shown in Figs. 8.22 and 8.23, respectively, are both
reflective types. That means the return loss of the OFF-state arm is poor. Terminated
switches can also be realized using p-i-n diodes and FETs by putting a resistor in
series with the shunt devices to achieve reference resistance Ro (50 ) of the shunt
path at ON condition. Schematic circuit diagram of a terminated SPDT switch using
FETs is shown in Fig. 8.24. The combined resistance of shunt FET at ON condition
and RS should be equal to the reference impedance Ro (=50 ).
Example 8.5 Calculate OFF-state arm isolation and return loss of a switch as shown
in Fig. 8.23. Consider ON-state and OFF-state impedances of the device are purely
resistive and values are 10 and 2000 , respectively.
RF1
RF2 RF3
RG RG
RG RG
Rs Rs
VG1
VG2
Fig. 8.24 FET-based terminated type SPDT Switch
Solution
With reference to the circuit of Fig. 8.23, suppose the signal path RF1 to RF2 is ON
and the path RF1 to RF3 is OFF. At this condition the equivalent circuit will be as
shown in Fig. 8.25.
When a signal of voltage vs and reference impedance 50 is applied to terminal
RF1 then the signal voltage vout at the RF3 port across a load of impedance 50
will be
vout 0.002212 vs
Thus, transmission coefficient from RF1 to RF3 port will be

2vout
S31 0.004425
vs
Thus the OFF-state isolation will be
20log(S31 ) 20 log(0.004425) −47.08 dB
With reference to Fig. 8.25c, the impedance seen by the RF3 port is given by
(58.78||50 + 2000)||10 9.951
Thus, the return loss of the RF3 port is

|9.951 − 50|
RL(port RF3) 20 log dB −3.50 dB
|9.951 + 50|
Fig. 8.25 Equivalent circuits for the switch of Example 8.5
Example 8.6 Calculate OFF-state isolation of a terminated switch as shown in

Fig. 8.24. Consider ON-state and OFF-state impedances of the device are purely
resistive and values are 10 and 2000 , respectively.
Solution
With reference to the circuit of Fig. 8.24, suppose the signal path RF1 to RF2 is ON
and the path RF1 to RF3 is OFF. At this condition, the equivalent circuit will be as
shown in Fig. 8.26. Here, the value of Rs will be 40 .
When a signal of voltage vs and reference impedance 50 is applied to terminal
RF1, then the signal voltage vout at the RF3 port across a load of impedance 50
will be
vout 0.006584vs
Thus, transmission coefficient from RF1 to RF3 port will be

2vout
S31 0.013169
vs
Fig. 8.26 Equivalent circuits for the switch of Example 8.6
Thus, the OFF-state isolation will be
20log(S31 ) 20 log(0.013169) −37.61 dB
With reference to Fig. 8.26c, the impedance seen by the RF3 port is given by
(58.80||50 + 2000)||50 48.796
Thus, the return loss of the RF3 port is

|48.796 − 50|
RL (port RF3) 20 log dB −38.28 dB
|48.796 + 50|
From the Examples 8.5 and 8.6, it is clear that in case of reflective-type switches,
return loss of the isolated port is poorer compared to terminated switch using same
number of shunt/series devices. To achieve better isolation in the OFF-state arm,
RF1
RF2 RF3
RG RG RG RG RG RG RG RG
Rs Rs Rs Rs
VG1 VG2
Fig. 8.27 FET-based terminated SPDT switch with better OFF-state isolation
VC1
Cb Cb
ZO D1 D2
D3 D4 ZO
Fig. 8.28 Schematic circuit diagram of π type switch using series-shunt p-i-n diodes
more number of series and shunt devices need to be added in the switch configura-
tions as shown in Fig. 8.27. In this case, extra isolation is achievable at the expense
of increased insertion loss in the ON-state arm.
Example 8.7 Determine the insertion loss and isolation of a π -type switch using
series and shunt p-i-n diodes as shown in Fig. 8.28 operating at 10 GHz for the
following parameters of the diodes: Rs 2 , Ri 2 , Lp 0.1 nH, Cj 0.02 pF
and Cp 0.01 pF.
Solution
Equivalent circuit diagram of the π -type switch of Fig. 8.28 is shown in Fig. 8.29
considering impedances of the series and shunt diodes are ZT 1 and ZT 2 respectively.
The transmission coefficient of the switch is given by
2Vout 2ZT 2 × ZT 2 ||Zo

S21 (8.28)
VS (Zo + ZT 2 )(ZT 2 ||Zo + 2ZT 1 + ZT 2 ||Zo )
ZT1 ZT1
ZO
ZT2 ZT2 Vout ZO
Fig. 8.29 Equivalent circuit diagram of π type switch
Therefore, insertion loss is given by

1 (Zo + ZT 2 )(Zo ZT 2 + Zo ZT 1 + ZT 1 ZT 2 )

S21 Zo ZT2 2
or
1 2ZT 1 Zo ZT 1 ZT 1 Zo
1+ + + + 2 (8.29)
S21 ZT 2 ZT 2 Zo ZT 2
Insertion loss of the π switch can be calculated by putting the values of ZT 1

and ZT 2 correspond to the low impedance and high impedance of the p-i-n diodes,
respectively, as

1
j2π fCT
Putting these values in (8.29), insertion loss of the switch at ON condition is given
by

1 RT 2XL Zo RT XL Zo 2RT Zo XL
1+ − − + j + + −
S21 Zo XC XC2 Zo XC XC XC2
Therefore,

RT 2XL Zo RT 2 XL Zo 2RT Zo XL 2
IL 1 + − − + + + − (8.30)
Zo XC XC2 Zo XC XC XC2

Putting ZO 50 , RT 4 , XL 2π fLp 2π × 10 × 0.1 6.283 and
XC 1000/(2π × 10 × 0.03) 530.516

4 2 × 6.283 50 × 4 2
IL 1 + − −
50 530.516 530.5162

6.283 50 2×4 50 × 6.283 2
+ + + −
50 530.516 530.516 530.5162
IL 1.1143 + 0.0547 1.169
or
IL (in dB) 10 log10 (1.169) 0.678 dB
At OFF condition the isolation of the π switch can be calculated by putting the
values of ZT 1 and ZT 2 corresponding to the high impedance and low impedances of
the p-i-n diodes, respectively, as
1
j2π fCT
Putting these values in (8.29), isolation of the switch at OFF condition is given
by

1 Zo RT − 2XC XL 2Zo RT XC XL
1+ − 2
S21 R2T + XL2 R2T − XL2 + (2RT XL )2

2RT XC + Zo XL XL Zo XC R2T − XL2
+j − − − 2
R2T + XL2 Zo R2T − XL2 + (2RT XL )2
Therefore,
2
Zo RT − 2XC XL 2Zo RT XC XL
Isolation 1 + − 2
R2T + XL2 R2T − XL2 + (2RT XL )2
2
+ − − − 2 (8.31)
R2T + XL2 Zo R2 − X 2 + (2RT XL )2
T L
Putting ZO 50 , RT 4 , XL 6.283 and XC 530.516
Isolation (1 − 116.56 − 433.2)2 + (−82.16 − 0.126 + 202.344)2

301141.9 + 14413.12 315555.05
or
Isolation (in dB) 10 log10 (318402.66) 55.0 dB
The ON-state insertion loss is 0.678 dB and OFF-state isolation is 55 dB.

Example 8.8 Determine the insertion loss and isolation of a π-type switch as shown
in Fig. 8.28 using series and shunt p-i-n diodes operating at 10 GHz for the following
parameters: Rs 2 , Ri 2 , Lp 0.2 nH, Cj 0.04 pF and Cp 0.02 pF.
Solution
For a π -type switch as shown in Fig. 8.28 using series and shunt p-i-n diodes, the
insertion loss at ON condition and isolation in OFF condition is given by (8.30) and
(8.31), respectively, as

IL 1 + − − + + + −
ZO XC XC2 ZO XC XC XC2
2
Isolation 1 + − 2
R2T + XL2 R2T − XL2 + (2RT XL )2
2
+ − − − 2
R2T + XL2 ZL R2 − X 2 + (2RT XL )2
T L
Here, RT (Ri + Rs ) 4 , and CT Cj + Cp 0.06 pF and Lp 0.2 nH.

Putting Zo 50 , RT 4 , XL 2π fLp 2π × 10 × 0.2 12.566 and
XC 1000/(2π × 10 × 0.06) 265.258 , the insertion loss is given by

4 2 × 12.566 50 × 4 2
IL 1 + − −
50 265.258 265.2582

12.566 50 2×4 50 × 12.566 2
+ + + −
50 265.258 265.258 265.2582
IL 0.9651 + 0.213 1.178
or
I L (in dB) 10 log10 (1.178) 0.710 dB

Isolation (1 − 37.183 − 44.083)2 + (−15.81 − 0.251 + 62.23)2
6442.67 + 2131.062 8573.73
or
Isolation (in dB) 10 log10 (8573.73) 39.33 dB
The ON-state insertion loss is 0.710 dB and OFF-state isolation is 39.33 dB.
Fig. 8.30 Insertion loss and 0.8 90

isolation of π type switch 80
over frequency
Isolation (dB)
0.7 70
I.L (dB)
60
0.6 50
40
0.5 30
0 2 4 6 8 10 12
Frequency (GHz)
Example 8.9 A π type switch as shown in Fig. 8.28 using series and shunt p-i-n
diodes for the following parameters: Rs 2 , Ri 2 , Lp 0.2 nH, Cj 0.04 pF
and Cp 0.02 pF. Plot the insertion loss and isolation of the attenuator over the
frequency range of 1–10 GHz.
Solution
For a π type switch as shown in Fig. 8.28 using series and shunt p-i-n diodes, the
insertion loss at ON condition and isolation in OFF condition is given by (8.30) and
(8.31), respectively, as

IL 1 + − − + + + −
Zo XC XC2 Zo XC XC XC2
2
Isolation 1 + − 2
R2T + XL2 R2T − XL2 + (2RT XL )2
2
+ − − − 2
R2T + XL2 Zo R2 − X 2 + (2RT XL )2
T L
Here, RT (Ri + Rs ) 4 , and CT Cj + Cp 0.06 pF and Lp 0.2 nH.

Putting Zo 50 , RT 4 , XL 2π fLp 2π × 10 × 0.2 12.566
and XC 1000/(2π × 10 × 0.06) 265.258 , the insertion loss and isolation is
calculated and shown in Fig. 8.30.
This plot shows that ON-state insertion loss increases and OFF-state isolation
decreases with the increase of frequency. This is due to the finite values of parasitic
capacitance and inductance of the diode.
Fig. 8.31 RF equivalent circuit diagram of SPDT switch of Example 8.10
Example 8.10 Determine S-parameter matrix of the FET-based SPDT switch as

shown in Fig. 8.27. Consider RF1 to RF2 path ON and RF1 to RF3 path open.
Assume ON-state and OFF-state impedances of the FETs are purely resistive of val-
ues 10 and 2000 , respectively, and Rs 40 . Also, calculate port return losses,
insertion loss and isolations.
Solution
RF equivalent circuit diagram of the SPDT switch is shown in Fig. 8.31 at the
condition RF1 to RF2 path ON and RF1 to RF3 path open.
From this equivalent circuit, the input impedance at port-1 can be written as
ZIN1 (((50||2040) + 10)||2040 + 10)||(((50||50) + 2000)||50 + 2000)

65.020
Therefore,
65.020 − 50 ◦
S11 0.13059 0
65.020 + 50
In a similar way,
65.025 − 50 ◦
S22 0.13062 0
65.025 + 50
48.809 − 50 ◦
S33 0.01206 180
48.809 + 50
Suppose a voltage signal vs1 with source impedance 50 is connected to port-1
and other two terminals are terminated with load impedances 50 , the output voltage
signal at port-2 will be
vout2 0.3993 vs1

Thus,
2vout2 ◦
S21 0.79863 0
vs1
Similarly, other transmission coefficient parameters can be calculated and the

S-parameter matrix of the SPDT switch is
⎡ ◦ ◦ ◦
⎤
0.13059 0 0.79863 0 0.00033 0
⎢ ⎥
[S] ⎣ 0.79863 0◦ 0.13062 0◦ 0.00023 0◦ ⎦
◦ ◦ ◦
0.00033 0 0.00023 0 0.01206 180
From the S-parameter matrix, the port return losses are

Return loss of port-1 20 log(0.13059) 17.680 dB
Insertion loss and isolations are
Insertion loss from port-1 to port-2 20 log(0.79863) 1.953 dB
Isolation in between port-1 to port-3 20 log(0.00033) 69.566 dB
Isolation in between port-2 to port-3 20 log(0.00023) 72.585 dB
References
1. Caulton M, Rosen A, Stabile A, Gombar A (1982) Pin diodes for low frequency high power
switching application. IEEE Trans Microw Theory Tech MTT-30:875–886
2. Caverly RH, Hiller G (1990) Establishing the minimum reverse bias for a p-i-n diode in a
high-power switch. IEEE Trans Microw Theory Tech 38:1938–1943
3. Drozdovskaia L (2001) RF and microwave frequency properties of a reverse-biased thick
switching p-i-n diode. IEEE Trans Microw Theory Tech 49(8):1370–1373
4. Bludov AV, Boltovets MS, Vassilevski KV, Zorenko AV, Zekentes K, Krivutsa VA, Kritskaya
TV, Lebedev AA (2004) Microwave switching, based on 4H-SiC p-i-n diodes. Tech Phys Lett
30(2):123–125
5. Camara N, Zekentes K, Romanov LP, Kirillov AV, Boltovets MS, Vassilevski KV, Haddad G
(2006) Microwave p-i-n diodes and switches based on 4H-SiC. IEEE Electron Device Lett
27(2):108–110
6. Gopinath A, Rankin JB (1985) GaAs FET RF switches. IEEE Trans Electron Devices ED-
32:1272–1278
7. Makioka S, Anda Y, Miyatsuji K, Ueda D (2001) Super self-aligned GaAs RF switch IC with
0.25 dB extremely low insertion loss for mobile communication systems. IEEE Trans Electron
Devices 48(8):1510–1514
8. Yang Z, Hu X, Yang J, Simin G, Shur M, Gaska R (2009) Maximum powers of low-loss
series-shunt FET RF switches. Solid-State Electron 53:117–119
9. Campbell C, Dumka D (2010) Wideband high power GaN on SiC SPDT switch MMICs. IEEE
MTT-S, IMS 2010, Anaheim, CA, USA, pp 145–148
10. Prasad A, Fager C, Thorsell M, Andersson CM, Yhland K (2014) Symmetrical large-signal
modeling of microwave switch FETs. IEEE Trans Microw Theory Tech 62(8):1590–1598
11. Simin G, Hu X, Ilinskaya N, Kumar A, Koudymov A, Zhang J, Khan MA, Gaska R, Shur
MS (2000) A 7.5 kW/mm current switch using AlGaN/GaN metal–oxide–semiconductor het-
erostructure field effect transistors on SiC substrates. Electron Lett 36:2043–2044
12. Ishida H, Hirose Y, Murata T, Kanda A, Ikeda Y, Matsuno T, Inoue K, Uemoto Y, Tanaka T,
Egawa T, Ueda D (2003) A high power Tx/Rx switch IC using AIGaN/GaN HFETs. In: IEDM
technical digest, pp 583–586
13. Koudymov A, Rai S, Adivarahan V, Gaevski M, Yang J, Simin G, Khan MA (2004) Monolith-
ically integrated high-power broad-band RF switch based on III-N insulated gate transistors.
IEEE Microw Wireless Compon Lett 14(12):560–562
14. Kaper V, Thompson R, Prunty T, Shealy JR (2004) Monolithic AlGaN/GaN HEMT SPDT
switch. In: Proceedings of 12th GAAS symposium, Amsterdam, The Netherlands, pp 84–86
15. Sattu A, Yang J, Shur M, Gaska R, Simin G (2010) AlGaN/GaN Microwave switch with hybrid
slow and fast gate design. IEEE Electron Device Lett 31(12):10389–1391
16. Sattu A, Billingsley D, Deng J, Yang J, Simin G, Shur M, Gaska R (2011) Low-loss AlInN/GaN
microwave switch. Electronics Lett 47(15)
Chapter 9
Microwave Attenuators
Abstract This chapter presents design and analysis of diode- and transistor-based
microwave attenuator circuits including their applications, performance variation and
compensation with operating temperature. Complete design details of series-, shunt-,
π -, TEE- and hybrid-configured attenuators with their merits and demerits are given.
Detailed design and analysis of driver circuits based on equiresistance curves of p-i-n
diodes for achieving temperature-insensitive attenuation of the attenuator circuits is
also presented.
9.1 Introduction
An attenuator provides known amount of attenuation when it is connected in a signal

path having known source and load impedances. These are used to control signal level
in the signal path. It may provide a fixed or variable/adjustable attenuation depend-
ing upon control voltage or current. Fixed attenuators are realized using combina-
tion of RF-resistive elements with fixed values, whereas voltage/current-controlled
attenuators are realized using combination of voltage/current-dependent resistors.
At RF/microwave frequency range, forward-biased p-i-n diodes, BJTs or FETs are
used as voltage/current-dependent RF resistors [1–13]. Voltage-variable p-i-n diode
attenuators are used to control signal level of communication systems by controlling
gain of the system [1–7]. In case of satellite transponders, p-i-n diode or FET-based
attenuators are used for on-board gain control of a communication transponder by
issuing command from ground or automatically by sensing RF power level.
Important performance parameters of a microwave attenuator are operating fre-
quency, attenuation range, insertion loss, port return losses, power handling capability
and DC power requirements. RF switches discussed in the previous section are also
attenuators with maximum attenuation at its OFF state and minimum attenuation at
its ON state. OFF-state reflective switch offers maximum attenuation by reflecting
the signal which is undesirable as function of attenuator. Variable attenuator provides
stepwise or continuous attenuation from its minimum to maximum attenuation range.
Known amount of attenuation and port matching over its entire attenuation range are

236 9 Microwave Attenuators
important parameters. When input and output ports of an attenuator are matched,
then attenuation offered by the attenuator is entirely due to the transmission loss. In
case of mismatch, some amount of the loss will be due to the reflection of the signal.
Variable attenuators using p-i-n diodes are mostly used due to their simplicity
in circuit and lower cost compared to transistor-based circuits. However, the RF
resistance of p-i-n diodes is very sensitive to its operating temperature [1–7]. Thus,
attenuation of a p-i-n diode-based attenuator is also a function of temperature, which
is undesirable. Some specially fabricated p-i-n diodes having zero temperature coef-
ficient of RF resistance can be used in an attenuator when biased at a constant
forward-bias current [1, 2]. However, this requirement cannot be met without sacri-
ficing other aspects of the p-i-n diode’s performance. Conventionally, analog as well
as digital compensation circuits with external temperature sensor is used to compen-
sate the attenuation. Another technique based on p-i-n diodes intrinsic property called
“optimum bias load-line technique” is used to achieve temperature-compensated con-
tinuous variable as well as stepwise variable attenuation [3–7].
GaAs MESFETs and HEMTs at zero-drain-bias condition are used as variable
resistors to construct microwave attenuators. The attenuation of transistor-based
attenuators is highly temperature-sensitive and is preferable only where continuously
variable attenuation is required. Stepwise variable attenuators are realized by switch-
ing fixed microwave attenuators (also called PADs) using transistors as microwave
switch. In this case, attenuation values remain nearly constant over a very wide range
of temperature since the attenuation values are mainly governed by the fixed PADs.
These are most suitable for realization in monolithic microwave integrated circuit
(MMIC) configuration.
9.2 Diode-Based Attenuator Circuits
RF resistance of p-i-n diode is function of its DC bias condition. By controlling the

bias current/voltage, its intrinsic layer resistance (Ri ) can be varied from few ohms
to several thousand ohms. The RF equivalent circuit of a p-i-n diode is shown in
Fig. 9.1 over the operating frequency for which f 1/2π τa , where τa is carrier
lifetime. Thus, RF impedance of the p-i-n diode can be written as

Z T (Ri ||Ci + Rs + j2π f L p )|| 1/j2π f C p (9.1)
Over the operating frequency range and bias condition where Ri 2π f L p and
also Ri 1/2π f C p , the RF equivalent circuit of the forward-biased p-i-n diode
can be represented as shown in Fig. 9.1b. In this case, the impedance Z T of the p-i-n
diode can be written as
Z T (Ri ||Ci ) + Rs (9.2)

9.2 Diode-Based Attenuator Circuits 237
(a) (b)
Lp
Rs
Rs
Cp
Ri Ci
Ri Ci
For
Fig. 9.1 Forward-biased equivalent circuit of p-i-n diode for f 1/2π τa
As discussed in Chap. 2, under forward-bias condition, the intrinsic layer resis-

tance Ri is current-controlled resistance, whereas other component values are prac-
tically independent of its DC bias. Thus, with the change of bias voltage/current
resistance, Ri of the forward-biased p-i-n diode acts as variable resistance. Using
this variable resistance, voltage/current variable microwave attenuators are realized.
9.3 Series- or Shunt-Connected Element
A simple voltage/current-dependent variable attenuator is a circuit where a p-i-n

diode is connected in series or shunt in the signal path as shown in Fig. 9.2. RF
equivalent circuits of the series-diode and shunt-diode attenuator circuits are shown
in Fig. 9.3a, b respectively. The circuits are same as series and shunt switch circuits;
however, here diodes are operated over its wide forward-bias range instead of only
high forward- and reverse-bias condition as in switch. The S-parameter matrix of the
series diode attenuator circuit can be written as (using 7.162c)
⎡ ⎤
ZT 2Z o
S11 S12
⎣ ⎦
Z T +2Z o Z T +2Z o
(9.3)
S21 S22 2Z o ZT
Z T +2Z o Z T +2Z o
Thus, attenuation of the series diode attenuator is given by

1 Z T
Attenuation(in dB) 20 log10 20 log10 1 + (9.4a)
|S21 | 2Z o
The input and output return losses of the series diode attenuator are given by
(a) VC (b) VC
Cbypass
RF
Cb choke Cb
Cb Cb
ZS RF ZS
choke ZL ZL
Fig. 9.2 Simple attenuator configuration a series diode and b shunt diode
(a) (b)
ZT
ZO ZO
ZO ZT ZO
Fig. 9.3 RF equivalent circuit of attenuators with a series diode b shunt diode

1 2Z o
RL(in dB) 20 log10 20 log10 1+ (9.4b)
S11 ZT
The variation of attenuation and return loss with the change of diode resistance
Ri is shown in Fig. 9.4 for typical values of other parameters of the diode. It shows
that with the increase of attenuation, return loss of the series attenuator degrades
drastically. This is the reason for not using attenuators with single series diode con-
figuration. At higher level of attenuation, major amount of attenuation is due to
reflection of power from the series element which is highly undesirable for prac-
tical systems. Reflected power due to the mismatch not only affects the preceding
system but also results in undesirable signal ripple in the frequency response due to
generation of standing wave.
For shunt-connected diode, the S-parameter matrix is given by (using 7.163d)
⎡ ⎤
−Z o 2Z T
S11 S12
⎣ ⎦
2Z T +Z o 2Z T +Z o
(9.5)
S21 S22 2Z T −Z o
2Z T +Z o 2Z T +Z o
Therefore, the attenuation of the shunt-connected diode attenuator is given by

9.3 Series- or Shunt-Connected Element 239
10 40
AƩenuaƟon
30
Return loss Return
Attenuation 5 20 loss
(dB) (dB)
10
0 0
0 50 100 150
(Ω)
Fig. 9.4 Series diode attenuator characteristic with variation of Ri
25 AƩenuaƟon 20
20 Return loss
15
15 Return
Attenuation 10 loss
(dB) 10
5
(dB)
5
0 0
0 50 100 150
(Ω)
Fig. 9.5 Shunt diode attenuator characteristic with variation of Ri

1 Z o
Attenuation(in dB) 20 log10
20 log10 1 + (9.6a)
|S21 | 2Z T
Similarly, the return losses (both input and output) of the shunt-connected diode
attenuator are given by

1 2Z T
RL(in dB) 20 log10
20 log10 1 − (9.6b)
S11 Zo
The plot of attenuation and return losses of the shunt-connected diode attenuator
is shown in Fig. 9.5. In this case also with the increase of attenuation, return loss
is decreasing. Therefore, this configuration of attenuators is also not suitable for
practical applications. Practically, it is always preferable to use input- and output-
matched attenuators, i.e. with good input and output return losses.
VC
RF
choke
λo/4 λo/4 λo/4
ZO
ZO
Fig. 9.6 Attenuator circuit using multiple shunt p-i-n diodes
Π network Π network
RF RF
λO/4 λO/4 λO/4
I/P O/P
ZT ZT ZT ZT
Fig. 9.7 Equivalent circuit of multiple shunt-connected p-i-n diode attenuator
9.4 Multiple Shunt-Connected Element
One way to achieve better input/output return losses is the use of multiple numbers
of shunt diodes with separation of λo /4 transmission line length that corresponds to
the operating frequency of f o as shown in Fig. 9.6. Return loss of this configuration
will be better due to the cancellation of signals reflected from the multiple diodes due
to their mutual phase difference of 180°. S-parameter elements of the attenuator can
be determined by matrix multiplication of ABCD parameter matrix of the individual
element of the RF equivalent circuit as shown in Fig. 9.7. The ABCD parameter
matrix of the attenuator is given by

A B Ad Bd Al Bl Ad Bd Al Bl Ad Bd Al Bl Ad Bd

C D C d Dd Cl Dl C d Dd Cl Dl C d Dd Cl Dl C d Dd
(9.7a)
Here, Z T is the forward-biased p-i-n diode impedance and is given by (9.2). The
ABCD parameter matrix of the shunt diode is given by
9.4 Multiple Shunt-Connected Element 241

Ad Bd 1 0
(9.7b)
C d Dd 1/Z T 1
And ABCD parameter matrix of a series transmission line of length l is given by

⎡ o o⎤
Al Bl cos πλ j Z o sin πλ
⎣ 1 πλo ⎦
2λ 2λ
πλo (9.7c)
Cl Dl j sin cos
Zo 2λ 2λ
Here, λ is the wavelength corresponding to any operating frequency f . Thus,

ABCD parameter of the λo /4 transmission line at operating frequency f o can be
determined by putting λ λo and can be written as
⎡ ⎤
Alo Blo 0 j Zo
⎣ 1 ⎦ (9.7d)
Clo Dlo j Zo 0
Thus, ABCD matrix of the attenuator can be written as

⎡ ⎤⎡ ⎤⎡ ⎤
j ZZTo j Z o j Zo
j Z
A B ⎢ ⎥⎣ 0 j Z o ⎦⎢ Z T o
⎥
⎣ Z 2 +Z 2 ⎦ ⎣ Z 2 +Z 2 Z ⎦ (9.7e)
C D Z
j Zo Z 2T j Z To j Z1o 0 j Zo Z 2T j Z To
o T o T
Here, first and last terms are ABCD parameter matrix of the network contain-
ing two shunt elements of impedance Z T and one series element of λo /4 length
transmission line as shown in Fig. 9.7.
The equation (9.7e) can be written as
⎡ ⎤
Z o Z T2 +Z o ( Z o2 +Z T2 ) Z o ( Z o2 +Z T2 )

A B ⎢−j Z T3
−j Z T2 ⎥
⎢⎣
⎥ (9.7f)
Z o2 Z T2 +( Z o2 +Z T2 ) Z o Z T2 +Z o ( Z o2 +Z T2 ) ⎦
2
C D
−j Z Z 4 − j Z 3
o T T
From this ABCD parameter matrix, S-parameters of the attenuator can be deter-
mined as follows (reciprocal matrix with symmetric structure):
S21 S12 2/(A + B/Z o + C Z o + D) (9.8a)
and
S11 S22 (B/Z o − C Z o )/(A + B/Z o + C Z o + D) (9.8b)
Putting the values of A, B, C and D, we can write

2Z T4
S21 S21 j (9.9a)
4Z o Z T3 + 4Z o2 Z T + 2Z o3 Z T + 2Z T4 + Z o4
2

Z o2 Z o2 + Z T2
S11 S22 (9.9b)
4Z o Z T3 + 4Z o2 Z T2 + 2Z o3 Z T + 2Z T4 + Z o4
Therefore, the attenuation at frequency f o of the four-element shunt-connected

attenuator is given by
Attenuation (in dB) 20 log10 (1/|S21 |)

4Z o Z T3 + 4Z o2 Z T2 + 2Z o3 Z T + 2Z T4 + Z o4
20 log10
2Z T4
(9.10a)
Similarly, the return losses at frequency f o of the four shunt-connected diode

attenuators are given by

1 1
RL(in dB) 20 log10 20 log10
S11 S22

4Z Z 3 + 4Z 2 Z 2 + 2Z 3 Z + 2Z 4 + Z 4
o T T o
20 log10 o T
o
T
(9.10b)
2 2
Zo Zo + ZT 2
The plot of attenuation and return losses at frequency f o for four shunt-connected
diode attenuators is shown in Fig. 9.8. It shows that the return losses of the attenuator
are better than that of the attenuator with single shunt-connected diode attenuator for
any value of attenuation. This can be further improved by connecting more numbers
of shunt-connected diodes with separation of λo /4 wavelength. Figure 9.9 shows
the attenuation and return losses of an attenuator with eight shunt-connected diodes
with separation of λo /4 wavelength. In this case, the return losses are better than
that of attenuator with four shunt-connected diodes. Figure 9.10 shows the return
losses for the four and eight shunt-connected diode attenuators with the attenuation
values. From this, it is clear that to realize higher value of attenuation with good
return losses, more number of diodes need to be used in case of shunt-configured
attenuators. A photograph of such an attenuator using eight shunt-connected diodes
with separation of λo /4 wavelength is shown in Fig. 9.11.
In case of multiple shunt-connected diode-based attenuators, the elements are
connected with separation of quarter wavelength transmission line corresponding
to a particular frequency f o . Therefore, at other frequencies, the separation will be
different from corresponding quarter wavelength transmission line. Thus, effective
cancellation of the reflected wave will go on decreasing with the frequency away
from f o . To achieve port matching over a broad frequency band, matched attenuator
circuits are designed.
9.4 Multiple Shunt-Connected Element 243
45 aƩenuaƟon
40
return loss
35
Attenuation 30
& 25
20
return loss 15
(dB) 10
5
0
5 55 105 155 205
(Ω)
Fig. 9.8 Attenuation and return losses of the four-element shunt-connected diode attenuator at
frequency f o
35
aƩenuaƟon
30
return loss
Attenuation 25
& 20
return loss 15
(dB) 10
5
0
25 75 125 175 225 275
(Ω)
Fig. 9.9 Attenuation and return losses of the shunt-connected eight-diode attenuator at frequency
fo
30
8 shunt diode
25 4 shunt diode
20
Return loss
15
(dB)
10
5
0
5 10 15 20 25 30 35
Attenuation (dB)
Fig. 9.10 Dependence of return losses with attenuation and number of diodes
Fig. 9.11 Photograph of an

attenuator with eight
shunt-connected p-i-n diodes
with separation of
quarter-wavelength
microstrip lines
9.5 Matched Attenuator Circuits
Matched attenuators ideally have matched input and output ports for all the values
of attenuation. Among various configurations, generally TEE- and π -type configu-
rations are used.
9.5.1 TEE Attenuator
Schematic circuit diagram of the diode-based TEE configuration attenuator is shown

in Fig. 9.12. Here, DC control voltage VC1 is for biasing the two series-connected
diodes and VC2 is for biasing the shunt diode. The bias configuration ensures same
amount of current to both the series diodes. Also the configuration ensures indepen-
dent control of currents for series and shunt diodes. Considering same characteristic
of both the series diodes, the RF impedance of both the series diodes will be same.
The RF equivalent circuit of the TEE-configured attenuator is shown in Fig. 9.13.
The conditions for the perfectly matching of input and output ports are
Z IN Z o and Z OUT Z o (9.11a)
Both the conditions lead to
Z o Z T 1 + Z T 2 ||(Z T 1 + Z o ) (9.11b)
or
Z o2 Z T2 1 + 2Z T 1 Z T 2 (9.11c)
9.5 Matched Attenuator Circuits 245
VC1
ZO
ZO
VC2
Fig. 9.12 Schematic circuit diagram of TEE attenuator
ZT1 ZT1
ZO
ZT2 vout ZO
Fig. 9.13 Shunt switch configuration for reflection coefficient determination
This is the required relationship between Z T 1 and Z T 2 for achieving perfect input
and output impedance matching.
Transmission coefficient of the TEE attenuator is given by
2vout 2Z o Z T 2
S21 (9.12a)
vs (Z o + Z T 1 )(Z o + Z T 1 + 2Z T 2 )
Combining (9.11c) and (9.12a),

2vout ZT 2
S21 (9.12b)
vs (Z o + Z T 1 + Z T 2 )

1
|S21 |

Zo + ZT 1 + ZT 2
20 log10 (9.13)
ZT 2
Considering all the impedances are resistive, assume,

Z T 1 R1 , Z T 2 R2 , and Z o Ro
Then, attenuation N is given by

Pavs 1
N
PL |S21 |2
Therefore, using (9.12b) attenuation can be expressed as

√ Ro + R1 + R2
N (9.14a)
R2
where
Ro2 R12 + 2R1 R2 (9.14b)
Combining (9.14a) and (9.14b), we can determine the values of R1 and R2 in

terms of required attenuation as
√
N −1
R1 Ro √ (9.15a)
N +1
and
√
2 N
R2 Ro (9.15b)
(N − 1)
Figure 9.14 shows the requirement of resistances for the TEE attenuator for dif-
ferent attenuation values satisfying perfect impedance matching of input and output
ports. This suggests that bias voltage/current requirements for series and shunt diodes
are different. In Fig. 9.12, a simple bias circuit is shown to provide independent bias
to the series- and shunt-connected diodes.
Example 9.1 Calculate resistance values of a perfectly matched TEE-type resistive
attenuator of attenuation 3-dB in a 50 system, and also write its S-parameter matrix.
Solution
Attenuation N is given by
Pavs
N
PL
Here, attenuation in dB is given by
10 × log(N ) 3 dB
60 500
R1 R2
50 400
40
300
(Ω) 30
(Ω)
200
20
10 100
0 0
0 10 20 30 40 50
Attenuation (dB)
Fig. 9.14 Values of R1 and R2 for different attenuation values with perfect port matching
Fig. 9.15 3-dB TEE-type

attenuator circuit 8.55 Ω 8.55 Ω
142Ω
Thus,
N 103/10
Ro 50
For a TEE-type attenuator, from (9.15a) and (9.15b)

√
3/20
N −1 10 −1
R1 Ro √ 50 × 8.55
N +1 103/20 + 1
and
√
2 N 2 × 103/20
R2 Ro 50 × 3/10 141.9
(N − 1) 10 −1
Thus, the resistance values of the 3-dB TEE-type attenuator are 8.55 and 141.9
as shown in Fig. 9.15.
For the matched attenuator S11 S22 0
√
S21 S12 1/ N 1/103/20 0.7079
VC1
Cb Cb
ZO
ZO
VC2
Fig. 9.16 Schematic circuit diagram of π attenuator
Thus, the S-parameters matrix is

S11 S12 0 0.7079

S21 S22 0.7079 0
9.5.2 π Attenuator
Schematic circuit diagram of a diode-based π configuration attenuator is shown in

Fig. 9.16. Here, DC control voltage VC1 is for biasing the two series-connected diodes
and VC2 is for biasing of the shunt diodes. The bias configuration ensures independent
current controllers for the series- and shunt-connected diodes. However, VC1 provides
same current for both the series diodes and VC2 provides same current to both the
shunt diodes. Considering Z T 1 is the impedance of each series pair diode and Z T 2 is
the impedance of each shunt pair diode, the RF equivalent circuit of the --π configured
attenuator is shown in Fig. 9.17. The conditions for the perfectly matching of input
and output ports are
Z IN Z o and Z OUT Z o (9.16a)
Both the conditions lead to
Z o Z T 2 ||(Z T 1 + Z T 2 ||Z o ) (9.16b)
or
ZT1 ZT1
ZO
ZT2 ZT2 vout ZO
Fig. 9.17 Equivalent RF circuit diagram of the π attenuator
Z T 1 Z T2 2
Z o2 (9.16c)
ZT 1 + ZT 2
This is the relationship in between Z T 1 and Z T 2 for achieving perfect input and
output impedance matching.
Transmission coefficient of the π attenuator is given by
2vout Z T2 2
S21 (9.17) (9.17a)
vs 2Z o Z T 2 + 2Z o Z T 1 + 2Z T 1 Z T 2 + Z T2 2

1
|S21 |

2Z o Z T 2 + 2Z o Z T 1 + 2Z T 1 Z T 2 + Z T2 2
20 log10 (9.18a)

Z T2 2
Considering all the impedances resistive, i.e.
Z T 1 R1 , Z T 2 R2 , and Z o Ro
Then, attenuation N is given by

Pavs 1
N
PL |S21 |2
Therefore, attenuation can be expressed as

√ 2Ro R2 + 2Ro R1 + 2R1 R2 + R22
N (9.18b)
R22
Combining (9.16c) and (9.18b), we can determine the value of R1 and R2 in terms
of required attenuation as
(N − 1)
R1 Ro √ (9.19a)
4 N
4500 1000
4000 900
3500 R1 R2 800
3000 700
600
2500
(Ω) 2000
500 (Ω)
400
1500 300
1000 200
500 100
0 0
0 10 20 30 40 50
Attenuation (dB)
Fig. 9.18 Values of R1 and R2 for different attenuation values with perfect port matching for
PI-type attenuator
and
√
N +1
R2 Ro √ (9.19b)
N −1
Figure 9.18 shows the requirement of resistance values of the π attenuator for
different attenuation values satisfying perfect impedance matching of input and out-
put ports. This suggests that bias voltage/current requirements for series and shunt
diodes are different. Figure 9.16 shows the bias circuit to provide independent bias
voltage/currents to the series- and shunt-connected diodes.
Example 9.2 Calculate resistance values of a perfectly matched π -type resistive

attenuator of attenuation 6-dB in a 50 system, and also write its S-parameter
matrix.
Solution
Attenuation N is given by
Pavs
N
PL
Here, attenuation in dB is given by
10 × log(N ) 6 dB
Thus,
N 106/10
Fig. 9.19 6-dB π -type

attenuator circuit 37.35 Ω
150.5Ω 150.5Ω
Ro 50
For a π-type attenuator, from (9.19a) and (9.19b)

6/10
(N − 1) 10 −1
R1 Ro √ 50 × 18.68
4 N 4 × 103/20
and
√ 6/20
N +1 10 +1
R2 Ro √ 50 × 6/20 150.5
N −1 10 −1
Thus, the resistance values of the 3-dB π -type attenuator are R1 18.68 and
R2 150.5 as shown in Fig. 9.19.
For the matched attenuator S11 S22 0
√
S21 S12 1/ N 1/106/20 0.5
Thus, the S-parameter matrix is

S11 S12 0.0 0.5

S21 S22 0.5 0.0
9.5.3 Quadrature Hybrid Matched Attenuator
The TEE- and π -type attenuators provide attenuation with ideally perfect input and
output impedance matching. To achieve this, bias circuit should provide current to
the diodes in such a way that it should maintain the required relationship in between
the series and shunt impedances. Practically, it is very difficult to meet this bias
requirement by using a simple driver circuit. Attenuator configuration using 3-dB
90° hybrid coupler solves this problem. In this case, two p-i-n diodes ideally with
same bias current can be used to achieve variable attenuation meeting the requirement
of port impedance matching.
VC
λ/4 line λ/4 line
3 dB 90o
1 4
ZO hybrid coupler
ZO
2 3
D1 D2
Fig. 9.20 Quadrature hybrid matched attenuator configuration with shunt DC path
Schematic circuit diagram of a p-i-n diode attenuator using 3-dB quadrature hybrid
is shown in Fig. 9.20. Here, control voltage VC is used to provide bias current to both
the diodes D1 and D2 . Both the diodes are connected in parallel as far as DC equivalent
circuit is concerned. Therefore, voltage applied by the bias circuit will be same for
both the diodes. Practically, it is very unlikely that the DC characteristic, i.e. I–V
characteristic of both the diodes, will be same. In such a situation, application of same
voltage cannot ensure equal currents through both the diodes leading to different
impedances. To ensure same currents through both the diodes, it is preferable to
connect both the diodes in DC series path as shown in Fig. 9.21.
The 3-dB quadrature hybrid has four ports with equal power sharing and 90°
phase difference in between the two output ports. When a signal fed to input port-1,
then half of the signal power will reach to the port-2 and remaining half power will
reach to the port-3. The phase difference between the two output signals will be 90°.
Fourth port is the isolated port with respect to the port-1. Therefore, there will be
no power at the port-4 when power is incident on the port-1. RF equivalent circuit
of the quadrature hybrid configuration p-i-n diode attenuator is shown in Fig. 9.22.
Here, Z T 1 and Z T 2 are the impedances of diodes D1 and D2 , respectively. Suppose an
RF source of amplitude vs is connected to the port-1 of the hybrid through resistive
impedance Z O and load impedance Z O is connected at the port-4. Thus, incident
voltage amplitude at port-1 will be
vs
V1+ (9.20a)
2
As per the characteristic of the 3-dB 90° hybrid coupler, incident voltage amplitude
to the diode D1 and D2 will respectively be
VC
λ/4 line λ/4 line
3 dB 90o
1 4
ZO hybrid coupler
ZO
2 3
D1 D2
Fig. 9.21 Quadrature hybrid matched attenuator configuration with series DC path
1 4
3 dB 90 o
ZO hybrid coupler
2 3
ZO
ZT1 D1 D2 ZT2
Fig. 9.22 Quadrature hybrid matched attenuator configuration
vs vs
√ θ ◦ and √ (θ − 90)◦ (9.20b)
2 2 2 2
Here, θ is the insertion phase of the signal from port-1 to port-2. The reflection
coefficients of the diodes with respect to the impedance Z o are given by
ZT 1 − Zo ZT 2 − Zo
Γ1 and Γ2 (9.20c)
ZT 1 + Zo ZT 2 + Zo
The reflected signal amplitude from the diodes D1 and D2 will be

vs Γ1 ◦ vs Γ2
√ θ and √ (θ − 90)◦ (9.20d)
2 2 2 2
The reflected signals from both the diodes will be the incident signal for the port-2
and port-3. The signal amplitude at port-1 and port-4 due to the incident signal on
port-2 will be
vs Γ1 vs Γ1
2θ ◦ and (2θ − 90)◦ (9.20e)
4 4
Similarly, the signal amplitude at port-1 and port-4 due to the incident signal on
port-3 will be
vs Γ2 vs Γ2
(2θ − 180)◦ and (2θ − 90)◦ respectively (9.20f)
4 4
Therefore, signal amplitude reflected from port-1 and transmitted to port-4
will respectively be
vs Γ1 vs Γ2
2θ ◦ + (2θ − 180)◦ (9.20g)
4 4
and
vs Γ1 vs Γ2
(2θ − 90)◦ + (2θ − 90)◦ (9.20h)
4 4
Thus, the reflection coefficient of this symmetric structured attenuator is given by
Γ1 2θ ◦ − Γ2 2θ ◦
S11 S22 (9.21a)
2
Therefore, return losses of the attenuator are given by

2
RL(in dB) 20 log10 (9.21b)
|Γ1 − Γ2 |
The reflected signal from the port-1, i.e. from the input port, will be non-zero if
both the diodes are not matched, i.e. if Γ1 Γ2 . However, for the matched pair of
diodes, Γ1 Γ2 for same bias condition and return loss will be zero, i.e. perfectly
matched.
The transmission coefficient of the attenuator can be calculated as follows:
Γ1 (2θ − 90)◦ + Γ2 (2θ − 90)◦

S21 S12 (9.22a)
2
Therefore, the attenuation of the attenuator can be written as

2
Attenuation(in dB) 20 log10 (9.22b)
|Γ1 + Γ2 |
Fig. 9.23 Photograph of a

multistage quadrature hybrid
attenuator
In case of perfectly matched pair of p-i-n diodes, i.e. for Γ1 Γ2 Γ , the

attenuation of the attenuator can be expressed as

1
Attenuation(in dB) 20 log10 (9.22c) (9.22b)
|Γ |
Photograph of a 3°dB 90° Lange coupler-based attenuator realized on a 20–mil-

thick alumina substrate is shown in Fig. 9.23.
Frequency bandwidth of the quadrature hybrid attenuator is limited by the band-
width of the coupler. At microwave frequency bands, Lange coupler is used for real-
ization of p-i-n diode attenuator that can operate over an octave bandwidth. TEE- and
π -type attenuators are basically very broadband attenuators due to the absence of
any frequency-limited elements like quarter wave transmission line or coupler. Thus,
frequency bandwidths of a TEE and π attenuators are limited by its bias circuit and
parasitics of p-i-n diode itself.
9.6 Driver Circuit for P-I-N Diode Attenuators
Microwave and RF variable attenuators based on p-i-n diodes are used in airborne and
space-based systems where temperature-insensitive attenuation is always demanded
over very wide operating temperature range. However, RF resistance of p-i-n diodes
that mainly determine the attenuation of the circuit is temperature dependent as
follows (2.17):
T n−m
Ri (T, Id ) z (9.23a)
[Ido (T, Vd )] p
Therefore, function of a driver circuit is not only to provide just DC voltage and
current to the diodes without affecting RF performance but also to regulate the bias
Id
RL
RS Id× RS
Vd
VS Ri Vdj
Ci
Fig. 9.24 A simple bias circuit of p-i-n diode
voltage/current over the operating temperature range to ensure temperature-invariant

attenuation.
From (9.23a), the temperature coefficient of RF resistance can be written as
∂ Ri /∂ T n−m p ∂ Id
Tc Ri − (9.23b)
Ri T Id ∂ T
Driver circuit in its simplest form consists of a DC supply voltage VS and a current
limiting series resistor R L as shown in Fig. 9.24. Thus, diode current Id is given by
VS − Vd j
Id (9.23c)
R L + Rs
Combining (9.23b) and (9.23c), the temperature coefficient of the RF resistance

is given by
n−m p ∂ Vd j
Tc Ri − − (9.23d)
T VS − Vd j ∂ T
Under sufficiently forward-bias condition, I–V relationship of a p-i-n diode can

be written as (2.16)

v q[Vd (T ) − Id (Vd , T ) × R S − E b ]
Id (Vd , T ) A T exp (9.23e)
ηkT
Combining (9.23d) and (9.23e), temperature coefficient of RF resistance of p-i-n

diodes can be expressed as
p ηkT
VS − Vd j + n−m E b − Vd j + pv+m−n
n−m q
Tc Ri (9.24)
ηkT
T
n−m
VS − Vd j O + q
9.6 Driver Circuit for P-I-N Diode Attenuators 257
Fig. 9.25 Temperature 1.0

coefficient of rf resistance of
p-i-n diode with different
supply voltages
0.0
ΓcRi(%/oC)
-1.0
VOPT
-2.0
0.5 1.0 1.5 2.0 2.5 3.0
VS (volt)
The plot of Tc Ri as a function of supply voltage VS , keeping Vdj constant, is shown

in Fig. 9.25 for typical parameter values of a p-i-n diode. From the plot and (9.24),
the following can be concluded:
1. Tc Ri is zero at a particular value of supply voltage VS . The supply voltage for
which the temperature coefficient of p-i-n diode RF resistance becomes zero is
called ‘Optimum Supply Voltage’ VOPT .
2. Tc Ri is positive for VS more than VOPT . When the diode is driven by a high
supply voltage (with high current limiting resistor), i.e. under constant-current-
bias condition, Tc Ri becomes constant and is given by (n − m)/T .
3. Tc Ri is negative for V S less than VOPT . Temperature coefficient of the RF resistance
becomes very high when the supply voltage VS goes towards Vdj , corresponding
to the constant-voltage-bias condition (i.e. with R L 0).
Based on this supply voltage-dependent temperature coefficient of a p-i-n diode
RF resistance, temperature-insensitive attenuation can be achieved from a simple
bias circuit shown in Fig. 9.24 with VS VOPT . The optimum supply voltage is
obtained from (9.24) by setting Tc Ri 0 at VS VOPT . Thus, VOPT is given by

p ηkTo pv
VOPT Vdo + (E b − Vdo ) + −1 (9.25a) (9.25)
n−m q n−m
This is the same expression as obtained (2.21) in Chap. 2. When the driver cir-
cuit is designed with the optimum supply voltage, the bias load lines follow the
equiresistance curve of the diodes and maintain temperature-invariant attenuation.
Figure 9.26 shows equiresistance curves and the corresponding optimum bias load
lines of a p-i-n diode at different bias currents. Here, all the load lines for different
currents corresponding to different attenuation levels, intercept the voltage axis at
nearly the same point at t 1.15 V; this is nothing but the bandgap potential in volts
of the p-i-n diode semiconductor, i.e. silicon. This is the basis for the design of p-i-n
diode attenuator driver circuit for achieving temperature-invariant attenuation.
I-V Curves
load lines
3.0 Equiresistance
80 oC 30 oC -20 oC curves
2.0
Id (mA)
1.0 VOPT
0.0
0.6 0.8 1.0 1.2
Vdj (V)
Fig. 9.26 Typical equiresistance curves with load lines of a p-i-n diode
Based on the straight line property of p-i-n diode equiresistance curves, sug-
gested p-i-n diode attenuator driver circuit is shown in Fig. 9.24 with VS VOPT and
the bias resistance R L R O given by
VOPT − VdO
RO (9.25b) (9.25)
IdO
Here, IdO and VdO are diode current and voltage corresponding to a particular
RF resistance RdO , i.e. attenuation. In case of attenuator circuits where N number of
diodes are connected in DC series path, the optimum supply voltage will be N ×VOPT .
Setting the diode current Id (VS − Vd )/R L , and differentiating RF resistance
Ri with respect to VS , the variation of RF resistance of p-i-n diode becomes
∂ Ri ∂ VS
−p (9.26)
Ri VS − Vd
In the driver circuit, VS VOPT is slightly greater than the voltage drop (Vd
Vdj + Id × Rs ) across the diode. Thus, it is required to use very stable voltage source
for the driver circuit. Figure 9.27 shows a driver circuit including a stable voltage
source. Here, VOPT is derived using potential divider constituted by resistor Ra and
Rb . Unity gain buffer is used to eliminate the loading effect of the potential divider.
Digital driver circuits are needed where stepwise variable attenuation is required.
A digital control circuit for temperature-invariant stepwise variable performance of
diode-based circuits is shown in Fig. 9.28. The optimum supply voltage (VOPT ) is
provided by a regulated voltage supply with the associated potential divider. Output of
the buffer circuit is connected to a bank of predetermined resistors (R1 to Rn ) through
an analog multiplexer. Depending upon the digital data to the analog multiplexer, one
9.6 Driver Circuit for P-I-N Diode Attenuators 259
VS
Ra OP AMP
VOPT
RO O/P
Cb Rb
p-i-n diode
attenuator
Fig. 9.27 Attenuator driver circuit with optimum load-line bias
R1
Digital data
Analog Mux
VS
Ra V VC
Voltage OPT
regulator
Cb
Rb
Rn
Fig. 9.28 Attenuator driver circuit for stepwise variable attenuation
resistor at a time will be selected from resistors (R1 to Rn ) to provide temperature-

invariant stepwise variable attenuation.
9.7 Effect of Nonideal Components in Driver Circuits
Deviation in VOPT , due to the setting tolerance of VS , will not only deviate the initial
settings of the attenuation but temperature coefficient of the attenuation will also
change as given by (9.26). Thus, setting of VOPT value is a critical aspect of the
driver circuits. Variation of RF resistance due to the tolerance of the bias resistor is
given by
∂ Ri ∂ RL
−p (9.27)
Ri RL
Thus, the variation of RF resistance is in the same magnitude as the tolerance

of the bias resistor (for p ∼ 1). Temperature coefficient of the RF resistance is
determined only by the setting of the open-circuit voltage (VOPT ) of the driver circuit.
Therefore, tolerance of the bias resistor (R L ) only contributes to the initial setting of
the attenuation value. It will not affect the temperature coefficient of the attenuation.
Temperature coefficient of VS and R L will affect the attenuation values. In case
of the digital driver circuit shown in Fig. 9.28, finite ON resistance of the analog
multiplexer is a part of the bias resistor having resistance of few tens of Ohms (of
74HC4051) to few hundreds of Ohms (of CD4051). Moreover, ON resistance of the
analog multiplexer has reasonably high-temperature coefficient. Thus, temperature
coefficient of the bias resistance and supply voltage must be taken into account
when control circuits are designed. By modifying VOPT and R O , it is possible to
eliminate or minimize the effect of temperature coefficient of the voltage source and
bias resistors [7].
Suppose modified optimum voltage and optimum resistor at ambient temperature
are VOPTM and ROM , respectively. Then, we can write
VOPT Vd + Id R O (9.28a)
VOPTM Vd + Id ROM (9.28b)
Here, Vd , Id , VOPTM and ROM all have finite temperature coefficients but VOPT
and R O are the temperature-invariant optimum voltage and bias resistance. Thus,
differentiating (9.28a) and (9.28b) with respect to T, we can write
∂ VOPT ∂ Vd ∂ Id
+ RO 0 (9.28c)
∂T ∂T ∂T
and
∂ VOPTM ∂ Vd ∂ Id ∂ ROM
+ ROM + Id (9.28d)
∂T ∂T ∂T ∂T
Combining (9.28c) and (9.28d) and eliminating Id from (9.28a) and (9.28b), we
can write

VOPT ∂∂VTd − ∂ V∂OPTM + Vd ∂ V∂OPTM − α(VOPT − Vd )
VOPTM T
∂ Vd
T
(9.29a)
∂T
− α(VOPT − Vd )
Here, the temperature coefficient of the bias resistor is α and is given by α

(∂ ROM /∂ T )/ROM .
In case of finite temperature coefficient of the bias resistor but temperature-
invariant optimum supply voltage,
∂ VOPTM /∂ T 0 (9.29b)
In this condition, the modified optimum voltage can be written as

9.7 Effect of Nonideal Components in Driver Circuits 261
Hot Amb Cold

Cold
Amb
Hot
Id (mA)
Required
(Vd,Id) Optimum Bias load line
Vd (Volt) VOPTM VOPT
Fig. 9.29 Graphical representation of V OPTM for positive temperature coefficient of bias resistor
VOPT ∂∂VTd − αVd (VOPT − Vd )

VOPTM ∂ Vd
(9.29c)
∂T
− α(VOPT − Vd )
Figure 9.29 shows graphical presentations of how the modified optimum voltage
provides temperature-invariant performance under positive temperature coefficient of
the bias resistor. The slope of the load lines decreases with the increase of temperature
due to the positive temperature coefficient of the bias resistor. Here, it is shown that
VOPTM with positive temperature coefficient bias resistor tracks the required diode
voltage and current (Vd , Id ) over the operating temperature range.
In case of finite temperature coefficient of the supply voltage but with the
temperature-invariant bias resistor, α 0 the modified optimum voltage can be
written as

VOPT ∂∂VTd − ∂ V∂OPTM + Vd ∂ V∂OPTM
VOPTM T
∂ Vd
T
(9.30)
∂T
Figure 9.30 shows the graphical presentations of how VOPTM provides

temperature-invariant attenuation under the positive temperature coefficient of the
supply voltage. Here, it is shown that VOPTM with positive temperature coefficient
tracks the required Vd and Id for temperature-invariant attenuation of the attenuator.
Here, all the load lines will remain parallel since α 0 but VOPTM increases with
the increase of temperature for its positive temperature coefficient.
Hot Amb Cold
Amb Hot
Cold
Id (mA)
Optimum
Required Bias load
(Vd,Id)
Vd (Volt) VOPT VOPTM
Fig. 9.30 Graphical representation of V OPTM for positive temperature coefficient of bias voltage
Temperature Controlled Chamber

IREF VRFF Other
temperature p-i-n diode
RL sensitive attenuator
C B elements A
Fig. 9.31 Schematic diagram of a measurement setup to include temperature coefficient of

temperature-sensitive elements
Under forward-bias condition, though temperature effect of Cd and Rs can be

considered negligible compared to the temperature effect of Rd , however, absolute
value of Rs should be taken into account to determine bias resistor (R O ) of the driver
circuits.
9.8 Experimental Determination of V OPTM
Figure 9.31 shows schematic block diagram of a scheme used to determine the
modified optimum voltage including the effect of temperature coefficient of the bias
resistor. To determine the optimum supply voltage of the driver circuit, RF circuit
of the attenuator including other temperature-sensitive elements such as multiplexer
and bias resistors are kept within the temperature-controlled chamber as shown in
Fig. 9.31. The voltage and current of the circuit are measured at point ‘C’ instead of
at the diode terminal point ‘A’. The value of the resistor R L is kept approximately
equal to the actual value required for the circuit.
9.8 Experimental Determination of V OPTM 263
S G D
D RF RSISTOR S
RF RF
IN OUT
n+ d n+
a
Lg G
substrate Control
voltage
Fig. 9.32 MESFETs are used as voltage-variable linear RF resistor for attenuator
In case of p-i-n diodes, the equiresistance curves are nearly straight lines and inter-
cept the voltage axis at same points for entire operating current condition. Therefore,
three-point (at three temperatures) measurement will give accurate V OPT .
To determine VOPTM , three sets of voltages and currents can be measured at three
different temperatures of the chamber, adjusting voltage/current to the circuit for
same attenuation value. The optimum load line will be the best-fitted straight line
passing all these three points. The intercept of the load line with the voltage axis is
the modified optimum open-circuit voltage VOPTM of the bias network.
9.9 FET-Based Attenuators
GaAs MESFETs at zero drain bias are used as variable resistors to construct
microwave attenuators. Mostly, n-channel depletion mode FET is used as variable
RF resistor, which has low source-to-drain resistance in the absence of gate-bias
voltage. With the application of negative voltage at gate terminal, the produced elec-
tric field below the gate causes the conduction channel to narrow, which increases
the resistance in between source and drain. Thus, MESFETs at zero drain bias are
used as variable RF resistance to realize voltage-variable attenuator (VVA). Gate-
bias voltage is used to control the channel resistance, i.e. resistance between drain
and source terminals which is used as linear RF resistor where the drain and source
terminals are used as RF ports as shown in Fig. 9.32. Ideally, the gate is isolated
from the channel, and thus there will be no current through the gate terminal; thus,
it dissipates essentially no DC power and can easily be designed into broadband
circuits. The RF signal flows from source to drain, while the RF isolated gate is used
for controlling the resistance value. The channel resistance (RC ) can be written as
(9.31)
1 Lg
Rc × (9.31)
q Nd μn W a − d Vg
sity within the channel, μn is the low field electron mobility, L g is the gate length
CDS
LS RS distributed channel RC RD LD
S D
CGS CGD
RG
LG
Fig. 9.33 Equivalent circuit of a zero-gate-bias FET
approximately equal
to the channel length, W is the gate periphery, a is the channel
depth and d Vg is the depletion layer depth which is function of the applied gate
control voltage. Linearity of the MESFETs as VVA is maintained as long as the
operating RF power level is very small compared to the applied gate control voltage.
Otherwise, depletion layer width will be modulated by the electric field due to the
RF signal amplitude between source–drain ports. Equivalent circuit model of a zero-
gate-biased MESFET is shown in Fig. 9.33. The voltage-variable channel resistance
can be considered as distributed RC network, where the resistances and capacitances
are voltage variable. Lumped elements are included in each terminal to take into
account the parasitics of the extrinsic MESFET. This FET model may be used to
design attenuator using CAD tool. However, at linear operating condition where the
drain-to-source resistance is proportional to the applied gate voltage, the equivalent
circuit of FETs can be represented as a parallel RC circuit with an effective resis-
tance and an effective capacitance. FET attenuator relies primarily on the variation
of drain-to-source resistance by varying the applied gate voltage. However, these
parasitic capacitances have a significant influence on the attenuator performance at
high frequencies.
Circuit schematic of a TEE-type attenuator using GaAs MESFETs is shown in
Fig. 9.34. RF resistance (Z T 1 ) of the two series-connected FETs are controlled by
same control voltage V G1 through gate-bias resistors RG , whereas RF resistance Z T 2
of the shunt-connected FET is controlled by control voltage V G2 . Here, the control
voltages V G1 and V G2 are dependent on each other for achieving the impedance
matching condition given by (9.11c) and attenuation given by (9.13) as a function
of the FET impedances. Circuit schematic of a π -type attenuator using GaAs MES-
FETs is shown in Fig. 9.35. RF resistance Z T 1 of the two shunt-connected FETs are
controlled by the same control voltage V G2 , whereas RF resistance Z T 1 of the series-
connected FET is controlled by control voltage V G1 . Here, the control voltages V G1
and V G2 are also dependent on each other for achieving the impedance matching
9.9 FET-Based Attenuators 265
ZT1 ZT1 RF ZT1 ZT1 RF

IN OUT
ZT2
ZT2 RG RG
RG
VG2
VG1
Fig. 9.34 FET-based TEE-type variable attenuator circuit diagram
RF ZT1 RF
IN OUT
ZT1
ZT2 ZT2
ZT2 ZT2
RG
RG RG
VG1
VG2
Fig. 9.35 FET-based π -type variable attenuator circuit diagram
condition given by (9.16c) and attenuation given by (9.18a) as a function of the FET
impedances.
These FET-based TEE- and π -type attenuators can be used as continuously vari-
able analog attenuator or as stepwise variable attenuator just by providing the gate-
bias voltages as analog or discrete stepwise control voltage. Stepwise variable digital
attenuators can also be realized by switching fixed attenuators (PADs) where FETs
or p-i-n diodes are used as switch. Schematic diagram of such type of attenuator
is shown in Fig. 9.36. The fixed attenuators are TEE- or π -type attenuators using
fixed RF resistors of proper resistance values for different attenuations. Switches are
used to select a fixed attenuation path or through path. Thus, the above configuration
can provide stepwise attenuation values ideally from 0 to 16 dB in steps of 1 dB by
using 4-bit digital control voltage. Practically, the through paths and switches have
finite insertion loss. Therefore, the 0-dB attenuation case will be practically insertion
loss of the attenuator when all the through paths are selected. Attenuation values of
VC VC VC VC
through through through through

path path path path
1 dB 2 dB 4 dB 8 dB
Fig. 9.36 Digital attenuator using fixed attenuators and FET as switch
the fixed attenuators should be selected taking into account the insertion loss of the
through paths and switches.
Example 9.3 S-parameter matrix in a 50 system of a 2-port attenuator is as follows:

0.243 − 125.93◦ 0.706 − 15.74◦
[S]
0.706 − 15.74◦ 0.243 − 125.93◦
Find the input return losses of the attenuator when the output port is terminated
with a load of (a) 50 , (b) 100 and (c) 10 .
Solution
(a) For output load termination of 50 , the input return loss of the attenuator in
dB is

1 1
RL(in dB) 20 log 20 log 12.288 dB
|S11 | 0.243
(b) For output load termination of 100 , the load reflection coefficient is
100 − 50 1
ΓL 0◦
100 + 50 3
For load reflection coefficient Γ L , the input reflection coefficient ΓIN is given
by
S12 S21 Γ L
ΓIN S11 +
1 − S22 Γ L
Putting the S-parameter values and Γ L , the input return loss in dB is given by
S12 S21 Γ L
ΓIN S11 +
1 − S22 Γ L
0.706 − 15.74◦ × 0.706 − 15.74◦ × 0.333 0◦

ΓIN 0.243 − 125.93◦ +
1 − 0.243 − 125.93◦ × 0.333 0◦
ΓIN 0.288 − 92.53◦
Therefore, the input return loss in dB is

1 1
20 log 20 log 10.824 dB
|Γ L | 0.288
(c) For output load termination of 10 , the load reflection coefficient is
10 − 50 2
ΓL − 0◦
10 + 50 3
For load reflection coefficient Γ L , the input reflection coefficient ΓIN is given by
S12 S21 Γ L
ΓIN S11 +
1 − S22 Γ L
Putting the S-parameter values and Γ L , the input return loss in dB is given by
S12 S21 Γ L
ΓIN S11 +
1 − S22 Γ L
0.706 − 15.74◦ × 0.706 − 15.74◦ × 0.667 0◦

ΓIN 0.243 − 125.93◦ +
1 − 0.243 − 125.93◦ × 0.667 0◦
ΓIN 0.480 6.396◦
Therefore, the input return loss in dB is

1 1
20 log 20 log 6.384 dB
|Γ L | 0.480
Example 9.4 Calculate S-parameter matrix of the π network as shown in Fig. 9.37a
in a 50 system, and also compute attenuation and return losses of the network.
Solution
The network is symmetrical. Thus, the input and output reflection coefficients for a
50 system are given by
(100||50 + 50)||100 − 50
S11 S22 −0.048 0.048 180◦
(100||50 + 50)||100 + 50
(a) (b)
50 Ω (100|| 50+50) Ω
100Ω 100Ω 100Ω
Fig. 9.37 A resistive π -type network
The forward and reverse transmission coefficients are given by

vo
S12 S21
vs /2
Equivalent circuit is shown in Fig. 9.37b to calculate the output voltage. The
output voltage is given by
2vs 100||50
vo
3 (100||50 + 50) + 100||50
Thus,
vo 4 100||50
S12 S21 0.381 0◦
vs /2 3 (100||50 + 50) + 100||50
Thus S-parameter matrix of the network is

S11 S12 0.048 180◦ 0.381 0◦

S21 S22 0.381 0◦ 0.048 180◦
The attenuation of the network is
20 log|S21 | 20 log(0.381) 8.383 dB
Input and output return losses are
−20 log|S11 | −20 log|S11 | −20 log(0.048) 26.44 dB
Example 9.5 Find the resistance values of a π - and TEE-type resistive matched
attenuator for maximum attenuation.
Solution
For a π -type resistive matched attenuator, the values of resistors are given by (9.19a)
and (9.19b) as
(a) (b)
Fig. 9.38 π - and TEE-matched attenuator circuits for maximum attenuation
√
(N − 1) N + 1
R1 Ro √ and R2 Ro √
4 N N −1
Here, N 1/|S21 |2 is the attenuation. For maximum attenuation, N is very high.

Thus,
√
(1 − 1/N ) 1 + 1/ N
R1 Ro √ ∝ and R2 Ro √ Ro
4/ N 1 − 1/ N
The π -type resistive attenuator circuit for maximum attenuation is shown in

Fig. 9.38a.
For a TEE-type resistive matched attenuator, the values of resistors are given by
(9.19a) and (9.19b) as
√
N −1 √
2 N
R1 Ro √ and R2 Ro
N +1 (N − 1)
Here, N 1/|S21 |2 is the attenuation. N is very high for maximum attenuation.

Thus,
√
1 − 1/ N √
2/ N
R1 Ro √ Ro and R2 Ro (1 − 1/N ) 0
1 + 1/ N
The TEE-type resistive attenuator circuit for maximum attenuation is shown in

Fig. 9.38b.
Example 9.6 Find the attenuation and return losses of a quadrature hybrid attenuator
at zero-bias condition in case any one of the two p-i-n diodes becomes (a) RF shorted
and (b) RF open (assume diode behaves as perfect RF open at no-bias condition).
Fig. 9.39 Quadrature hybrid 1 2

matched attenuator
configuration
D1 D2
Solution
Attenuation and return losses of the 3-dB quadrature hybrid matched attenuator as
shown in Fig. 9.39 are given by (9.21b, 9.22b)

2
RL(in dB) 20 log10
|Γ1 − Γ2 |

2
|Γ1 + Γ2 |
Here, Γ1 and Γ2 are the reflection coefficients of the diodes D1 and D2 , respec-
tively.
(a) Suppose diode D1 becomes RF shorted, and thus, Γ1 −1. Under no DC bias
condition, impedance of the diode D2 is very high and can be considered as RF
open. Thus, Γ2 +1.

2
Return losses(in dB) 20 log10 0 dB
|−1 − 1|

2
Attenuation(in dB) 20 log10 ∝
|−1 + 1|
This result shows that full RF power will be reflected to the input port and it will
provide maximum attenuation.
(b) Suppose diode D1 becomes RF open, and thus, Γ1 +1. Under no DC bias
condition, impedance of the diode D2 is very high and can be considered as RF
open. Thus, Γ2 +1.

2
Return losses(in dB) 20 log10 ∝
|+1 − 1|
(a)
1 2 (b)
RS
D1 D2 Ri Ci

2
Attenuation(in dB) 20 log10 0 dB
|+1 + 1|
This result shows that in case one diode is RF open, loss of the attenuator is zero
at no-bias condition.
Example 9.7 Compute and plot attenuation of a quadrature hybrid balanced attenu-
ator as shown in Fig. 9.40a at operating frequency 10 GHz with respect to intrinsic
resistance Ri of the diode over the range of 20–100 . Consider equivalent circuit of
the p-i-n diodes as shown in Fig. 9.40b with RS 5 , C i 0.10 pF.
Solution
The RF impedance of the diodes is given by
Z T (Ri ||Ci ) + Rs
Diodes are batch matched and thus

(Ri ||Ci ) + Rs − Ro
Γ1 Γ2 Γ
(Ri ||Ci ) + Rs + Ro
or
1/2
(Ri + Rs − Ro )2 + (2π f Ci Ri (Rs − Ro ))2
|Γ |
(Ri + Rs + Ro )2 + (2π f Ci Ri (Rs + Ro ))2
Putting R S 5 , Ro 50 , f 10 GHz
1/2
(Ri − 45)2 + (0.28274Ri )2
|Γ |
(Ri + 55)2 + (0.34557Ri )2
Thus, attenuation of the attenuator in dB is given by

43.1 Ω
20 18.11 dB
15
Attn
(dB)
10
5
10 20 30 40 50 60 70 80 90 100
Ri (Ω)
or,

1 (Ri + 55)2 + (0.34557Ri )2
Attenuation (in dB) 20 log10 10 log10
|Γ | (Ri − 45)2 + (0.28274Ri )2
Figure 9.41 shows the plot of attenuation in dB with the variation of resistance
Ri .
Here, it is to be noted that (a) the maximum attenuation is only 18.11 dB. The
lower value of maximum attenuation is due to the finite capacitive reactance of
the diode. (b) The maximum attenuation is at Ri 43.1 instead of 50 . This
is due to the effects of diode series resistance of 5 and capacitive reactance. (c)
However, the return losses of the attenuator are ideally infinite in spite of the parasitic
series resistance and capacitive reactance of the diodes. This is due to the balanced
attenuator configuration with matched pair diodes.
Example 9.8 Find the attainable maximum attenuation of the balanced attenuator in
Example 9.6 by optimizing the bias-dependent resistor value Ri . What will be the
value of Ri at this condition? What will be the attenuation of the attenuator if one
diode becomes failed as open at the maximum attenuation bias condition?
Solution
In Fig. 9.41, it is shown that the maximum attainable attenuation is 18.11 dB. At the
maximum attenuation condition, Ri 43.1 .
In case of diode D1 failed as open, the reflection coefficient of the diode is
Γ1 +1
For the diode D2 , the reflection coefficient is

(Ri ||Ci ) + Rs − Ro
Γ2
(Ri ||Ci ) + Rs + Ro
Therefore, the attenuation of the attenuator is

2
|1 + Γ2 |
or
1/2
(2Ri + 2Rs )2 + (4π f Ci Ri Rs )2
|1 + Γ2 |
(Ri + Rs + Ro )2 + (2π f Ci Ri (Rs + Ro ))2

1/2
(Ri + 55)2 + (0.34557 Ri )2
Attenuation (in dB) 20 log10 2
(2Ri + 10)2 + (0.06283 Ri )2
Putting Ri 43 corresponding to maximum attenuation bias condition:
Attenuation (in dB) 6.286 dB
Example 9.9 An attenuator consists of a p-i-n diode. Current-dependent RF resis-

tance Rd and diode current–voltage I d –V d relationship of the p-i-n diode are

aT Vd − E b
Rd and Id bT × exp
2
Id cT
temperature coefficient of attenuation of the attenuator.
Solution
For achieving zero temperature coefficient of attenuation of the attenuator using the
p-i-n diodes, the RF resistance of the diodes should remain temperature-independent.
Thus,
∂ Rd a aT ∂ Id
− 2 0 (9.32a)
∂T Id Id ∂ T
or
∂ Id Id
(9.32b)
∂T T
Suppose the diode is biased through a bias resistance Rb and supply voltage VS .
VS Id Rb + Vd (9.32c)
VS
Rb
λo/4 λo/4 λo/4

ZO
ZO

VS − I d Rb − E b
Id bT × exp
2
(9.32d)
cT

∂ Id Id Rb 2Id Id (VS − Id Rb − E b )
1+ − (9.32e)
∂T cT T cT 2
Suppose for zero temperature coefficient VS VOPT and Id Ido . Putting these

Ido Ido Rb 2Ido Ido (VOPT − Ido Rb − E b )
1+ − (9.32f)
T cT T cT 2
of RF resistance as
VOPT E b + cT (9.32g)
Example 9.10 An attenuator as shown in Fig. 9.42 consists of n-numbers of batch

matched p-i-n diodes in shunt configuration. Current-dependent RF resistance, Rd ,
and current–voltage (Id − Vd ) relationship of the p-i-n diodes are

aT Vd − E b
Rd and Id bT × exp
2
Id cT
Solution
Thus,
∂ Rd a aT ∂ Id
− 2 0 (9.33a)
∂T Id Id ∂ T
or
∂ Id Id
(9.33b)
∂T T
supply voltage VS . Here, DC current Id flows through each diode, and voltage across
VS n Id Rb + Vd (9.33c)

VS − n I d Rb − E b
Id bT × exp
2
(9.33d)
cT
Differentiating (9.33d) w.r.t. T ,

∂ Id n Id Rb 2Id Id (VS − n Id Rb − E b )
1+ − (9.33e)
∂T cT T cT 2
Suppose, for zero temperature coefficient, VS VOPT and Id Ido . Putting these

Ido n Ido Rb 2Id Ido (VOPT − n Ido Rb − E b )
1+ − (9.33f)
T cT T cT 2
of RF resistance as
VOPT E b + cT (9.33g)
This shows that the optimum supply voltage is independent of the number of
diodes connected in parallel DC path.
Example 9.11 An attenuator as shown in Fig. 9.43 consists of two matched pair p-
i-n diodes connected in DC series path. Current-dependent RF resistance, Rd , and
current–voltage (Id − Vd ) relationship of the p-i-n diodes are
VS
Rb
ZO
ZO
D1 D2

aT Vd − E b
Rd and Id bT 2 × exp
Id cT
Solution
Thus,
∂ Rd a aT ∂ Id
− 2 0 (9.34a)
∂T Id Id ∂ T
or
∂ Id Id
(9.34b)
∂T T
supply voltage VS . Here, DC current Id flows through each diode and voltage across
VS Id Rb + 2Vd (9.34c)

VS − Id Rb − 2E b
Id bT 2 × exp (9.34d)
2cT
Differentiating (9.34d) w.r.t. T ,

∂ Id Id Rb 2Id Id (VS − Id Rb − 2E b )
1+ − (9.34e)
∂T 2cT T 2cT 2
Suppose, for zero temperature coefficient, VS VOPT and Id Ido . Putting these

Ido Ido Rb 2Ido Ido (VOPT − Ido Rb − 2E b )
1+ − (9.34f)
T 2cT T 2cT 2
of RF resistance as
VOPT 2(E b + cT ) (9.34g)
This shows that the optimum supply voltage for two DC series-connected diodes
is two times of the optimum supply voltage of one diode. In general, optimum supply
voltage for n-numbers of series-connected diodes is n-times of the optimum voltage
required for one diode.
References
1. Caverly RH, Hiller G (1993) Temperature insensitive PIN diode attenuators. Appl Microw
Summer 88
2. Caverly RH, Hiller G (1995) Temperature effect on PIN diode forward bias resistance. Solid
State Electron 38(11):1879–1885
3. Bera SC, Bharadhwaj PS (2005) Insight into PIN diode behavior leads to improved control
circuit. IEEE Trans Circ Syst-II 52(1):1–4
4. Bera SC, Singh RV, Garg VK (2005) A temperature dependent PIN diode model leads to simple
temperature invariant attenuator circuit. Microw J 48(02):104–116
5. Bera SC, Singh RV, Garg VK, Sharma SB (2007) Optimum bias load-line compensates tempera-
ture variation of junction diode’s RF resistance. IEEE Trans Microw Theor Tech 55(2):215–221
6. Bera SC, Bharadhwaj PS, Singh RV (2009) Control circuit for diode based RF circuits.
EP1652020, 2008; IN229296, 2009; US7538612, 2009; JP4851792, 2011
7. Bera SC, Singh RV, Garg VK (2009) Driver circuits for temperature-invariant performance of
junction diodes. IET Circ Dev Syst 3(4):143–152
8. Tajima Y, Tsukii T, Mozzi R, Tong E, Hanes L, Wrona B (1982) GaAs monolithic wideband
(2–18 GHz) variable attenuators. IEEE MTT-S Dig 479–481
9. Mondal JP, Milnes AG, Oakes JG (1985) Theoretical analysis for microwave T- and π -type
attenuator circuits using MESFETs. Int J Electron 58(2):231–247
10. Lizama G, Andrade T, Benton R (1987) 1–6 GHz GaAs MMIC linear attenuator with integral
drivers. IEEE Microw mm-wave Monolithic Circ Symp Digest 105–107
11. Bachert PS (1996) FET attenuator, 0–1 GHz. Appl Microw Wirel 57–64
12. Dindo S, Meierer R, North R (1997) Computer-aided design of MMIC variable attenuators.
Microw J 134–140
13. Daoud SM, Shastry PN (2005) A novel wideband MMIC voltage controlled attenuator with a
bandpass filter topology. 13th GaAs Symp 629–632
Chapter 10
Microwave Phase Shifters
Abstract This chapter presents design, analysis and applications of various

microwave phase shifters based on varactor diodes, p-i-n diodes and switched delay
lines. Design and analysis of continuous variable and constant insertion loss of four-
quadrant microwave phase shifters based on p-i-n diodes are described. The chapter
also describes control circuit for temperature-insensitive stepwise variable phase
shifters.
10.1 Introduction
Phase shifter provides known amount of controllable phase shift when RF/microwave
signal passes through it. Variable phase shifter is a linear two-port device, which
changes the phase of the signal in response to an external analog or digital command
[1–10]. It is realized in its simplest form by connecting a reactive element in series
or shunt with a transmission line [3, 4]. Phase shifters are used for the realization of
phased array antenna systems, linearizers for high power amplifiers, various types of
phase modulators, etc. True time delay (TTD) lines and phase shifters (PS) are the
basic functional blocks for complex signal processing used in variety of applications
including arbitrary waveform generation, filtering, antenna beamforming and recon-
figuration ability in communication systems. Important performance parameters for
a phase shifter are frequency range, total phase shift, accuracy, resolution, bandwidth,
insertion loss, port return losses, PM-to-AM conversion, power handling, switching
speed, etc.
10.2 Phase Shift and Time Delay
Under sinusoidal excitation of angular frequency ω, steady-state voltage and current

wave along a transmission line at l is given by (7.3, 7.8)

280 10 Microwave Phase Shifters

v(l, t) Re (V + e−γ l + V − e+γ l )eiωt (10.1a)
+
V −γ l V − +γ l iωt
i(l, t) Re e − e e (10.1b)
Zo Zo
where γ is the propagation constant and is a complex number, which can be expressed
as
γ α + jβ (10.2)
Here, β is the phase constant and is given by β ω/c 2π/λg . The real part α is
known as attenuation constant and expressed in neper/unit length, and the imaginary
part β is the phase constant and expressed in radian/unit length. The attenuation
constant α determines how the signal amplitude reduces as it propagates along the
transmission line. The phase constant β determines the difference of signal phase
in between the input end and the point of interest. A signal experiences a phase
shift of 360° (2π radians) when it travels through a transmission line of length β.
Phase shift may also be seen as time delay. As a sinusoidal signal travels through a
transmission line, it experiences time delay and associated phase shift as shown in
Fig. 10.1. Suppose a signal of frequency f ( ω/2π ) travelled through a distance x
with propagation velocity v p , then time delay will be
x x βx
Time delay (10.3)
vp f λg ω
Therefore, RF/microwave phase shifters can be realized by switching several

transmission lines of different physical lengths. Phase shift of 360° (2π radians) is
equivalent to delay of one time period, T 1/ f . Time delay can be expressed in
terms of phase shift (Φ) as
Φ
Time delay Φ in radian (10.4)
2π f
Phase shift Φ experienced by a signal travelling through a distance l can be

expressed as
l
Φ 2π (10.5)
λg
Thus by diverting a signal from one path to another of different lengths, phase shift
can be obtained. In this case, the phase shift is dependent on guide wavelength λg .
Therefore, phase shift value obtained by changing the line length deviates linearly
from the intended value as the signal frequency deviates in either direction from
10.2 Phase Shift and Time Delay 281
Fig. 10.1 Phase shift/delay experienced by a sinusoidal wave travelling through a transmission line
the nominal centre frequency. However, time delay will remain same as the signal
frequency changes from the centre frequency. Therefore, switching transmission
lines, constant phase shift cannot be obtained over a wide frequency band. Some
other techniques are used for broadband phase shifter that will be discussed in the
following sections.
10.3 Types of Phase Shifter
There are two distinct types of phase shifters with two distinct characteristics as
shown in Fig. 10.2. In both the cases, insertion phase in each setting is linearly
proportional to the frequency. In the first case, Fig. 10.2a, insertion phase slope
changes from one setting to another which implies that time delay is different from
one setting to another. Therefore, phase shifters of this category are also known as
time shifter. In the second case, Fig. 10.2b, insertion phase slope is constant for all the
phase setting, i.e. phase with respect to frequency characteristics which are parallel,
(a) (b)
Fig. 10.2 Phase shift with frequency, relative phase shift a changes with frequency, b constant over
the frequency
and implies that time delay remains constant from one setting to another. Therefore,
this type of phase shifter changes the phase of the incoming signal maintaining
constant time delay.
The first types of phase shifters are generally realized by switching different
lengths of transmission line. Thus, these types of phase shifters are inherently wide
bandwidth, limited only by the limitation of switching circuit’s bandwidth. On the
other hand, special networks are required for the second types of phase shifters to
obtain the desired phase characteristic. These types of phase shifters are inherently
narrowband in nature. Wider bandwidths are obtainable but at a higher complexity
of the networks.
At a single frequency, both types of phase shifters can provide identical phase shift.
When the frequency is varied, difference between these two types of phase shifters
becomes apparent. Therefore, for very narrowband applications, both types of phase
shifters can perform the same function. However, for broadband applications, types
of phase shifters are to be properly chosen depending upon the applications.
Depending on the control mechanism, phase shifters can be categorized into two
types, analog and digital phase shifters. Analog phase shifters are voltage-variable
devices offering continuously variable phase shift. The phase control is achieved
by analog control voltage. In general, the control voltage varies the capacitance of
varactor diodes to change the reactance of the circuit for achieving variable phase shift
of the RF/microwave signal. Digital phase shifters are realized by using p-i-n diodes
or FETs. The diodes and FETs are used as variable impedance or high-frequency
switches to achieve desired phase shifter performance.
10.4 Realization of Phase Shifters 283
10.4 Realization of Phase Shifters
Depending on the characteristics of the phase shifters and control signals, different
techniques are used to realize it. The subsequent sections will provide the design and
realization details of various phase shifters.
10.4.1 Switched Transmission Line Phase Shifter
Switched transmission line phase shifter design is the most straightforward because
it uses the time delay difference by switching the signal path in between two trans-
mission lines to provide the desired phase shift. The switching elements used for the
digital phase shifters may be mechanical or electronic. In case of electronic digital
phase shifters, p-i-n diodes, field effect transistors (FET), or microelectromechan-
ical systems (MEMS) are used. Among all the switching devices, p-i-n diodes are
commonly used in phase shifters for their relatively simple bias circuit, high-speed
switching and acceptable lower loss. As discussed in the previous chapter, the change
of RF resistance from few ohms to several kilo-ohms of a p-i-n diode from its forward
bias to reverse bias makes it suitable as RF switching device.
The standard switched-line phase shifter switches transmission line segments of
different electrical lengths to obtain predetermined amount of phase shift. Thus, the
phase shift is dependent only on the lengths of the transmission lines used. One of the
two transmission lines is used as reference line, and the other is used as a delay line.
The most important characteristic of the switched transmission line phase shifter is
that the phase shift is approximately a linear function of frequency, which means
phase shift is not constant over the operating frequency range; however, time shift is
constant over the frequency range from one phase shift setting to another. Here, the
phase shift created is dependent only by the length of the transmission lines, making
the phase shifter very stable over time and temperature. The temperature-dependent
RF resistances of the p-i-n diodes affect the insertion loss of the phase shifter but it
does not affect the phase shift.
10.4.1.1 Design of Switched-Line Phase Shifter
Figure 10.3 shows basic configuration of the switched-line phase shifter with
RF source of impedance Z o and also loads of impedance Z o . Here, two single-
pole double-throw (SPDT) switches are used to select one of the two transmis-
sion lines at a time, either reference line of length L R or delay line of length L 1 .
The switches are mechanical or p-i-n diode/transistor/MEM-based electronic RF
switches. Practically, isolation of the RF switches should be better very high (prac-
tically >25 dB) to avoid phase errors. The insertion loss of the phase shifter is the
combination of losses due to the switches and transmission line. Insertion loss vari-
Fig. 10.3 Basic configuration of switched-line phase shifter
Fig. 10.4 Switched transmission line digital phase shifter
ation from one phase shift to another is equal to insertion loss difference in between
the reference line and delay line.
The phase shift of the switched transmission line phase shifter is given by (10.5)
(L 1 − L R ) (L 1 − L R )
Φ( f ) 2π 2π f (10.6)
λg vp
Here, λg is the guide wavelength at a particular operating f v p /λg . From

(10.6), it is clear that the phase shift value deviates from the intended value linearly
as the frequency of the signal deviates in either direction from the nominal (centre)
frequency. To obtain phase shift of 180°, 90° and 45°, the required path length L 1
is a half, quarter and one-eight wavelength longer than the reference line length L R ,
respectively.
Schematic configuration of a switched transmission line 3-bit digital phase shifter
is shown in Fig. 10.4. Here, three switched-line phase shifters are connected in
cascade. Using proper combination of ON/OFF states of four switches (SW-1 to
SW-4), a 3-bit phase shifter can provide discrete phase shifts from 0° to 360° in
steps of 360◦ /8 45◦ . By connecting more number of switched-line phase shifter
in cascade configuration, step size of the phase shifter can be reduced.
10.4.2 Varactor Diode-Based Analog Phase Shifters
Phase shift of an analog phase shifter provides predetermined continuously variable

phase shift depending on the applied analog control voltage. Thus, analog phase
shifter provides very high-resolution variable phase shift. Most commonly, varactor
diodes are used as control element for achieving continuously variable phase shift.
Capacitance of reverse-biased varactor diodes is used as variable reactance for analog
phase shifters [3, 4].
Configuration of an analog phase shifter using varactor diode as variable reactance
is shown in Fig. 10.5. Two output ports (port-2 and port-3) of a 3-dB 90° hybrid
coupler are terminated with the varactor diodes to ground. Bandwidth of these types of
phase shifter depends on the bandwidth of the quadrature coupler itself. The coupler
can be a Lange coupler, branch line hybrid coupler, or a rat-race coupler on microstrip
or stripline configuration. The coupler splits the input signal at port-1 to two equal
amplitudes with a phase difference of 90° and incident on the port-2 and port-3. The
incident signals are reflected from the diodes and combined at the isolated port (port-
4) which is the output port of the phase shifter. Considering identical characteristic
of both the diodes (matched pair), the magnitudes and angles of reflected signals due
to the reflections from the diodes are equal. These two reflected signals will combine
at the output port and cancel at the input port. Thus, the transmission coefficient of
the phase shifter is given by (using 9.22a)

d (2θ − 90)◦ +
d (2θ − 90)◦
S21 (10.7a)
2
Fig. 10.5 Analog phase shifter using varactor diodes

Fig. 10.6 phase characteristic of the varactor diode-based voltage-variable phase shifter for three
different settings of the capacitance

d (2θ − 90)◦ (10.7b)
Here, θ is the insertion phase of the signal in between port-1 and port-2.
d is
the reflection coefficient of the diodes with respect to reference impedance Z o and
is given by
Zd − Zo

d (10.8a)
Zd + Zo
Here, Z d is the varactor diode impedance as shown in Fig. 10.5. Neglecting the
effect of package inductance L p , capacitance C p and reverse-bias resistance R p , the
equivalent diode impedance is simplified to a series combination of voltage-variable
capacitance Ci and resistance Rs . Thus, reflection coefficient of the diode can be
written as
(Rs − Z o ) − j X c

d (10.8b)
(Rs + Z o ) − j X c
Here, X c 1/(2π f C j ). Therefore, the insertion phase of the phase shifter is

given by

Xc −X c
S21 tan−1 − tan−1 + (2θ − 90)◦ (10.9)
Z o − Rs Rs + Z o
The capacitance C j can be controlled by changing the reverse-bias voltage of the

varactor diodes. Typical phase shift relative to the insertion phase of the reflection-
type phase shifter is shown in Fig. 10.6. It provides a voltage-variable phase shift
with range from 0° close to 180°.
10.4.3 Four-Quadrant Continuously Variable Phase Shifter
Vector modulator-based continuously variable four-quadrant phase shifter can be

designed using p-i-n diodes [5–10]. Bias-controlled RF resistance of p-i-n diodes is
used to realize the phase shifter. Schematic block diagram of a phase shifter is shown
in Fig. 10.7. Here, input signal is divided into two equal amplitudes with phase
difference of 90° by a 3-dB 90° power divider. Both the signals are selectively atten-
uated and phase shifted by an amount of 0° or 180° using voltage/current-controlled
variable attenuator and finally combined by an in-phase power combiner. As shown
in the vector diagram in Fig. 10.8 that by continuously varying the amplitude of
both the vectors simultaneously and switching the phase of the vectors by 180°, four
quadrants, i.e. 360° phase shift can be achieved. As shown in the vector diagram,
the output signal (vo ) will be the vector addition of the two controlled signals v1 and
v2 . To maintain output signal level constant over the full phase shift range, both the
vectors need amplitude control (av1 and bv2 ). Thus, signal level at the output is given
by
Fig. 10.7 Block diagram of vector modulator phase shifter
Fig. 10.8 Vector modulator diagram for the four-quadrant phase shifter

vo (av1 )2 + (bv2 )2 (10.10)
Ideally, to achieve constant attenuation over the entire phase shift of 0°–360°,
the loss is 6 dB. 3 dB loss is due to the combination of two equally divided signals
with 90° phase shift, and another 3 dB loss is due to the selective control of both the
divided signals over the full phase shift range to achieve constant amplitude of the
combined signal at the output.
10.4.3.1 Variable Attenuator with 180° Phase Switching
The variable attenuator with 180° phase switching capability can be achieved from a
p-i-n diode-based reflection-type attenuator using 3-dB 90° coupler. The attenuator is
same as discussed in Chap. 9 and is shown in Fig. 10.9. Considering both the diodes
are having matched characteristic, the transmission coefficient of the attenuator is
given by (10.7b)
S21
pd (2θ − 90)◦ (10.11a)
Here, θ is the insertion phase of the signal in between input and coupled ports and
pd is the reflection coefficient of the ground-connected p-i-n diode with respect to

impedance Z o and is given by
Z pd − Z o

pd (10.11b)
Z pd + Z o
Fig. 10.9 Quadrature hybrid matched attenuator configuration with series DC path
Here, Z pd is the p-i-n diode’s impedance neglecting the package inductance L p ,

capacitance C p as shown in Fig. 10.9. Thus, reflection coefficient of the grounded
(Ri ||Ci + Rs ) − Z o

pd
(Ri ||Ci + Rs ) + Z o
or
(Ri + Rs − Z o ) − j2π f Ci Ri (Z o − Rs )

pd (10.11c)
(Ri + Rs + Z o ) + j2π f Ci Ri (Z o + Rs )
Combining (10.11a) and (10.11c), amplitude and phase of the transmission coef-
ficient can be written as
(Ri + Rs − Z o )2 + (2π f Ci Ri (Z o − Rs ))2

|S21 | (10.12a)
(Ri + Rs + Z o )2 + (2π f Ci Ri (Z o + Rs ))2

−1 4π f Ci Ri2 Z o
S21 − tan
(Ri + Rs + Z o )(Ri + Rs − Z o ) − (2π f Ci Ri )2 (Z o2 − Rs2 )
(10.12b)
Amplitude (in dB) and phase characteristics of the attenuator based on (10.12a)
and (10.12b) are shown in Fig. 10.10, over the diode’s resistance Ri 5 –2 k
considering f 4 GHz, Rs 3 and Ci 0.02 pF. It shows that phase of the
signal is switched by 180° when the diode resistance crosses the value equal to
the characteristic impedance Z o 50 Ω. The phase deviation from 180° is due
to the capacitive reactance of the diode. This can be eliminated by implementing
inductance in parallel with the diode. The value of the inductance will be such that
parallel resonance frequency will be the operating frequency of the attenuator. The
reflection-type p-i-n diode attenuator using 3-dB 90° coupler is used to achieve high
level of attenuation with phase reversal characteristic at its maximum attenuation
level.
10.4.3.2 Four-Quadrant Phase Shifter
Schematic RF circuit diagram of the four-quadrant phase shifter is shown in

Fig. 10.11. The most important circuit module is the variable attenuator with all
possible attenuation range and with the provision of 180° phase switching. To real-
ize this phase shifter, p-i-n diode based on two variable attenuators with 180° phase
switching is used. Input 3-dB 90° coupler can be implemented by Lange coupler to
achieve amplitude and phase performance over broad frequency band. Output 3-dB
Fig. 10.10 Amplitude and phase characteristics of the attenuator with bias-dependent diode resis-
tance
Fig. 10.11 Schematic circuit diagram of the four-quadrant phase shifter using p-i-n diodes
in-phase power combiner can be realized using Wilkinson-type power divider. Here,
input signal is divided into equal amplitudes in between the direct and coupled ports,
with a 90° phase difference. These two divided signals are selectively attenuated and
phase shifted by the variable attenuators and finally added by the Wilkinson power
divider. From Fig. 10.11, eliminating the common phase shift in both the signal arms,
the transmission coefficient of the phase shifter can be written as

pd1 v4s +
pd2 v4s − 90◦
S21 vs (10.13a)
2
or

pd1
pd2
S21 + − 90◦ (10.13b)
2 2
where
pd1 and
pd2 are the reflection coefficient given by (10.11c) of the grounded
p-i-n diodes in attn-1 and attn-2, respectively, as shown in Fig. 10.11.
Suppose

pd1 ρ pd1 θ pd1 and
pd2 ρ pd2 θ pd2 (10.13c)
Then,

2 ρ
2 21
ρ pd1 pd2 1
|S21 | + + ρ pd1 ρ pd2 cos θ (10.14a)
2 2 2
and

ρ pd2 sin θ
S21 tan−1 (10.14b)
ρ pd1 + ρ pd2 cos θ
where θ is the phase angle in between the two signals combined at the output port
of the phase shifter and is given by
θ θ pd2 − θ pd1 − 90◦ (10.14c)
Ideally, the phase angles θ pd1 and θ pd2 will be 0 or 180°. Therefore, angle θ
between the two signal vectors coming to the output of the phase shifter will be
always 90°. Thus, in ideal case putting θ 90◦ in (10.14a), the minimum insertion
loss is 3 dB. This loss is due to the 90° phase difference in between the two signals
combined at the output. The maximum loss will happen when one of the attenuators
will give maximum attenuation. Thus putting ρpd1 or ρpd1 0 in (10.14a), the
maximum attenuation becomes 6 dB. Therefore, to maintain constant amplitude for
the entire phase shift range of 360°, proper biasing to the attenuators is required to
keep fixed insertion loss of the phase shifter ideally at 6 dB.
The photograph of a S-band four-quadrant phase shifter is shown in Fig. 10.12.
The circuit elements are realized with MIC technology on 20-mil-thick alumina
substrate (εr 9.9) in microstrip line configuration. Lange couplers are used for
both the attenuators and input 3 dB 90° coupler. Wilkinson power combiner is used
for the output combiner. Beam-lead p-i-n diodes are used for its negligible parasitic
inductive and capacitive reactances at S-band frequency. Biasing scheme of the phase
shifter is shown in Fig. 10.12. Resistors R1 and R2 are used to set optimum open-
circuit voltage (VOPT ) to achieve temperature-invariant attenuation of the attenuators
[11, 12]. Variable resistors Rb1 and Rb2 are used to control the bias of both the
attenuators independently. Thus, adjusting the values of the resistors Rb1 and Rb2 ,
Fig. 10.12 Schematic phase shifter circuit with biasing scheme and photograph of a four-quadrant
continuously variable phase shifter
Fig. 10.13 Phase characteristic of the four-quadrant phase shifter for various settings of attenua-
tions of attn-1 and attn-2
any phase from 0 to 360° can be obtained as shown in Fig. 7.13 by maintaining the
insertion loss constant. Without any bias to the p-i-n diodes, practically the insertion
loss of the phase shifter is about 4.5 dB, and extra 1.5 dB is due to the loss in
power divider, combiner and attenuator circuits. In practice, it is possible to achieve
continuously variable phase shift with constant loss of 7.5 dB.
10.4.4 Four-Quadrant Digital Phase Shifter
Four-quadrant digitally controlled phase shifter can be realized using the vector
modulator configuration discussed in the previous section. Here, the bias voltages
to the diodes of both the attenuators are controlled digitally. Figure 10.14 shows the
digital bias control circuit for the phase shifter. Biases of the two attenuators are
controlled simultaneously by selecting the bias resistors using digitally controlled
analog switch (multiplexer) circuits. VOPT1 and VOPT2 are the optimum open-circuit
Fig. 10.14 Four-quadrant

digital phase shifter using
vector modulator
configuration
voltages of the p-i-n diode-based attenuator-1 and attenuator-2, respectively, to obtain

temperature-invariant attenuation of the attenuators that is required for achieving
temperature-invariant phase shift.
10.5 Applications of RF/Microwave Phase Shifter
RF and microwave phase shifters have many applications in communication systems

such as phase discriminators, linearization of high power amplifiers, beamforming
networks and phased array antennas.
Generally, the loaded line phase shifters are popularly used in phased array anten-
nas. In case of electronically steered antenna arrays, the phase shifters are used to
set proper phase of the signals from individual antenna elements to steer the antenna
beam in the desired direction without physically rotating the antenna. Each radiating
element requires a phase shifter that applies the necessary phase shift to steer the
antenna beam. The phase of each element is so adjusted that individual signals line
up in the desired direction. In this case, the phase shift settings should not change
with frequency to keep the beam in the desired direction over the range of operating
frequency band.
Array beamforming techniques are used to form simultaneously available multiple
beams. Array beamforming is used to achieve low side lobe level with high gain and
controlled beam width. In case of analog array beamforming system, systematically
cable lengths to the elements are varied to control the signal phase. In this case of
array beamforming technique, both phase and amplitude need to be controlled by
the individual signals to achieve adjustment of side lobe level and to steer nulls.
References
1. Garver RV (1969) 360° varactor linear phase modulator. IEEE Trans Microwave Theory Tech
17(3):137–146
2. Upshur JI, Geller BD (1990) Low-loss 360° X-band analog phase shifter. In: IEEE MTT-S
international microwave symposium digest, 8–10 May 1990
3. Ulriksson B (1979) Continuous varactor-diode phase shifter with optimum frequency response.
IEEE Transactions on Microwave Theory and Techniques, vol. MTT-27, pp 650–654, July 1979
4. Niehenke EC, DiMarco VV, Friedberg A, (1985) Linear analog Hyperabrupt varactor diode
phase shifters. IEEE MTT-S Digest, 657–660
5. Yoo TW, Song JH, Park MS (1997) 360° reflection-type analogue phase shifter implemented
with a single 90° branch-line coupler. Electron Lett 33(3):224–226
6. Liew YH, Joe J, Leong MS (1999) A Novel 360° analog phase shifter with linear voltage phase
relationship. Microwave Conf Asia Pac 1:17–20
7. Bera SC, Raval BM, Singh RV, Garg VK (2000) Electronically controlled continuously variable
four quadrant microwave phase shifter. In: 31st mid term symposium on emerging wireless
communication technology and systems. IETE, 8–9 Apr 2000
8. Lin C-S, Chang S-F, Hsiao W-C (2008) A full-360° reflection-type phase shifter with constant
insertion loss. IEEE Microwave Wirel Compon Lett 18(2):106–108
9. Lin C-S, Chang S-F, Chang C-C, Shu Y-H (2007) Design of a reflection-type phase shifter
with wide relative phase shift and constant insertion loss. Microwave Theory Tech IEEE Trans
55:1862–1868
10. Xiao MX, Cheung SW, Yuk TI (2009) A wideband 360° analog phase shifter design. In:
Proceedings of 22nd IEEE Canadian conference on electrical and computer engineering, pp
524–527
11. Bera SC, Bharadhwaj PS, Singh RV (2011) Control circuit for diode based RF circuits,
EP1652020, 2008; IN229296, 2009; US7538612, 2009; JP4851792
12. Bera SC, Singh RV, Garg VK (2009) Driver circuits for temperature-invariant performance of
junction diodes. IET Circuits Devices Syst 3(4):143–152
Chapter 11
Microwave Modulators
Abstract Direct carrier modulation at RF/microwave frequency avoids the use of

upconverter and downconverter assemblies reducing hardware complexity and cost.
This chapter presents design, analysis and applications of various direct microwave
carrier modulators. Here, the design and analysis of p-i-n diode and transistor-based
amplitude modulator, bi-phase-balanced modulator, I–Q vector modulator and other
phase modulators are discussed.
11.1 Introduction
Modulators are three-port networks. One of the ports is for modulating signal and
the other two ports are RF ports. One of the RF ports is for the un-modulated carrier,
i.e. local oscillator (LO) input and other is for the RF-modulated output. A complex
chain of downconverter, mixer, filter, and upconverter is used to design a conventional
modulator. Direct carrier modulation at RF/microwave frequency avoids the use of
upconverter and downconverter assemblies leading to less hardware complexity, cost
and compact realization in MMIC technology [1, 2]. In case of direct modulation
techniques, p-i-n diode or FET-based voltage/current-controlled RF resistance are
used to modulate the amplitude and phase of the carriers by using modulating signal
as control voltage/current of the device. Variable RF resistance of p-i-n diodes and
180° phase reversal of reflection-type attenuators around 50 makes the p-i-n diodes
suitable for various types of modulators. Most frequently used digital modulation
techniques are the discrete phase modulation, i.e. M-array phase shift keying (PSK).
BPSK and QPSK are the most popularly used digital phase modulations techniques
for wireless and satellite communication systems. In case of BPSK and QPSK mod-
ulators, p-i-n diodes are used as a switch with 180° phase switching characteristic by
using a 3-dB hybrid coupler. However, in the case of M-array PSK modulators, p-i-n
diodes are used as a variable attenuator with 180° phase switching characteristic in
reflection type configuration. I–Q modulator is the most important element for the
realization of broadband direct carrier digital modulators. Broadband microwave I–Q

296 11 Microwave Modulators
modulator modules typically cover an instantaneous bandwidth of up to 2-octaves,

e.g. 1–4 GHz. Such modules are usually based on diodes or microwave transistors
for amplitude and phase control.
11.2 Amplitude Modulators
Amplitude modulation (AM) has been the most historical and popular method for
transmitting information. It modifies the amplitude of a carrier signal in accordance
with a modulating signal. The simplest way to realize amplitude modulator is to
control insertion loss of a voltage/current-controlled attenuator using the modulating
signal as the control voltage/current. A basic circuit diagram of a p-i-n-diode-based
simple direct amplitude modulator circuit is shown in Fig. 11.1. Analog or digital
modulating signal is applied to the modulating input of the modulator. Modulating
input terminal of the modulator is isolated from the RF signal path by using high
impedance line and a low impedance radial stub of length λ/4 corresponds to the RF
frequency. Level of the modulating signal is so adjusted that at its maximum value the
RF resistance of the diode remains more than 50 to avoid the phase and attenuation
characteristic reversal. Another way to avoid phase and attenuation reversal is the use
of 50 RF resistors in series with the p-i-n diodes at the coupled (port-2) and direct
(port-3) ports of the coupler. The series 50 resistor will ensure the impedance of
both the ports more than 50 for any level of modulating signal. A DC bias for the p-
i-n diodes is required to obtain a specific attenuation level over which time-varying
modulating signal is superimposed. One important requirement for the amplitude
modulators is linearity. The modulator must translate the modulating signal with the
introduction of minimal distortion. One drawback of this simple amplitude modulator
circuit is that the amplitude of the modulated RF carrier is not linearly proportional
to the control voltage. Various linearization techniques are used to minimize the AM
signal distortion.
The p-i-n-diode-based amplitude modulator, as shown in Fig. 11.1, can be used
as a pulse amplitude modulation circuit, i.e. as a switching circuit for RF power
levels. Pulse modulation is achieved by applying forward bias up to its maximum
attenuation level. After attaining the maximum attenuation level, the RF power can
be pulsed “ON” by applying a high positive constant bias to drive the p-i-n diodes to
its low impedance state. At the end of the required pulse width, the DC bias voltage
is switched corresponding to the maximum attenuation level.
Another way of implementing the amplitude modulator is by summing the mod-
ulating signal into an automatic level control (ALC) loop. This type of simple mod-
ulator using p-i-n diode is used for controlling RF signal level in a closed-loop ALC
systems. In the ALC system, RF signal is sampled, detected and amplified to gen-
erate a control signal that is applied to the control input of the modulator to provide
constant RF level output. ALC-based amplitude modulation scheme offers better
linearity limited by the available ALC dynamic range, which is mainly governed
by the dynamic range of the detector. In this case, the maximum modulating signal
11.2 Amplitude Modulators 297
Fig. 11.1 Schematic diagram of p-i-n diode based direct amplitude modulator
frequency is lower compared to the open-loop alternatives due to the settling time of
the closed-loop ALC system. Where, the settling time of an ALC system is governed
by the loop stability requirement.
Considering the coupler as an ideal coupler and zero insertion phases from its
input to the coupled port, the S-parameter of the modulator of Fig. 11.1 is given by

0 − jΓT
[S] (11.1)
− jΓT 0
Here, ΓT is the voltage reflection coefficient of the diode used as reflection ter-
mination as shown in Fig. 11.1. From the equivalent circuit of the p-i-n diodes, the
magnitude of the transmission coefficient of the modulator can be written as (10.12a)

|S21 | |ΓT | (11.2)
Maximum frequency of operation will be governed by the effect of the parasitic

capacitance Ci . For higher microwave and mmwave frequency range, devices such
as MESFET, pHEMT and HBTs capable to work at higher frequency are preferable.
11.3 Phase Modulators
Most frequently used digital modulation techniques are the binary phase-shift key-
ing (BPSK), different variances of quadrature phase-shift keying (QPSK, OQPSK,
DQPSK), M-array phase-shift keying (M-PSK) and M-array quadrature amplitude
modulations (M-QAM). Among all the modulation techniques, BPSK and QPSK are
the most frequently used digital phase modulation techniques for satellite commu-
nication systems. Due to the advancement of high-efficiency low distortion power
amplifiers technology, higher order modulation techniques such as 8APSK, 16APSK
are also being used in satellite communication. Microwave transistors such as MES-
FET, pHEMT, HBT or p-i-n diodes are used for the realization of direct microwave
phase modulators [3–7]. Transistors are preferable to realize phase modulators
at higher microwave and mmwave frequency range for ease in realization using
MMIC technology [8–14]; whereas, p-i-n diodes are used at a comparatively lower
microwave frequency range for its simpler operation. Bi-phase modulator is the basic
element to realize BPSK, QPSK and other higher order phase modulators.
11.3.1 Bi-Phase Modulators
Bi-phase modulator is the basic phase modulator used to realize higher order phase
modulators. This is also directly used as binary phase-shift keying (BPSK) modulator.
Bi-phase modulators are also used to realize vector modulator, which is used in
communication systems. For example, in a linearizer for power amplifiers, a vector
modulator is preferred to generate simultaneous amplitude and phase nonlinearities
to compensate power amplifiers nonlinearities, instead of using a variable attenuator
cascaded with a variable phase shifter. A vector modulator is also widely used as
I–Q modulator.
For bi-phase modulator, a reflection-type variable resistance configuration is used
to control the attenuation level. The circuit diagram of a bi-phase modulator using p-i-
n diodes is the same as shown in Fig. 11.1. This is an analog reflection-type attenuator
(RTA), where p-i-n diodes are used as variable RF resistance. For higher microwave
frequency range, Schottky diodes are also used as a control device. For an ideal
bi-phase modulator using reflection-type attenuator, RF impedance should be pure
resistance and vary from zero to infinity to achieve all possible signal amplitudes with
0° and 180° phase shift. However, due to the p-i-n diodes parasitic series resistance
(Rs ) and parallel capacitance (Ci ), bi-phase modulators performance deviate with
the increase of frequency. The amplitude of the modulator is governed by (11.2),
whereas the phase of the signal is given by (10.12b)

S21 − tan
(Ri + Rs + Z o )(Ri + Rs − Z o ) − (2π f Ci Ri )2 Z o2 − Rs2
(11.3)
11.3 Phase Modulators 299
Fig. 11.2 Typical amplitude and phase characteristic of p-i-n-diode-based bi-phase modulator
Taking typical values of Rs 5 and Ci 0.2 pF for a p-i-n diode, the amplitude
and phase characteristic of the bi-phase modulator is shown in Fig. 11.2 at 4 GHz. The
plot shows that the insertion phase change deviates from its ideal value of 180° for
the RF resistance around 50 (just lower and higher than 50 ). Also, the insertion
loss is finite, i.e. deviates from the ideal value of 0 dB. One way to minimize the phase
deviation from its ideal value of 180° is the use of external inductance in parallel
with the diode, which minimizes the effect of parasitic capacitance. However, it is
possible only over a very narrow band of frequency. Another drawback of using p-i-n
diode at its low RF resistance is the requirement of high forward current. This leads
to the requirement of more DC power from the control signal.
Because of the near-zero DC control power requirement and better performance
at higher microwave frequency, transistors are preferable than p-i-n diodes for the
realization of bi-phase modulator. Another advantage of using transistors for the mod-
ulator is that it can be realized in a very small size using MMIC technology. Schematic
circuit diagram of a bi-phase modulator using transistor is shown in Fig. 11.3. In this
case also 3-dB 90° coupler is used to realize transistor-based RTA and the working
principle is the same as the p-i-n-diode-based bi-phase modulator. The control signal
is applied to the gate terminal of the transistors through high-value bias resistors.
Thus, the requirement of current from the control signal is very less for the entire
range of RF resistance. The cold FETs, i.e. FETs with no bias voltages are used as
variable resistors [3, 4]. The FET’s dynamic range of variable resistors is smaller
and parasitic elements are larger compared to that of diodes. Therefore, practically
amplitude and phase errors are introduced by the parasitic elements (feed induc-
tance, junction capacitances, etc.) that deviate from the required performance of
a modulator. To minimize the effect of parasitic components, the proper value of
inductance in series with the gate or drain and a resistance in parallel with the tran-
sistor is used.
A simple bi-phase modulator allows passing the carrier signal to the modulated
output port, known as carrier leakage. Thus, some of the transmitted power is shared
Fig. 11.3 Bi-phase modulator using transistors
by the un-modulated carrier which is undesirable. To alleviate the problem due to the
parasitic of the devices and to suppress the un-modulated carrier transmission from
the output of the modulator, balanced modulator configuration is used.
11.3.2 Bi-Phase-Balanced Modulators
The balanced structure is used for bi-phase modulators to overcome the problem of
amplitude and phase deviation due to the parasitic effects of p-i-n diode and tran-
sistors. The balanced or push–pull configuration also eliminates the carrier leakage
problem i.e. removes the carrier signal from the modulated output signal in addition
to the elimination of the amplitude and phase errors caused by the parasitic of the
used active devices. The balanced structure gives good constellation symmetry due
to the use of the second attenuator operated in anti-phase.
Schematic circuit diagrams for the balanced bi-phase modulators using p-i-n
diodes and transistors are shown in Figs. 11.4 and 11.5, respectively. Here, two
branches are used with 180° phase difference using two 3-dB 90° couplers. Com-
plementary baseband control signals (C and C) are applied to the control inputs of
the branches. Therefore, if transistors or p-i-n diodes of one arm provides more than
50 RF resistance then the diodes/transistors of the other arm will provide less than
50 RF resistance for achieving ideally equal amplitude but opposite phase signals
due to the reflection from the devices.
11.3 Phase Modulators 301
Fig. 11.4 Balanced bi-phase modulator using p-i-n diodes
Fig. 11.5 Balanced bi-phase modulator using transistors
The transmission coefficient of the balanced bi-phase modulator is given by
ΓT 1 v4s + ΓT 2 v4s −180◦

S21 vs (11.4a)
2
1
or, S21 (ΓT 1 − ΓT 2 ) (11.4b)
2
Here ΓT 1 and ΓT 2 are the reflection coefficients of the attenuators of direct and
complementary arms, respectively. Suppose, ΓC and ΓC are the reflection coefficients
of the attenuators at signal level C and complementary signal level C, respectively.
Therefore, if the direct path attenuator is having signal C and the complementary
path is having C then the transmission coefficient is
1
S21 ΓC − ΓC (11.4c)
2
Γbpm (say) (11.4d)
Whereas, if C and C are applied to the direct path attenuator and complementary
path, respectively, then the transmission coefficient of the modulator is given by
1 1
S21 ΓC − ΓC − ΓC − ΓC (11.4e)
2 2
−Γbpm (11.4f)
From (11.4d) and (11.4f), it is clear that there is 180° phase difference in between
the two states and this phase difference is independent of the amount of phase and
amplitude deviations from its ideal values. The 180° phase difference is coming from
the 90° phase difference from the input coupler and another 90° phase difference
coming from the output coupler. The phase change of 180° is solely determined by
the input and output couplers. Many broadband coupler technologies are available.
Therefore, broadband bi-phase modulators can be realized using this configuration.
Another advantage of the balanced bi-phase modulator is that this configuration
provides carrier suppression at the output, which is not possible from the simple
bi-phase modulator.
In comparison with the reflection-type attenuators of Figs. 11.1 and 11.3, the
balanced bi-phase modulator needs three additional 3-dB 90° couplers, two additional
active device (p-i-n
diodes or transistors) and an additional complementary baseband
control signal C . A larger area for the circuit is required to realize the transistor-
based configuration in MMIC due to increase in circuit complexity.
11.4 I–Q Vector Modulators
In communication systems, a vector modulator is an important element. In digital

communication applications, to generate I–Q modulated signals a vector modulator is
widely used. In case of microwave linearizer for power amplifier, a vector modulator
is used to generate simultaneously required amplitude and phase distortion to com-
pensate for power amplifier nonlinearity. Therefore, vector modulators can replace
11.4 I–Q Vector Modulators 303
the use of a variable attenuator cascaded with a variable phase shifter for linearizer
realization. Vector modulators are also used in phased-array antenna applications for
an adaptive beam-forming network.
Traditionally, vector modulators are designed by two main configurations for
microwave and mmwave applications. In the first approach, the RF power is divided
into two equal amplitudes and 90° phase shift. Then the signal is combined with a
3-dB in-phase power combiner after passing through two bi-phase modulators. This
configuration has inherent 3-dB insertion loss due to the addition of two signals with
90° phase difference.
In the second approach, a voltage/current-controlled attenuator cascaded with
a 360° voltage/current-controlled variable phase shifter is used. In this configura-
tion, there is no inherent insertion loss. However, this configuration needs a variable
attenuator with a constant phase and a variable phase shifter with constant inser-
tion loss, which is very difficult to design without complicated amplitude and phase
control techniques. In terms of realization, the first configuration is more attractive
especially in microwave and mmwave applications. The system is composed of a
bi-phase variable attenuator, which is already discussed in the previous section.
Block diagram of IQ modulator circuit using two bi-phase (BPSK) balanced mod-
ulators is shown in Fig. 11.6. Here, the RF signal is divided into two equal amplitude
outputs, with a 90° phase difference in between them using a 3-dB quadrature hybrid.
The in-phase or 0° channel is designated as the I channel and the quadrature or 90°
channel is designated as the Q channel. Each signal passes through a bi-phase mod-
ulator which controls the signal amplitude and selects 0° or 180° phase state for
both I and Q paths. The phase state of the bi-phase modulators determines the phase
quadrant of the output signal. The amplitude and phase of the resultant vector are
determined by the magnitude of each of the signals which are combined at the output.
Thus, by varying the amplitude of both the paths and selection of the either 0° or
180° phase state of the bi-phase modulator, the resultant can be placed anywhere
within the bounded dotted square area shown in Fig. 11.7.
The transmission coefficient (S21 ) of the I–Q vector modulator can be written as
Γbpm1 v4s + Γbpm2 v4s − 90◦

S21 vs (11.5a)
2
Γbpm1 Γbpm2
or, S21 + − 90◦ (11.5b)
2 2
Two bi-phase modulators are used for the vector modulator. In its simplest con-
figuration, the bi-phase modulators can be designed using single 3-dB 90° coupler
with p-i-n diode or transistor as shown in Figs. 11.1 and 11.3, respectively. There-
fore, overall three numbers of 3-dB 90° couplers and one Wilkinson-type equal power
combiner is required to implement the vector modulator. The configuration is exactly
the same as the 360° phase shifter as discussed in the previous chapter. However,
to implement the vector modulator using balanced bi-phase modulators, eight active
devices, nine hybrid couplers, one Wilkinson power combiner, and four baseband
signal channels are required. Thus, to realize the vector modulator using balanced
Fig. 11.6 I–Q modulator circuit using two bi-phase (BPSK) balanced modulator
Fig. 11.7 Vector signal

diagram of I-Q modulator
bi-phase modulator a large chip area is needed in monolithic microwave integrated

circuit (MMIC) technology. The 3-dB quadrature hybrid is an important element
to determine the operating frequency range of an I–Q vector modulator. It should
have a very low quadrature phase error, i.e. phase difference between the two output
deviations should be very small from its ideal value of 90°. To meet this requirement
over a broad frequency band, Lange coupler is preferable. For the in-phase output
combiner, a Wilkinson power combiner is used for its better phase and amplitude
balance performance. In addition to the couplers performances, I and Q channel’s
transmission line length must also be phase matched.
11.5 PSK Modulators 305
Table 11.1 Signal levels for

States I I Q Q
QPSK modulator
(0, 0) Low High Low High
(0, 1) Low High High Low
(1, 0) High Low Low High
(1, 1) High Low High Low
11.5 PSK Modulators
Balanced bi-phase modulator as shown in Figs. 11.4 and 11.5 using diodes and
transistors, respectively, can be used as BPSK modulator. For BPSK modulation,
there are same amplitudes with two phase states 0 and 180°. The two states are
represented as state-0 (C is low and C is high) and state-1 (C is high and C is low).
Here digital modulating signals are applied to the control input C and comple-
mentary signal input C. When C is in the high state and C is in the low state, the phase
delay of the signal path will be 180° compared with the condition when C is low and
C is high. It is required to generate complementary signal level from the modulating
signal. Thus, the bi-phase balanced modulator acts as BPSK modulator when the
modulating signal applied to its control input C and signal level complementary to
the modulating signal applied to the input C.
The I/Q vector modulator circuit as shown in Fig. 11.6 is the basic functional
block for realization of multilevel digital modulation schemes, for example QPSK,
M-APSK and M-QAM. There are four phase states with the same amplitude in the
case of QPSK modulation. The states are represented as states of (0, 0), (0, 1), (1, 0)
and (1, 1). The signal levels to the four control inputs are shown in Table 11.1.
In all the four states, the amplitude of the modulated signal remains the same only
if the signal phase of the four states corresponds to four quadrants.
11.6 QAM Modulators
Higher order modulations such as multilevel QAM schemes are required in high data
rate communication systems to utilize spectrum more effectively. QAM modulators
are also realized using the I/Q vector modulator circuit as shown in Fig. 11.6 [2].
In this case, each bi-phase modulator operates in different amplitude settings each
with 0 and 180° phase shift. For example, two amplitude settings are required for
16-QAM and eight amplitude settings are required for 256-QAM with each 0° and
180° phase offset. Therefore, driver circuits generate control voltages from digital
bits to control phase as well as the amplitude of the bi-phase modulator. Any devi-
ation in the amplitude of the bi-phase modulators, due to temperature variation or
inherent imbalance of the modulator, leads to the imperfection of constellation. To
solve the problem of temperature variation of amplitude in case of p-i-n diode bi-
phase modulator, optimum bias load line technique can be used for the driver circuit
realization.
References
1. Boveda A, Orilgoso F, Alonso JI (1993) A 0.7–3 GHz GaAs QPSK/QAM direct modulator.
IEEE J Solid State Circuits 28:1340–1349
2. Ashitiani AE, Nam S, d’Espona A, Lucyszyn S, Robertson ID (1998) Direct multilevel carrier
modulation using millimeter-wave balanced vector modulators. IEEE Trans Microw Theory
Tech 46:2611–2619
3. Lodhi T, Edgar DL, McLelland H, Ferguson S, Elgaid K, Stanley CR, Thayne IG (2000) A
77 GHz coplanar waveguide MMIC BPSK vector modulator realized using InP technology.
In: IEEE gallium arsenide integrated circuit symposium digest, pp 183–186
4. Chongcheawchamnan M, Bunnjaweht S, Kpogla D, Lee D, Robertson ID (2002) Microwave
I-Q vector modulator using a simple technique for compensation of FET parasitics. IEEE Trans
Microw Theory Tech 50:1642–1646
5. Chang H-Y, Wang H, Wang Y-C, Chao P-C, Chen C-H, Niu D-C (2002) A 45–90 GHz BPSK
modulator using HBT technology. In: Proceedings of Asia–Pacific microwave conference,
Kyoto, Japan, vol 1, pp 60–62
6. Chang H, Huang T, Wang H, Wang Y, Chao P, Chen C (2004) Broad-band HBT BPSK and IQ
modulator MMICs and millimeter-wave vector signal characterization. IEEE Trans Microw
Theory Tech 52(3):908–919
7. Alavi MS, Viswesaran A, Staszewski RB, de Vreede LCN, Long JR, Akhnoukh A (2011)
A 2 GHz digital I/Q modulator in 65 nm CMOS. In: Proceedings of IEEE Asian solid-state
circuits conference, pp 277–280
8. Devlin LM, Minnis BJ (1990) A versatile vector modulator design for MMIC. In: IEEE MTT-S
international microwave symposium digest, pp 519–522
9. Norris GB, Boire DC, St. Onge G, Wutke C, Barratt C, Coughlin W, Chickanosky J (1990) A
fully monolithic 4 to 18 GHz digital vector modulator. In: IEEE MTT-S international sympo-
sium digest, pp 789–792
10. Telliez I, Couturier A-M, Rumelhard C, Versnaeyen C, Champion P, Fayol D (1991) A compact
monolithic microwave demodulator-modulator for 64-QAM digital radio links. IEEE Trans
Microw Theory Tech 39:1947–1954
11. Gokdemir T, Nam S, Ashtiani AE, Robertson ID, Karacaoglu U (1998) Millimeter wave
monolithic-balanced BPSK modulator using a miniaturized backward-wave coupler. In: IEEE
MTT-S international symposium digest, pp 877–880
12. Desrosiers R, Cowles J, Hornbuckle C, Gutierrez-Aitken A, Becker J (1998) Monolithic 14 GHz
widerband InP HBT BPSK modulator. In: Proceedings of gallium arsenide integrated circuit
symposium, pp 135–138
13. Nam S, Shala N, Ang KS, Ashtiani AE, Gokdemir T, Robertson ID, Marsh SP (1999) Monolithic
millimeter-wave balanced bi-phase amplitude modulator in GaAs/InGaP HBT technology. In:
IEEE MTT-S international microwave symposium digest, vol 1, pp 243–2461
14. Ashtiani AE, Gokdemir T, Vilches A, Hu Z, Robertson ID, Marsh SP (2000) Monolithic
GaAs/InGaP HBT balanced vector modulator for millimeter-wave wireless systems. In: IEEE
radio frequency integrated circuits symposium digest, pp 187–190
Chapter 12
Amplitude Tilt Microwave Equalizers
Abstract Equalizers are used in cascade with other circuits and systems to improve
the overall gain flatness over its operating frequency range. This chapter presents
design and analysis of fixed and adjustable amplitude tilt microwave equalizers. Pas-
sive components, e.g. resistor-, capacitor- and inductor-based fixed slope equalizers
and p-i-n diode-based adjustable slope equalizers are discussed. Versatile equaliz-
ers with adjustable positive, negative slope and adjustable insertion loss are also
presented to compensate gain variation over frequency and temperature range of
microwave systems.
12.1 Introduction
Amplitude tilt equalizer provides predetermined gain/loss (|S 21 |) slope over the
operating frequency range [1–8]. It is mainly used in cascade with other circuits or
systems to improve the overall gain response over its operating frequency range.
Variable gain slope microwave equalizers are used for compensation of gain roll-off
of microwave and RF circuits and systems using the provision of slope adjustability.
To meet the increasing demand of broadband applications such as internet and
multimedia, microwave and RF broadband amplifiers are required. Due to the
typically falling-off gain of transistors with the increase of frequency, modern
microwave receivers uses equalizers to compensate for the excess passband negative
gain slope. In this case, equalizer with positive gain slope is needed to cascade
with the transistor-based amplifiers for achieving broadband gain performance.
Wideband travelling wave tube amplifiers (TWTAs) have parabolic gain response
with maximum gain at the centre frequency and fall-off gradually at the band edges,
sometimes by 11 dB or more. In this case, the solution is to use parabolic gain
equalizer with maximum loss at the centre frequency to get an overall flat response
over the required frequency band. Broadband pre-distortion linearizers require
equalizers with very high positive and also very high negative gain slopes to achieve
required amplitude and phase distortion over a wide operating frequency band.

308 12 Amplitude Tilt Microwave Equalizers
12.2 Different Configurations
There are several configurations of gain slope equalizer circuits which are based on
lumped resistors, capacitors and/or inductors connected as series and shunt elements
as shown in Fig. 12.1a [1, 2]. In case of multiple shunt networks configuration as
shown in Fig. 12.1b, the shunt networks are connected quarter-wavelength apart to
achieve better input and output port return losses. Both the equalizers provide positive
gain slopes with the increase of frequency. However, these equalizers are not much
suitable at microwave frequency range due to parasitic effects of the used lumped
elements. Substrate integrated waveguide (SIW) and photonic bandgap (PBG)-based
positive gain slope equalizers are also used. However, these are having disadvantages
of packaging due to their definite conductor pattern on the ground plane.
Most suitable microwave gain slope equalizers are based on combination
of lumped and distributed components. In these cases, multiple shunt elements con-
taining one or more fixed resistors and an open- or short-circuited transmission lines
are used to realize the equalizer as shown in Fig. 12.2. Here, shunt elements are
connected with separation of quarter-wavelength transmission line for better ports
return losses.
Example 12.1 Compute S-parameter of a shunt network consists of a resistance Ra

shunted by a transmission line (Z o ) of length l as shown in Fig. 12.3. Also plot
transmission coefficient and reflection coefficient in dB of the network over the
(a) R1 (b)
C1 R2 R1 R2 R3
L1 L1 L2 L3
Fig. 12.1 Configurations of amplitude tilt equalizers using lumped components
Fig. 12.2 Configurations of

amplitude tilt equalizers
using transmission line
elements
12.2 Different Configurations 309
Fig. 12.3 The shunt network consists of a resistance and a transmission line of Example 12.1
Fig. 12.4 The two port network consists of the shunt network of Example 12.1
frequency range of 1–20 GHz for Ra 30 and l λo /8, where λo is free space
wavelength at 10 GHz.
Solution
The shunt network consists of a transmission line of length l and a resistor Ra as
shown in Fig. 12.3. Using (7.79), putting Z L 0 the input impedance Z a of the
shunt network can be written as

2π
Z sn Ra + j Z o tan l
λ
Here, Z o is the characteristic impedance of the shunt transmission line. The S-

parameters of the shunt network can be determined as follows from (7.134) using
Fig. 12.4:
b1
S11 , when a2 0
a1
V−
1+ , when V2+ 0
V1
−Z o /(Z o + 2Z sn )
It is a symmetric network, therefore:
S11 S22 −Z o /(Z o + 2Z sn )

Similarly from (7.79) and using (7.55) and (7.136), S21 can be determined as
√
b2 V − Zo V2−
S21 √2 , when V2+ 0
a1 Z o V1+ V1+
2Z sn /(Z o + 2Z sn ) S12

⎡ ⎤
−Z o 2Z sn
S11 S12 ⎦, where Z sn Ra + j Z o tan 2π l
⎣
Z o +2Z sn Z o +2Z sn
S21 S22 2Z sn −Z o λ
The transmission coefficient of the shunt network is given by

2Z sn 2Ra + j2Z o tan 2π
λ
l
S21
2π
Z o + 2Z sn Z o + 2Ra + j2Z o tan λ l
Transmission coefficient in dB can be written as

4Ra2 + 4Z o2 tan2 2π
λ
l
20 × log(|S21 |) 10 × log

(Z o + 2Ra )2 + 4Z o2 tan2 2π
λ
l
Putting Ra 30 , Z o 50 , l λo /8, the transmission coefficient can be

written as

π
3600 + 10,000 tan2 40 f GHz
Transmission coefficient 10 × log
π dB
12,100 + 10,000 tan2 40 f GHz

π
36 + 100 tan2 40 f GHz
10 × log
π dB
121 + 100 tan2 40 f GHz
Similarly, the reflection coefficient in dB can be written as

2500
Reflection coefficient 10 × log
π dB
12,100 + 10,000 tan2 40 f GHz

25
10 × log
π dB
121 + 100 tan2 40 f GHz
The variation of transmission and reflection coefficients over the frequency range
of 1–20 GHz is plotted in Fig. 12.5 using the above expressions. This network
provides decrease of insertion loss with the increase of frequency.
12.3 Equalizer with Adjustable Amplitude Slope 311
Fig. 12.5 Transmission and reflection coefficients of Example 12.1
12.3 Equalizer with Adjustable Amplitude Slope
Frequency-dependent microwave impedances are used in shunt configuration to a

main transmission line to generate gain/loss slope over operating frequency range.
Practically, most useful microwave equalizers are configured for generating var-
ious types of gain slopes by adjusting one or more circuit elements to provide
monotonous negative and positive, parabolic and multi-peak gain slopes over the
operating frequency range [3–6]. For active equalizers, p-i-n diodes are used to vary
the impedances of the networks to achieve various types of gain slopes over the oper-
ating frequency range. A short-circuited shunt network consists of a transmission line
in between two resistors Ra and Rb are used as shown in Fig. 12.6. The length of the
transmission line is integer numbers (n) of one-eighth wavelength corresponding to
the centre frequency ( f o ) of operation. In case of adjustable gain slope requirements,
the resistors Ra and/or Rb is synthesized using p-i-n diodes. The input impedance
Z sn of the shunt network of Fig. 12.6 with transmission line length of nλo /8 is given
by (7.79):

o
Rb + j Z o tan nπλ
Z sn Ra + Z o
4λ o (12.1)
Z o + j Rb tan nπλ
4λ
Fig. 12.6 A short-circuited network consists of a transmission line of length nλo /8 in between two
resistors
Fig. 12.7 Equalizer with two shunt networks with separation of λo /4
Here, Z o is the shunt transmission line’s characteristic impedance, n is an integer

(1, 2, 3, …), λo and λ are the guide wavelengths that correspond to the operating
centre frequency f o and any other operating frequency f , respectively.
For Rb Z o , the value of the input impedance Z sn is Ra + Z o . This implies
that the input impedance Z sn is independent on frequency. However, for Rb Z o ,
depending upon the value of Rb , the shunt network’s input impedance may increase
or decrease with the frequency.
In practice, multiple numbers of shunt networks are used in a separation of λo /4
length transmission line for realization of equalizer to achieve more gain slope and
better port return losses over the interested frequency band. The schematic circuit
diagram of an equalizer with two identical shunt networks is shown in Fig. 12.7. The
ABCD parameter of the equalizer of Fig. 12.7 is given by the matrix multiplication
of the three elements: the first shunt network, the series transmission line of length
λo /4 and the second shunt network. Therefore, the ABCD parameter of the equalizer
with two shunt elements separated by the λo /4 transmission line is given by
⎡
o
o ⎤
1 0 cos πλ j Z o sin πλ 1 0

AC ⎣ 2λ 2λ
⎦
1

1 (12.2)
B D Z sn
1 j 1 sin πλo
Zo 2λ
cos πλo 2λ Z sn
1
Thus, transforming ABCD parameter matrix (12.2), the scattering parameters of

the equalizer can be written as

o Z2
o
− 2Z o
Z sn
cos πλ 2λ
− j Z 2o sin πλ
2λ
S11 S22
πλo sn

(12.3)
2 + Z sn cos 2λ + j 2 + Z sn + Z 2 sin πλ
2Z o 2Z o Z o2 o
2λ
sn
2
S21 S12

πλo (12.4)
cos πλ
Ro2
2 + 2Z o
Z sn 2λ
o
+ j 2+ 2Z o
Z sn
+ 2
Z sn
sin 2λ
Figure 12.8 shows S 21 response of the equalizer for n 1 over the frequency
range of 5–15 GHz for a fixed value of the resistor Ra 50 and different values of
the resistance Rb 15, 50 and 150 . For n 1, the length of the shunt transmission
line is λo /8 and the achieved gain slope monotonously increases (Rb < 50 ),
12.3 Equalizer with Adjustable Amplitude Slope 313
Fig. 12.8 Frequency response of the equalizer for n 1 and Ra 50 . Positive gain slope for Rb
15, zero gain slope for Rb 50 and negative gain slope for Rb 150
Fig. 12.9 Frequency response (|S 21 |) of the equalizer for n 2, and Ra 50 . Maximum at centre
frequency for Rb 15 , zero gain slope for Rb 50 and minimum at centre frequency for Rb
150
decreases (Rb > 50 ) or remains constant (Rb 50 ) over the frequency range. In
case of solid-state systems such as receivers, driver amplifiers and power amplifiers,
gain decreases with the increase of frequency due to inherent characteristic of the
solid-state devices. To compensate the gain slope of these systems, the equalizer
particularly with positive gain slope (Rb < 50 ) is used.
Frequency response (S 21 ) for n 2, i.e. for the length of the shunt transmission
line of λo /4 is shown in Fig. 12.9. It shows that the equalizer provides parabolic gain
slope over frequency range of 5–15 GHz. To compensate parabolic gain variation, this
type of equalizers can be used. For example, gains of broadband TWTAs are higher
at its centre frequency and decreases with the increase and decrease of frequency.
Fig. 12.10 Frequency response of the equalizer for n 3, Ra 50 . Maximum is near lower
frequency for Rb 15 , zero gain slope for Rb 50 and minimum is near higher frequency for
Rb 150
Here, the equalizer with shunt transmission line of length λo /4 and with Rb > 50
is suitable for the gain slope compensation.
Gain response with multiple peaks is required to equalize systems where several
circuits with different gain versus frequency responses are cascaded to achieve sys-
tem performance. Figures 12.10 and 12.11 show the other two types of gain slope
characteristics which correspond to n 3 and 4, i.e. for the length of shunt transmis-
sion line of 3λo /8 and λo /2, respectively. Thus, with the increase of n, i.e. length of
the transmission line of the shunt network, more and more number of peaks in the
gain response is achievable. It also shows that with the increase of length, same gain
slope is achieved over less frequency range that is required to equalize gain slope of
relatively narrow band systems.
To optimize gain flatness of various circuits and systems having different gain
slopes, slope adjustability of a gain slope equalizer is very much required. In practice,
at the design stage, it may not be known the requirement of amount as well as
whether positive or negative gain slope is required. Thus, equalizer with provision
of adjustable positive and negative gain slopes is preferable to adjust the gain slopes
during production phase. In these cases, resistor Rb is synthesized using p-i-n diodes.
Whereas, the cases where either adjustable positive or adjustable negative gain slopes
are required, p-i-n diodes are used to realize variable resistors Ra .
12.4 Equalizer with Adjustable Parabolic Gain Slope for Broadband MPM 315
Fig. 12.11 Frequency response of the equalizer for n 4, Ra 50 and Rb 15, 50 and 150
12.4 Equalizer with Adjustable Parabolic Gain Slope

for Broadband MPM
Microwave power module (MPM) uses short length TWT to achieve higher output
power with lesser gain. Remaining gain is realized by cascading solid-state ampli-
fiers (SSAs) at its input to make the overall system with smaller size and lesser
mass. Broadband TWTs suffer from parabolic gain slope, sometimes nearly 11 dB
with maximum gain around centre frequency. To achieve flat gain response over
the operating frequency range, gain slope equalizer is incorporated in the solid-state
amplifier cascaded with the TWT. An equalizer circuit having parabolic gain slope
with maximum loss at the centre frequency is required to serve this purpose. Circuit
diagram of such an equalizer is shown in Fig. 12.12. Here, the resistor Rb is fixed
but the resistor Ra is variable. The variable resistor Ra is synthesized using p-i-n
diode to achieve electronically adjustable gain slope. RF equivalent circuit of the
p-i-n diode is parallel combination of RF resistor rd and parasitic capacitor cd as
shown in Fig. 12.12. Figure 12.13 shows frequency response of the equalizer for n
4, i.e. length of the shunt transmission line of λo /2. The traces are for different
values of p-i-n diode resistance which corresponds to different bias currents of the
diodes, taking Cd 0.017 pF. The plots show that the parabolic gain slope up to
10 dB with minimum loss of about 2 dB at the edges is achievable with the change of
diode current by maintaining return losses better than 10 dB for all the settings. To
achieve more gain slope and/or better return losses, more number of identical shunt
networks with the separation of λo /4 transmission line length may be used.
Fig. 12.12 Equalizer circuit with p-i-n diodes as variable resistor to achieve adjustable either pos-
itive or negative gain slope over frequency
Fig. 12.13 Equalizer circuit with p-i-n diodes as variable resistor to achieve adjustable parabolic
gain slope over frequency for Ra 15
12.5 Equalizer with Adjustable Positive Gain Slopes

for Solid-State Circuits
Broadband solid-state RF and microwave systems such as receivers, power ampli-

fiers, etc. suffer from negative gain slope over operating frequency range due to the
inherent behaviour of the used solid-state devices. For satellite communication, the
allocated bandwidth for each frequency band is not very high. For example, the full
Ku-band downlink frequency band is from 10.7 to 12.75 GHz, i.e. about 2 GHz
bandwidth. However, the gain slope requirement may be very large for the Ku-band
subsystems which use many circuits in cascade configuration.
Circuit diagram of an equalizer operating over the frequency range of
10.7–12.8 GHz, to improve the gain flatness of solid-state subsystem, will be the
same as shown in Fig. 12.12 with different values of the circuit components. The
design goal of the equalizer is to provide positive gain slope over this frequency range
12.5 Equalizer with Adjustable Positive Gain Slopes for Solid-State Circuits 317
Fig. 12.14 Photograph of a Ku-band amplitude tilt equalizer
Fig. 12.15 Frequency response of the equalizer for n 7 and Rb 250
with minimum insertion loss at the highest frequency of operation, i.e. at 12.8 GHz,
where the gain of the subsystem is minimum. To meet this requirement, the length
of the shunt transmission line of the equalizer is selected as 7λo /8 (n 7), λo is
corresponding to the centre frequency of 11.75 GHz. Two identical shunt elements
are used with the separation of λo /4 to achieve better input and output port matching.
The equalizer is designed in microstrip configuration on a 10 mil-thick alumina sub-
strate (εr 9.9). Photograph of the equalizer is shown in Fig. 12.14. Thin film chip
resistors are used for Rb to perform at Ku-band frequency. Bended shunt transmis-
sion lines are used to make the circuit on a small size substrate of 0.5 mm × 0.5 mm.
Beam lead p-i-n diodes are used for their negligible parasitic inductance and very
small parasitic capacitance (only 0.017 pF). High-impedance λo /4 line with radial
stub is used to provide bias of the p-i-n diodes.
Frequency response of the equalizer is shown in Fig. 12.15. Equalizer provides
variable gain slope response with the variation of circuit current Id corresponding to
diode resistance 25, 40, 60, 100 and 200 for the fixed resistance values of Rb
250 . Return losses of the circuit are better than 10 dB over the entire bandwidth
for all the gain slope settings.
Fig. 12.16 Equalizer circuit with p-i-n diodes as variable resistor
12.6 Equalizer with Adjustable Positive as Well as Negative

Gain Slopes
In case of complex microwave systems and subsystems, it is required to incorporate

an equalizer which provides adjustable positive as well as negative gain slope over the
frequency of operation. Circuit diagram of such an equalizer is shown in Fig. 12.16.
Here, the resistors Ra are fixed but the resistors Rb are variable. The variable
resistors Rb are synthesized using p-i-n diodes to achieve electronically adjustable
gain slope. Figure 12.17 shows frequency response of the equalizer for n 7, i.e.
length of the shunt transmission line of 7λo /8 and Ra 40 . The traces are for
different values of p-i-n diode resistance corresponding to different bias current of
the diodes, taking Cd 0.017 pF. The traces show that the gain slopes of the equalizer
are positive as well as negative depending on the diode bias current. For all the gain
slope settings, the port return losses are better than 12 dB. In this case also, to increase
the maximum gain slope with better port impedance matching, i.e. return loss, more
numbers of shunt networks are to be used. Another way to increase the maximum
gain slope is the use of low value of resistors Ra . In this case, penalty will be the
poor port return losses.
12.7 Versatile Equalizer with Variable Gain Slope

and Insertion Loss
The gain of microwave circuits and systems are function of operating frequency as
well as its operating temperature. To compensate the gain variation over the frequency
as well as temperature, a versatile variable gain slope equalizer with the provision of
adjustable gain (i.e. insertion loss in this case) is required [7, 8]. To achieve adjustable
positive and negative gain slopes with the provision of adjustable insertion loss, both
12.7 Versatile Equalizer with Variable Gain Slope and Insertion Loss 319
Fig. 12.17 Frequency response of the equalizer circuit of Fig. 10.13 for n 7 and Ra 40
Fig. 12.18 Equalizer circuit with p-i-n diodes as variable resistor to achieve adjustable gain slope
over frequency
the resistors Ra and Rb should be independently adjustable. An equalizer circuit

diagram with resistors in each shunt network synthesized by p-i-n diodes is shown
in Fig. 12.18. Here, four shunt networks are used with λo /4 separation to achieve
more amounts of gain slopes with better ports return losses. Biases of the diodes are
fed from a single DC voltage source V s with resistive networks. Inductors (L) are
used with all the shunt transmission lines to provide DC bias isolation in between
the diodes (D11 –D14 ) and diodes (D21 –D24 ).
Frequency response (|S 21 |) of the equalizer circuit, as shown in Fig. 12.18, is
shown in Fig. 12.19 over the frequency range of 3–5 GHz, with n = 3 and Ro
50 . The different curves are correspond to different values of diode currents. Here,
resistance Ra is for the diodes D11 , D12 , D13 and D14 . Whereas, resistance Rb is for
the diodes D21 , D22 , D23 and D24 . For Rb > 50 , |S 21 | increases with frequency. But
Fig. 12.19 Simulated |S 21 | as a function of frequency of the equalizer of Fig. 12.7 for different
values of Ra and Rb
for Rb < 50 , |S 21 | decreases. It can also be shown that for n 1, the characteristic
will be reversed. For a fixed value of Rb , with the increase of Ra , attenuation |S 21 |
decreases without much affecting the slope of |S 21 | over the frequency range as shown
in Fig. 12.19. Thus, by changing the value of resistor Rb of the equalizer, attenuation
slope over the frequency range can be made positive, negative or zero. Whereas,
attenuation at the centre frequency can be determined by changing the value of the
resistor Ra .
To achieve better port matching, i.e. for better input and output return losses of
the equalizer, more numbers of shunt networks can be used with the separation of
λo /4 (λo corresponds to f o ). For example, the circuit shown in Fig. 12.18 consists
of four shunt networks with separation of λo /4 corresponding to 4 GHz.
In Fig. 12.20, photograph of the equalizer circuit operating over the frequency
range of 3–5 GHz is shown. 25 mil thick alumina substrate is used to realize the
circuit using microstrip transmission line configuration. High-impedance λo /4 trans-
mission lines are used to realize required inductors. Meander transmission lines are
used for series and shunt transmission lines to realize the circuit on a 0.5 mm × 0.5 mm
size of the substrate. Due to the negligible parasitic capacitance and inductance, beam
lead p-i-n diodes are used to operate over the frequency range of 3–5 GHz.
and open-circuit transmission line (Z o ) stub of length l as shown in Fig. 12.21. Also
plot transmission coefficient and reflection coefficient of the network in dB over the
frequency range of 1–20 GHz for Ra 30 and l λo /8, where λo is free space
wavelength at 10 GHz.
Solution
The shunt network consists of a transmission line of length l and a resistor Ra as
shown in Fig. 12.21. Using (7.80), the input impedance Z sn of the shunt network can
be written as
Fig. 12.20 Photograph of C-band equalizer with four shunt networks
Fig. 12.21 The shunt network consists of a resistor and an open-circuit transmission line of Exam-
ple 12.2
Fig. 12.22 The two-port network consists of the shunt network of Example 12.2
Z sn Ra − j Z o cot(βl)
Here, Z o is the characteristic impedance of the shunt transmission line. The S-

parameters of the shunt network can be determined from Fig. 12.22 using (7.134) as
follows:
b1
S11 , when a2 0
a1
V−
1+ , when V2+ 0
V1
−Z o /(Z o + 2Z sn )
S11 S22 −Z o /(Z o + 2Z sn )
Similarly from (7.79) and using (7.136), S21 can be determined as

√
b2 V2− Z o V2−
S21 √ , when V2+ 0
a1 Z o V1+ V1+
2Z sn /(Z o + 2Z sn ) S12

⎡ ⎤
−Z o 2Z sn
S11 S12
⎣ ⎦, where Z sn Ra − j Z o cot(βl)
S21 S22 2Z sn −Z o
The transmission coefficient of the shunt network is given by

2Z sn 2Ra − j2Z o cot(βl)
S21
Z o + 2Z sn Z o + 2Ra − j2Z o cot(βl)

4Ra2 + 4Z o2 cot 2 (βl)
10 × log
(Z o + 2Ra )2 + 4Z o2 cot 2 (βl)
Putting Ra 30 , Z o 50 , l λo /8, the transmission coefficient can be

written as

π
3600 + 10,000 cot 2 40 f GHz
Transmission coefficient 10 × log
π dB
12,100 + 2500 cot 2 40 f GHz

π
36 + 100 cot2 40 f GHz
10 × log
π dB
121 + 25 cot 2 40 f GHz

−Z o −Z o
S21
Z o + 2Z sn (Z o + 2Ra ) − j2Z o cot(βl)
Fig. 12.24 The shunt

network consists of a
transmission line between
two resistors of Example
12.3
Reflection coefficient in dB can be written as

Z o2
10 × log
(Z o + 2Ra )2 + 4Z o2 cot 2 (βl)

2500
Reflection coefficient 10 × log
π dB
12,100 + 10,000 cot 2 40 f GHz

25
10 × log
π dB
121 + 100 cot 2 40 f GHz
The variation of transmission and reflection coefficients over the frequency range
of 1–20 GHz is plotted in Fig. 12.23 using the above expressions. This network
provides increase of insertion loss with the increase of frequency.
Example 12.3 Compute S-parameter of a shunt network consists of transmission
line (Z o ) of length l in between two resistances Ra and Rb as shown in Fig. 12.24.
Also plot transmission coefficient of the network in dB over the frequency range of
5–15 GHz for l λo /4, where λo is free space wavelength at 10 GHz, Ra 20 ,
and Rb as the parameter of values 20 , 30 , 50 , 70 and 100 .
Solution
The shunt network consists of a transmission line of length l in between the resistors
Ra and Rb is shown in Fig. 7.2.4. Using (7.16), the input impedance Z sn of the shunt
network can be written as
Fig. 12.25 The two port network consists of the shunt network of Example 12.3
Rb + j Z o tan(βl)
Z sn Ra + Z o
Z o + j Rb tan(βl)
Here, Z o is the characteristic impedance of the shunt transmission line. The two-
port network consists of the shunt network is shown in Fig. 12.25. The S-parameters
of the two port network can be determined using (7.134) as follows:
b1
S11 , when a2 0
a1
V−
1+ , when V2+ 0
V1
−Z o /(Z o + 2Z sn )
S11 S22 −Z o /(Z o + 2Z sn )
Similarly from (7.79) and using (7.136), S21 can be determined as

√
b2 V2− Z o V2−
S21 √ + , when V2+ 0
a1 Z o V1 V1+
2Z sn /(Z o + 2Z sn ) S12

⎡ ⎤
−Z o 2Z sn
S11 S12
⎣
⎦, where Z sn Ra + Z o Rb + j Z o tan(βl)
S21 S22 2Z sn −Z o Z o + j Rb tan(βl)
The transmission coefficient of the shunt network is given by:


2Z sn 2Z o (Ra + Rb ) + j2 Ra Rb + Z o2 tan(βl)
S21

Z o + 2Z sn Z o (Z o + 2Ra + 2Rb ) + j Rb (Z o + 2Ra ) + 2Z o2 tan(βl)

2
4Z o2 (Ra + Rb )2 + 4 Ra Rb + Z o2 tan2 (βl)
10 × log
2
Z o2 (Z o + 2Ra + 2Rb )2 + Rb (Z o + 2Ra ) + 2Z o2 tan2 (βl)
Similarly, the reflection coefficient can be written as
−Z o −Z o2 − j Z o Rb tan(βl)
S21

Z o + 2Z sn Z o (Z o + 2Ra + 2Rb ) + j Rb (Z o + 2Ra ) + 2Z o2 tan(βl)
Reflection coefficient in dB can be written as

Z o4 + (Z o Rb )2 tan2 (βl)
10 × log
2
Z o2 (Z o + 2Ra + 2Rb )2 + Rb (Z o + 2Ra ) + 2Z o2 tan2 (βl)
Putting Z o 50 , Ra 20 , βl (π f GHz /20), the transmission and reflec-

tion coefficients in terms of parameter Rb can be written as

π
4(20 + Rb )2 + 4(0.4Rb + 50)2 tan2 20 f GHz
|S21 |dB 10 × log
π dB
(90 + 2Rb )2 + (1.8Rb + 100)2 tan2 20 f GHz

π
2500 + Rb2 tan2 20 f GHz
|S11 |dB 10 × log
π dB
(90 + 2Rb )2 + (1.8Rb + 100)2 tan2 20 f GHz
Plot of the transmission and reflection coefficients of the network for different
values of Rb are shown in Figs. 12.26 and 12.27.

and inductance L a connected in series as shown in Fig. 12.28. Also plot transmission
and reflection coefficients of the network in dB over the frequency range of 1–20 GHz
for Ra 40 and L a 0.5 nH.
Solution
The shunt network consists of a resistance Ra and capacitance L a connected in series
as shown in Fig. 12.28. The input impedance Z sn of the shunt network can be written
as
Z sn Ra + j2π f L a
Fig. 12.26 Transmission coefficients of Example 12.3
Fig. 12.27 Reflection coefficients of Example 12.3

network of Example 12.4
From (7.163d), matrix form S-parameters of the shunt network can be written as
⎡ ⎤
−Z o 2Z sn
S11 S12
⎣ ⎦, where Z sn Ra + j2π f L a
S21 S22 2Z sn −Z o
Putting Z sn Ra + j2π f L a , the S-parameters matrix of the network can be

written as
⎡ ⎤
−Z o 2Ra + j4π f L a
S11 S12 ⎢ (Z o +2Ra )+ j4π f L a (Z o +2Ra )+ j4π f L a ⎥
⎣ ⎦
S21 S22 2Ra + j4π f L a −Z o
(Z o +2Ra )+ j4π f L a (Z o +2Ra )+ j4π f L a
Therefore, the transmission coefficient in dB can be written as

4Ra2 + (4π f L a )2
|S21 |dB 10 × log
(Z o + 2Ra )2 + (4π f L a )2
Putting Z o 50 , Ra 40 and L a 0.5 nH:

1600 + (π f GHZ )2
|S21 |dB 10 × log
4225 + (π f GHZ )2

Z o2
|S11 |dB 10 × log
(Z o + 2Ra )2 + (4π f L a )2
Putting Z o 50 , Ra 40 and L a 0.5 nH:

625
|S11 |dB 10 × log
4225 + (π f GHZ )2
Transmission and reflection coefficients of the shunt network of Fig. 12.28 are
shown in Fig. 12.29 over the frequency range of 1–20 GHz.

network of Example 12.5

and capacitance Ca connected in series as shown in Fig. 12.30. Also plot transmission
and reflection coefficients of the network in dB over the frequency range of 5–15 GHz
for Ra 50 and Ca 0.25 pF.
Solution
The shunt network consists of a resistance Ra and capacitance Ca connected in series
as shown in Fig. 12.29. The input impedance Z sn of the shunt network can be written
as
Z sn Ra − j/(2π f Ca )
From (7.163d), S-parameters matrix of the shunt network can be written as

⎡ ⎤
−Z o 2Z sn
S11 S12
⎣ ⎦, where Z sn Ra − j/(2π f Ca )
S21 S22 2Z sn −Z o
Putting Z sn Ra − j/(2π f Ca ), the S-parameters matrix of the network can be

written as
⎡ ⎤
π f Z o Ca 2π f Ca Ra − j
S11 S12 ⎢ (Z o +2Ra )π f Ca − j (Z o +2Ra )π f Ca − j ⎥
⎣ ⎦
S21 S22 2π f Ca Ra − j π f Z o Ca
(Z o +2Ra )π f Ca − j (Z o +2Ra )π f Ca − j
Therefore, the transmission coefficient in dB can be written as:

(2π f Ca Ra )2 + 1
|S21 |dB 10 × log
((Z o + 2Ra )π f Ca )2 + 1
Putting Z o 50 , Ra 50 and Ca 0.25 pF:

(0.025π f GHZ )2 + 1
|S21 |dB 10 × log
(0.0375π f GHZ )2 + 1
Fig. 12.32 The shunt network of Example 12.6

(π f Z o Ca )2
|S11 |dB 10 × log
((Z o + 2Ra )π f Ca )2 + 1
Putting Z o 50 , Ra 50 and Ca 0.25 pF:

(0.0125π f )2
|S11 |dB 10 × log
(0.0375π f GHZ )2 + 1
Example 12.6 Compute S-parameter of a shunt network consists of a resistance Ra ,

capacitance Ca and inductance L a connected in series as shown in Fig. 12.32. Also
plot transmission and reflection coefficients of the network in dB over the frequency
range of 5–15 GHz for Ra 50 , L a 1.25 nH and Ca 0.25 pF.
Solution
The shunt network consists of a resistance Ra , inductance L a and capacitance Ca
connected in series as shown in Fig. 12.32. The input impedance Z sn of the shunt
network can be written as
Z sn Ra + j2π f L a − j/(2π f Ca )
From (7.163d), S-parameters of the shunt network can be written as:

⎡ ⎤
−Z o 2Z sn
S11 S12
⎣ ⎦, here Z sn Ra + j2π f L a − j/(2π f Ca )
S21 S22 2Z sn −Z o
Putting Z sn Ra + j2π f L a − j/(2π f Ca ), the S-parameters matrix of the network

can be written as
⎡ ⎤
−Z o π f Ca 2π f Ca Ra + j (4π 2 f 2 Ca L a −1)
S11 S12 ⎢ (Z o +2Ra )π f Ca + j (4π 2 f 2 Ca L a −1) (Z o +2Ra )π f Ca + j (4π 2 f 2 Ca L a −1) ⎥
⎢⎣ 2π f Ca Ra + j (4π 2 f 2 Ca L a −1)
⎥
⎦
S21 S22 −Z o f Ca
(Z o +2Ra )π f Ca + j (4π 2 f 2 Ca L a −1) (Z o +2Ra )π f Ca + j (4π 2 f 2 Ca L a −1)
Therefore, the transmission coefficient in dB can be written as

2
(2π f Ca Ra )2 + 4π 2 f 2 Ca L a − 1
|S21 |dB 10 × log
2
((Z o + 2Ra )π f Ca )2 + 4π 2 f 2 Ca L a − 1
Putting Z o 50 , Ra 50 , L a 1.25 nH and Ca 0.25 pF:

2
(0.025π f GHZ )2 + 0.00125π 2 f GHZ
2
−1
|S21 |dB 10 × log
2
(0.0375π f GHZ )2 + 0.00125π 2 f GHZ
2
−1

(Z o π f Ca )2
|S11 |dB 10 × log
2
((Z o + 2Ra )π f Ca )2 + 4π 2 f 2 Ca L a − 1
Putting Z o 50 , Ra 50 , L a 1.25 nH and Ca 0.25 pF:


(0.0125π f )2
|S11 |dB 10 × log
2
(0.0375π f GHZ )2 + 0.00125π 2 f GHZ
2
−1
References
1. Morgan M, Newton T, Hayward B, Boyd T (2007) Non-reflective transmission-line filters for

gain slope equalization. In: IEEE MTT-S international microwave symposium digest, June 2007,
pp 545–548
2. Ma Q, Ma M (2008) Broadband amplifier gain slope equalization filter. In: Proceedings of
progress in electromagnetic research symposium, Hangzhou, China, Mar 2008, pp 21–25
3. Madjar A, Gerfer EE (1991) A novel 2–4 GHz multi pass band tunable and gain controlled
miniature active equalizer/filter. In: IEEE MTT-S international microwave symposium digest,
July 1991, pp 451–454
4. Vassilev V, Angelov I, Belitsky V (1999) Design and performance of a 3.4 to 4.6 GHz active
equalizer with controlled gain slope. Appl Microw Wireless 11(12):51–53
5. Vassilev V, Angelov I, Kovtonyuk S, Belitsky V (2000) Low-noise band selective amplifier
and active equalizer with controlled gain-slope for 3.4–4.6 GHz. In: Proceedings of GHz’00
symposium of digest, Goteborg, Sweden, Mar 13–14, 2000, pp 177–180
6. Fejzuli A, Kaarsberg R, Roldan N (2006) Broadband amplifier gain slope equalization with a
single passive component. High Freq Electron 5(6):22–26
7. Bera SC, Jain VK, Singh RV, Garg VK (2009) Circuit for compensating gain variation over
operating frequency and/or temperature range. India Patent, IN284294
8. Bera SC (2011) Amplitude tilt active equalizer for frequency and temperature compensation.
IEEE Microwave Wirel Compon Lett 21(7):344–346
Chapter 13
Microwave Detectors
Abstract This chapter presents design and analysis of various detectors for
microwave power level measurements. Basic concepts of average, RMS, peak and
envelope power are discussed, and methods of their measurements are presented. Use
of Schottky barrier diodes and backward diodes as detectors; with their advantages
and disadvantages are also presented.
13.1 Introduction
Signal power level is an important performance parameter of any microwave com-

munication system. Output and input power levels are critical parameters for design
of any circuit, system or component. Power level in a microwave system and subsys-
tem is so important for its proper function that each stage in the system must receive
predetermined proper signal power level from the previous stage and also it should
provide proper output power level which will be the input power level of the next
stage. At lower frequencies, signal power is calculated from measured voltage and
current. Direct measurement of power is more convenient and accurate starting from
the microwave frequency range to optical spectrum. Thus, as the frequency increases,
power is measured by which voltage and current are calculated. Microwave power
detectors are used to detect/measure the power level.
Schottky barrier and tunnel (backward) diodes are two popularly used devices
which are used as microwave power level detector. Depending upon the signal struc-
ture and applications average, envelope, peak or pulse powers are measured. Selection
of detector circuit also depends upon the bandwidth and dynamic power range of the
signal.
The average or root-mean-square (RMS) power level is important in case where the
total power is important. In systems such as radar, where basic performance mainly
depends on the pulse power, it is important to measure pulse power of the microwave
signal. The instantaneous envelope output power is used for RF system protection
against excessive RF power overdrive. This is also used for power amplifier lineariza-
tion and realization of enhanced efficiency power amplifier by envelope tracking. In

334 13 Microwave Detectors
order to adjust automatic gain control circuits in an RF chain, received signal strength
of a communication system is measured to maintain the desired output signal fed to
a transmitter. It protects the sensitive power amplifiers from excessive power over-
drive. It also ensures stringent requirements for the transmitted output power from a
radio transmitter meeting the specified communications standards and legal regula-
tions. For microwave and mmwave signal amplitude measurements, Schottky barrier
detector diodes are used for its faster response compared to p–n junction diodes [1,
4]. Faster response of Schottky barrier diodes is due to the absence of minority
carrier storage delay time in Schottky junction, i.e. due to its operation with major-
ity carrier only. Therefore, Schottky barrier diode can switch from its conduction to
non-conduction condition almost instantaneously with the reversal of driving signals
polarity. Thus, the Schottky diode’s impedance can change almost instantaneously
with the change of driving signals polarity that leads to its very high rectification
efficiency extending to microwave and mmwave frequency range.
13.2 Microwave Power
Before designing a microwave detector, it is important to know the significance of

instantaneous, average, envelope and pulse power. Suppose RF voltage v(t) across
a load and current i(t) through the load at a time t are given by
v(t) vm cos(ωt) (13.1a)
and
i(t) i m cos(ωt + θ) (13.1b)
Here, θ is the phase difference in between the voltage and current. The instanta-
neous power across the load is given by
p(t) v(t) × i(t) vm cos(ωt) × i m cos(ωt + θ ) (13.2a)
This expression gives the power at any instant of time. Figure 13.1 shows the
instantaneous voltage, current and instantaneous power. From (13.2a), the instanta-
neous power can also be written as
or
1 1
p(t) vm i m cos(θ ) + vm i m cos(2ωt + θ ) (13.2b)
2 2
The first term is the DC component, i.e. independent of time. The second term
is the time-dependent sinusoidal component of the instantaneous power with double
the frequency of voltage and current waveform.
13.2 Microwave Power 335
p(t)
t
v(t)
0, 0
i(t)
Fig. 13.1 Instantaneous RF voltage, current and power
Average RF/microwave power is the area under the power versus time plot. There-
fore, it is determined by integrating and dividing the length of integration time. The
length of the integration time should be an integer number of periods. Thus, the
average RF power is given by
1 nTo
pav ∫ p(t)dt (13.3a)
nTo 0
Putting the value of p(t) from (13.2b) to (13.3a), the average power can be written
as
1 1 nTo 1 1 nTo
pav vm i m cos(θ ) ∫ dt + vm i m ∫ cos(2ωt + θ)dt (13.3a)
2 nTo 0 2 nTo 0
1
vm i m cos(θ ) + 0 (13.3b)
2
vm i m
√ √ cos(θ ) (13.3c)
2 2
Therefore, the first term of (13.2b) is the average RF power. In case of resistive
load, the phase difference in between the voltage v(t) and the current i(t) is zero,
i.e. θ 0. Thus, for resistive load, the average RF power is given by
vm i m
pav √ √ (13.3d)
2 2
vrms × i rms in case of sinusoidal signal (13.3e)
Instantaneous RF power and its two components, i.e. average power and time-
dependent sinusoidal power, are shown in Fig. 13.2. Ideally, for measurement of
average RF power, it should be integrated over integer numbers of RF cycles. In
practice, integration time is taken over many RF periods to avoid the restriction of
integration over exact integer multiple cycles to avoid any significant error. For a
continuous wave signal (i.e. sinusoidal waveform), the lowest and highest frequency
p(t)
p(t) - pav
pav
t
T
0, 0 T/2
Fig. 13.2 Instantaneous RF power with its two components
are same; however, for a multicarrier and amplitude-modulated wave, it is required

to average the power over many periods. Thus, in case of amplitude-modulated or
multicarrier signal to measure the average power, it is required to take average over
many periods of the lowest frequency involved.
In communication systems, it is very much convenient to express RF power level
in dB with respect to a known reference power level. The unit dBm is used when the
power level expressed with respect to 1 mW power and unit dBW is used for power
level when it is expressed with respect to 1 W power level.

absolute power
Power(in dBm) 10 × log (13.4a)
1 mW

absolute power
Power(in dBW) 10 × log (13.4b)
1W
The use of dB scale leads to easy calculation of gain/attenuation for a system just
by algebraic addition among its output and input power level when both are expressed
in dB. Overall gain/attenuation of a system consists of several number of cascaded
matched networks is just the algebraic sum of individual network’s gain. In this case,
just addition of power gain in dB for each network replaces the multiplication of
numeric gain.
Example 13.1 Convert 100 mW RF power in dBm and dBW.
Solution
Using (13.4a),

100 mW
100 mW Power(in dBm) 10 × log
1 mW
2
10 × log 10 dBm
20 dBm
Using (13.4b),

100 mW
100 mW Power(in dBW) 10 × log dBW
1W
2
10
10 × log dBW
103
−10 dBW
Example 13.2 Convert 30 dBm and 20 dBW RF power to mW and W respectively.
Solution
From (13.4a), absolute power can be expressed as
Power in dBm
Absolute power 10 10 mW
Thus,
30
30 dBm power 10 10 mW
103 mW
1000 mW
Similarly, from (13.4b), absolute power can be expressed as
Power in dBW
Absolute power 10 10 W
Thus,
20
20 dBW power 10 10 mW
102 W
100 W
Example 13.3 An attenuator of 15 dB attenuation is fed with 100 W input. Find the
output power of the attenuator in dBW and Watt.
ATTN
15 dB
PIN = 100 W =20 dBW POUT = 5 dBW = 3.1623 W
Fig. 13.3 Power levels of an attenuator
ATTN
13 dB
PIN = 400 W =26 dBW POUT = 20 W = 13 dBW
Fig. 13.4 Power levels of an attenuator
Solution

100 W
100 W Power(in dBW) 10 × log dBW 20 dBW
1W
The output of the attenuator will be (Fig. 13.3)
POUT (dBW) PIN (dBW) − ATTN(dB)

(20 − 15) dBW
5 dBW
POUT (W) 10 POUT (dBW)/10
100.5 dBW
3.1623 W
Example 13.4 The input power given to an attenuator is 400 W. The power output
of the attenuator is 20 W. Calculate the attenuation value of the attenuator in dB.
Solution
Input power of the attenuator is 400 W (Fig. 13.4).

400 W
1W
The output power of the attenuator is 20 W.

Power
TP
P
PP
Pavg
time
Fig. 13.5 Pulsed RF power

20 W
1W
The attenuation of the attenuator in dB is given by
Attenuation (dB) PIN (dBW) − POUT (dBW)

(26 − 13) dB
13 dB
Though the average power is very popular for communication systems; however,
the basic performance of radar, navigation and TDMA wireless communication sys-
tems depend primarily on the peak power radiation. The time constants of peak
power instruments and sensors should be in the sub-microsecond region to measure
the ‘pulse power’ and ‘peak envelope power’. The pulsed RF power waveform is
shown in Fig. 13.5. The power is averaged over the time duration of pulse width, τ P ,
for pulse power. Thus, the pulse power PP can be represented as
i τp
PP ∫ v(t) × i(t)dt (13.5a)
τ p t0
The pulse power is average power over the pulse duration. Thus, it averages out
any overshoot or ringing within this duration. The pulse power also can be expressed
in terms of its duty cycle as
Pavg
PP (13.5b)
duty cycle
Here, Pavg is the RF power averaged over integer numbers of full pulse cycles.
The duty cycle is the pulse width times the repetition frequency. From Fig. 13.5, the
duty cycle of the pulsed power can be written as
τP
Duty Cycle (13.6)
TP
Therefore, the pulse power can be expressed as

Pavg TP
PP Pavg × (13.7)
duty cycle τP
Using this definition of pulse power, calculation of pulse power can be done by
the measurement of the duty cycle and the average power. This technique avoids the
use of more complicated and expensive direct pulse power measurement scheme.
Example 13.5 The pulse power of an RF pulse power source is 1 KW. The duty cycle
is 10−2 . Find its average power.
Solution
The average RF power can be expressed in terms of its pulse power as (13.7)
Pavg pulse power × duty cycle

1 kW × 10−2
10 W
Example 13.6 A microwave pulse has an average power of 200 W and duration is
5 µs. The time interval between pulses is 2 ms. Determine the value of pulse power.
Solution
Pulse power is given by (13.7)
Pavg TP
PP Pavg ×
duty cycle τP
2 ms
200 W ×
5 μs
80 kW
13.3 Diode Detectors
Point contact diodes were the first high-frequency diodes though they were extremely
fragile and did not have repeatable performance [1]. For microwave frequency power
measurement, low-barrier Schottky contact technology is used to construct diodes
with metal–semiconductor junctions [2, 4]. These diodes are very rugged and pro-
vide very consistent performance. These diodes are able to detect and measure very
13.3 Diode Detectors 341
low power up to very high frequencies. Metal–semiconductor diode also known as

Schottky barrier diode is very popular as detector diode for its operation with major-
ity carriers only. With the change in polarity of the driving signal, the impedance of
a Schottky diode changes almost instantaneously. This property makes the Schot-
tky diode suitable as rectifier not only at microwave frequency range but also used
in mmwave and terahertz frequency range. By suitable construction, it is possible
to achieve very low cut-in voltage of the Schottky barrier diodes. This makes the
Schottky barrier diode as very sensitive detector to measure low level of microwave
power compared to the p–n junction diodes.
Under forward-bias condition of junction voltage Vd j , the diode current is given
by

q Vd j
Id Io exp −1 (13.8)
ηkT
Here, the ideality factor η ∼

1.0–1.2, because of the negligible recombination in
the depletion region, and Io is the reverse saturation current. Due to the absence
of charge storage delay effect of the Schottky diodes, its DC transfer characteristic
(13.8) is also valid for AC transfer characteristic up to very high frequency.
Suppose, RF signal is applied in a finite biased Schottky diode. The diode voltage
can be represented by
Vd j Vo + Vs cos(ωt) (13.9)
Combining (13.8) and (13.9), the diode current can be written as

Vo + Vs cos(ωt)
Id Io exp −1 (13.10)
Vt

Vo Vs cos(ωt)
Io exp exp −1 (13.11)
Vt Vt
where Vt (ηkT /q). Expanding the exponential term containing AC part by series
expansion, the diode current can be written as

Vo Vo
Id Io exp − 1 + Io exp
Vt Vt

Vs Vs2 V3 V4
× cos(ωt) + 2
cos2 (ωt) + s 3 cos3 (ωt) + s 4 cos4 (ωt) + · · · (13.12)
Vt 2Vt 6Vt 24Vt
This can be written as

Id ( A)
20
10
Vd (mV)
-40 -30 -20 -10 10 20 30 40
-10
-20
Fig. 13.6 Square-law characteristic region of Schottky barrier diode

Vo
Id Io exp − 1 ⇒ this term is DC current due to the bias voltage
Vt

Vo Vs
+ Io exp cos(ωt) ⇒ is due to the fundamental AC component
Vt Vt
2
Vo Vs Vs2
+ Io exp + cos(2ωt) ⇒ rectified dc and second harmonic
Vt 4Vt2 4Vt2
3
Vo Vs Vs3
+ Io exp cos(ωt) + cos(3ωt) ⇒ fundamental and third harmonic
Vt 8Vt2 24Vt2
4
Vo Vs Vs4 Vs4
+ Io exp + cos(2ωt) + cos(4ωt) ⇒ dc + second and fourth
Vt 64Vt4 48Vt4 192Vt4
+ ··· (13.13)
This expression is written under the assumption that the exponential I–V rela-
tionship is also valid under RF signal operation. This assumption is true if V s V t .
Otherwise, exponential I–V characteristic of the Schottky diode will not remail valid
due to the significant rectification of the RF signal. This expression shows that second
and all other even-order nonlinearity produces DC components due to the rectifica-
tion effect of Schottky diodes. The property of generating DC term related to RF
signal by the Schottky diode is used as power detector.
The region of I–V characteristic around the zero voltage across a low-barrier
Schottky diode is shown in Fig. 13.6. Over this region, the I–V characteristic follows
the square law, i.e. output current of the diode is proportional to the square of the
input signal voltage. For a typical Schottky diode, the range of RF power from noise
level to about −20 dBm considered as small signal level over which the diode’s I–V
characteristic obeys the square law. It is to be noted that for 50 system, about
22.4 mV (RMS) corresponds to −20 dBm (10 µW) RF power. Figure 13.7 shows
the input signal voltage and output diode current waveform over the operating square
low region.
From (13.13), the DC components of the diode current can be written as
Id ( A)
Detected
voltage Diode current
Vd (mV)
Input voltage
Fig. 13.7 Working of diode detector over square-law region
Fig. 13.8 Equivalent circuits of diode detector with output capacitor as filter
2 4
Vo Vo Vs Vo Vs
Idc Io exp − 1 + Io exp 2
+ I o exp + ···
Vt Vt 4Vt Vt 64Vt4
(13.14)
Here, the first term is the DC bias current, second term is the rectified DC compo-
nent proportional to the square of the RF input voltage generated due to the second-
order nonlinearity and third term is the rectified DC component generated due to the
fourth-order nonlinearity, and so on.
At small-signal condition, i.e. for Vs Vt , up to the second-order term can be
considered as significant. In this region, the diode current is proportional to the square
of the RF input voltage. So the diode is said to be operating in the square-law region.
In this condition, the DC components of the current can be written as
2
Vo Vo Vs
Idc Io exp − 1 + Io exp × (13.15)
Vt Vt 4Vt2
The presence of bias-dependent exp(Vo /Vt ) in the rectified DC voltage increases

the sensitivity of the detector diode. The DC current can be rewritten as
(a) (b) Rd
Lp Rs
Matching
Network
Ro
Cb
Cp Cd
Fig. 13.9 Detector circuit diagram with matching network and the diode equivalent circuit

Vo V2
Idc Io exp −1 + s (13.16)
Vt 4rd Vt
where r d is the RF resistance of the diode and is given by

∂ Id Vt
rd 1/ (13.17)
∂ Vd j Io exp(Vo /Vt )
Therefore, the detector diode with output capacitor to filter out the RF components
can be considered as a current source driving a shunt resistor. The shunt resistor is
equal to the diode resistance rd . The basic equivalent circuits of the diode detector
are shown in Fig. 13.8a, b considering the detector diode as current and voltage
source respectively. Here, the fixed DC bias current of the diode is not shown for
simplicity. The bypass capacitor Cb acts as low-pass filter which filters out the RF
signals passing through the diode.
At higher power level of signal, the diode cannot behave as square low device due
to the significant fourth-order and higher order component of Vs . The device then
rectifies the signal in a quasi-square-law I–V region which is called the transition
region. Above that range, the exponential I–V relationship will no longer valid and
practically the detector relationship moves into the linear detection region where
detected output voltage/current will be proportional to the input RF voltage/current.
To transfer maximum power from source to the detector diode, the RF resistance
of the diode must be matched to the generator source resistance. The detector diode
will detect RF voltage as long as the RF voltage falls across the diode. Practically,
the RF resistance of diode is typically much larger than 50 . To avoid reflection
of RF power to the source, a separate matching network is used to set the detector’s
input termination impedance matched to the source resistance. Circuit diagram of
a diode detector circuit with a matching network is shown in Fig. 13.9a, and the
complete equivalent circuit diagram of a Schottky diode is also shown in Fig. 13.9b.
Matching networks can be either resistive or reactive. A shunt resistor Rm , as shown
in Fig. 13.10, is used as resistive matching network to achieve broadband matching
with low reflection and flat frequency response. As discussed in Chap. 3, the RF
resistance of Schottky diodes is a function of its operating temperature. Thus, the
sensitivity of the diode detector and its reflection coefficient are also strong functions
Fig. 13.10 Resistive

matching network of the
detector
Ro
Rm Cb RL
of temperature. The RF impedance of the diodes should be very high compared to

the matching resistor Rm to achieve its less temperature dependence. In that case,
the matching resistor Rm acts as the termination of the generator due to very high
RF impedance of the diodes. If RF resistance of the diode becomes too large, there
would be poor power conversion from RF signal to detected DC voltage. The inverse
dependency of diodes RF resistance on reverse saturation current leads to decrease
of detector sensitivity with the increase of diode impedance. A compromise between
sensitivity to RF signals and temperature dependency of the performance can be
achieved by suitable construction of the diode using suitable materials to achieve
low potential barrier across the junction. A low-barrier Schottky barrier diode can
meet this requirement.
In case of reactively matching network, detector works over only moderate band-
width. It also has drawback of poor matching compared to resistive matching and
the performance of the circuit is also influenced by the change of diode character-
istic. However, the advantage of using reactive matching network is that it provides
better sensitivity to the RF signal. Sometimes combination of resistive and reactive
components is used for the matching network to obtain better sensitivity to RF signal
with lesser influence by the change of diode characteristic.
Tunnel and backward diodes have inherent I–V characteristic which leads to
higher rectification current sensitivity [4]. This enables the tunnelling effect diodes
to rectify small signals of amplitude which is smaller than the forward-bias cut-in
voltage. The current is due to the tunnelling of the carriers, and it is fast in the
absence of minority carriers charge storage effect which leads to broadband per-
formance capability. Tunnelling current is insensitive to the change of temperature.
Therefore, the performance of the backward and tunnel diodes is also not sensitive
to its operating temperature. Therefore, tunnel and backward diode detectors have
excellent temperature stability, very fast pulse response time and broadband fre-
quency flatness. Open-circuit voltage sensitivity and high-power burnout are lesser
than silicon-based Schottky detectors, but the tunnel detector’s relatively low video
impedance with no DC bias requirement enables DC and AC coupling with video
and log video post-amplifiers.
Build-in detectors are required for embedded on-chip measurement of internal
signal levels in analog and mixed-signal circuits. In this case mostly, RMS power
detectors are more useful than peak or envelope detectors, because RMS detection
provides signal levels independent of signal shape, i.e. modulation, number of car-
riers, etc. To realize on-chip detectors, it is required to develop detectors using the
same technology in which full circuit/systems are made such as Si [2–4], GaAs [5]
and GaN-based different types of FETs [6–8]. GaN-based Schottky barrier diodes
offer superior performance in terms of high-power and high-frequency operation
over a wide range of operating temperature due to their higher bandgap and higher
breakdown electric field [6–8]. The GaN-based Schottky barrier diodes can operate
as microwave detector with non-zero DC biasing due to its 0-V turn-on voltage,
thus easy to implement in MMIC technology without any bulky passive compo-
nents. High reverse breakdown voltage of the GaN-based diodes provides very wide
dynamic range which is desired as RF power detector.
13.4 RMS Power (Average) Detector
One of the major applications of diode detectors is to measure the true RMS power
which is nothing but the average signal power. A true RMS detector should be
designed to measure the actual RMS power of the input signal, ideally independent
of the modulation complexity or the modulation scheme. Generally, a RMS detec-
tor consists of Schottky barrier diodes operating over its square-law region and an
integrator acting as low-pass filter. The RMS output signal is directly proportional to
the time-averaged of the square of the input signal voltage. Under the small-signal
detector operation, the output of the detector is determined by the slope of the I–V
characteristic of the diode, and thus the detector output current is proportional to the
square of the input voltage and thus proportional to the input power.
From the power series expansion of the diode characteristic, it is already discussed
that the rectified output voltage is a function of the square of the input signal voltage.
This relationship is valid up to a power level of about −20 dBm for practically used
Schottky diode detectors. That means the rectified output voltage is proportional to
the RF signal power regardless of the signal content. That means, under the small-
signal diode operation region, as long as square-law detection is valid, the detected
output is proportional to the input power which is independent of the signal content,
i.e. complexity and modulation scheme.
As the power level increases above about −20 dBm, the higher order terms of
the power series will be comparable with the second-order term and the rectification
process becomes more and more linear. Thus, under large signal condition, the output
voltage becomes proportional to the input voltage. For complex signals of power level
corresponding to large signal condition, the output is then dependent upon the phase
relationships among the various components of the input signal. Thus, the measured
output will not be the true RMS power representation of the complex input signal.
In case of very small power levels, i.e. below −50 dBm, the diode detector input is
about 700 µV (RMS) which may correspond to detector DC output voltage of about
700 nV. This low signal level requires low-noise amplifier and chopper circuit to
measure reliably in the presence of noise.
13.4 RMS Power (Average) Detector 347
Schottky diode-based sophisticated detector circuit is capable to operate as true

RMS power detector over the power range of about −70 to −20 dBm. Beyond
−20 dBm, power level correction factor may be used to enhance the dynamic range.
Though the use of corrections for the CW signals is applicable, however, for other
signals which are modulated or multicarrier, the use of correction is not reliable. In
this case with the expense of detection sensitivity, in place of single diode stacks (mul-
tiple numbers) of detector diodes, it can be used to extend the square-law operation
extended to higher power levels. In practice, multiple stacks of a different number of
detector diodes are used. For low-level power measurement, stacks of lesser number
of diodes and for higher power measurement stacks of more number of diodes are
used. Electronic RF switches are used to select the RF paths based on the RF power
level to operate the diode over its square-law region over lowest to extended power
levels.
Thermal-based detectors are also used for microwave power detection [9]. Using
resistive components, it first converts AC power to thermal power, and then using
a temperature sensor, e.g. thermistor, it measures the change of temperature. The
change of temperature is proportional to the thermal power or the dissipated AC
power, which gives the measured RMS power level. These kinds of detectors have
the advantages of good accuracy, wide bandwidth and independent of signal shapes,
i.e. independent of number of carriers and modulations. However, these types of
detectors are not suitable to implement in chip level.
13.5 Envelope and Peak Power Detector
For multicarrier signals and complex digital-modulated single-carrier signals, such

as wideband code division multiple access (WCDMA), instantaneous power mea-
surement is required. This enables to measure the peak power or peak-to-average
power ratio (PAPR). Envelope power can be measured using a diode detector in its
square-law range.
The maximum value of the envelope power is the peak envelope power. In order to
measure envelope power accurately for a fast-changing envelope power, the detector
should be fast enough to track the envelope power. To provide this capability for
the measurement of peak envelope power, the detector circuit should have sufficient
video bandwidth. However, with the increase of the video bandwidth, there will be
a reduction in dynamic range of the detector.
13.6 Applications
Microwave diode detectors are used for high accuracy RF signal measurement and
control applications [10, 11]. The envelope power detectors can also be used to create
fast, excessive RF power protection, power amplifier linearization and efficiency-
enhancing envelope tracking power amplifier realizations [12]. Diode detector cir-
cuits are the building block of many microwave instruments like power meter, spec-
trum analyzer, network analyzer, etc. One of the main differences among different
power measuring instruments is the frequency selectivity. In case of frequency selec-
tive measurement, the RF power within a predetermined bandwidth is measured. This
is generally done by spectrum analyzer instrument. Generally, microwave power
meters are not frequency selective. Power meter measures the average power over
the full frequency range of the detector including present noise and any generated
harmonics. However, spectrum analyzer measures power over a selected frequency
band. Frequency selectivity in spectrum analyzer makes it enable to measure much
lower power level compared to general power meters.
References
1. Bose JC (1904) Detector for electrical disturbances. US Patent 755840

2. Wu Y (1996) Design of Schottky barrier detector diode for silicon MMIC’s. Int Rep, NISRC,
Belfast, Northern Ireland
3. Wu Y, Armstrong BM, Gamble HS, Hu Z, Chen Q, Yang S, Fusco VF, Stewart JAC (1998)
Microwave PtSi–Si Schottky-barrier-detector diode fabrication using an implanted active layer
on high-resistivity silicon substrate. IEEE Trans Microw Theor Tech 46(5):641–646
4. Park SY, Yu R, Chung SY, Berger PR, Thompson PE, Fay P (2007) Sensitivity of Si-based
zero-bias backward diodes for microwave detection. Electron Lett 43(5):53–54
5. De Bo W, Ping LX, Liu T (2012) A novel thermoelectric and capacitive power sensor with
improved dynamic range based on GaAs MMIC technology. IEEE Electron Device Lett
33(2):269–271
6. Zhang AP, Johnson JW, Luo B, Ren F, Pearton SJ, Park SS, Park YJ, Chyi JI (2001) Vertical
and lateral GaN rectifiers on freestanding GaN substrates. Appl Phys Lett 79:1555–1557
7. Zhou Y, Wang D, Ahyi C, Che C-T, Williams J, Park M, Williams NM, Hanser A (2006) High
breakdown voltage Schottky rectifier fabricated on bulk n-GaN substrate. Solid State Electron
50(11/12):1744–1747
8. Zhou Q, Wong K-Y, Chen W, Chen KJ (2010) Wide-dynamic-range zero-bias microwave
detector using AlGaN/GaN heterojunction field-effect diode. IEEE Microwave Wirel Compon
Lett 20(5):277–279
9. Yi Z, Liao X (2012) Packaging test-fixture research for 10–12 GHz frequency detector based
on thermoelectric microwave power sensor. Electron Lett 48(2):103
10. Bera SC, Singh RV (2004) A temperature-compensated closed loop overdrive level controller
for microwave solid-state power amplifiers. Microw J 47(4):114–122
11. Bera SC, Singh RV, Garg VK (2006) Design and temperature compensation of a Ku-Band
channel amplifier with ALC for a satellite transponder. Microw J 49(4):68–82
12. Bera SC, Singh RV, Garg VK (2008) Diode-based predistortion linearizer for power amplifiers.
Electron Lett 44(2):125–126
Chapter 14
Microwave Solid-State Amplifiers
Abstract This chapter presents design and analysis of low-noise, high-gain and high
power microwave solid-state amplifiers. Stability analysis and design of matching
networks for all the types of amplifiers using Smith chart is presented. The dynamic
range of amplifiers and its dependence on noise figure and nonlinearities of the
amplifiers are discussed. Using conditional and unconditional stable devices, design
examples for transducer power gain, operating power gain and available power gain
are also presented with various examples. Design and analysis of linear and nonlinear
Class-A, Class-B, Class-C and Class-F high power amplifiers are discussed in detail.
14.1 Introduction
A microwave amplifier is a circuit that takes a signal of low or intermediate level over
a range of frequency band and significantly boosts its power level to a predetermined
level. An amplifier is one of the most common elements in any communication sys-
tem to boost the signal level wherever it is required. In addition to amplification of
the desired signal, a practical amplifier-added noise is generated within it and also
distorts the amplified signal. Based on the signal power level to be amplified, there
are three types of microwave amplifiers: low-noise amplifier, high-gain amplifier and
high power amplifier. Microwave communication systems include receiver and trans-
mitter. Most of the elements of a transmitter and receiver are amplifiers operating
at different power levels. Low-noise amplifiers (LNA) are used as the front end of a
receiver where the strength of the desired signal is very low. High-gain amplifiers are
used in the intermediate stages of a transmitter–receiver section to boost the signal
strength to a desired level to drive other circuitry, e.g., high power amplifiers of a
transmitter. The High Power Amplifier (HPA) is used as the final power amplifying
stage of a transmitter. The design of matching networks and biasing criterions are
different for LNA, high gain and high power amplifiers. However, there are common
requirements for all types of the amplifiers such as bandwidth, gain and gain flat-
ness over frequency, stability, efficiency, reliability, etc. In most of the cases, design

350 14 Microwave Solid-State Amplifiers
trade-offs are required to achieve any one parameter compared to the other to meet
the specific requirements of a system.
Based on the devices used for the realization of an amplifier, there are two general
classes of microwave solid-state amplifiers (SSAs): diode based and transistor based.
Due to the simplicity of fabrication and lower cost, two terminals diode-based SSAs
were most popular. Varactor-diode-based parametric amplification, transferred elec-
tron device, i.e., Gunn diode’s negative resistance based amplification and avalanche
transit time diode (IMPATT, TRAPATT, BARITT) based pulsed power amplifications
were used for various applications. With the advancement of microwave transistors
fabrication technologies, today solid-state amplification is dominated by the use of
three terminal devices, transistors. Among the microwave transistors, all the low-
noise amplifiers for communication systems are dominated by psuedomorphic and
metamorphic high electron mobility transistors (pHEMTs and mHEMTs). High gain
with moderate output power amplifiers are also dominated by MESFET and HFETs.
With the advancement of high bandgap semiconductor materials such as GaN on SiC,
single-ended high power amplification beyond 100 W is achievable using solid-state
technology.
At low frequencies, the design of an amplifier is trivial which involves designing a
proper DC bias circuit for maximum power output. However, at microwave frequency
range, in addition to the design of proper DC bias circuit, it involves the design
of proper input and output matching networks based on the requirements of noise
figure, gain and output power over the desired band of frequency. A typical block
diagram of a transistor-based single-stage microwave amplifier is shown in Fig. 14.1.
It consists of bias networks and input and output matching networks. The bias network
provides proper terminal voltages/currents to the device without affecting the in-
band RF performance of the amplifier. It also provides out-off band stability to the
amplifier. Input and output matching networks provide proper impedance matching
of the device for achieving the required RF performance of the amplifier. Based
on the performance requirements, i.e. low noise figure, high gain or high power,
the bias voltages/currents as well as impedances offered by the matching networks
Biasing networks
DEVICE
ZS Input Output
Matching Matching ZL
Network Network
Fig. 14.1 Microwave amplifier with its associated networks

14.1 Introduction 351
are different. Designing a stable amplifier is the prime requirement for any type of
amplifier irrespective of its other performance requirements.
For designing an amplifier which will operate in its linear operating region, it is
convenient to use S-parameter model of the device. Front-end amplifiers of a receiver
such as LNA and other intermediate amplifiers such as IF amplifier used in the
satellite receiver and channel amplifiers operate in their linear operating region. These
amplifiers can be designed using S-parameter model of the transistors. However, the
final stage of a transmitter section, in general, operates in its nonlinear operating
region to achieve high output power with higher DC to RF efficiency. Thus, for
designing the power amplifiers, it is required to use the nonlinear model of the
device. Design methodology for different types of amplifiers will be discussed in
their respective sections.
14.2 Types of Microwave Amplifiers
Design philosophy of a microwave amplifier depends on the performance require-

ments and its operating condition. Design of an amplifier operating in small signal
condition is different from the design of an amplifier operating in large signal condi-
tion. It is very important to understand the difference between the terms large signal
and high power, similarly the difference between small signal and low power. In case
of FET-based amplifiers, large signal refers to a power level in which the RF signal
level (voltage) across the gate-to-source layer is a significant fraction of the applied
DC gate bias voltage. High power refers to an arbitrary power level which is indepen-
dent of any amplifier, whereas large signal refers to a particular power level related
to the particular amplifier. For example, achieving one watt output power from an
amplifier of maximum output power capability of one watt is considered as the large
signal operation of the amplifier. Whereas, the same one watt output power taken
from an amplifier of maximum output power capability of ten watts is considered
as small signal operation. Even an amplifier of maximum output power capability
of few mW operates in large signal condition when it is operated with output power
level of the few mW. The operating condition of an amplifier is determined by the
signal level handled by the amplifier compared to its maximum power capability. The
output power capability of an amplifier is determined by the applied bias voltage and
current to the device which is limited by the absolute maximum rating of applied
bias voltage (VDS ) and current (IDSS ) or power dissipation of the device (in terms of
channel temperature).
An amplifier operating in large signal operating condition is considered operating
in nonlinear region because of the nonlinear relationship between the input power and
output power level. Thus, gain and other performance parameters become a function
of input power level and the device performance cannot be modelled by a single
set of S-parameter values. There is a technique of using large signal S-parameters
for designing amplifiers operating in large signal operating conditions. However, it
RECEIVER TRANSMITTER
Mixer
6 GHz Low High High 4 GHz
Noise Gain Power
Amp Amp Amp
Rx Tx
2 GHz
Antenna Antenna
Local
Oscillator
Fig. 14.2 Schematic diagram of a simple transceiver
is more convenient to use large signal circuit model or load-pull measurement and
design methods for an amplifier operating in large signal condition.
Block diagram of a simple transceiver (transmitter + receiver) for a communication
system is shown in Fig. 14.2. In a communication system, basically, three types
of amplifiers: low-noise, high-gain and high power amplifiers are used. Low-noise
amplifier is used at the front end of a receiver to amplify very weak signal received by
the receive antenna. The high-gain amplifier further amplifies the signal to a moderate
power level required to drive the final high power amplifier for signal transmission.
The low-noise amplifiers are used where the signal levels are very small. Thus,
low-noise amplifiers should have the capability to amplify very weak signal without
much addition of its own noise. Generally, LNAs and high gain amplifiers are always
operated in small signal condition. Whereas, high power amplifiers need to operate
in large signal condition to provide high output power with high DC to RF efficiency.
Input and output matching networks as well as DC bias conditions of a transistor
are different which depends on the operating condition of the amplifier. As long as
the transistors operate in the linear region, i.e. operate in small signal condition, the
output is a linear reproduction of the input. However, when the transistors operate
in large signal condition, output no longer changes linearly with the input. This
nonlinear behaviour is the main cause of signal distortion in microwave circuits.
Other effects of nonlinearity are gain reduction and phase change in an amplifier
resulting in the modulation transfer from one signal to another and generation of
various frequencies related to frequencies of the input signals.
14.2 Types of Microwave Amplifiers 353
Output
Power
1 dB
Minimum
detectable
signal level
Noise
floor 1-dB gain
compression
point
Input
Dynamic Range Power
Fig. 14.3 Typical input–output characteristic of amplifier
14.2.1 Dynamic Range of Amplifier
An amplifier amplifies signal only over a certain range of its input power level
maintaining acceptable linearity. Beyond this region, the signal output level does not
follow the acceptable linearity with the input and thus the amplified signal becomes
distorted beyond an acceptable limit. Amplifier also adds noise to the amplified
signal during amplification. Therefore, noise dominates if the level of the input signal
becomes below a certain level. The minimum input signal level above the noise level
that can be detected is considered as the lower limit of the dynamic range.
Dynamic range of an amplifier is considered as the range of power level over
which an amplifier amplifies the signal maintaining specified linearity. The lower
level of the dynamic range is limited by the contribution of amplifiers noise , whereas
the higher level is limited by the allowable nonlinearity of the amplifier. A typical
input–output characteristic of an amplifier is shown in Fig. 14.3. Above the minimum
detectable power level, amplifier amplifies the signal linearly, i.e. 1 dB increase
of input power level results in 1 dB increase of output power level. This linear
relationship continues until the RF output power reaches to a certain value, and
beyond that power level, the rate of increase of output power level decreases due to
the reduction of gain of the amplifier. Over this nonlinear characteristic, the output
signal vo of an amplifier can be represented by the power series function of its input
signal vi as
vo a0 + a1 vi + a2 vi2 + a3 vi3 + · · · (14.1)

Here, a0 , a1 , a2 , a3 , . . . are constants having amplitude and phase term. Suppose,

a sinusoidal signal vi A cos ωt is applied to input of the amplifier, and considering
the nonlinearities up to cubic term, the output signal can be written as
vo a0 + a1 A cos ωt + a2 A2 cos2 ωt + a3 A3 cos3 ωt (14.2)

2 3
a2 A a3 A
a0 + a1 A cos ωt + (1 + cos 2ωt) + (3 cos ωt − cos 3ωt)
2 4

a2 A2 3a3 A3 a2 A2 a3 A3
a0 + + Aa1 + cos ωt + cos 2ωt + cos 3ωt (14.3)
2 4 2 4
The output signal level corresponding to the fundamental frequency component

is given by

3a3 A2
Aa1 1 +
4a1
Therefore, gain of the amplifier corresponding to the fundamental frequency is a

function of the input signal level and can be written as

3a3 A2
G(A) a1 1 + (14.4)
4a1
For a3 /a1 < 0, gain of the amplifier decreases with the increase of input signal
level. With further increase of input power level, output of the amplifier saturates.
The reduction of gain and saturation of the output power of an amplifier is due to
the limitation of bias voltage and current to the amplifier which is determined by
the maximum absolute rating of applied voltage and current handling capability of
the device used for the amplifier. Generally, the upper limit of dynamic range is
considered as the power level that causes 1 dB gain reduction with respect to the
linear gain of the amplifier.
The noise is generated within a solid-state device itself as well as comes from an
external source. To detect a signal reliably, the signal strength should be much higher
(generally considered as 3 dB) than the noise floor. The noise source in solid-state
devices may be categorized into three types: thermal or Johnson noise, shot noise
and flicker noise.
The thermal noise is generated due to the random fluctuations of signal produced
by thermal agitation of the bound charges in the device. The RMS value of open-
circuit thermal noise voltage Vn with resistance R over a frequency band of B is given
by
√
Vn 4kTBR (14.5)
Thus, a noisy resistor can be modelled as a Thevenin circuit consisting of a

noiseless resistance R together with a Thevenin voltage source given by (14.5). Thus,
the available noise power at the terminal of any resistor is simply given by
Vn2
Pn kTB (14.6)
4R
The spectrum of Johnson noise is white because the power spectral density is
independent of frequency.
Shot noise is due to the fluctuation in kinetic energy of the current carrying discrete
electrons in a device. The energy fluctuation modulates the arrival rate of electrons at
a node where the current is measured. The shot noise is also considered as constant
power spectral density. The Flicker noise is called 1/f noise or pink noise. The power
spectral density of pink noise increases with the decrease of frequency, and at low
enough frequencies it dominates Johnson (thermal) and shot noise. Though the pink
noise is largely responsible for the “phase noise” in oscillators, it is negligible in case
of FET-based microwave amplifiers.
The noise generated in an amplifier is quantitatively expressed by noise factor
(NF) and noise factor expressed in dB called noise figure (NFdB ). Thus, the noise
figure can be written as
Noise Figure (NFdB ) 10 × log(Noise Factor)
The noise factor is defined as

Total available output noise power
NF
Available output noise power due to the input source
Pno
(14.7a)
GA × Pni
where Pno is the total available output noise power, GA is the available gain of the
amplifier and Pni is the available noise power of the source.
Thus, the total available output noise power can be written as
Pno GA × Pni × NF (14.7b)
Replacing the available gain by the ratio of signal power available from the net-
work PAVN and signal power available from the source PAVS , the noise factor can be
written as
PAVS /Pni
NF (14.7c)
PAVN /Pno
Thus,
Available input signal-to-noise power ratio

NF (14.8)
Available output signal-to-noise power ratio
Therefore, the noise factor is the signal-to-noise ratio (SNR) of the amplifier at
its input divided by the SNR at its output.
Noise figure of a system can also be represented by an effective input noise
temperature Te . It is the thermodynamic temperature of a resistance which is placed
at the input of the system, and establishes the same available noise power at the output
of the element as the actual element assuming the system is noise free. Suppose the
system is driven by a reference source of noise temperature To , then the total available
noise power at the output of the system will be
Pno (k Te B + k To B)GA (14.9a)
And, the available noise power at the input will be
Pni k To B (14.9b)
Thus, noise factor of the system can be represented by

(k Te B + k To B)GA Te
NF 1+
k To BGA To
Therefore, noise temperature Te of the system can be expressed as
Te (NF − 1)To (14.9c)
Example 14.1 Calculate output noise power over 4.0 kHz bandwidth of an amplifier.
The available gain and noise figure of the amplifier are 50, 3 dB, respectively.
Solution
From (14.7b), the output noise power is given by
Pno Pni × GA × NF
Suppose, the amplifier is driven by a signal source consisting of a Thevenin resis-

tance R, which is held at the standard room temperature of 300 K. In this condition,
the available output noise power density is given by
kTo × GA × NF 10 log(k) + 10 log(To ) + 10 log(GA ) + 10 log(NF)

10 log(1.38 × 10−23 ) + 10 log(300) + 50 + 3 dBW/Hz
[−228.6 + 24.77 + 53] dBW/Hz
≈ [−204 + 53] dBW/Hz −151 dBW/Hz
−121 dBm/Hz
Therefore, the total output noise power over the 4.0 kHz bandwidth is as follows:
−121 + 10 log(4.0 × 103 ) dBm

−85 dBm
An input signal of an amplifier is detectable only if its amplified output signal

power is above the noise power level by certain amount (considered as 3 dB). Suppose,
the minimum detectable signal (MDS) level is X-dB above the noise power level.
Thus, the minimum detectable output power level (PoMDS ) can be written as
PoMDS (in dBm) Pno (in dBm) + X (in dB)

10 log(kTo ) + 10 log(B) + 10 log(GA ) + 10 log(NF) + X (in dB)
Therefore, the minimum detectable input signal level PiMDS can be expressed as
PiMDS (in dBm) 10 log(kTo ) + 10 log(B) + 10 log(NF) + X (in dB)

−174 dBm + 10 log(B) + 10 log(NF) + X (in dB) (14.10)
This is the lower limit of input power level for the dynamic range of an amplifier
of noise factor NF and bandwidth B in Hz. Generally, for estimation of dynamic
range, X-dB is considered as 3-dB.
In general, input power level corresponding to 1-dB gain compression point is
considered as the higher limit of the input dynamic range. The input power level
Pi1dB corresponding to 1-dB gain compression point can be written as
Pi1dB (in dBm) Po1dB (in dBm) − G1dB (in dB) (14.11)
Here, Pi1dB , Po1dB and G1dB are the input power level, output power level and gain
of the amplifier corresponding to 1-dB gain compression point, respectively. G1dB is
nothing but the 1-dB less gain compared to the amplifier’s linear gain GL .
Therefore, the dynamic range (DR) of an amplifier can be written as
DR(in dB) Pi1dB (in dBm) − PiMDS (in dBm) (14.12a)
Using (14.10), (14.11) and considering X (in dB) 3 dB, the dynamic range can
be written as
DR(in dB) Po1dB (in dBm) − [GL (in dB) − 1]

− 10 log(kTo ) + 10 log(B) + 10 log(NF) + 3 (14.12b)
Example 14.2 Calculate the dynamic range of an amplifier of output power 1-W
at 1-dB gain compression point. The available linear gain and noise figure of the
amplifier is 44 and 5 dB, respectively. The bandwidth of the amplifier is 40 MHz.
Solution
The higher limit on input power level of the dynamic range:
Pi1dB (in dBm) Po1dB (in dBm) − [GL (in dB) − 1]

10 log(103 ) − [44 − 1]
30 − 43 −13
The lower limit of the dynamic range:

PiMDS (in Bm) 10 log(kTo ) + 10 log(B) + 10 log(NF) + 3
−174 + 10 log(40 × 106 ) + 5 + 3
−174 + 76 + 5 + 3 −90
Therefore,
DR(in dB) Pi1dB (in dBm) − PiMDS (in dBm) −13 − (−90) 77
Therefore, dynamic range of the amplifier is 77 dB.
14.2.2 Spurious-Free Dynamic Range
Another term spurious-free dynamic range (SFDR) is also used to specify the useful
operating region of an amplifier and other microwave communication circuits. The
spurious-free dynamic range measures the available dynamic range of a signal, which
is above the spurious as well as the noise floor. A signal with the spurious products
is shown in Fig. 14.4. In an amplifier spurious level increases with the increase of
output power level. Thus, in a communication system, the largest output power from
an amplifier is limited by the allowable spurious power level. Therefore, the higher
limit of the SFDR is determined by the spurious acting as noise as shown in Fig. 14.4.
In case of intermodulation distortion-limited applications, the largest output sig-
nal level of an amplifier is determined by the signal level for which the intermodula-
tion spurious level determines the minimum detectable signal as shown in Fig. 14.5.
The third-order spurious signal level is proportional to the cubic term of the input
signal level. Therefore, for each dB increase of input power level, there will be a
3-dB increase of third-order frequency power level as shown in Fig. 14.5. Generally,
spurious-free dynamic range corresponding to the third-order intermodulation lim-
ited case is smaller than the DR range considering 1-dB gain compression point as
the higher power level.
The spurious-free dynamic range (SFDR) for third-order intermodulation limited
system can be computed considering the input–output relationship of an amplifier
which is governed by the following nonlinear equation up to cubic term:
Signal
Noise SFDR Spurious

Signal floor components
level
Frequency
Fig. 14.4 Spurious-free dynamic range (SFDR)
fundamental
Output
Power MDS
3rd order
Input
SFDR Power
vo a0 + a1 vi + a2 vi2 + a3 vi3
Here, a0 , a1 , a2 , a3 are the constants. Suppose the following signal with two dif-
ferent frequencies ω1 2π f1 and ω2 2π f2 are applied to the amplifier.
vi A cos ω1 t + B cos ω2 t
Here, A and B are amplitudes of the signal. The output is given by
vo a0 + a1 (A cos ω1 t + B cos ω2 t) + a2 (A cos ω1 t + B cos ω2 t)2

+ a3 (A cos ω1 t + B cos ω2 t)3
or,
vo a0 + a1 (A cos ω1 t + B cos ω2 t)

+ a2 A2 cos2 ω1 t + B2 cos2 ω2 t + 2AB cos ω1 t cos ω2 t

+ a3 A3 cos3 ω1 t + B3 cos3 ω2 t

+3A2 B cos2 ω1 t cos2 t + 3AB2 cos ω1 t cos2 ω2 t
or,
vo a0 + a1 (A cos ω1 t + B cos ω2 t)
a2 A2 a2 B2
+ (1 + cos 2ω1 t) + (1 + cos 2ω2 t)
2 2
+ a2 AB[cos(ω1 + ω2 )t + cos(ω1 − ω2 )t]
a3 A3 a3 B3
+ (3 cos ω1 t − cos 3ω1 t) + (3 cos ω2 t − cos 3ω2 t)
4 4
2
3a3 A B
+ (1 + cos 2ω1 t) cos ω2 t
2
3a3 AB2
+ (1 + cos 2ω2 t) cos ω1 t
2
or,
a2 A2 a2 B2
vo a0 + +
2 2

3a3 A3 3a3 AB2 3a3 B3 3a3 A2 B
+ a1 A + + cos ω1 t + a1 B + + cos ω2 t
4 2 4 2
a2 A2 a2 B2
+ cos 2ω1 t + cos 2ω2 t
2 2
+ a2 AB[cos(ω1 + ω2 )t + cos(ω1 − ω2 )t]
a3 AB a3 B3
− cos 3ω1 t + cos 3ω2 t
4 4
3a3 AB
+ [A cos(2ω1 + ω2 )t + B cos(2ω2 + ω1 )t]
4
3a3 AB
+ [A cos(2ω1 − ω2 )t + B cos(2ω2 − ω1 )t] (14.13)
4
From (14.13), it is clear that the nonlinearity of the amplifier generates har-
monic components 2ω1 , 2ω2 , 3ω1 , 3ω2 as well as intermodulation components
(ω1 + ω2 ), (ω1 − ω2 ), (2ω1 + ω2 ), (2ω2 + ω1 ), (2ω1 − ω2 ), (2ω2 + ω1 ). Among all
these, third-order intermodulation components (2ω1 − ω2 ) and (2ω2 + ω1 ) are most
concerned due to their higher amplitudes compared to other intermodulation prod-

ucts and its existence within the pass band of the communication systems which
cannot be rejected by using any filter. In case of equal level of the two input carriers,
i.e., A B, the level of each third-order intermodulation components is derived from
(14.13) as follows:
3a3 A3
3rd order intermodulation level (14.14)
4
This shows that the levels of the third-order intermodulation products (IM3) are
proportional to the cubic power of the input signal level. There will be 3-dB increase
of absolute power level of IM3 products for each dB increase of input signal level.
This implies that there will be 2-dBc increase of IM3 level for 1-dB increase of input
power level.
Input–output power relationships of fundamental and third-order harmonics are
shown in Fig. 14.5. From the slope relationship, it can be written that

slope of IM3 power 3 × (slope of fundamental power)
From Fig. 14.6

PoIP3 − Po2f1 −f2 PoIP3 − Pof1
3
Pin Pin
Here, PoIP3 is the third-order intercept output power level, Po2f1 −f2 is the minimum
output detectable power level and Pof1 is the fundamental component output power
level corresponding to Po2f1 −f2 as shown is Fig. 14.6.

2
or, Pof1 − Po2f1 −f2 PoIP3 − Po2f1 −f2 (14.15a)
3
or, Pof1 − Po2f1 −f2 2 PoIP3 − Pof1 (14.15b)
Therefore, for third-order intermodulation limited system, putting PoMDS

Po2f1 −f2 in (14.15a), the SFDR can be written as
2
SFDR Pof1 − Po2f1 −f2 (PoIP3 − PoMDS ) (14.16)
3
Putting the value of PoMDS from (14.10), the SFDR can be written as
2
SFDR (PoIP3 − 10 log(kTo ) − 10 log B − 10 log GA − 10 log NF − X )
3
(14.17)
A communication transponder consists of several amplifiers as shown in Fig. 14.2.

The front-end amplifier receives and amplifies very weak signal. Thus, the front-
end amplifier is designed with minimum noise figure to achieve minimum possible
detectable signal level. Intermediate amplifiers in a communication system operate
over a moderate power level. Thus, these amplifiers are designed with maximum
achievable power gain with moderate noise figure and output power capability. Final
amplifier in a communication system is the high power amplifier to increase higher
level of the dynamic range with the target of highest DC to RF efficiency.
Example 14.3 Calculate SFDR of an amplifier with output third-order intercept point
40 dBm. The available linear gain and noise figure of the amplifier is 44 and 5 dB,
respectively. The bandwidth of the amplifier is 40 MHz.
Solution
Here,
PoIP3 40 dBm
10 log(kTo ) −174 dBm
10 log B 76 dBHz
10 log GA 44 dB
10 log NF 5 dB
Considering X 3 dB
2
SFDR (PoIP3 − 10 log(kTo ) − 10 log B − 10 log GA − 10 log NF − 3)
3
2
(40 + 174 − 76 − 44 − 5 − 3) dB 57.3 dB
3
14.3 Stability of Microwave Amplifier 363
14.3 Stability of Microwave Amplifier
While designing an amplifier, one important requirement is to ensure the stability of

the amplifier, i.e., to get rid of any oscillations. Stability in a microwave circuit can be
determined by its S-parameters. For a two-port network described by S-parameters,
the input reflection coefficient ΓIN and the output reflection coefficient ΓOUT with the
arbitrary load of reflection coefficient ΓL and arbitrary source of reflection coefficient
ΓS can be written as (from 7.177a, 7.177b):
ΓL S12 S21
ΓIN S11 + (14.18a)
1 − ΓL S22
ΓS S12 S21
ΓOUT S22 + (14.18b)
1 − ΓS S11
A two-port network will oscillate if either or both of the input and the output
ports have negative resistance. That means the network will oscillate if |ΓIN | >
1 or |ΓOUT | > 1 or both greater are than one. Though any passive network has
the magnitude of the reflection coefficients less than one but for any active device
with positive gain there is a fair possibility of having the magnitude of reflection
coefficients greater than one.
There are two types of stability for a network, unconditional stability and condi-
tional stability or potential unstability. A network is unconditional stable if the real
part of the input and output impedances of the network are positive for all passive
load and source impedances. Whereas, a network is said to be conditionally stable or
potentially unstable if the real part of the input or output impedances of the network
is negative for at least one passive load or source impedances. Also, the stability
should be checked not only over the operating frequency range but also for all other
frequencies.
The condition for unconditional stability of a network for any arbitrary passive
load (| L |<1) and passive source (| S |<1) impedances is given by
|ΓIN | < 1 (14.19a)
and
|ΓOUT | < 1 (14.19b)
From (14.19a) and (14.19b) the condition for unconditional stability for a two-
port network can be derived as [1–3]
1 − |S11 |2 − |S22 |2 + ||2

K >1 (14.20a)
2|S12 S21 |
and,
|| < 1 (14.20b)
where
|| |S11 S22 − S12 S21 | (14.20c)
This is known as the Rollett’s condition for stability of a two-port network. These
two conditions are the necessary and sufficient conditions for knowing the stability
of a two-port network. These set of conditions are valid for any passive source and
load impedances, i.e. source and load reflection coefficients of magnitude less than
one. However, it should be remembered that these conditions are not necessarily
sufficient for a source and load that have reflection gain, i.e. reflection coefficients
greater than one.
Alternatively, condition for unconditional stability that involves only one condi-
tion of single parameter is the geometrically derived criterion as [4]
1 − |S22 |2
μ1 >1 (14.21a)
S11 − S ∗ + |S12 S21 |
22
or
1 − |S11 |2
μ2 >1 (14.21b)
S22 − S ∗ + |S12 S21 |
11
Either μ1 or μ2 alone greater than unity will ensure unconditional stability of

a two-port network for any arbitrary passive source and load impedances. These
stability factors measure the distance from the centre of the Smith chart to the nearest
unstable regions.
For an unilateral network, S12 0 and thus from (14.18a) and (14.18b), the
condition for unconditional stability of a unilateral network will be |S11 | < 0 and
|S22 | < 0. Practically, there is no such transistor having S12 0, though it is very
small. Thus, |S11 | < 0 and |S22 | < 0 are not sufficient conditions for unconditional
stability of any practical device.
Practically, most microwave transistors used for amplifiers are potentially unstable
due to the existence of internal feedback. The existence of nonzero-valued gate-to-
drain capacitance Cgd in the FET and collector-to-base capacitance Ccb in a BJT
provides the internal feedback which make the device potentially unstable.
A typical S-parameter file of a pHEMT device operating in small signal condi-
tion is shown in Table 14.1 over the frequency range of 0.5–18 GHz for the device
DC bias of VDS 2V, ID 15 mA. We can check the stability of the device at
different frequencies.
Example 14.4 Check the stability at frequency 6.0 GHz of the device for which
S-parameters are given in Table 14.1.
Solution
From Table 14.1, the S-parameters of the device at 6.0 GHz are
Table 14.1 A typical S-parameter of a pHEMT device under bias condition: VDS 2 V, ID
15 mA, reference impedance: Zo 50
Freq. S11 S11 S21 S21 S12 S12 S22 S22
(GHz) (MAG) (DEG) (MAG) (DEG) (MAG) (DEG) (MAG) (DEG)
0.5 0.962 −13 6.112 166 0.0111 76 0.539 −15
1.0 0.937 −22 5.956 159 0.0211 69 0.525 −19
1.5 0.913 −33 5.810 150 0.0302 64 0.511 −24
2.0 0.889 −41 5.690 142 0.0394 58 0.498 −30
2.5 0.86 −51 5.522 133 0.0484 53 0.484 −36
3.0 0.834 −61 5.386 124 0.0567 48 0.469 −43
3.5 0.810 −71 5.236 116 0.0637 43 0.456 −49
4.0 0.784 −80 5.067 107 0.0702 38 0.440 −55
4.5 0.761 −90 4.911 99 0.0760 33 0.423 −61
5.0 0.740 −99 4.752 91 0.0809 28 0.410 −67
5.5 0.720 −107 4.586 84 0.0851 24 0.397 −73
6.0 0.701 −116 4.420 76 0.0889 19 0.385 −79
6.5 0.682 −124 4.260 69 0.0918 15 0.373 −84
7.0 0.663 −131 4.107 62 0.0941 11 0.362 −89
7.5 0.644 −139 3.974 55 0.0962 7 0.351 −93
8.0 0.627 −148 3.852 49 0.0980 3 0.343 −98
8.5 0.611 −157 3.747 42 0.0995 −1 0.333 −102
9.0 0.595 −165 3.659 35 0.1008 −5 0.323 −107
9.5 0.581 −173 3.571 29 0.1022 −9 0.313 −112
10.0 0.567 178 3.497 22 0.1039 −13 0.303 −116
10.5 0.556 170 3.430 16 0.1049 −17 0.293 −121
11.0 0.546 163 3.368 9 0.1064 −21 0.284 −127
11.5 0.537 155 3.317 3 0.1078 −26 0.274 −131
12.0 0.528 149 3.265 −4 0.1093 −30 0.265 −135
12.5 0.52 142 3.216 −10 0.1105 −35 0.255 −139
13.0 0.513 135 3.169 −17 0.1116 −39 0.246 −143
13.5 0.506 128 3.120 −24 0.1126 −44 0.235 −146
14.0 0.498 121 3.080 −30 0.1137 −49 0.225 −150
14.5 0.492 113 3.044 −37 0.1151 −54 0.215 −155
15.0 0.489 106 3.014 −44 0.1160 −59 0.207 −159
15.5 0.484 98 2.990 −51 0.1171 −65 0.200 −163
16.0 0.485 91 2.967 −58 0.1185 −71 0.193 −167
16.5 0.485 83 2.945 −65 0.1197 −77 0.187 −171
17.0 0.485 75 2.947 −71 0.1206 −82 0.182 −175
17.5 0.487 69 2.961 −77 0.1215 −87 0.177 −178
18.0 0.490 64 2.979 −81 0.1230 −90 0.174 179
S11 0.701 − 116◦

S12 0.0889 19◦
S21 4.420 76◦
S22 0.385 − 79◦
To calculate ||:
S11 S22 − S12 S21

0.701 − 116◦ × 0.385 79◦ − 0.0889 19◦ × 4.420 76◦
0.393 − 125.2◦
Thus, || 0.393 < 1

To calculate Rollett’s stability factor K:
1 − |S11 |2 − |S22 |2 + ||2

K
2|S12 S21 |
1 − (0.701)2 − (0.385)2 + (0.393)2

2(0.0889 × 4.420)
0.655
Thus, K 0.655 < 1

From (14.20a) and (14.20b), it is clear that the device at the given bias condition
is potentially unstable or conditionally stable at 6.0 GHz. This can also be verified
by computing either μ1 or μ2 as given below.
To compute μ1 :
1 − |S22 |2
μ1
S11 − S ∗ + |S12 S21 |
22
1 − (0.385)2

|0.701 − 116◦ − 0.385 79◦ × 0.393 − 125.2◦ | + |0.0889 × 4.420|
0.806 < 1
μ1 < 1 confirms the potential instability of the device at 6.0 GHz.

To compute μ2 :
1 − |S11 |2
μ2
S22 − S ∗ + |S12 S21 |
11
1 − (0.701)2

|0.385 − 79◦ − 0.701 116◦ × 0.393 − 125.2◦ | + |0.0889 × 4.420|
0.651 < 1
μ2 < 1 also confirms the potential instability of the device at 6.0 GHz.
Example 14.5 Check the stability at frequency 14.0 GHz of the same device using S-
parameters given in Table 14.1.
Solution
S11 0.498 121◦

S12 0.1137 − 49◦
S21 3.080 − 30◦
S22 0.225 − 150◦
To calculate ||:
S11 S22 − S12 S21

0.498 121◦ × 0.225 − 150◦ − 0.1137 − 49◦ × 3.080 − 30◦
0.291 83.9◦
Thus, || 0.393 < 1

To calculate Rollett’s stability factor K:
1 − |S11 |2 − |S22 |2 + ||2

K
2|S12 S21 |
1 − (0.498)2 − (0.225)2 + (0.291)2

2(0.1137 × 3.080)
1.122
Thus, K 1.122 > 1.

From (14.21a) and (14.21b), it is clear that the device at the given bias condition
is unconditionally stable at 14.0 GHz. This can also be verified by computing μ1 and
μ2 as given below.
To compute μ1 at 14.0 GHz:
1 − |S22 |2
μ1
S11 − S ∗ + |S12 S21 |
22
1 − (0.225)2

|0.498 121◦ − 0.225 150◦ × 0.291 83.9◦ | + |0.1137 × 3.080|
1.082 > 1
μ1 > 1, confirms that the device is unconditionally stable at 14.0 GHz.

To compute μ2 at 14.0 GHz:
1 − |S11 |2
μ2
S22 − S ∗ + |S2 S21 |
11
1 − (0.498)2

|0.225 − 150◦ − 0.498 − 121◦ × 0.393 − 125.2◦ | + |0.0889 × 4.420|
1.137 > 1
μ2 > 1 also confirms that the device is unconditionally stable at 14.0 GHz.
These two examples show that, though the device is potentially unstable at
6.0 GHz, the same device at the same bias condition is unconditionally stable at
14 GHz. Practically, all the microwave devices at lower operating frequency are
potentially unstable and the same device is unconditionally stable at its higher oper-
ating frequency range. Figure 14.7 shows the plot of different stability factors of the
device from 0.5 to 18 GHz based on its S-parameters written in the Table 14.1. It shows
that the device is potentially unstable over the frequency range below 11.5 GHz and
the same device is unconditionally stable over the frequency range above 11.5 GHz.
The stability condition is frequency dependent, as it is shown in the above dis-
cussions for certain matching networks it is possible for an amplifier to be stable at
its design frequency and still be unstable at other frequencies. Thus, care must be
taken to design the matching networks to ensure stability not only over the operating
frequency range but also outside the operating frequencies. A stable amplifier can
be designed using potentially unstable or conditional stable device only for certain
range of passive source and load impedances, i.e. for the amplifier both the conditions
|ΓIN | < 1 and |ΓOUT | < 1 can be satisfied.
To address the stability problems, graphical analysis is also used to determine
the regions in the Smith chart where the values of load impedances (ΓL ) and source
impedances (ΓS ) produce stable amplifier, i.e. |ΓIN | < 1 and |ΓOUT | < 1. Based on
1.5
K | |
|delta| mu1
1 mu22 2
1
1.0
11.5 GHz
0.5
| |
0.0
0 5 10 15 20
Frequency (GHz)
Fig. 14.7 Plot of stability factors of the pHEMT device over frequency
these two conditions, i.e. |ΓIN | < 1 and |ΓOUT | < 1, there are two stability circles,
source stability circle and load stability circle. The source or input stability circle is
defined as the loci of source reflection coefficient (ΓS ) for which |ΓOUT | 1. And
the load or output stability circle is defined as the loci of load reflection coefficient
(ΓL ) for which |ΓIN | 1. The input and output stability circles define the boundary
between stable and unstable regions for ΓS and ΓL , respectively.
To determine the input stability circle, the magnitude of the output reflection
coefficient will be equal to one for the source reflection coefficient. Thus, it can be
written as

ΓS S12 S21

|ΓOUT | S22 + 1 (14.22a)
1 − ΓS S11
or
|S22 (1 − ΓS S11 ) + ΓS S12 S21 | |1 − ΓS S11 |
or
|S22 − ΓS | |1 − ΓS S11 |
Squaring both the sides, it can be written as

∗
∗
|S22 |2 + ||2 |ΓS |2 − ΓS S22 + ∗ ΓS∗ S22 1 + |S11 |2 |ΓS |2 − S11 ΓS + ΓS S11
or
∗
|S11 |2 − ||2 ΓS ΓS∗ − S11 − S22
∗
ΓS − S11 − ∗ S22 ΓS∗ |S22 |2 − 1
or,
∗
∗
S11 − S22 ΓS + S11 − ∗ S22 ΓS∗ |S22 |2 − 1
ΓS ΓS∗ −
|S11 |2 − ||2 |S11 |2 − ||2
|S −S ∗ |2
now adding 11 2 222 2 to both the sides:
(|S11 | −|| )
2
∗ ∗
S11 − S22 |S22 |2 − 1 S11 − S ∗ 2
22
Γ S − + 2
|S11 |2 − ||2 |S11 |2 − ||2 |S11 |2 − ||2
or

∗ ∗
S11 − S22
S12 S21
Γ S − (14.22b)
|S11 | − || |S11 | − ||
2 2 2 2
This is an equation of circle in the ΓS plane with centre cS and radius rS as follows:

∗ ∗
S11 − S22
cS (14.23a)
|S |2 − ||2
11
S12 S21
rS (14.23b)
|S11 |2 − ||2
In a similar way, centre and radius of the output stability circle can be derived. To
determine the output stability circle, the magnitude of the input reflection coefficient
will be equal to one for the load reflection coefficient. Thus, it can be written as

ΓL S12 S21

|ΓIN | S11 + 1 (14.24a)
1 − ΓL S22
or
|S11 (1 − ΓL S22 ) + ΓL S12 S21 | |1 − ΓL S22 |
or,
|S11 − ΓL | |1 − ΓL S22 |
Squaring both the sides, it can be written as

∗
|S11 |2 + ||2 |ΓL |2 − ΓL S11 + ∗ ΓL∗ S11
∗

1 + |S22 |2 |ΓL |2 − S22 ΓL∗ + ΓL S22
or
∗
|S22 |2 − ||2 ΓL ΓL∗ − S22 − S11
∗
ΓL − S22 − ∗ S11 ΓL∗ |S11 |2 − 1
or
∗
∗
S22 − S11 ΓL + S22 − ∗ S11 ΓL∗ |S11 |2 − 1
ΓL ΓL∗ −
|S22 | − ||
2 2
|S22 |2 − ||2
|S − S ∗ |2
now adding 22 2 112 2 to both the sides:
(|S22 | −|| )
2
S22 − S11∗ ∗
|S11 |2 − 1 S22 − S ∗ 2
11
ΓL − + 2
|S22 |2 − ||2 |S22 |2 − ||2 |S22 |2 − ||2
or

∗ ∗
S22 − S11
S12 S21
ΓL − (14.24b)
|S22 |2 − ||2 |S22 |2 − ||2
This is an equation of circle in the ΓL plane with centre cL and radius rL as follows:

∗ ∗
S22 − S11
cL (14.25a)
|S22 |2 − ||2

S12 S21
rL (14.25b)
|S22 |2 − ||2
Using Eqs. (14.23a), (14.23b) and (14.25a), (14.25b), input and output stability
circles can be drawn and stable and unstable regions in the ΓS and ΓL plane can be
determined.
For a conditionally stable device, input and output stability circles are shown in
Fig. 14.8a, b, respectively. In the ΓS plane, on one side of the boundary of the input
stability circle, |ΓOUT | < 1 and on the other side of the boundary, |ΓOUT | > 1.
At the origin of the Smith chart, ΓS 0 thus from (14.18b) the output reflection
coefficient, ΓOUT S22 . Therefore, if |S22 | < 1 then |ΓOUT | < 1, i.e. the origin of
the Smith chart represents a stable operating point. On the other hand, if |S22 | > 1
then |ΓOUT | > 1, i.e. the origin of the Smith chart represents an unstable operating
point. This is illustrated in Fig. 14.8a.
Similarly, in the ΓL plane, on one side of the input stability circle boundary
|ΓIN | < 1 and on the other side of the boundary |ΓIN | > 1. At the origin of the Smith
chart, ΓL 0, and thus from (14.18a) the input reflection coefficient ΓIN S11 .
Therefore, if |S11 | < 1 then |ΓIN | < 1, i.e. the origin of the Smith chart represents
a stable operating point. On the other hand, if |S11 | > 1 then |ΓIN | > 1, i.e. the
origin of the Smith chart represents an unstable operating point. This is illustrated in
Fig. 14.8( b).
For an unconditionally stable device, any passive load and source impedance
produce stable amplifier. For an unconditionally stable device, |S11 | > 1 or |S22 | > 1
is not possible. Otherwise, the source and load impedance corresponding to ΓS 0
or ΓL 0 will produce |ΓOUT | > 1 or |ΓIN | >1, which is contradictory to the
condition for unconditional stable device.
Therefore, for |S11 | < 1 and |S22 | < 1, the input and the output stability circles
should fall completely outside the Smith chart or to completely enclose the Smith
chart as shown in Fig. 14.9. Therefore, the conditions for unconditional stability for
all passive source and load impedances can be written as
||cS | − rS | > 1 for |S22 | < 1 (14.26a)

||cL | − rL | > 1 for |S11 | < 1 (14.26b)
Example 14.6 Compute the stability circles at 6.0 GHz of the device for S-parameters
which are given in Table 14.1.
Fig. 14.8 Stability circles for conditionally stable device, a ΓS plane and b ΓL plane
Solution
For input stability circle at 6.0 GHz:

∗ ∗
S11 − S22
cS
|S11 |2 − ||2
(0.701 − 116◦ − 0.393 − 125.2◦ × 0.385 79◦ )∗

(0.701)2 − |0.393|2
(0.701 − 116◦ − 0.151 − 46.2◦ )∗

0.337
0.664 128.4◦
1.973 128.4◦
0.337
Fig. 14.9 Stability circles for unconditionally stable devices, a ΓS plane and b ΓL plane

S12 S21

rS
|S11 |2 − ||2
0.0889 × 4.420

(0.701)2 − |0.393|2
0.3929
1.167
0.337
Therefore, ||cS | − rS | 1.973 − 1.167 0.806 ( μ1 (!!) Example 14.4).
The input stability circle and stable region on the Smith chart is shown in
Fig. 14.10a.
For output stability circle at 6.0 GHz:

∗ ∗
S22 − S11
cL
|S22 |2 − ||2
(0.385 − 79◦ − 0.393 − 125.2◦ × 0.701 116◦ )∗

(0.385)2 − |0.393|2
−60.063 120.7◦

S12 S21
rL
|S |2 − ||2 22
0.0889 × 4.420

(0.385)2 − |0.393|2
0.3929
60.713
0.00622
Therefore, ||cL | − rL | 0.651 ( μ2 (!!) Example 14.4).
The output stability circle and the stable region on the Smith chart are shown in
Fig. 14.10b.
Example 14.7 Compute the stability circles at 14.0 GHz of the device for S-
parameters which are given in Table 14.1.
Solution
For input stability circle at 14.0 GHz:

∗ ∗
S11 − S22
cS
|S11 |2 − ||2
(a) Input stability circle (b)

= 1.973 128.4o Output stability circle
= 60.063 59.3o
= 1.167
1.0 j 1.0=j 60.713
2.0 j 0.5 j 2.0 j

0.5 j
0.0 0.5 1.0 0.0 0.5 1.0 2.0

2.0
| |< 1 | |< 1
-2.0 j -0.5 j -2.0 j

-0.5 j
-1.0 j -1.0 j
Fig. 14.10 Stable and unstable regions of the device

(0.498 121◦ − 0.291 83.9◦ × 0.225 150◦ )∗

(0.498)2 − |0.291|2
0.527 − 114.42◦
3.227 − 114.42◦
0.163
S12 S21

rS
|S11 |2 − ||2
0.1137 × 3.080

(0.498)2 − |0.291|2
0.3502
2.145
0.163
||cS |−rS | 3.227 − 2.145 1.082 ( μ1 (!!) Example 14.5).
The input stability circle is shown in Fig. 14.11a. It shows that for the uncondi-
tionally stable device, the input stability circle is outside the Smith chart.
For output stability circle at 14.0 GHz:

∗ ∗
S22 − S11
cL
|S22 |2 − ||2
(0.225 − 150◦ − 0.291 83.9◦ × 0.498 − 121◦ )∗

(0.225)2 − |0.291|2
−9.126 175.4◦

S12 S21
rL
|S22 |2 − ||2

0.1137 × 3.080

(0.225)2 − |0.291|2
0.350
10.263
0.034
(a) (b)
= 2.145
= 3.227 − 114.42o
= 9.126 − 4.6o
= 10.263
Input stability circle
Output stability circle
Fig. 14.11 Stability circles for unconditionally stable devices

Therefore, ||cL | − rL | 1.137 ( μ2 (!!) Example 14.5).

The output stability circle is shown in Fig. 14.11( b). It shows that for the uncon-
ditionally stable device, the output stability circle encloses the Smith chart.
14.4 Single-Stage Amplifier Design
Knowing the stability of a device, input and output matching networks are designed
to ensure unconditional stability of the amplifier over its entire frequency range
where the device has gain. Stability of an amplifier using an unconditionally stable
device is not an issue as long as the passive matching networks are used and there
is no positive feedback. To design an amplifier using an unconditional stable device
with maximum gain, complex conjugate matching at its input and output is done.
However, to realize a stable amplifier care must be taken to avoid any unwanted
positive feedback due to microwave radiation from the output network that couples
to its input network.
There are many ways to design a stable amplifier using a potentially unstable
device. As already discussed, use of lossless appropriate input and output matching
networks that ensure the source and load impedances outside the unstable region in
the Smith chart makes the amplifier stable. Other ways to design stable amplifier are
the use of resistive loading at the device input and/or output, and use of negative
feedback.
Transducer power gain GT is defined as the ratio of power delivered to the load
(PL ) and power available (PAVS ) from the source. This is already derived in Chap. 7
and is written here as follows:
PL 1 − |ΓS |2 2 1 − |ΓL |
2
GT |S21 | (14.27)
PAVS |1 − ΓS ΓIN |2 |1 − S22 ΓL |2
Thus, the transducer power gain depends on both the source and the load reflection
coefficients. With reference to Fig. 14.1 and (14.27), the transducer power gain GT
of an amplifier can be expressed as separate gain factors as
GT GS × Go × GL (14.28)
where GS , the first term, is gain factor due to the input (source) matching network,
Go , the middle term, is the gain factor for the transistor itself, and GL , the last term,
is the gain factor for the output (load) matching network. For a given transistor with
particular bias condition, while input of the device is fed by a source of reference
impedance (Ro ) and output is terminated with the reference load impedance (Ro ), then
the gain of the device is fixed and is given by, GT Go |S21 |2 . Thus, the overall
gain of the amplifier is controlled by the gains GS and GL which are determined by
the input and output matching networks. Maximum gain may be realized when the
matching networks provide a conjugate match to the input and output of the device.
14.4 Single-Stage Amplifier Design 377
However, the conjugate matching at both the terminals are possible only when the
device is unconditionally stable at the operating frequency band. If the device is
not unconditionally stable, then the conjugate matching will not be permissible at
both the device terminals. In this case, input and output matching networks must be
selected to ensure load and source impedances within the respective stable regions
of the Smith chart.
14.4.1 Amplifiers Using Unconditionally Stable Device
A transistor over its unconditionally stable frequency range provides stable gain for
any passive source and load impedances. Therefore, depending on the performance
requirements, the input and output matching networks can be designed without any
concern of stability of the amplifier. For example, to design a low-noise amplifier, the
input matching network should provide input termination to the device corresponding
to its minimum noise figure requirement and output can be conjugate matched for
achieving maximum possible gain. In case of high-gain amplifier, both the input
and output ports of the device can be conjugately matched. Therefore, to achieve
maximum gain, simultaneous conjugate matching is done for an unconditional stable
transistor. The conditions for simultaneous conjugate match at input and output are
ΓIN ΓS∗ , and ΓOUT ΓL∗ (14.29)
Therefore, from (14.18a) and (14.18b):

S11 − ΓS∗ (1 − S22 ΓL ) + S12 S21 ΓL 0 (14.30a)

S22 − ΓL∗ (1 − S11 ΓS ) + S12 S21 ΓS 0 (14.30b)
Or,
ΓS∗ − S11
ΓL (14.31a)
ΓS∗ S22 −
Γ ∗ − S22
ΓS ∗L (14.31b)
ΓL S11 −
Computing ΓL∗ using (14.31a) and putting it into (14.30b), the source reflection
coefficient ΓS ΓSM , corresponding to the simultaneous conjugate matching of
both the ports can be written as
∗
2 ∗
S11 − S22 ΓSM − 1 + |S11 |2 − |S22 |2 − ||2 ΓSM + S11 − S22 ∗ 0
or,
C1 ΓSM
2
− B1 ΓSM + C1∗ 0 (14.32) (14.32)
where
∗
B1 1 + |S11 |2 − |S22 |2 − ||2 and C1 S11 − S22
Similarly, computing ΓS∗ using (14.31b) and putting it into (14.30a), the load
reflection coefficient ΓL ΓLM , corresponding to the simultaneous conjugate match-
ing of both the ports can be written as
∗
2 ∗
S22 − S11 ΓLM − 1 + |S22 |2 − |S11 |2 − ||2 ΓLM + S22 − S11 ∗ 0
or,
C2 ΓLM
2
− B2 ΓLM + C2∗ 0 (14.33) (14.33)
where
∗
B2 1 + |S22 |2 − |S11 |2 − ||2 and C2 S22 − S11
From (14.32) and (14.33), values of source reflection coefficient ΓSM and load
reflection coefficients ΓLM corresponding to simultaneous conjugate matching at
input and output can be written as

B1 |C1 | B1 2
ΓSM ± −1 (14.34a)
2C1 C1 2|C1 |

B2 |C2 | B2 2
ΓLM ± −1 (14.34b)
2C2 C2 2|C2 |
For passive matching networks at both the input and output, |ΓSM | < 1
and |ΓLM | < 1. The plus (+) or minus (−) sign in the expressions should be
selected accordingly.
Putting the value of ΓIN from (14.18a) in (14.27), the transducer power gain can
be written as

1 − |ΓS |2 |S21 |2 1 − |ΓL |2
GT (14.35)
|(1 − S11 ΓS )(1 − S22 ΓL ) − S12 S21 ΓS ΓL |2
Under simultaneous conjugate math conditions (ΓS ΓSM and ΓL ΓLM ), the
maximum transducer gain can be expressed as

1 − |ΓSM |2 |S21 |2 1 − |ΓLM |2
GT max (14.36)
|(1 − S11 ΓSM )(1 − S22 ΓLM ) − S12 S21 ΓSM ΓLM |2
1 − |ΓSM |2 2 1 − |Γ LM |
2
|S21 |
1 − |ΓSM |2 2 |1 − S22 ΓLM |2
2 1 − |ΓLM |
2
1
|S21 | GP max (14.37)
1 − |ΓSM |2 |1 − S22 ΓLM |2
Comparing with (7.182), this expression (14.37) is nothing but the maximum
operating power gain GP max . This can also be expressed as
1 − |ΓSM |2 1
GT max |S21 |2 GA max (14.38)
|1 − S11 ΓSM |2 1 − |ΓLM |2
Comparing with (7.183), this is the expression of maximum available gain GA max .
Therefore, a two-port network under conjugate match conditions at its both the ports
provides maximum available gain. The maximum available gain for an unconditional
stable device can also be expressed in terms of its stability factor as (to be proved in
Sect. 14.5.3):
|S21 |

GT max k − k2 − 1 (14.39)
|S12 |
This can be expressed in dB as

|S21 |
GA max (dB) 10 log + 10 log k − k 2 − 1 (14.40)
|S12 |
For unilateral transistor, S12 0 which leads to

∗ ∗
ΓSM S11 and ΓLM S22
Therefore, from (14.38) the maximum available gain for a unilateral device can
be expressed as
1 1
(GA max )UNILATERAL |S21 |2 (14.41)
1 − |S11 |2 1 − |S22 |2
Example 14.8 Calculate maximum available gain at 14.0 GHz of the device of S-
parameters as in Table 14.1.
Solution
From (14.39), the maximum available gain:
|S21 |

GT max k − k2 − 1
|S12 |
S11 0.498 121◦

S12 0.1137 − 49◦
S21 3.080 − 30◦
S22 0.225 − 150◦
Therefore,
S11 S22 − S12 S21

0.498 121◦ × 0.225 − 150◦ − 0.1137 − 49◦ × 3.080 − 30◦
0.291 83.9◦
Thus, stability factor:
1 − |S11 |2 − |S22 |2 + ||2

K
2|S12 S21 |
1 − (0.498)2 − (0.225)2 + (0.291)2

2(0.1137 × 3.080)
1.122

3.080
GA max (dB) 10 log + 10 log 1.122 − 1.1222 − 1
0.1137
14.43 dB − 1.922 dB 12.5 dB
14.4.2 Amplifier Using Conditionally Stable Device
Conjugate matching is not allowed for an amplifier at a frequency where the device
is potentially unstable. Either input or output or both ports of the device should be
terminated with an impedance which is not corresponding to conjugate match. The
load and source impedances should be within the stable region of the Smith chart.
Maximum stable gain from a device can be achieved when the stability factor k 1.
Thus, the maximum stable gain MSG can be expressed as
|S21 |
MSG (14.42a)
|S12 |
This can be expressed in dB as:

|S21 |
MSG(dB) 10 log (14.42b)
|S12 |
Example 14.9 Compute maximum stable gain at 6.0 GHz of the device (Ref.
Table 14.1).
Solution
S12 0.0889 19◦

S21 4.420 76◦
From (14.42b), the maximum stable gain at 6 GHz:

4.420
MSG(dB) 10 log
0.0889
17.32 dB
Figure 14.12 shows the plot of gain over frequency for a pHEMT device whose S-
parameters are written in Table 14.1. The maximum stable gain (MSG) of the device
is shown over the frequency range where it is conditionally stable and maximum
available gain (MAG) is written over the unconditionally stable frequency range.
The plot shows an important characteristic of decrease of gain with increase of
operating frequency. This creates a challenge for an amplifier designer to design an
amplifier with a flat gain response over a broad range of frequency bands.
Gain (dB)
30
MSG (dB) MAG (dB)
25 < 1
20 MSG
> 1
15 MAG
10
0 5 10 15 20
Frequency (GHz)
Fig. 14.12 Plot of maximum stable gain (MSG) and maximum available gain (MAG) of the pHEMT
device (S-parameter Table 14.1)
14.5 Amplifier with Specific Gain
Under both input and output conjugate matched condition, a transistor over its uncon-
ditional stable frequency region provides maximum gain. Similarly, over its condi-
tional stable frequency range, the transistor provides maximum gain for the input and
output matching condition for which stability factor of the amplifier becomes one. In
many cases, it is desirable to design an amplifier for less than the maximum obtainable
gain to improve other performance such as operating bandwidth, minimization of the
effect of device parameters variations on amplifier’s performance, noise figure, out-
put power capability, stability, etc. Therefore, in many cases, it is required to design
the input and/or output matching networks which purposely introduce impedance
mismatches to reduce the overall gain. For designing a low-noise amplifier, it is
required to match the input terminal of a transistor other than corresponding to max-
imum gain, i.e. it is required to intentionally introduce input mismatch to achieve
the best noise figure. Similarly, conditions for achieving highest output power and
highest DC to RF efficiency from a transistor are different from its output matching
network corresponding to achieving maximum gain. Therefore, in case of power
amplifier design, it is required to intentionally introduce some mismatch at device
output. The design for achieving specific gain can be done by plotting constant gain
circles on a Smith chart, by representing loci of the specific fixed values of gain for
the input and output sections.
14.5.1 Amplifier with Specific Transducer Power Gain
The transducer power gain GT is given by (14.27) and (14.28):
1 − |ΓS |2 1 − |ΓL |2
GT × |S21 | 2
× GS × Go × GL
|1 − ΓS ΓIN |2 |1 − S22 ΓL |2
where GS is the source gain factor and is given by
1 − |ΓS |2
GS (14.43a)
|1 − ΓS ΓIN |2
GL is the load gain factor and is given by
1 − |ΓL |2
GL (14.43b)
|1 − S22 ΓL |2
and Go |S21 |2 is the device gain factor. It is the gain of the device when source and
load impedances are the reference impedance (Ro ).
14.5 Amplifier with Specific Gain 383
The amplifier will provide maximum gain when the source gain and load gain
factors become maximum. Therefore, the maximum obtainable gain will be
GTmax GSmax × Go × GLmax (14.44)
The transducer power gain GT depends upon both the source and load impedances.
Thus, to achieve a specific gain, which is less than the maximum achievable gain,
the source gain factor (GS ) and load gain factor (GL ) need to be adjusted lower than
their maximum values GSmax and GLmax , respectively. Practically all the transistors
are bilateral, i.e. S12 0. In this case, the input and output reflection coefficient,
ΓIN and ΓOUT depend on the load and source impedances, respectively. Therefore, for
a specific transducer power gain using a bilateral transistor, it is difficult to determine
the source and load impedances. To simplify the procedure here we consider the
device as unilateral device, i.e. S12 0.
For a unilateral transistor, ΓIN S11 and ΓOUT S22 . Therefore, transducer
power gain in unilateral case GTU can be written as (14.35):
1 − |ΓS |2 1 − |ΓL |2
GTU × |S21 | 2
× GSU × Go × GLU (14.45)
|1 − S11 ΓS |2 |1 − S22 ΓL |2
Thus, for unconditional stable unilateral device, the maximum transducer power
∗ ∗
gain GTUmax will be achieved when ΓS S11 and ΓL S22 and is given by
1 1
GTUmax × |S21 |2 × GSUmax × Go × GLUmax
1 − |S11 | 2
1 − |S22 |2
The normalized gain factors gSU and gLU can be written as
GSU 1 − |ΓS |2
gSU × 1 − |S11 |2 (14.46a)
GSUmax |1 − S11 ΓS |2
GLU 1 − |ΓL |2
gLU × 1 − |S22 |2 (14.46b)
GLUmax |1 − S22 ΓL | 2
Expanding (14.46a):

gSU |1 − S11 ΓS |2 1 − |ΓS |2 × 1 − |S11 |2
or
∗

gSU |S11 |2 + 1 − |S11 |2 |ΓS |2 − gSU S11 ΓS + S11 ΓS∗ 1 − |S11 |2 − gSU
or
∗

gSU S11 ΓS + S11 ΓS∗ 1 − |S11 |2 − gSU
ΓS ΓS∗ −
1 − (1 − gSU )|S11 | 2
1 − (1 − gSU )|S11 |2
or
2 √ 2
g S ∗ 1 − gSU 1 − |S11 |2
Γ S − SU 11 (14.47)
1 − (1 − gSU )|S11 |2 1 − (1 − gSU )|S11 |2
This is the form of |ΓS − CS |2 |RS |2 , which is equation of a circle.

Therefore, the expression (14.47) represents circle, called constant source gain
circle, in the ΓS plane for a fixed value of source gain factor gSU . The centre CS and
radius RS of the constant gain circle in the ΓS plane are given by
∗
gSU S11
CS (14.48a)
1 − (1 − gSU )|S11 |2
√
1 − gSU 1 − |S11 |2
RS (14.48b)
1 − (1 − gSU )|S11 |2
In a similar way expanding (14.46b), it can be written as

2 √ 2
g S ∗ 1 − gLU 1 − |S22 |2
ΓL − LU 22 (14.49)
1 − (1 − gLU )|S22 |2 1 − (1 − gLU )|S22 |2
The expression (14.49) also represents circle, called constant load gain circle, in
the ΓL plane for a fixed value of load gain factor gLU . The centre CL and radius RL
of the constant gain circle in the ΓL plane are given by
∗
gLU S22
CL (14.50a)
1 − (1 − gLU )|S22 |2
√
1 − gLU 1 − |S22 |2
RL (14.50b)
1 − (1 − gLU )|S22 |2
From (14.48a) and (14.50a), it is clear that the centres of each family of circles
corresponding to different values of gSU , and gLU lie along straight lines given by
∗ ∗
the angle of S11 and S22 , respectively. Corresponding to the maximum source gain
factor, gSU 1, the radius of the constant input gain circle RS 0 and centre of
∗
the circle become CS S11 . Similarly, for the maximum- load gain factor, gLU 1
the radius of the constant output gain circle RL 0 and centre of the circle become
∗
CL S22 . It can also be shown that the circles correspond to 0 dB source and load
gain factors will pass through the centre of the Smith chart.
Example 14.10 Computation of constant transducer power gain circles and plotting
them in Smith charts of a unilateral device with following S-parameters:
S11 0.498 121◦

S12 0
S21 3.080 − 30◦

S22 0.225 − 150◦
Solution
1 1
GSUmax 1.33 1.238 dB
1 − |S11 | 2 1 − 0.4982
Go |S21 |2 3.0802 9.486 9.771 dB
1 1
GLUmax 1.053 0.226 dB
1 − |S22 | 2 1 − 0.2252
Thus,
GTUmax GSUmax × Go × GLUmax

1.33 × 9.486 × 1.953 13.288
(1.238 + 9.771 + 0.226) dB
11.234 dB
Thus, the maximum obtainable gain from the device is 11.234 dB when both input
and outputs are conjugate matched. To achieve specific gain which is other than this
maximum gain, it is required to introduce mismatches at the input and/or output of
the device. We can draw constant input and output gain circles for different values
of input and output gain factors.
Tables 14.2 and 14.3 show the centre and radius of circles that correspond to
different values of source and load gain factors, respectively. Constant source gain
circles are shown in Fig. 14.13 on a ΓS plane corresponding to specific source gain
of 1.238, 1.0, 0.5, 0.229, 0.0, −0.5 and −1.0 dB using the parameters shown in
Table 14.2. The 0 dB gain circle passes through the centre of the Smith chart. The
centre of all the circles lies on a straight line (dotted line) passing through the centre
of the Smith chart. For a specific source gain, the source reflection coefficient may
be selected from values lying on the corresponding circle. In the same way, constant
load gain circles corresponding to 0.226, 0, −0.5, −1.0 and −2.0 dB are drawn in
the ΓL plane as shown in Fig. 14.14.
To obtain a specific transducer gain from the unilateral device, the source and
load reflection coefficients should be selected from the specific values of constant
gain circles. For example, to achieve 10 dB transducer gain, GSU 0.229 dB and
GLU 0 dB can be selected. So that the transducer gain will be
GTU (0.229 + 9.771 + 0) dB 10.0 dB
To achieve this, the source reflection coefficient should be chosen any value on
the 0.229 dB source gain factor circle and load reflection coefficient will be zero.
Table 14.2 Parameters for specific transducer power gain circles corresponding to specific source
gain factors of a unilateral device : S11 0.498 121◦ , S12 0, S21 3.080 − 30◦ , S22
0.225 − 150◦
∗
gSU S11
GSU (dB) GSU (ratio) g SU CS 1−(1−g S√
R
SU )|S 11 |2

GSU 1−gSU 1−|S11 |2
GSUmax 1−(1−gSU )|S11 |2
√
1 × 0.498 − 121◦ 1 − 1 1 − 0.4982
1.238 1.33 1.0 1 − (1 − 1)0.4982 1 − (1 − 1)0.4982
0.498 − 121◦ 0
√
0.947 × 0.498 − 121◦ 1 − 0.947 1 − 0.4982
1.0 1.259 0.947 1 − (1 − 0.947)0.4982 1 − (1 − 0.947)0.4982
0.478 − 121◦ 0.176
√
0.844 × 0.498 − 121◦ 1 − 0.844 1 − 0.4982
0.5 1.122 0.844 1 − (1 − 0.844)0.4982 1 − (1 − 0.844)0.4982
0.437 − 121◦ 0.309
√
0.793 × 0.498 − 121◦ 1 − 0.793 1 − 0.4982
0.229 1.054 0.793 1 − (1 − 0.793)0.4982 1 − (1 − 0.793)0.4982
0.416 − 121◦ 0.361
√
0.752 × 0.498 − 121◦ 1 − 0.752 1 − 0.4982
0 1.0 0.752 1 − (1 − 0.752)0.4982 1 − (1 − 0.752)0.4982
0.399 − 121◦ 0.399
√
0.670 × 0.498 − 121◦ 1 − 0.670 1 − 0.4982
−0.5 0.891 0.670 1 − (1 − 0.670)0.4982 1 − (1 − 0.670)0.4982
0.363 − 121◦ 0.470
√
1 × 0.597 − 121◦ 1 − 597 1 − 0.4982
−1.0 0.794 0.597 1 − (1 − 597)0.4982 1 − (1 − 597)0.4982
0.330 − 121◦ 0.530
However, for achieving broadband gain performance, it is preferable to select the

source and load reflection coefficients on the corresponding gain circles near the
centre of the Smith chart.
14.5.2 Amplifier with Specific Available Power Gain
Transducer power gain depends on both load and source matching networks. Due
to the inter-dependency of the load and source impedances in case of bilateral tran-
sistors (S21 0), it is practically difficult to design an amplifier with specific gain.
It becomes more complicated in case of a conditionally stable device. The available
Table 14.3 Parameters for specific transducer power gain circles corresponding to specific load
gain factors of a unilateral device: S11 0.498 121◦ , S12 0, S21 3.080 − 30◦ , S22
0.225 − 150◦
∗
gLU S22
GLU (dB) GLU (ratio) g LU CL 1−(1−g L√
R
LU )|S 22 |2

GLU 1−gLU 1−|S22 |
2
GLUmax 1−(1−gLU )|S22 |2
√
1 × 0.225 150◦ 1 − 1 1 − 0.2252
0.226 1.053 1.0 1 − (1 − 1)0.2252 1 − (1 − 1)0.2252
0.225 150◦ 0
√
0.949 × 0.225 150◦ 1 − 0.949 1 − 0.2252
0 1.0 0.949 1 − (1 − 949)0.2252 1 − (1 − 949)0.2252
0.214 150◦ 0.214
√
0.846 × 0.225 150◦ 1 − 0.846 1 − 0.2252
−0.5 0.891 0.846 1 − (1 − 0.846)0.2252 1 − (1 − 0.846)0.2252
0.192 150◦ 0.375
√
0.754 × 0.225 150◦ 1 − 0.754 1 − 0.2252
−1.0 0.794 0.754 1 − (1 − 0.754)0.2252 1 − (1 − 0.754)0.2252
0.172 150◦ 0.477
√
0.599 × 0.225 150◦ 1 − 0.599 1 − 0.2252
−2.0 0.631 0.599 1 − (1 − 0.599)0.2252 1 − (1 − 0.599)0.2252
0.138 150◦ 0.614
power gain is independent of the load impedance. Thus, it is convenient to use avail-
able power gain for designing a low-noise amplifier where specific source impedance
is required for achieving a minimum noise figure.
Available power gain GA of a network is defined as the ratio of power available
from the network PAVN and power available from the source PAVS . Thus for a two-port
network, from (7.183), the available power gain can be rewritten as
PAVN 1 − |ΓS |2 1
GA |S21 |2 (14.51a)
PAVS |1 − S11 ΓS |2 1 − |ΓOUT |2
The available power gain depends on the input matching condition, ΓS because
actual power transferred to the transistor is not necessarily the same as available power
from the source which happens only for conjugate match at the input. However, GA
is independent of the input reflection coefficient ΓIN and load impedance ΓL . The
available power gain GA can be written as
GA gA |S21 |2 (14.51b)
where
1.0 j
0.5 j
2.0 j
Source Gain
factor (dB )
-1.0
-0.5
0.0 0.5 1.0 2.0
0.229
0.5
1.0
1.238
Locus of centre
of circles
-0.5 j -2.0 j
-1.0 j
Fig. 14.13 Circles on ΓS plane corresponding to constant source gain factors
1 − |ΓS |2
gA
|1 − S11 ΓS |2 1 − |ΓOUT |2
1 − |ΓS |2

S22 − ΓS 2
|1 − S11 ΓS | 1 − 1−S11 ΓS
2
1 − |ΓS |2

|1 − S11 ΓS |2 |S22 − ΓS |2
1 − |ΓS |2
(14.52)
1 − |S22 |2 + |ΓS |2 |S11 |2 − ||2 − 2 Re(ΓS C1 )
Then (14.52) can be expressed as

2
gA C1∗

Γ S −
1 + gA |S11 |2 − ||2
1 2

1 − 2k gA |S12 S21 | + gA2 |S12 S21 |2 2

(14.53)
1 + gA |S11 |2 − ||2
1.0 j
Locus of centre
2.0 j
of circles 0.5 j
Load Gain
factor (dB)
0.226
0.5 1.0 2.0
0
0.0
-0.5
-1.0
-2.0
-0.5 j
-2.0 j
-1.0 j
Fig. 14.14 Circles on ΓL plane corresponding to constant load gain factors
The expression (14.53) represents circle in the ΓS plane for a fixed value of
normalized available gain gA . The centre CA and radius RA of the constant gain circle
in the ΓS plane are given by
gA C1∗
CA (14.54a)
1 + gA |S11 |2 − ||2
1
1 − 2kgA |S12 S21 | + gA2 |S12 S21 |2 2
RA (14.54b)
1 + gA |S11 |2 − ||2
The centres of the constant gA circles are always on the line drawn between C1∗
and the origin of the ΓS plane.
In case of unconditionally stable device, RA becomes zero when gA reaches to its
maximum value.
In case of potentially unstable device, when gA is infinite, the constant gain circle
become input stability circle and its centre and radius are given by
C1∗
CA (14.55a)
|S11 |2 − ||2
|S12 S21 |
RA (14.55b)
|S11 |2 − ||2
Example 14.11 Computation of constant available power gain circles of a bilateral

device with S-parameters: S11 0.498 121◦ , S12 0.1137 −49◦ , S21 3.080 −
30◦ , S22 0.225 − 150◦ .
Solution
0.498 121◦ × 0.225 − 150◦ − 0.1137 − 49◦ × 3.080 − 30◦

0.291 83.9◦ , Thus, || 0.291 < 1
The stability factor:
1 − (0.498)2 − (0.225)2 + (0.291)2

K 1.122 > 1
2(0.1137 × 3.080)
Thus, the device is unconditional stable. The maximum available gain:

3.080
GA max (dB) 10 log + 10 log 1.122 − 1.1222 − 1
0.1137
14.328 dB − 2.127 dB 12.2 dB
∗
C1 S11 − S22
0.498 121◦ − (0.291 83.9◦ ) × (0.225 150◦ )
0.498 121◦ − 0.291 83.9◦ × 0.225 150◦ 0.527 114.4◦
C1∗ 0.527 − 114.4◦
And,
|S12 S21 | |0.1137 − 49◦ × 3.080 − 30◦ | 0.350

|S11 |2 − ||2 0.4982 − 0.2912 0.163
Using the values of these parameters, centre and radius of constant available
power gain circles given by (14.54a) and (14.54b) corresponding to 12.2, 10.0, 8.0
and 6.0 dB are calculated as shown in Table 14.4. It shows that for maximum available
gain, the radius of the constant gain circle becomes zero and angle of the centre of
all the circles are same as the angle of C1∗ .
The constant available gain circles of the device are plotted on Smith chart shown
in Fig. 14.15. The centres of all the circles are always on the line drawn between
C1∗ and origin of the ΓS plane as shown by a dashed line. The radii of constant gain
circles are getting smaller with the increase of gain.
To achieve constant available gain, the input matching should provide any source
reflection coefficient ΓS which lie on that constant available gain circle.
Table 14.4 Parameters for available power gain circles of a bilateral device: S11
0.498 121◦ , S12 0.1137 − 49◦ , S21 3.080 − 30◦ , S22 0.225 − 150◦

1
1−2kgA |S12 S21 |+gA2 |S12 S21 |2 2
GA (dB) g A A
C RA
1+gA |S11 |2 −||2

GA g C∗
|S21 |2 A 1
1+gA |S11 |2 −||2
16.596 0.922 − 114.4◦

12.2 3.0802 1.286 0
1.286 0
1.749 0.717 − 114.4◦
10 0.555 − 114.4◦
10.0 3.0802 1.172 0.555
1.172 0.473
1.054 0.474 − 114.4◦
6.310 0.350 − 114.4◦
8.0 3.0802 1.109 0.729
1.109 0.658
0.665 0.316 − 114.4◦
3.981 0.221 − 114.4◦
6.0 3.0802 1.069 0.832
1.069 0.778
0.420 0.207 − 114.4◦
1.0 j
0.5 j 2.0 j
Available
gain (dB)
0.5 1.0 2.0
0
6.0
8.0
10.0
12.2
Locus of
centre of -0.5 j
-2.0 j
the circles
-1.0 j
Fig. 14.15 Constant available gain circles on ΓS plane

Example 14.12 Compute available gain circles of the device at frequency 6.0 GHz,
where S-parameters are given by S11 0.701 − 116◦ , S12 0.0889 19◦ , S21
4.420 76◦ , S22 0.385 − 79◦
Solution
0.701 − 116◦ × 0.385 − 79◦ − 0.0889 19◦ × 4.420 76◦

0.393 − 125.2◦
Thus, || 0.393 < 1
1 − (0.701)2 − (0.385)2 + (0.393)2

K 0.655 < 1
2(0.0889 × 4.420)
Thus the device is conditionally stable. It is required to draw the input stability
circle for the device. The source and load reflection coefficients should be in the
stable region on ΓS and ΓL plane, respectively.
The centre and radius of the input stability circle is given by
(0.701 − 116◦ − 0.393 − 125.2◦ × 0.385 79◦ )∗

cS
(0.701)2 − |0.393|2
◦
1.973 128.4
0.0889 × 4.420
rS 1.167
(0.701)2 − |0.393|2
The maximum available gain of the device:

4.420
GA max (dB) 10 log 16.97 dB
0.0889
∗
C1 S11 − S22
0.701 116◦ − (0.393 − 125.2◦ ) × (0.385 − 79◦ )
0.664 − 128.4◦
C1∗ 0.664 128.4◦
And,
|S12 S21 | |0.0889 19◦ × 4.420 76◦ | 0.393

|S11 |2 − ||2 0.7012 − 0.3932 0.337
Using the values of these parameters, centre and radius of the constant available
power gain circles given by (14.54a) and (14.54b) corresponding to 16.0, 14.0, 12.0,
10.0 and 8.0 dB are tabulated in Table 14.5. It shows that angle of the centre of all
the constant available gain circles are the same as angle of C1∗ .
Table 14.5 Parameters for available power gain circles of a bilateral device: S11 0.701 −
116◦ , S12 0.0889 19◦ , S21 4.420 76◦ , S22 0.385 − 79◦
GA (dB) gA CA RA
1.354 128.4◦
16.0 39.811
4.4202
2.038 1.686 0.803 128.4◦ 0.770
1.686 0.457
0.854 128.4◦
14.0 25.119
4.4202
1.286 1.433 0.596 128.4◦ 0.771
1.433 0.538
0.539 128.4◦
12.0 15.849
4.4202
0.811 1.273 0.423 128.4◦ 0.827
1.273 0.650
0.340 128.4◦
10.0 10.000
4.4202
0.512 1.172 0.290 128.4◦ 0.882
1.172 0.752
0.215 128.4◦
8.0 6.310
4.4202
0.323 1.109 0.193 128.4◦ 0.922
1.109 0.832
Input
1.0 j
stability
circle
0.5 j 2.0 j
Locus of
centre of
the circles
Available
gain (dB)
16.0
0.5 1.0 2.0
14.0
12.0
10.0
8.0
-0.5 j
-2.0 j
-1.0 j
Fig. 14.16 Available gain circles on ΓS plane for a potential unstable device
The constant available gain circles of the device are plotted on Smith chart as
shown in Fig. 14.16. The centres of all the circles are always on the line drawn
between C1∗ and origin of the ΓS plane as shown by a dotted straight line. The radii of
constant gain circles are getting larger with the decrease of available gain.
The input stability circle is also shown in the same ΓS plane. The stability circle
and all the constant gain circles intersect at two same points on the |ΓS | 1 circle
of the Smith chart.
To achieve constant available gain, the input matching network should provide
a source reflection coefficient ΓS which lie on that corresponding constant available
gain circle and the point should be within the stable region in the ΓS plane.
14.5.3 Amplifier with Specific Operating Power Gain
Operating power gain GP of a network is defined as the ratio of power delivered to

the load, PL and power input, PIN to the network. Thus, for a two-port network, from
(7.182) GP can be rewritten as
2 1 − |ΓL |
2
PL 1
GP |S21 | (14.56)
PIN 1 − |ΓIN |2 |1 − S22 ΓL |2
Thus, the operating power gain depends on the load impedance ΓL ; however,
it is independent of the source impedance ΓS and output impedance ΓOUT . There-
fore, for designing a high power amplifier, output of the device should be matched
corresponding to achieving maximum output power or efficiency or linearity.
The operating power gain GP can be written as
GP gP |S21 |2 (14.57)
where
1 − |ΓL |2
gP
1 − |ΓIN |2 |1 − S22 ΓL |2
1 − |ΓL |2

11 − ΓL 2
1 − S1−S 22 ΓL
|1 − S22 ΓL |2
1 − |ΓL |2

|1 − S22 ΓL |2 − |S11 − ΓL |
1 − |ΓL |2
(14.58)
1 − |S11 |2 + |ΓL |2 |S22 |2 − ||2 − 2 Re(ΓL C2 )
Thus (14.58) can be expressed as

2
gP C2∗

Γ L −
1 + gP |S22 |2 − ||2
1 2

1 − 2kgP |S12 S21 | + gP2 |S12 S21 |2 2

(14.59)
1 + gP |S22 |2 − ||2
The expression (14.59) represents circle in the ΓL plane for a fixed value of
normalized operating power gain gP . The centre CP and radius RP of the constant
gain circle in the ΓL plane are given by
gP C2∗
CP (14.60a)
1 + gP |S22 |2 − ||2
21
1 − 2kgP |S12 S21 | + gP2 |S12 S21 |2
RP (14.60b)
1 + gP |S22 |2 − ||2
The centres of the constant gP circles are always on the line drawn between C2∗
and the origin of the ΓL plane.
In case of unconditionally stable device, when output is matched for maximum
gain, i.e. conjugate matched, the radius of the constant gain circle RP becomes zero.
Thus from (14.60b), the normalized maximum operating power gain gP max can be
expressed as
1 − 2kgP max |S12 S21 | + gP2 |S12 S21 |2 0 (14.61)
or
√
k − k2 − 1
gP max (14.62)
|S12 S21 |
Therefore, the maximum operating power gain GP max becomes

|S21 |

GP max k − k2 − 1 (14.63)
|S12 |
This is the same result as mentioned in (14.39). In case of potentially unstable

device, gP becomes infinite. The constant operating power gain circles become output
stability circles with centre and radius given by (14.60a and 14.60b):
C2∗
CP (14.64a)
|S22 | − ||2
2
|S12 S21 |
RP (14.64b)
|S22 |2 − ||2
Example 14.13 Computation of constant operating power gain circles of a bilateral

device with S-parameters: S11 0.498 121◦ , S12 0.1137 −49◦ , S21 3.080 −
30◦ , S22 0.225 − 150◦ .
Solution
0.498 121◦ × 0.225 − 150◦ − 0.1137 − 49◦ × 3.080 − 30◦

0.291 83.9◦ , Thus, || 0.291 < 1
The stability factor:
1 − (0.498)2 − (0.225)2 + (0.291)2

K 1.122 > 1
2(0.1137 × 3.080)
Thus, the device is unconditional stable. The maximum operating power gain:

3.080
GP max (dB) 10 log + 10 log 1.122 − 1.1222 − 1
0.1137
14.328 dB − 2.127 dB 12.2 dB
∗
C2 S22 − S11
0.225 150◦ − (0.291 83.9◦ ) × (0.498 − 121◦ )
0.311 − 175.4◦
C2∗ 0.311 + 175.4◦
And
|S12 S21 | |0.1137 − 49◦ × 3.080 − 30◦ | 0.350

|S11 |2 − ||2 0.4982 − 0.2912 0.163
Using the values of these parameters, centre and radius of constant operating
and 6.0 dB are calculated as tabulated in Table 14.6. It shows that for maximum
operating power gain (12.2 dB), the radius of the constant gain circle becomes zero
and angle of the centre of all the circles are the same as angle of C2∗ .
The constant operating power gain circles of the device are plotted on Smith chart
as shown in Fig. 14.17. The centres of all the circles are always on the line drawn
between C2∗ and origin of the ΓL plane as shown by a dotted line. The radii of the
constant gain circles are getting smaller with the increase of gain.
To achieve specific operating power gain, the output matching network should
provide load reflection coefficients ΓL which lie on that constant operating power
gain circle.
Example 14.14 Computation of operating gain circles of the device at frequency

6.0 GHz, with S-parameter as S11 0.701 − 116◦ , S12 0.0889 19◦ , S21
4.420 76◦ , S22 0.385 − 79◦
Solution
0.701 − 116◦ × 0.385 − 79◦ − 0.0889 19◦ × 4.420 76◦

0.393 − 125.2◦
Table 14.6 Parameters for operating power gain circles of a bilateral device: S11
0.498 121◦ , S12 0.1137 − 49◦ , S21 3.080 − 30◦ , S22 0.225 − 150◦
⎡
1 ⎤
1−2kgP |S12 S21 |+gp2 |S12 S21 |2 2
GP
gP |S |2 ⎣ ⎦
GP (dB)
21
P
C RP

|S |2
1+gp

2
11 −||
g C∗
P 2
1+gP |S22 |2 −||2
16.596 0.545 175.4◦

12.2 3.0802 0.940 0
0.940 0
1.749 0.579 175.4◦
10 0.328 175.4◦
10.0 3.0802 0.964 0.555
0.964 0.575
1.054 0.341 175.4◦
6.310 0.207 175.4◦
8.0 3.0802 0.977 0.729
0.977 0.746
0.665 0.212 175.4◦
3.981 0.131 175.4◦
6.0 3.0802 0.986 0.832
0.986 0.844
0.420 0.133 175.4◦
Operating 1.0 j
gain (dB)
0.5 j 2.0 j
6.0
8.0
10.0
12.2
0.5 1.0 2.0

0
Locus of centre
of the constant
GP circles -0.5 j
-2.0 j
-1.0 j
Fig. 14.17 Constant operating power gain circles on ΓL plane

Thus, || 0.393 < 1
1 − (0.701)2 − (0.385)2 + (0.393)2

K 0.655 < 1
2(0.0889 × 4.420)
Thus the device is conditionally stable. It is required to draw the output stability
circle for the device. The source and load reflection coefficients should be within the
stable region on ΓS and ΓL plane, respectively.
The centre and radius of the output stability circle is given by
(0.385 − 79◦ − 0.393 − 125.2◦ × 0.701 116◦ )∗

cL
(0.385)2 − |0.393|2
60.063 59.3◦
0.0889 × 4.420
rL 60.713
(0.385)2 − |0.393|2
The maximum stable gain MSG of the device (14.42b):

4.420
MSG(dB) 10 log 16.97 dB
0.0889
∗
C2 S22 − S11
0.385 − 79◦ − (0.393 − 125.2◦ ) × (0.701 116◦ )
0.389 − 120.7◦
Therefore,
C2∗ 0.389 120.7◦
And
|S12 S21 | |0.0889 19◦ × 4.420 76◦ | 0.393

|S11 |2 − ||2 0.7012 − 0.3932 0.337
Using the values of these parameters, centre and radius of constant operating
10.0 and 8.0 dB are tabulated in Table 14.7. It shows that angle of the centre of all
the constant operating power gain circles are the same as angle of C2∗ i.e. 120.7°.
The constant operating power gain circles of the device are plotted on Smith chart
as shown in Fig. 14.18. The centres of all the circles are always on the line drawn
between C2∗ and origin of the ΓL plane of the Smith chart as shown by a dotted straight
line. The radii of the constant gain circles are getting higher with the decrease of
gain.
Table 14.7 Parameters for operating power gain circles of a bilateral device: S11
0.498 121◦ , S12 0.1137 − 49◦ , S21 3.080 − 30◦ , S22 0.225 − 150◦
GP (dB) gP CP RP
0.792 120.7◦
16.0 39.811
4.4202
2.038 0.987 0.803 120.7◦ 0.769
0.987 0.779
0.500 120.7◦
14.0 25.119
4.4202
1.286 0.992 0.504 120.7◦ 0.770
0.992 0.777
0.315 120.7◦
12.0 15.849
4.4202
0.811 0.995 0.317 120.7◦ 0.827
0.995 0.831
0.199 120.7◦
10.0 10.000
4.4202
0.512 0.997 0.200 120.7◦ 0.881
0.997 0.884
0.126 120.7◦
8.0 6.310
4.4202
0.323 0.998 0.126 120.7◦ 0.922
0.998 0.924
Locus of
centre of
the circles Output
1.0 j
stability
2.0 j
circle
0.5 j
operating
gain (dB)
0.5 1.0 2.0
0
16.0
14.0
12.0
10.0
8.0 -0.5 j
-2.0 j
-1.0 j
Fig. 14.18 Operating gain circles on ΓL plane for a potential unstable device
The output stability circle is also shown in the same ΓL plane. The output stability
circle and all the constant operating power gain circles intersect at same two points
on the |ΓL | 1 circle of Smith chart.
To achieve constant operating power gain, the output matching network should
provide any load reflection coefficient ΓL which lie on that corresponding constant
operating power gain circle and the point should be within the stable region in the
ΓL plane.
14.6 Small Signal Amplifiers
Small signal amplifiers operate in their linear region. S-parameter model is valid to
model the devices in this operating region. The following sections will present the
design of low-noise and high-gain amplifiers based on S-parameter models which
operate in their linear region.
14.6.1 Low Noise Amplifier (LNA) Design
Front end of any receiver handles most weak signal in a communication system.
Therefore, the receiver needs a front-end amplifier with as low a noise figure as
possible. Practically, input matching circuits are different to obtain minimum noise
figure and maximum gain for an amplifier [5–9]. Some sort of compromise must be
made with the gain of the amplifiers to obtain the lowest noise figure. This can be
done by using constant available gain circles and circles of constant noise figure to
select a usable trade-off between noise figure and gain.
For microwave amplifiers, the noise factor can be written as [5]
RN
NF NFmin + |YSource − YOPT |2 (14.65)
GSource
Here, YSource is the source admittance presented to the transistor, YOPT is the
optimum source admittance that results in minimum noise figure NFmin . RN is the
equivalent noise resistance of the transistor and GSource is the real part of the source
admittance, i.e. YSource GSource + jBSource . The equivalent noise resistance, RN ,
determines the sensitivity or dependency of the noise factor, NF, with the deviation
of source admittance, YSource from its optimum value, YOPT , corresponding to the
minimum noise figure, NFmin . The parameters NFmin , YOPT and RN are characteris-
tics of a particular transistor and known as noise parameters of the device. Noise
parameters of a pHEMT device are shown in Table 14.8. The S-parameters of the
same device are shown in Table 14.1.
The admittance parameters YSource and YOPT of (13.65) can be replaced by the
corresponding reflection coefficients ΓS and ΓOPT as:
1 1 − ΓS
YSource (14.66a)
Zo 1 + ΓS
14.6 Small Signal Amplifiers 401
Table 14.8 Typical noise Freq. NFmin Γ OPT RN

parameter of the pHEMT
device, bias condition: V DS (GHz) (dB) (MAG) (DEG) ()
2 V, I D 15 mA, reference 1.0 0.29 0.756 14 15.60
impedance: Z o 50 2.0 0.30 0.690 28 14.65
3.0 0.34 0.643 43 13.56
4.0 0.38 0.606 58 12.10
5.0 0.41 0.578 73 10.53
6.0 0.46 0.553 87 8.86
7.0 0.50 0.534 102 7.16
8.0 0.55 0.518 116 5.62
9.0 0.60 0.505 131 4.29
10.0 0.64 0.495 145 3.23
11.0 0.69 0.486 159 2.53
12.0 0.73 0.476 173 2.22
13.0 0.78 0.467 −173 2.37
14.0 0.84 0.455 −160 2.96
15.0 0.88 0.443 −146 4.01
16.0 0.93 0.428 −132 5.47
17.0 0.99 0.412 −118 7.26
18.0 1.05 0.394 −103 9.61
1 1 − ΓOPT
YOPT (14.66b)
Zo 1 + ΓOPT
Using (14.66a) and (14.66b), expression (14.65) of noise figure can be written as
4RN |ΓS − ΓOPT |2

NF NFmin + (14.67)
GSource Zo2 |1 − ΓS |2 |1 − ΓOPT |2
Expressing GSource , the real part of YSource in terms of the reflection coefficient and
substituting in (14.67):

NF NFmin + (14.68)
Zo 1 − |ΓS |2 |1 + ΓOPT |2
Defining the noise figure parameter N, as
|ΓS − ΓOPT |2 Zo (NF − NFmin )

N |1 + ΓOPT |2 (14.69)
1 − |ΓS | 2 4RN
This is a constant, for a given noise figure and set of noise parameters NFmin for
ΓOPT . Thus, (14.69) can be written as

ΓS ΓS∗ + ΓOPT ΓOPT
∗ ∗
− ΓS ΓOPT − ΓS∗ ΓOPT N 1 − ΓS ΓS∗ (14.70)
or,
(1 + N )ΓS ΓS∗ + ΓOPT ΓOPT

∗ ∗
− ΓS ΓOPT − ΓS∗ ΓOPT N
or,
∗
ΓOPT ΓOPT ΓS∗ ΓOPT ∗
ΓS ΓOPT N ∗
ΓOPT ΓOPT
ΓS ΓS∗ + − − + ∗
− ΓOPT ΓOPT
(N + 1)2 N +1 N +1 N +1 (N + 1)2
or,
2
2 N N + 1 − |Γ |2
Γ OPT
ΓS − OPT (14.71)

N +1 N +1
This is equation of a circle of constant noise figures. The centre CF and radius RF
are given by
ΓOPT
CF (14.72a)
N +1

N N + 1 − |ΓOPT |2
RF (14.72b)
N +1
For special case of minimum noise figure, NF NFmin and ΓS ΓOPT , thus,
N 0. In this condition, the centre and radius of the noise circle become CF ΓOPT
and RF 0 respectively. Thus, for minimum noise figure, the circle becomes a point.
Example 14.15 Noise parameters of a device in 50 system are as follows: NFmin
0.84 dB, ΓOPT 0.455 − 160◦ , RN 3.0 .
Compute the noise figure of the amplifier in case of the following source reflection
coefficients: (i) ΓS ΓOPT (ii) ΓS 0 (iii) ΓS 0.718 − 114◦ .
Solution
NFmin 0.84 dB 1.213

ΓOPT 0.455 − 160◦ −0.428 − j0.156
RN 3.0
The noise factor is given by (14.68):

NF NFmin +
Zo 1 − |ΓS |2 |1 + ΓOPT |2
(i) In case of ΓS ΓOPT , the noise factor
4RN |ΓOPT − ΓOPT |2

NF NFmin + NFmin 0.84 dB
Zo 1 − |ΓOPT |2 |1 + ΓOPT |2
(ii) In case of ΓS 0, the noise factor
4 × 3.0 |0 − 0.455 − 160◦ |2

NF 1.213 +
50 1 − |0|2 |1 + 0.455 − 160◦ |2
12 0.207
1.213 + × 1.354
50 0.352
1.32 dB
(iii) In case of ΓS 0.718 − 114◦ , the noise factor
4 × 3.0 |0.718 − 114◦ − 0.455 − 160◦ |2

NF 1.213 +
50 1 − |0.718 − 114◦ |2 |1 + 0.455 − 160◦ |2
12 0.269
1.213 + × 1.592
50 0.170
2.02 dB
Example 14.16 Noise parameters of a device in 50 system are as follows: NFmin

0.84 dB, ΓOPT 0.455 − 160◦ , RN 9.0 .
Compute the noise figure of the amplifier in case of following source reflection
coefficients: (i) ΓS ΓOPT (ii) ΓS 0 (iii) ΓS 0.718 − 114◦ .
Solution
NFmin 0.84 dB 1.213

ΓOPT 0.455 − 160◦ −0.428 − j0.156
RN 9.0
The noise factor is given by (14.68):

NF NFmin +
Zo 1 − |ΓS |2 |1 + ΓOPT |2
(i) In case of ΓS ΓOPT , the noise factor
4RN |ΓOPT − ΓOPT |2

NF NFmin + NFmin 0.84 dB
Zo 1 − |ΓOPT |2 |1 + ΓOPT |2
(ii) In case of ΓS 0, the noise factor
4 × 9.0 |0 − 0.455 − 160◦ |2

NF 1.213 +
50 1 − |0|2 |1 + 0.455 − 160◦ |2
36 0.207
1.213 + × 1.637
50 0.352
2.14 dB
(iii) In case of ΓS 0.718 − 114◦ , the noise factor
4 × 3.0 |0.718 − 114◦ − 0.455 − 160◦ |2

NF 1.213 +
50 1 − |0.718 − 114◦ |2 |1 + 0.455 − 160◦ |2
36 0.269
1.213 + × 2.349
50 0.170
3.71 dB
Examples 14.15 and 14.16 show the effects of equivalent noise resistance RN on
the noise figure when the source matching deviates from its optimum value. Though
RN does not affect the minimum noise figure corresponding to ΓOPT , it affects the
noise figure when the source impedance is other than the value of ΓOPT . Therefore,
it is always desirable to have the value of RN as low as possible.
Example 14.17 Compute and draw the noise figure circles for NF NFmin , 1.0, 1.5
and 2.0 dB at 14 GHz of the device with S-parameters and noise parameters tabulated
in Tables 14.1 and 14.8 respectively.
Solution
The noise parameters of the device at 14 GHz are (from Table 14.8):
NFmin 0.84 dB 1.213

ΓOPT 0.455 − 160◦ −0.428 − j0.156
RN 2.96
Zo (NF − NFmin ) 50(NF − 1.213)

N |1 + ΓOPT |2 |1 − 0.428 − j0.156|2
4RN 4 × 2.96
or,
N 8.708 × (NF − 1.213)
Therefore, from (14.72a) and (14.72b):

0.455 − 160◦
CF
8.708 × (NF − 1.213) + 1
1.0 j
0.5 j 2.0 j
Noise
Figure
(dB)
2.0
1.5
0.5
1.0 0
1.0 2.0
0.84
Locus of centre -0.5 j

of the noise -2.0 j
figure circles
-1.0 j
Fig. 14.19 Noise figure circles on ΓS plane

N N + 1 − |ΓOPT |2
RF
√ N +1
8.708 × (NF − 1.213)(8.708 × (NF − 1.213) + 0.793)

8.708 × (NF − 1.213) + 1
Putting the values of NF in the above expressions of center and radius of the noise
figure circles:
for, NF NFmin , N 0, RF 0, CF 0.455 − 160◦

for, NF 1 dB, N 0.068, RF 0.226, CF 0.426 − 160◦
for, NF 1.5 dB, N 0.296, RF 0.438, CF 0.351 − 160◦
for, NF 2.0 dB, N 0.552, RF 0.555, CF 0.293 − 160◦
Using these data, the constant noise figure circles of the device are plotted on
Smith chart as shown in Fig. 14.19. It shows that angle of the centre of all the
constant noise figure circles from centre of Smith chart are the same as the angle of
ΓOPT , i.e. −160◦ .
Any source reflection coefficient on a particular constant noise figure circle will
provide the same noise figure. And any source reflection coefficient inside a constant
noise figure circle will provide better noise figure. In general, it is not possible to
achieve maximum gain and minimum noise figure simultaneously from a device, the
constant noise figure circles and constant available gain circles are used to select a
source reflection coefficient for particular noise figure and available gain.
Example 14.18 Determine the source and load reflection coefficients and also the
source and load impedances to design a LNA at 14 GHz, using a pHEMT having S-
parameters and noise parameters given as follows:
S11 0.498 121◦ , S12 0.1137 − 49◦

S21 3.080 − 30◦ , S22 0.225 − 150◦
NFmin 0.84 dB, ΓOPT 0.455 − 160◦ , RN 2.96
Solution
From the S-parameters,
0.291 83.9◦ , and K 1.122
Here || < 1 and K > 1, therefore, the device is unconditional stable. The
maximum available gain:

3.080
GAmax (dB) 10 log + 10 log 1.122 − 1.1222 − 1 12.2 dB
0.1137
Constant associated gain circles for constant gain of 12.2, 11.6, 10.65, 10, 8 and
6 dB are shown in Fig. 14.20 on ΓS plane (Refer Example 14.11). Constant noise
figure circles for noise figure of 0.84, 1.0, 1.32, 1.5 and 2.0 dB are also drawn in
the same ΓS plane (Refer Example 14.16). Here, the source reflection coefficient for
minimum noise figures, ΓOPT 0.455 − 160◦ (marked as A), is different from
the source reflection coefficient for maximum available gain, i.e. 0.717 − 114.4◦
(marked as B). Therefore, it is not possible to design the amplifier with simultaneous
minimum noise figure and maximum gain. From the figure it is clear that, if the
amplifier is designed with source reflection coefficient of ΓOPT 0.455 − 160◦ ,
noise figure of the amplifier will be minimum, i.e. 0.84 dB, however the available
gain of the amplifier will be 10.65 dB. On the other hand, if the amplifier is designed
with source reflection coefficient of 0.717 − 114.4◦ , available gain of the amplifier
will be maximum, i.e. 12.2 dB; however noise figure of the amplifier will be 2.0 dB.
For a source reflection coefficient within the shaded region, as shown in Fig. 14.20,
the noise figure of the amplifier will be within 0.84–1 dB and available gain of the
amplifier will be within 10.65–11.6 dB.
Depending on the requirement of noise figure and gain of the amplifier, the source
reflection coefficient can be determined. Since the device is unconditionally stable at
this frequency, the output can be complex conjugate matched for attaining maximum
possible gain.
1.0 j
0.5 j 2.0 j
Noise
Figure
(dB) 2.0 dB Available
Gain
2.0 (dB)
1.5
1.32 6.0
0.5
1.0 0
1.0 2.0
8.0
A
0.84
10.0
10.65
11.6
12.2
B
-0.5 j
-2.0 j
10.6 < GA < 11.6
0.84 < NF < 1 -1.0 j
Fig. 14.20 Constant available gain and noise figure circles on ΓS plane
Designing the amplifier with minimum noise figure, the source reflection coeffi-
cient should be ΓOPT . Thus, noise figure will be 0.84 dB. And gain of the amplifier
will be 10.65 dB.
ΓS ΓOPT 0.455 − 160◦
Thus, the source impedance ZS will be given by

1 − ΓS 1 − 0.455 − 160◦
ZS Zo 50 ×
1 + ΓS 1 + 0.455 − 160◦
0.572 − j0.156
50 × 19.227 − j7.546
1.428 + j0.156
The output reflection coefficient will be

ΓS S12 S21
ΓOUT S22 +
1 − ΓS S11
0.455 − 160◦ × 0.1137 − 49◦ × 3.080 − 30◦
0.225 − 150◦ +
1 − 0.455 − 160◦ × 0.498 121◦
−0.082 + j0.137
(−0.195 − j0.113) +
0.824 + j0.143
Table 14.9 Noise figure and gain for different matching conditions
Γ S (mag, ang) Z S () NF (dB) ∗
Γ L Γout Z L () Gain (GHz)
(mag, ang)
0.455 − 160◦ 19.2 − j7.55 0.84 0.272 − 166◦ 28.9 − j4.07 10.65
0.718 − 114◦ 11.5 − j31.0 2.0 0.580 175◦ 13.3 + j1.84 12.2
0 50 − j0 1.32 0.225 150◦ 32.9 + j7.81 10.0
(−0.195 − j0.113) + (−0.069 + j0.178)

−0.264 + j0.065
For complex conjugate match of output port:

∗
ΓL ΓOUT
−0.264 − j0.065
0.272 − 166◦
Thus, the load impedance ZL will be given be

1 − ΓL 1 − 0.272 − 166◦
ZL Zo 50 ×
1 + ΓL 1 + 0.272 − 166◦
1.264 + j0.065
50 × 28.922 − j4.071
0.736 − j0.065
The source and load reflection coefficients for minimum noise figure as well as
other two cases are shown in Table 14.9.
One of the other two cases is the complex conjugate match for both the ports,
to achieve a maximum available gain of 12.2 dB. Since both the ports are complex
conjugate matched at this frequency, there will be no reflection of power from both
the ports, i.e. VSWR of both the ports will be one at this frequency. However, in this
case, the noise figure of the amplifier becomes 2.0 dB.
Just to show the effect of matching network, another case is shown where the
source impedance seen by the device is just 50 . In Fig. 14.20, it is shown that the
noise figure and gain of the amplifier become 1.32 and 10 dB, respectively. In this
∗ ∗
case, the load reflection coefficientΓL ΓOUT S22 0.225 150◦ .
Example 14.19 Synthesize input and output matching networks using series and
shunt transmission lines for the amplifier of Example 14.18, where the desired source
impedance ZSdesired and load impedance ZLdesired are
ZSdesired (11.5 − 31j)

ZLdesired (13.3 + 1.84j)
Solution
Synthesis of input matching network:
The desired normalized source impedance is
zSdesired (0.23 − 0.62j)
Reflection coefficient corresponding to this desired normalized source

impedance is
ΓSdesired 0.718 − 114◦
The desired source reflection coefficient is plotted in Fig. 14.21a on a Smith

chart and marked as P. The function of the matching network is to transfer the
reference source impedance say 50 (marked as A in the Smith chart) to the desired
normalized source impedance zSdesired .
Though there are many combinations of transmission lines to synthesize a match-
ing network, here the considered input matching network using series and shunt
transmission lines is shown in Fig. 14.21a. The lengths and characteristic impedances
(here it is chosen as 50 ) of the lines are to be properly selected to transfer the
impedance from point A to point P. A series transmission line transfers impedance
along a constant VSWR circle. The constant VSWR circle is drawn passing through
the point P. The shunt stub must transfer the 50 impedance to a point on the con-
stant VSWR circle. Here, a short-circuited shunt stub is chosen to transfer 50 to
a point B on the constant VSWR circle along the constant conductance circle. The
electrical length of the short-circuit stub is 25.9° (0.072λ) as shown. The function of
the series transmission line is to transfer the point B to the impedance point P moving
along the constant VSWR circle as shown in the diagram. The electrical length of
the series transmission line is 125° (0.347λ).
Synthesis of output matching network:
The desired normalized load impedance is
zLdesired (0.266 + 0.0368j)
Reflection coefficient corresponding to this desired normalized load impedance

is
ΓLdesired 0.58 + 175◦
The desired load reflection coefficient is plotted in Fig. 14.21b on a Smith chart
and marked as P. The function of the desired matching network is to transfer the
reference load impedance say 50 (marked as A in the Smith chart) to the desired
load impedance zLdesired .
The output matching network using series and shunt transmission lines is shown
in Fig. 14.21b. The lengths and characteristic impedances (here it is chosen as 50 )
of the lines are to be properly selected to transfer the impedance from point A to point
Fig. 14.21 a Synthesis of 2

input matching network, b (a)
synthesis of output matching
1 50
network, c amplifier circuit
with input and output
matching networks
0.5 2
lS1 = 25.9o B
lS2 = 125o
0 0.5 1 2
A
Constant
VSWR
Circle
Z Chart P
0.5 2
Y Chart
1
2
(b)
1 50
1
Constant
0.5 2 VSWR
Circle
P
0 0.5 1 2
A
lL2 =29.6o
B
lL1=54.9
0.5 2
1 Z Chart
Y Chart
VG VD
(c)
RG RD
0.082
0.347
0.153
0.072
= 1 2 3
AMP1 AMP2 AMP3
1 2 3
1 2 3
Fig. 14.22 Schematic diagram of a multistage amplifier
P. A constant VSWR circle is drawn passing through the point P. Thus, the shunt stub
must transfer the 50 impedance to a point on this constant VSWR circle. An open-
circuit stub is chosen to transfer the 50 to a point B on the constant VSWR circle
along the constant conductance circle. The electrical length of the open-circuit stub
is 54.9° (0.153λ) as shown. The function of the series transmission line is to transfer
the point B to the desired impedance point P moving along the constant VSWR circle
as shown in the diagram. The electrical length of the series transmission line is 29.6°
(0.082λ). The input and output matching networks along with the bias networks of
the device are shown in Fig. 14.21c.
14.6.2 High-Gain Amplifier Design
In the previous section, it is shown that the matching requirements for minimum
noise figure and maximum gain are different. Thus, practically high-gain ampli-
fiers have comparatively higher noise figure than that of an LNA using the same
device. Amplifiers are designed with highest possible gain compromising its noise
figure performance for applications where poor NF of the amplifier does not affect
the overall system performance. Design philosophy of all high-gain amplifiers is
same irrespective of their operating frequency and realization technology [10–13].
A communication receiver consists of multistage amplifier to achieve required gain
as shown in Fig. 14.22. Suppose, gain and noise factors of the first, second and third
amplifiers are G1 and NF1 , G2 and NF2 and G3 and NF3 , respectively, and bandwidth
of each amplifiers are the same, i.e., B. Suppose the system is driven by a reference
source of noise temperature To . Then the available noise power at the input of the
first amplifier is
Pni kTo B
Here, k is the Boltzmann constant. Thus, from (14.7) the noise power output, Pno1
of the first amplifier can be written as
Pno1 (Pni ) × (G1 NF1 ) (kTo B) × (G1 NF1 ) (14.73a)

kTo BG1 + (NF1 − 1)kTo BG1 (14.73b)
The first term is the noise power due to the presence of input noise power Pni and
the second term is the noise power which is added by the first amplifier. In a similar
way, the total noise power at the output of the second amplifier can be calculated
considering Pno1 as the noise input to the second amplifier. Thus, the noise power
output from the second amplifier will be
Pno2 (kTo BG1 NF1 )G2 + (NF2 − 1)(kTo B)G2 (14.74)
The noise power Pno2 is the input noise power for the third amplifier. Thus, the
total noise power at the output of the third amplifier will be
Pno3 [(kTo BG1 NF1 )G2 + (NF2 − 1)(kTo B)G2 ]G3 + (NF3 − 1)(kTo B)G3
kTo BG1 G2 G3 NF1 + (NF2 − 1)kTo BG2 G3 + (NF3 − 1)kTo BG3 (14.75)
Suppose, NF123 is the overall noise figure of the three stage amplifier. The overall
gain of the three-stage amplifier is G1 × G2 × G3 . Thus, the noise power output of
the third stage amplifier will be
Pno3 kTo BG1 G2 G3 NF123 (14.76)
From (14.75) and (14.76), the overall noise figure can be expressed as
(NF2 − 1) (NF3 − 1)
NF123 NF1 + +
G1 G1 G2
In this way, the analysis can be extended for multistage amplifiers and the overall
noise figure can be expressed as
(NF2 − 1) (NF3 − 1) (NFm − 1)
NF123...m NF1 + + + ··· + (14.77)
G1 G1 G2 G1 G2 G3 . . . Gm−1
This is known as Friis’s formula for overall noise figure for a cascaded multistage
network. This formula shows that the overall noise factor will be dominated by
the noise factor of the first stage if the gain of the first stage is sufficiently high.
Contribution in the overall noise factor by other stages will be reduced by the overall
gain of the preceding stages.
Example 14.20 Determine the overall noise figure of a cascaded three-stage ampli-
fier. The individual noise figure and gain of the three stages are as follows:
NF1 0.84 dB, G1 10.65 dB

NF2 2.0 dB, G2 12.2 dB
NF3 2.0 dB, G1 12.2 dB
Solution
Noise and gain in factors of the individual amplifiers are
NF1 1.21, G1 11.61

NF2 1.58, G2 16.60
NF3 1.58, G1 16.60
The overall noise figure:

(1.58 − 1) (1.58 − 1)
NF123 1.21 + +
11.61 11.6 × 16.6
1.21 (contribution from first stage)
+ 0.0504 (contribution from second stage)
+ 0.003 (contribution from third stage)
1.267
Therefore the overall noise figure of the cascaded three-stage amplifier in dB is
NF123 (in dB) 10 log(1.267) dB 1.027 dB
The above example shows how the contribution in overall noise figure of the
following stages diminishes. In communication transponders, high-gain amplifiers
are used after a low-noise amplifier of sufficiently high gain so that poor noise figure
of the following stages does not affect the overall noise performance.
Example 14.21 Noise figure and gain of two amplifiers AMP1 and AMP2 are as
follows:
NF1 2 dB, G1 6 dB
NF2 2.2 dB, G2 15 dB
Determine which combination of these two amplifiers will provide better cascaded
noise figure.
Solution
Noise and gain in factors of the individual amplifiers are
NF1 1.5849, G1 3.98

NF2 1.6596, G2 31.62
Two possible combinations are shown in Fig. 14.23a, b. The cascaded noise factor
for the first combination (AMP1 as input amplifier):
(1.6596 − 1)
NF12 1.5849 +
3.98
Fig. 14.23 Cascaded (a)

two-stage amplifiers
AMP1 AMP2
1 = 6 2 = 15
1 = 2 2 = 2.2
(b)
AMP2 AMP1
2 = 15 1 = 6
2 = 2.2 1= 2
1.5849 + 0.1657
1.7506
2.432 dB
The cascaded noise factor for the second combination (AMP2 as input amplifier):
(1.5849 − 1)
NF21 1.6596 +
31.62
1.6596 + 0.0185
1.6781
2.248 dB
The second combination, i.e. the amplifier of higher noise figure as input stage
of the cascaded two-stage amplifier provides better overall noise figure. This is due
to the lower gain of the AMP1. Thus, to achieve overall lowest noise figure from a
multistage amplifier, not only noise figure of the first stage, its gain is also important.
The gain of the first stage amplifier should not be too low.
For high-gain amplifiers, it is always targeted for a complex conjugate match
for both the ports. However, this is only possible at the frequency where the device
is unconditional stable. For the frequency where the device is conditionally stable,
conjugate matching is not possible. With the help of input and output stability circles,
the source and load reflection coefficients are selected to ensure the design of amplifier
with stable gain.
Example 14.22 Determine source and load reflection coefficient for a high gain
amplifier of gain 16 dB at frequency 4.0 GHz, where S-parameters of the device are
S11 0.784 − 80◦ , S12 0.0702 38◦ , S21 5.067 107◦ , S22 0.440 − 55◦ .
Also estimate the noise figure of the amplifier for the following noise parameters:
NFmin 0.38 dB, ΓOPT 0.606 58◦ , RN 12.1 .
Solution
For the device:
0.784 − 80◦ × 0.440 − 55◦ − 0.0702 38◦ × 5.067 107◦

0.450 − 83.95◦
1 − (0.784)2 − (0.440)2 + (0.450)2
K 0.555
2(0.0702 × 5.067)
The device is conditionally stable because || 0.450 < 1 and K 0.555 < 1.
Therefore, it is required to draw the stability circles for the device. The source and
load reflection coefficients should lie within the stable region on ΓS and ΓL plane,
respectively.
The centre and the radius of the input stability circle are given by

∗ ∗
S11 − S22
cS
|S11 |2 − ||2
(0.784 − 80◦ − 0.450 − 83.95◦ × 0.440 55◦ )∗
cS
(0.784)2 − |0.450|2
(0.784 − 80 − 0.198 − 28.95◦ )∗
◦

0.412
◦
1.645 93.16

S12 S21
rS
|S |2 − ||2
11
0.0702 × 5.067

(0.784)2 − |0.450|2
0.356
0.864
0.412
The centre and the radius of the output stability circle are given by

∗ ∗
S22 − S11
cL
|S22 |2 − ||2
(0.440 − 55◦ − 0.450 − 83.95◦ × 0.784 80◦ )∗

(0.440)2 − |0.450|2
◦
−37.67 106.6

S12 S21
rL
|S |2 − ||2
22
0.0702 × 5.067

(0.440)2 − |0.450|2
Fig. 14.24 Source and load reflection coefficients for the conditional stable device using constant
available gain circle
0.356
38.2
0.0093
The input and output stability circles are shown in Fig. 14.24. Since S11 and S22
are both less than 1, the region containing the centre of the Smith chart are the stable
regions.
The maximum stable gain of the device:

S21 5.067
MSG 10 log 10 log 18.54 dB
S12 0.0702
To design a stable amplifier using a conditional stable transistor, it is required to

mismatch either input or output or both the ports of the device. There are two options
to select the source and load reflection coefficients. The targeted design gain of the
amplifier is 16 dB. In one option, mismatch can be introduced in the input section
and complex conjugate match can be provided to the output port. In other option,
mismatch can be introduced in the output terminal and input port can be complex
conjugate matched. For the former case, constant available gain circles approach and
for the latter case constant operating gain circle approach should be used.
Constant Available Gain Approach:
In this approach, source reflection coefficient will be selected and the load reflection
coefficient will be selected for complex conjugate matching of the output port. Thus,
amplifier will provide better return loss at the output port compared to the input port.
A constant available gain circle is drawn for GA 16 dB as shown in Fig. 14.24.
Source reflection coefficient ΓS on any point on the circle will provide the available
gain of 16 dB. To ensure better stability, it is advisable to select the ΓS as far as
possible from the input stability circle. Here, ΓS 0.141 83.88◦ is selected on the
constant available gain circle as shown. Thus, the source impedance ZS is given by
1 − ΓS 1 − 0.141 83.88◦
ZS Zo 50 × 49.51 + j14.14
1 + ΓS 1 + 0.141 83.88◦
Thus, output reflection coefficient:

ΓS S12 S21
ΓOUT S22 +
1 − ΓS S11
0.141 83.88◦ × 0.0702 38◦ × 5.067 107◦
0.440 − 55◦ +
1 − 0.141 83.88◦ × 0.784 − 80◦
0.457 − 61.82
For complex conjugate match of output port:

∗
ΓL ΓOUT 0.457 61.82◦
The load reflection coefficient ΓL 0.457 61.82◦ is shown in Fig. 14.24. Since
it is within the stable region on the ΓL plane, the complex conjugate match for the
output port is feasible. The load impedance ZL is given by
1 − ΓL 1 − 0.457 61.82
ZL Zo 50 × 50.89 + j51.82
1 + ΓL 1 + 0.457 61.82
Here, noise figure circle is also drawn. It is shown that for the selected source
reflection coefficient, the noise figure is 0.82 dB which is more than the minimum
noise figure of 0.38 dB.
Constant Operating Gain Approach:
In this approach, load reflection coefficient is selected at first then the source reflec-
tion coefficient is selected for complex conjugate matching of the input port. Thus,
amplifier provides better return loss at the input port compared to the output port. A
constant operating gain circle is drawn for GP 16 dB as shown in Fig. 14.25. The
load reflection coefficient ΓL on any point on the circle will provide the operating
Input stability
circle Output stability
circle
1.0 j
Constant
gain circle 2.0 j
0.5 j
GP =16 dB Noise Figure
circle
(0.55 dB)
0.5 1.0 2.0

0
-0.5 j
-2.0 j
-1.0 j
Fig. 14.25 Source and load reflection coefficients for conditional stable device using constant
operating gain approach
gain of 16 dB. To ensure better stability, it is preferable to select the ΓL as far as

possible from the output stability circle. Here, ΓL 0.288 − 72.82 is selected on
the constant operating gain circle as shown. Thus, the load impedance ZL will be
given by
1 − ΓL 1 − 0.288 − 72.82◦
ZL Zo 50 × 50.24 − j30.122
1 + ΓL 1 + 0.288 − 72.82◦
Thus the input reflection coefficient will be

ΓL S12 S21
ΓIN S11 +
1 − ΓL S11
0.288 − 72.82◦ × 0.0702 38◦ × 5.067 107◦
0.784 − 80◦ +
1 − 0.288 − 72.82◦ × 0.440 − 55◦
0.324 − 57.61◦
For complex conjugate match of input port:

∗
ΓS ΓIN 0.324 57.61◦
The source reflection coefficient ΓS 0.324 57.61◦ is shown in Fig. 14.25. Since
it is within the stable region on the ΓS plane, the complex conjugate match for the
input port is feasible. The source impedance ZS is given by
1 − ΓS 1 − 0.324 57.61◦
ZS Zo 50 × 59.05 + j36.1
1 + ΓS 1 + 0.324 57.61◦
Here, noise figure circle passing through the selected source reflection coefficient
is also drawn. It is shown that for the selected source reflection coefficient, the noise
figure is 0.55 dB which is more than the minimum noise figure of 0.38 dB.
Example 14.23 Synthesize input and output matching networks using series and
shunt transmission lines for the amplifier where the desired source impedance ZSdesired
and load impedance ZLdesired are
ZSdesired (59.05 + j36.1)

ZLdesired (50.24 − j30.122)
Solution
Synthesis of input matching network:
The desired normalized source impedance corresponding to the ZSdesired is
zSdesired (1.181 + 0.722j)
The source reflection coefficient corresponding to this source impedance is
ΓSdesired 0.324 57.61◦
The desired source reflection coefficient is plotted in Fig. 14.26a on a Smith chart
and marked as P. The function of the matching network is to transfer the reference
source impedance, say 50 (marked as A in the Smith chart) to the desired source
reflection coefficient ΓSdesired (marked as P in the Smith chart).
Series transmission line transfers impedance along a constant VSWR circle. The
constant VSWR circle is drawn in the Smith chart passing through the point P. Thus,
the shunt stub must transfer the 50 impedance to a point on the constant VSWR
circle. Here, a short-circuited shunt stub is chosen to transfer 50 to a point B on the
constant VSWR circle along the constant conductance circle. The electrical length
of the short-circuit stub is 55.59° (0.154λ) as shown. Now, the function of the series
transmission line is to transfer the point B to the desired impedance point P moving
the series transmission line is 25.65° (0.0713λ).
Synthesis of output matching network:

The desired normalized load reflection coefficient corresponding to ZLdesired is
zLdesired (1.005 − 0.6024j)
The load reflection coefficient corresponding to this load impedance is
ΓLdesired 0.288 − 72.82
The load reflection coefficient is plotted in Fig. 14.26b on a Smith chart and
marked as P. The function of the desired network is to transfer the load impedance
say 50 (marked as A in the Smith chart) to the desired load impedance zLdesired .
The output matching network using series and shunt transmission lines is shown
in Fig. 14.26b. A constant VSWR circle is drawn passing through the point P. Thus,
the shunt stub must transfer the 50 impedance to a point on this constant VSWR
circle. Here, a short-circuited stub is chosen to transfer the 50 to a point B on the
constant VSWR circle along the constant conductance circle. The electrical length
of the short-circuit stub is 58.98° (0.164λ) as shown. Now, the function of the series
transmission line is to transfer the point B to the desired impedance point P moving
the series transmission line is 89.78° (0.249λ).
The input and output matching networks along with the gate and drain bias net-
works of a device are shown in Fig. 14.26c.
14.7 Large Signal Amplifiers
Power amplifiers are designed to deliver maximum power output for a given device
with selected DC bias condition. To achieve optimum output power, the output of
the device needs to be matched other than the complex conjugate matching. Thus,
a device cannot be matched for maximum power output with maximum gain. In
case of a power amplifier, the target is to get maximum output power with the sac-
rifice of some amount of gain to maximize efficiency. However, when the signal
gain drops below a certain value the required extra RF drive power will often cancel
out any efficiency advantage. Power amplifiers are designed to operate under the
large signal condition to achieve high power. A typical DC I–V characteristic of a
MESFET/HEMT is shown in Fig. 14.27. It is shown how the transconductance gain,
gm , and drain-to-source conductance gds varies with the gate-to-source voltage. The
negative value of gds at higher Vgs and Vds is mainly due to the heating effect of the
device at a high current level. Other than these resistive equivalent circuit compo-
nents, reactive components such as gate-to-source capacitance Cgs and gate–to-drain
capacitance Cgd are also have nonlinear characteristic. Under small amplitude of
sinusoidal signal, the output voltage and current waveforms also remain sinusoidal.
However, under large signal operation, the output voltage and current of the device
14.7 Large Signal Amplifiers 421
Fig. 14.26 a Synthesis of 2

input matching network, b
synthesis of output matching
(a)
1 50
network, c amplifier circuit
with input and output
matching networks
lS2 = 25.65o
1
0.5 2
o
lS1 = 55.59
P
B
0 0.5 1 2
A
Constant
VSWR
Circle
Z Chart 2
0.5
Y Chart
1
2
(b)
1 50
1
Constant
o 0.5 2 VSWR
lL1=58.98 Circle
B lL2 =89.78o
0 0.5 1 2
A
0.5 2
1 Z Chart
Y Chart
(c) VG VD
RG RD
0.249
0.0713
0.164
0.154
Non-linear
Low
=
<0
Increasing
Linear
approximation
of vs High
characteristic
>0
Pinch off
voltage
dependency
Low on
Fig. 14.27 A typical DC I–V characteristics of power MESFET/HEMT
swings over a wide range of I–V characteristics of the device. Traditionally, a load
line is used to determine output signal waveforms under large signal operating con-
dition. In a realistic scenario, the load curve is an ellipse due to the reactive elements
at the output of the device. The reactive elements may be due to the equivalent reac-
tive elements of the device output or part of the load impedance. The load curve
remains elliptic as long as the signal amplitude is small. The inverse slope of the
major axis of the elliptic curve is the load resistance and the minor axis is related
to the reactive part of the load impedance. The dynamic load line under large signal
condition is clipped at both the ends and the shape deviates from elliptic shape as
shown in Fig. 14.28. However, considering the reactive elements within the device
output as part of the output matching network of the amplifier, the load curve can be
considered as a straight line as shown in the figure as a dotted line.
At large signal operating condition, the output voltage and current of the device
swing over a wide range of I–V characteristics of the device. This results in distorted
output voltage and current waveforms as shown in Fig. 14.29. This is the reason for the
power amplifiers even in Class-A operation cannot be considered as a linear amplifier.
The output current waveform is not only non-sinusoidal it is also unsymmetrical
which contributes to DC current thus changes the device operating point.
A power amplifier delivers the highest output power only when the output volt-
age as well as the current swings become maximum as shown in Fig. 14.29. The
maximum voltage and current swings are determined by the bias voltage, cur-
rent and selection of load line of an amplifier. It is also very important that under
large signal operating condition, the drain current Id and gate-to-drain voltage Vds
remain within the allowable maximum ratings of the device. The drain voltage of a
Large signal
non elliptic
load curve
Small signal
elliptic load
curve
Fig. 14.28 A typical dynamic load line of power amplifiers
( , )
0
−
Fig. 14.29 Signal amplification MESFET/HEMT device

MESFET/HEMT device is mainly limited by its gate-to-drain breakdown voltage

VdgB . The maximum drain current IF flows when the channel of the device is fully
open which occurs at small forward gate bias voltage VF as shown in Fig. 14.29. In
case of many high power devices, the maximum allowable drain current may be lim-
ited by the maximum allowable temperature of the device due to thermal dissipation
and to keep the channel temperature of the device within allowable limit. The lower
limit of the voltage swing is limited by the knee voltage Vk as shown in Fig. 14.29. To
achieve maximum output RF power for the Class-A amplifier, the load line, i.e. out-
put impedance of the FET device should be chosen so that the drain voltage becomes
maximum when the gate voltage is at its pinch-off point Vgso and the drain voltage
becomes minimum when the drain current is at its maximum allowable value.
14.7.1 Linear Power Amplifier
To derive the output power and other performance parameters of a Class-A power
amplifier, consider an amplifier circuit using an ideal FET of I–V characteristic
with infinite input impedance and output conductance as shown in Fig. 14.30. Also
consider that the inductor and the DC blocking capacitor provide infinite and zero
impedances, respectively at the operating RF frequency. Thus, to achieve maximum
RF voltage and current swing the drain bias voltage and drain current of the device
will be
VdgB − Vgso − VK
VdQ VS + VK
2
VdgB − Vgso + VK
(14.78a)
2
and
IF
IdQ (14.78b)
2
At this condition the gate bias voltage will be
Vgso + VF
Vgs − + VF
2
−Vgso + VF
(14.78c)
2
Thus, the DC power delivered by the supply to the device is given by
PDC VdQ × IdQ

VdgB − Vgso + VK IF
× (14.79a)
2 2
( , )
− 0
−
Fig. 14.30 Signal amplification through an ideal linear amplifier
This is independent of the RF power level. Thus, without any RF input power, the
amplifier will consume this amount of DC power which is one major drawback of
Class-A amplifiers.
The optimum load impedance RLopt for which the voltage and current swings are
maximum is given by
RLopt (14.79b)
IF
Suppose, idrf and vdsrf are the peak of the drain current and drain-to-source voltage
of the device. The output RF power can be expressed as
vdsrf idrf
Prf √ × √ (14.80a)
2 2
The maximum voltage and current swings of a Class-A amplifier is shown in

Figs. 14.30 and 14.31.
The maximum amplitude of the output voltage:
vdsrf (max) (14.80b)
2
( , )
− 0
−
= < >
Fig. 14.31 Input output waveforms under different load impedances
The maximum amplitude of the output current:

IF
idrf (max) (14.80c)
2
Thus, the maximum RF output power of Class-A amplifier is given by
vdsrf (max) idrf (max)
Prf (max) √ × √
2 2

IF VdgB − Vgso − VK
(14.81)
8
This is the maximum linear output power from a FET based Class-A amplifier
under the assumption of idealized I–V characteristic of the device.
To account the effect of gain of an amplifier, power added efficiency is used in
place of simple efficiency, which is the ratio of output RF power over the DC power
consumption. The power added efficiency ηadd of an amplifier is defined as
Output RF power − Input RF power

ηadd
DC power consumption
or

1 Output RF power
ηadd 1− (14.82)
G DC power consumption
Here, G is the ratio of output RF power to the input RF power, i.e. RF gain of the
amplifier. With the increase of output RF power, the efficiency of Class-A amplifier
increases. From (14.79a) and (14.80a) the power added efficiency can be written as

1 2 × vdsrf × idrf
ηadd 1 − (14.83)
G IF VdgB − Vgso + VK
The power added efficiency is maximum at the maximum output RF power con-
dition. Putting the maximum values of vdsrf and idrf from (14.80b) and (14.80c), the
maximum power added efficiency ηadd (max) can be expressed as

1 1 VdgB − Vgso − VK
ηadd (max) × 1 − × (14.84)
2 G VdgB − Vgso + VK
Therefore, classical maximum power added efficiency of 50% from a Class-A

amplifier is only possible in the limiting case of infinite (practically very high) gain
(G ∞) and zero Knee voltage (VK 0).
Example 14.24 Calculate the maximum output power and power added efficiency
from a GaAs FET power amplifier operating in Class-A with optimum load
impedance for achieving maximum output RF power. Where VK 1.5 V, IF 1A,
VdgB 24 V, Vgso 2.5 V and G 13 dB. Also, calculate the optimum output
load impedance of the device for achieving maximum voltage and current swings.
Solution
Here, G 13 dB 1013/10 20
The maximum RF output power from (14.81):

IF VdgB − Vgso − VK
Prf (max)
8
1 × (24 − 2.5 − 1.5)
W
8
2.5 W
The DC power consumption from (14.79a):

IF × VdgB − Vgso + VK
PDC
4
1 × (24 − 2.5 + 1.5)

4
5.75 W
Therefore, the maximum power added efficiency:

1 2.5
ηadd (max) 1 −
20 5.75
41.3%
The optimum output load impedance is given by (14.79b)

RLopt
IF
24 − 2.5 − 1.5
20
1
The voltage and current swings are maximum when the output load impedance is
optimum to provide maximum RF output power. When the output load impedance
is lesser than the optimum load impedance (i.e. RL < RLopt ) then though the current
swing becomes maximum, the voltage swing becomes lesser than the maximum
possible voltage swing as shown in Fig. 14.31. On the other hand when the output
of the device is terminated with load impedance which is greater than the optimum
load impedance (i.e. RL > RLopt ) then to keep the voltage swing within the allowable
limit, the current swing should be less than the values obtained in optimum load
condition as shown in Fig. 14.31.
Example 14.25 Calculate the maximum output power and gain from a GaAs FET
power amplifier operating in Class-A when (a) the amplifier is matched for maximum
RF output power, (b) the amplifier is matched for maximum RF gain. The equivalent
circuit of the device is shown in Fig. 14.32. Where, 1/2π fcgs Rs , rds 100 ,
other parameters are IF 1 A, VdgB 20 V, Vgso 4 V, VK
VdgB , VF
Vgso , gm 0.25 mho.
Solution
(a) To match the device for maximum RF output power, the output impedance to
be selected for maximum drain-to-source voltage and drain current swings:
VdgB − Vgso 20 − 4
vdsrf (max) V8V
2 2
IF 1
idrf (max) A
2 2
The optimum output impedance is
= >
Fig. 14.32 Signal amplification through an ideal linear amplifier
RoOPT
IF
20 − 4
16
1
The optimum load impedance is
rds × RoOPT
RLOPT
rds − RoOPT
100 × 16
19.05
100 − 16
Therefore, the maximum RF current through the load resistance:

idrf (max) × rds 100
iLrf (max) 0.42 A
rds + RLopt 2 × 119.05
vdsrf (max) iLrf (max)
2 2
8 0.42
√ × √ W 1.68 W
2 2
With reference to Fig. 14.32, the available RF power from the source is
vs2
Pavs
8 Rs
Since, 1/2π fcgs Rs , it can be considered that voltage across the capacitor cgs
will be vs . Thus, the drain current idrf can be written as
idrf gm vs
The RF load current can be written as

rds
iLrf gm vs
rds + RLOPT
The output RF power can be written as

iLrf 2
Prf √ × RLOPT
2
2
gm vs rds
√ × RLOPT
2 rds + RLOPT
The gain of the amplifier is given by

2
Prf rds
G 4RLOPT Rs gm
Pavs rds + RLOPT
2
100
4 × 19.05 × 50 0.25
100 + 19.05
168.6 22.25 dB
(b) To match the device for maximum gain, both the output and input should be
conjugate matched. Thus, the load impedance RL will be equal to the output
impedance rds , i.e. 100 .
VdgB − Vgso 20 − 4
vdsrf (max) V8V
2 2
vdsrf (max) 8
idrf (max) A
RL rds 50
Therefore, the maximum RF current through the load resistance:

idrf (max) × rds 8 × 100
iLrf (max) A
rds + RL 50 × 200
vdsrf (max) iLrf (max)
2 2
8 8
√ × √ W 0.32 W
2 100 2
Which is 10 log(1.68/0.32) 7.2 dB less power compared to the output power

obtained by optimum output impedance matching.
Thus, gain of the amplifier will be given by
( )
Output matching for maximum output power

Output matching for maximum power gain
( )
Fig. 14.33 Transfer characteristic for two different output matching conditions
2
rds
G 4RL Rs gm
rds + RL
2
100
4 × 100 × 50 0.25 ×
100 + 100
312.5 24.9 dB
This gain is about 2.65 dB more than the gain when the device is matched for
maximum output RF power.
Input–output transfer characteristic of two amplifiers one with output matching
corresponding to maximum output power and another with output matching corre-
sponding to maximum gain is shown in Fig. 14.33.
The linear analysis of the amplifier operating in large signal condition is done
considering that the linear characteristic of the devices holds good over its entire
operating regions. Equivalent circuit elements of practical transistors not only depend
on the DC bias level, these are also variations with the instantaneous signal levels.
The practical dependence of equivalent circuit parameters on the signal level pro-
duces distortion in the output signal waveform before starting of cut-off and clipping
behaviour of the transistor.
The Class-A power amplifier operating up to 1–2 dB gain compression point
may be due to the nonlinearities of the circuit elements and the output signal can be
expressed by a power series of the input signal up to the third-order terms as
vo a0 + a1 vi + a2 vi2 + a3 vi3
Suppose, vi A cos ωt, then the output voltage can be written as
vo a0 + a1 A cos ωt + a2 A2 cos2 ωt + a3 A3 cos3 ωt

a2 A2 a3 A3
a0 + a1 A cos ωt + (1 + cos 2ωt) + (3 cos ωt − cos 3ωt)
2 4
2 3
a2 A 3a3 A a2 A2 a3 A3
a0 + + Aa1 + cos ωt + cos 2ωt + cos 3ωt
2 4 2 4
Thus, the output signal level corresponding to fundamental frequency is given by

3a3 A3 3a3 A2
Aa1 + Aa1 1 +
4 4a1
Here, the second term is contribution from the third-order distortion of the output
signal. This term causes gain compression of the amplifier with the increase of input
signal level. Suppose, the gain of the amplifier at 1-dB gain compression point (G1dB )
is corresponding to input signal level A1dB , then we can write

3a3 A21dB
20 log(a1 A1dB ) 20 log a1 A1dB 1 + +1
4a1
or

3a3 A21dB
20 log(a1 A1dB ) 20 log(a1 A1dB ) + 20 log 1 + +1
4a1
or

3a3 A21dB
20 log 1 + −1
4a1
or
3a3 A21dB −1
1+ 10 20 0.89
4a1
or
3a3 A21dB
0.89 − 1 −0.11
4a1
or

4a1 √
|A1dB | × 0.11
3a3
or

4a1
|A1dB |(in dB) 20 log − 9.63 (14.85)
3a3
The third-order output intercept point is defined as the output power level at which
the third-order intermodulation power level becomes equal to the power level corre-
sponding to the fundamental frequency. From (14.14), the condition for third-order
intercept point can be written as
3a3 A3IP3
AIP3 a1
4
or

4a1
AIP3 (14.86)
3a3
Therefore, input power in dB corresponding to third-order intercept point PiIP3 is

given by

4a1
PiIP3 20 log (14.87)
3a3
Now, the output power at 1-dB gain compression point Po1dB can be expressed as
Po1dB A1dB + G1dB A1dB + GLdB − 1 (14.88)
Here, GLdB is the linear gain of the amplifier. Combining (14.85), (14.87) and
(14.88), the Po1dB can be expressed as
Po1dB PiIP3 + GLdB − 10.63 (14.89)
or
Po1dB PoIP3 − 10.63 (14.90)
This is a very useful expression and frequently used by manufactures and engineers
to define the output power capability of power devices and amplifiers.
Example 14.26 Calculate the output power at 1-dB gain compression point of an
amplifier having third order output intercept point 1-W. Also, calculate the third order
intermodulation when the amplifier operates at its 1-dB gain compression point and
10 dB input back-off compared to its P1dB (consider the amplifier characteristic can
be expressed by a power series up to third order).
Solution
Output power at 1-dB gain compression point is given by
Po1dB PoIP3 − 10.63
Here, PoIP3 1 W 30 dBm.

Thus, Po1dB (30 − 10.63) 19.37 dBm.
From the definition, at the third-order intercept point the power of fundamental
frequency and third-order intermodulation is the same and each dB decrease/increase
of input power level corresponding to 1-dB decrease/increase of fundamental output
power and 3 dB decrease/increase of third-order intermodulation output power level.
Since Po1dB is 10.63 dB less than the third-order intercept point, the power level
of the third-order intermodulation product at Po1dB will be (30 − 3 × 10.63) dBm
−1.89 dBm, which is (19.37 + 1.89) dBc 21.26 dBc less compared to fundamental
power level.
Thus, IM3 (at Po1dB ) −21.26 dBc.
Similarly the third order intermodulation product at 10-dB input back-off com-
pared to 1-dB gain compression point will be
IM3 (at Po1dB − 10 dB) (−21.26 − 2 × 10) dBc −41.26 dBc.
It should be kept in mind that this calculation is valid up to the power level
over which the transfer characteristic is expressed by a power series up to third-order
term. For Class-A power amplifiers, operating regions up to 1–2 dB gain compression
point are considered to be expressed by a power series of the input signal up to the
third-order terms.
14.7.2 Load-Pull Characterization Technique
Load-pull technique was developed to practically tune the output termination to

determine the required output impedance for maximum output power from a device.
This technique is not only used for output tuning, but also used to characterize
the device as well as to determine source and load impedances for targeted out-
put power, gain, noise figure, efficiency, etc. The load-pull characterization involves
tuning the input and output impedances using tuners, recording performance and
impedances and then plotting contours of constant performances such as noise
figure, output power, efficiency, etc., on the impedance plane. To determine opti-
mum load impedance of microwave power transistor corresponding to maximum
output power, the load-pull technique is used. In case of power amplifiers, this tech-
nique is used for large signal characterization and determination of load as well
as source impedances of microwave power transistors to map contours of constant
power and efficiency on a Smith chart. Schematic diagram of a typical load-pull
test setup is shown in Fig. 14.34. Here, output RF power is delivered to the load
Input Reflected Output

Power Power Power
PM PM PM
DUT
INPUT OUTPUT
TUNER TUNER
VECTOR NETWORK
ANALYZER
Fig. 14.34 Load-pull setup for power device characterization
and input power to the amplifier and power reflected from the amplifier is measured
using RF power meter. In load-pull measurement technique for obtaining maximum
output power/efficiency, the load impedance is modified/adjusted using a stub tuner
until the output RF power/efficiency becomes maximum. A vector network analyzer
(VNA) is used to measure the real and imaginary parts of the tuned load impedance
(complex load) with the help of RF switch. The gain of the amplifier is maximized
by adjusting the input tuner. For a bidirectional device, it is required to adjust both
the tuner iteratively to maximize output power/efficiency and gain of the amplifier
due to the dependence of output impedance on its input source impedance.
The requirement of load impedances for achieving maximum output power and
maximum efficiency is different. Constant output power contours are the loci of load
impedances for constant output power. The constant output power contours are circles
when the amplifier operates at small signal conditions as discussed and computed
using S-parameters of the device earlier in this chapter. With the increase of power
level, the constant output power contours deviate from circles and look like potato
shapes as shown in Fig. 14.35. The output impedance contours consist of several
curves on a Smith chart representing constant output power and become a point for
the maximum output power. In practical case, constant efficiency contours are also
derived by experiment and drawn on the same Smith chart with the constant power
contours. Depending on applications, proper load impedance is selected for targeted
output power and efficiency.
14.7.3 Amplifiers with Reduced Conduction Angle
One of the techniques to improve power amplifiers efficiency is to operate the device
for reduced conduction angle. Power amplifiers are categorized into Class-A, Class-
AB, Class-B and Class-C based on the conduction angle. Efficient RF/microwave
1.0 j
0.5 j 2.0 j
Maximum
efficiency
0.5 1.0 2.0

0
Maximum
output power
-0.5 j
-2.0 j
-1.0 j
Constant power Constant efficiency
Fig. 14.35 Constant output power contours
Table 14.10 Amplifier Class of amplifier Bias current Conduction angle

classes
Class-A I F /2 2π
Class-AB I F /2–0 2π –π
Class-B 0 π
Class-C 0 π –0
amplifiers are designed by biasing the transistor to a low quiescent current, thus
allowing the device for conduction over lesser angle as shown in Fig. 14.36. Beyond
Class-A operation, the device is biased with a quiescent point more towards cut-off.
The conduction angle and operating bias conditions for different classes of ampli-
fiers are shown in Table 14.10. It is clear from Fig. 14.36, that larger amplitude of RF
drive is required for the power amplifiers beyond Class-A operation to achieve max-
imum current swing. Therefore, the gain of the amplifier reduces when the device
operates beyond Class-A condition.
The current waveform with reduced conduction angle is shown in Fig. 14.37. The
clipped drain current waveform idrf can be expressed as Fourier series:
∞
∞

idrf (t) a0 + an cos nωt + bn sin nωt (14.91)
n1 n1
Here, ω is the angular frequency of the RF signal, a0 , an and bn are constants and
Class-A Class-AB
/2
< /2
0 0
Class-B Class-C
0 0
Fig. 14.36 Device current for different classes of amplifiers
0,0 2 5 /2 3 4
2 1
3 /2 7 /2
Fig. 14.37 Current waveform with reduced conduction angle (2θ1 )

2π
1
a0 idrf (θ )dθ (14.92a)
2π
0
2π
1
an idrf (θ ) cos nθ dθ (14.92b)
π
0
2π
1
bn idrf (θ ) sin nθ dθ (14.92c)
π
0
Here, ωt θ . Suppose, conduction angle of the drain current waveform is 2θ1 as

shown in Fig. 14.37. The coefficients can be determined by performing the integration
over any 2π radians without loss of any generality. Here, for ease of mathematical
derivation the integration will be performed over the period of 3π/2 to 7π/2 where
the function can be expressed as
idrf (θ ) 0, over 3π/2 ≤ θ ≤ 5π/2 − θ1

IdQ + Idp sin θ over 5π/2 − θ1 ≤ θ ≤ 5π/2 + θ1 (14.93)
0, over 5π/2 − θ1 ≤ θ ≤ 7π/2
From Fig. 14.37, Idp can be expressed as
IF IdQ + Idp (14.94a)
Putting idrf 0 for θ 5π/2 − θ1 in (14.93):
IdQ + IdP sin(5π/2 − θ1 ) 0 (14.94b)
Therefore,
IF −IF cos θ1
Idp and, IdQ (14.95)
1 − cos θ1 1 − cos θ1
The coefficients can be determined as follows:

7π/2
1
a0 idrf (θ )dθ
2π
3π/2
1
5π/2+θ
1
IdQ + Idp sin θ dθ
2π
5π/2−θ1
1 5π/2+θ
IdQ θ − Idp cos θ 5π/2−θ11
2π
1
2IdQ θ1 − 2Idp sin θ
2π
Putting the values of Idp and IdQ from (14.95)
IF (sin θ1 − θ1 cos θ1 )
a0 (14.96)
π (1 − cos θ1 )
This gives the DC component of the current waveform.
2π
1
an idrf (θ ) cos nθ dθ 0
π
0
1
5π/2+θ
1
IdQ + Idp sin θ cos nθ dθ 0
π
5π/2−θ1
0
And,
1
5π/2+θ
1
bn IdQ + Idp sin θ sin nθ dθ
π
5π/2−θ1
Therefore, the fundamental component of the current waveform can be expressed

as
1
5π/2+θ
1 Idp
b1 IdQ sin θ + (1 − cos 2θ) dθ
π 2
5π/2−θ1
5π/2+θ1
1 Idp Idp
−IdQ cos θ + θ− sin 2θ
π 2 4 5π/2−θ1

1 Idp
2IdQ sin θ1 + Idp θ1 + sin 2θ1
π 2
Putting the values of Idp and IdQ from (14.95)

IF (2θ1 − sin 2θ1 )
b1 (14.97)
2π (1 − cos θ1 )
The maximum output power and efficiency of the amplifier for different con-
duction angles can be calculated using the DC component a0 and amplitude of the
fundamental component b1 of the drain current waveform. To compute these param-
eters, it will be assumed that the harmonic of the current waveforms are perfectly
short and the current swing is linear and up to the maximum value IF . It will also
be assumed that FET is used as the amplifying device of zero Knee voltage and the
load resistance RL will remain constant with the following value:
2VDC
RL (14.98)
IF
Here, VDC is the drain supply voltage of the FET device.

For Class-A amplifier configuration, there will be no clipping of the current wave-
form, thus conduction angle is 2θ1 2π . Thus, from (14.96) and (14.97) the DC
current (IDC ) and amplitude of the fundamental component (I d1 ) of the drain current
waveform can be written as
IF (sin π − π cos π) IF
IDC a0 (14.99a)
π (1 − cos π ) 2
IF (2π − sin 2π ) IF
Id 1 b1 (14.99b)
2π (1 − cos π) 2
Therefore, the maximum RF output power:

2
IF 2VDC IF VDC
PArf max √ (14.99c)
2 2 IF 4
The maximum efficiency ηmax will be

Prf max 1
ηA max 50% (14.99d)
VDC IDC 2
Maximum RF output power and efficiency for conduction angle of 2π to zero

which covers Class-A to Class-C amplifiers are derived based on (14.96), (14.97)
and (14.98) and are shown in Fig. 14.38. Here, output RF power is shown in dB
with respect to maximum RF output power corresponding to Class-A condition. It
shows that efficiency of the amplifier increases while configuring the amplifier from
Class-A towards Class-C amplifier. However, RF output power drastically reduces
beyond Class-B amplifier configuration
For Class-B amplifier configuration, the conduction angle 2θ1 π . Thus, from
(14.96) and (14.97), the DC current (IDC ) and amplitude of the fundamental compo-
nent (Id 1 ) of the drain current waveform can be written as

IF sin π2 − π2 cos π2 IF
IDC a0 π (14.100a)
π 1 − cos 2 π
IF (π − sin π ) IF
Id 1 b1 (14.100b)
2π (1 − cos(π/2)) 2
Therefore, the maximum RF output power:

A Class - AB B Class - C
1 100
0
90
Efficiency (%)
-1
-2 80
PRF (dB)
-3
-4 70
-5
60
-6
-7 50
2 3 /2 /2 0
Conduction Angle (2 1)
Fig. 14.38 RF power and efficiency of different classes of amplifier
2
IF 2VDC IF VDC
PBrf max √ (14.100c)
2 2 IF 4
The maximum output RF power of a Class-B amplifier is same as Class-A ampli-

fier. The maximum efficiency of Class-B amplifier ηB max is given by
PBrf max π
ηB max 78.5%
VDC IDC 4
The amplifier operating in Class-B condition delivers the same output power as
of Class A condition at their maximum current and voltage swings. However, the
DC current is reduced to IF /π , which is reduced by a factor of π/2 compared to DC
current in Class A amplifier.
The conduction angle of the device operating in Class-C condition is further
reduced which thus increases efficiency. However, with the reduction of conduction
angle, RF output power decreases substantially. For the reduced conduction angle,
the current waveform becomes pulse like which contain high level of harmonic com-
ponents. Thus, in addition to the generation of high level of harmonic components,
the drastic reduction of RF output power makes Class-C amplifier not favourable for
RF/microwave communication applications.
14.7.3.1 Back-off Operation of Class-A and Class-B Amplifiers
Depending on the applications, power amplifiers are also operated under different
output power conditions. For example, power amplifiers are operated under power
back-off condition under multicarrier as well as the applications where higher mod-
ulation codes are used for communication. Thus, it is important to know the output
/2
0 0
Class-A Class-B
Fig. 14.39 Signal waveforms of Class-A and Class-B amplifier
RF power and implication on the efficiency of Class-A and Class-B amplifiers when
they are operated under power back-off condition. Input voltage and output current
swings for Class-A and Class-B amplifiers at the maximum input and back-off swing
conditions are shown in Fig. 14.39.
Suppose FET device of gate-to-drain breakdown voltage VdgB and Knee voltage
Vk is used for the power amplifier. The gate-to-source pinch-off voltage of the FET
is VgsO . Bias conditions for Class-A amplifiers to achieve maximum drain voltage
and current swings under maximum RF operating condition:

Vgs − VgsO − VF /2
IdQ IF /2

VdQ VdgB − VgsO + VK /2
The load resistance:

RL VdgB − VgsO − VK /IF
The RF output power for Class-A amplifier is given by

2 2
idrf idrf VdgB − VgsO − VK
PArf √ RL (14.101a)
2 2IF
The DC power:

VdgB − VgsO + VK
Pdc VdQ × IdQ IF × (14.101b)
4
Thus, the power added efficiency:
2
1 idrf VdgB − VgsO − VK
ηadd 2 1− (14.101c)
G IF VdgB − VgsO + VK
Considering, VdgB VgsO and VdgB VK , the power added efficiency of the
Class-A amplifier of very high gain can be approximated as
2
idrf
ηadd 2 (14.101d)
IF
Thus, the efficiency of Class-A amplifier increases with the increase of RF current
swing and it becomes the maximum of value 50% when the RF current swing becomes
IF /2.
Bias conditions for Class-B amplifiers to achieve maximum drain voltage and
current swings under maximum RF operating condition is

Vgs − VgsO
IdQ 0
and

VdQ VdgB − 2VgsO + VK − VF /2
Thus load resistance will be

RL VdgB − 2VgsO − VK − VF /IF
For Class-B amplifier, the current waveform is half sinusoidal. Suppose, peak
value of the half sinusoidal waveform is idrf . From (14.100b), the amplitude of the
fundamental component of the half sinusoidal waveform of peak value idrf is idrf /2.
Thus, the RF output power PBrf of Class-B amplifier:
2 2
idrf idrf VdgB − 2VgsO − VK − VF
PBrf √ RL (14.102a)
2 2 8IF
From (14.100a), the DC component Idc of the half sinusoidal waveform of peak
value idrf will be
idrf
IDC
π
Thus, DC power will be

idrf VdgB − 2VgsO + VK − VF
Pdc IDC × VdQ (14.103a)
π 2
The power added efficiency:
Efficiency (%)
80
78.5%
60
50 %
40
Class-B
20 Class-A
0
-20 -15 -10 -5 0
Output power Back-off (dB)
Fig. 14.40 Efficiency of Class-A and Class-B power amplifier with output power back-off

1 π idrf VdgB − 2VgsO − VK − VF
ηadd 1− (14.103b)
G 4 IF VdgB − 2VgsO + VK − VF
Considering VdgB VgsO , VdgB VK and VdgB VF the power added efficiency
of Class-B amplifier with vary high gain can be approximated as

π idrf
ηadd (14.103c)
4 IF
Thus, efficiency of Class-B amplifier also increases with the increase of RF current
swing and it reaches a maximum value of 78.54% when the RF current swing becomes
equal to value of IF .
From (14.101d) and (14.103c), it is clear that power-added efficiency is linearly
proportional to idrf for Class-B amplifier whereas proportional to the square of idrf
for Class-A amplifier. Figure 14.40 shows the efficiency variation of Class-A and
Class-B amplifiers with output power back-off. Output power back-off is defined as
the decrease of RF output power compared to the maximum RF output power.
Thus, for Class-A amplifier the RF output power back-off (OBO) in dB can be
expressed from (14.101a) as
2
idrf 2 × idrf
OBOClass-A 10 × log10 20 × log10
idrf max IF
Similarly, for Class-B amplifier the RF output power back-off in dB can be

expressed as
Fig. 14.41 Drain current

waveform of overdriven FET
power amplifier
2 1
2
0
2 2
2
idrf idrf
OBOClass-B 10 × log10 20 × log10
idrf max IF
Figure 14.40 shows that the efficiency of Class-A amplifier decreases sharply
compared to Class-B amplifier with output power back-off. This is one of the draw-
backs of Class-A amplifier compared to other classes of amplifier.
14.7.4 Nonlinear Power Amplifiers
Practically, it is required to maintain linearity of an amplifier for communication

systems. To maintain the linearity under the assumption of an ideal characteristic of
the device, it is required to restrict the input signal level of a power amplifier to avoid
clipping of the output voltage and current waveforms. However, there are certain
applications in communication systems where linearity can be traded for achieving
better efficiency of the power amplifiers. Therefore, Class-A and Class-B amplifiers
are also operated under higher input signal level which drives the amplifiers towards
saturation.
Under overdriven condition, both upper and lower portion of the drain current
waveform may clip due to the limitation of current from the device. A typical clipped
RF output current waveform is shown in Fig. 14.41. This periodic output current
waveform can be expressed with Fourier series as
∞
∞

idrf (t) a0 + an cos nωt + bn sin nωt (14.104)
n1 n1
Here, ω is the angular frequency of the RF signal a0 , an and bn are constants and
2π
1
a0 idrf (θ )dθ (14.105a)
2π
0
2π
1
an idrf (θ ) cos nθ dθ (14.105b)
π
0
2π
1
bn idrf (θ ) sin nθ dθ (14.105c)
π
0
As shown in the figure, the clippings of the current waveform are over 2θ1 and
2θ2 . Therefore, the drain current waveform can be expressed as
idrf (θ ) IdQ + Idp sin θ, over 0 ≤ θ ≤ π/2 − θ1

IF , over π/2 − θ1 ≤ θ ≤ π/2 + θ1
IdQ + Idp sin θ, over π/2 + θ1 ≤ θ ≤ 3π/2 − θ2 (14.106)
0, over 3π/2 − θ2 ≤ θ ≤ 3π/2 + θ2
IdQ + Idp sin θ, over 3π/2 + θ2 ≤ θ ≤ 2π
The constants can be determined as follows:
2π
1
a0 idrf (θ )dθ
2π
0
⎡ π π
−θ1
2 2 +θ1
1 ⎢
⎣ IdQ + Idp sin θ dθ + IF dθ
2π
π
0 2 −θ1
⎤
−θ2
3π
2 2π
⎥
+ IdQ + Idp sin θ dθ + IdQ + Idp sin θ dθ ⎦
π 3π
2 +θ1 +θ2
2
or
π − θ1 − θ2 sin θ2 − sin θ1 θ1
a0 IQ + Idp + IF (14.107)
π π π
Similarly,
2π
1
a1 idrf cos θ dθ
π
0
⎡ π π
−θ1
2 2 +θ1
1⎢
⎣ IdQ + Idp sin θ cos θ dθ + IF cos θ dθ
π
π
0 2 −θ1
⎤
−θ2
3π
2 2π
⎥
+ IdQ + Idp sin θ cos θ dθ + IdQ + Idp sin θ cos θ dθ ⎦
π 3π
2 +θ1 +θ2
2
or
a1 0 (14.108)
2π
1
b1 idrf (θ ) sin θ dθ
π
0
⎡ π π
−θ1
2 2 +θ1
1⎢
⎣ IdQ + Idp sin θ sin θ dθ + IF sin θ dθ
π
π
0 2 −θ1
⎤
−θ2
3π
2 2π
⎥
+ IdQ + Idp sin θ sin θ dθ + IdQ + Idp sin θ sin θ dθ ⎦
π 3π
2 +θ1 +θ2
2
or,

1 1
b1 π − θ1 − θ2 − (sin 2θ1 + sin 2θ2 ) Idp
π 2
2 sin θ2 − 2 sin θ1 2 sin θ1
+ IQ + IF (14.109a)
π π
Using these expressions, output power and efficiency of Class-A and Class-B
amplifiers when operated under overdriven condition can be derived.
14.7.4.1 Overdriven Class-A Power Amplifiers
Class-A amplifiers are operated over the RF power level in which output current of
the device conducts over full RF cycle. However, when the Class-A amplifier operates
in RF overdriven condition, output current waveform of the device clips at extreme
levels. Input and output signal current waveforms of a Class-A amplifier when it is
operated under such large RF signal level condition, is shown in Fig. 14.42. Here,
it is considered that the device is biased to its exactly midpoint of the current and
( , )
0
−
Fig. 14.42 Current waveform of overdriven power amplifier
voltage capacity which ensures symmetric clipping of the output current waveform,
i.e.
θ1 θ2 α (say)
IQ IF /2
Putting this condition to (14.107), the DC current can be expressed as

π − 2α IF α
IDC a0 + IF
π 2 π
IF

2
Putting the values of θ1 θ2 α and IQ IF /2 to (14.109a), the fundamental
component of RF output current can be written as
or

1 1 2 sin α
idrf b1 π − 2α − (sin 2α + sin 2α) Idp + IF (14.109b)
π 2 π
Putting idrf (π/2 − θ1 ) IF in (14.106):

IF − IQ IF
Idp (14.110)
cos α 2 cos α
Putting the value of Idp in (14.109b):
⁄ (dB) Efficiency (%)

2.5 100
2.0 90
1.5 80
1.0 70
0.5 60
0.0 50
0 45 90 135 180
Clipped angle, 2 (deg)
Fig. 14.43 RF output power and efficiency for clipped sinusoidal output (overdriven Class-A ampli-
fier)

IF π − 2α + sin 2α
idrf (14.111)
π 2 cos α
Therefore, the output RF power and efficiency can be written as

idrf 2
Prf √ RL
2

IF π − 2α + sin 2α 2
VdgB − VgsO − VK (14.112)
2π 2 2 cos α

2 π − 2α + sin 2α 2 VdgB − VgsO − VK
η 2 (14.113)
π 2 cos α VdgB − VgsO + VK
RF output power and efficiency of overdriven Class-A amplifier is shown in

Fig. 14.43. The saturated RF output power, i.e. power corresponding to α approaching
90°, is about 2-dB more than the maximum RF output power compared to unclipped
Class-A operation. Under this condition, the efficiency will be about 81%.
It is very important and interesting, that the output power as well as efficiency of
an overdriven Class-A amplifier improves significantly. However, to achieve square
wave output waveform, injection of a very high level of input is required which not
only decreases the gain of the amplifier but also degrades device reliability due to
high gate current for excessive forward gate voltage. One practical way for achieving
high power and efficiency from the Class-A amplifier is the use of square wave input
signal in place of the sinusoidal signal waveform. Practically, this is done to some
−
0
( −2 + − )/2
Fig. 14.44 Current waveform of overdriven Class-B power amplifier
extent by clipping the signal waveform in the preceding amplifier stage before feeding
it to the high power amplifier.
14.7.4.2 Overdriven Class-B Power Amplifiers
Input and output signal waveforms of a class-B amplifier when operating under higher
signal level condition which drives the amplifier to clip the output RF current is shown
in Fig. 14.44. With reference to Fig. 14.41, for current waveform of Fig. 14.44:
θ2 π/2, IQ 0
and
θ1 α (say)
Putting idrf (π/2 − α) IF in (14.106):

IF
Idp (14.114)
cos α
Putting this condition to (14.114), the DC current can be expressed as

IF α cos α − sin α + 1
IDC a0 (14.115)
π cos α
3.0 90
Efficiency (%)
2.0 85
PRF (dB)
1.0 80
0.0 75
0 45 90 135 180
Clipped Angle (2 1)
Fig. 14.45 RF power and efficiency of overdriven Class-B power amplifier
From (14.109a), the fundamental component of RF output current can be written

as
IF (π − 2α + sin 2α)
idrf b1 (14.116)
π 2 cos α
Therefore, output RF power Prf and efficiency η can be written as

idrf 2
Prf √ RL
2

IF π − 2α + sin 2α 2
VdgB − 2VgsO − VK − VF (14.117)
2π 2 2 cos α
1 (π − 2α + sin 2α)2
η (14.118)
4π cos α (α cos α − sin α + 1)
RF output power and efficiency of overdriven Class-B amplifier is shown in

Fig. 14.45. The saturated RF output power is about 2 dB more than the maximum RF
output power. At this condition, the efficiency is about 81%. However, it is clear from
the plot that the required amount of waveform clipping is not same for the maximum
output power and maximum efficiency i.e. efficiency does not go on increasing as in
case of overdriven Class-A amplifier.
14.7.5 Class-F Power Amplifier
In the case of amplifiers operating in an ideal Class-B operation mode with the load
tuned at the fundamental frequency, the maximum achievable efficiency is 78.5%. At
Fig. 14.46 Device output ,

voltage and current
waveforms for Class-F
amplifier
0 2
this condition, the device output voltage waveform is sinusoidal and current wave-
form is half-sinusoidal. In case of Class-F amplifier, efficiency can be maximized
up to the theoretical limit of 100%, by shaping the device voltage and current wave-
forms by putting additional resonators which are tuned to the corresponding harmonic
components [14–29].
A Class-F power amplifier provides improved efficiency with higher RF power
output compared to Class-A and Class-B configuration by proper termination of
harmonics at the device output which shapes its voltage and current waveforms. Pur-
pose of loading for harmonics is to shape the device output voltage in a way to avoid
overlap or to minimize the overlap with its current, thus greatly reducing the power
dissipation within the device, and hence increasing the efficiency. Here, amplitude
of voltage and current harmonics are appropriately added to the respective amplitude
of fundamental components to keep resultant voltage and current swings within the
allowable limit of the device. This ensures to operate the device within its alloable
voltage and current limit and without driving the amplifier into hard compression.
In general, either voltage or current waveform (e.g. voltage) approximates a square
wave, while the other (e.g. current) approximates a half sine wave with minimum
overlap over the time by providing proper termination to the harmonics at the out-
put of the device. The ideal voltage and current waveforms of a device operating
in Class-F configuration is shown in Fig. 14.46. Zero overlap between voltage and
current of the device signifies that the power dissipation within the device is zero.
Thus, efficiency of the power amplifier operating in Class-F condition is 100%. The
voltage waveform vdrf of the device can be expressed using Fourier series as follows:
∞
∞

vdrf (t) a0 + an cos nωt + bn sin nωt (14.119)
n1 n1
Here, ω is the angular frequency of the RF signal a0 , an and bn are constants and
2π
1
a0 vdrf (θ )dθ
2π
0
2π
1
an vdrf (θ ) cos nθ dθ
π
0
2π
1
bn vdrf (θ ) sin nθ dθ
π
0
Here, the drain voltage waveform can be expressed as
vdrf (θ ) 0, over 0 ≤ θ ≤ π
vdp , over π ≤ θ ≤ 2π
Therefore, constants can be determined as follows:
2π
1
a0 vdrf (θ )dθ
2π
0
2π
1
vdp dθ
2π
π
or
vdp
a0
2
Similarly,
2π
1
an vdp cos nθ dθ
π
π
vdp 2π
vdp sin nθ π 0 for all values of n
nπ
Similarly,
2π
1
bn vdp sin nθ d θ
π
π
vdp
[− cos nθ ]2π
π
nπ
or
2vdp
bn − nπ
for all odd values of n
0 for all even values of n
Therefore, the output voltage waveform can be expressed as
∞
vdp 2vdp 2vdp
vdrf (t) − sin ωt − sin nωt (14.120)
2 π n1,3,5,...
nπ
The first term of (14.120) is the DC component, second term is the fundamental
component and others are all odd harmonic components. Therefore, the normal-
ized amplitude of fundamental and harmonic components with respect to the DC
components can be expressed as
4
vn (14.121)
nπ
The half sinusoidal output current waveform is given by
idrf (θ ) Idp sin θ, over 0 ≤ θ ≤ π

0, over π ≤ θ ≤ 2π
Thus, the current waveform can be expressed in Fourier series as

∞

Idp Idp 2Idp cos nωt
idrf (t) + sin ωt − (14.122)
π 2 π n2,4,6,...
n2 − 1
Here, the first and second terms are the DC and fundamental components, respec-
tively. Other terms are all the even-order harmonic components.
Current and voltage waveforms have never simultaneous nonzero values, and
only the fundamental frequency component is present in both voltage and current
waveforms and thus power is generated only in fundamental frequency. Whereas
the current contains only even harmonics, the voltage contains only odd harmonics,
therefore there will be no generation of harmonic power.
Output matching network and bias condition of a Class-F power amplifier should
be such that it supports square wave voltage waveform and half sinusoidal output
current wave with no overlapping in the time axis. A generalized Class-F power
amplifier using FET and its drain voltage vdrf and drain current idrf waveforms are
shown in Fig. 14.47 when the device is biased as for the case of Class-B amplifier
configuration. From the above analysis, it is clear that the drain voltage and current
waveform of the fundamental frequency are shifted in phase by 180 from each other.
Also, the drain voltage waveform consists of all odd harmonics (no even harmonics)
to build its square wave shape whereas the drain current waveform adds all even har-
at all odd harmonics

0 at all even harmonics
Harmonics
Resonator
Fig. 14.47 Dain voltage and current waveform of Class-F amplifier
monics (no odd harmonics) to build its half sine wave shape. Thus, there will be no
power dissipation within the device and also there will be no power generation at the
harmonic frequencies since there is either no voltage or no current present at a given
harmonic. To achieve these voltage and current waveforms, output impedance must
provide either zero or infinite impedances at the harmonic frequencies. In case of
an ideal Class-F power amplifier with square shape voltage waveform and half sinu-
soidal current waveform, the output impedance should provide an infinite impedance
for all odd harmonics and zero impedance to all even harmonics and optimum load
impedance corresponding to maximum allowable voltage and current swings at the
fundamental frequency. Under this condition, the ratio of voltage amplitude of fun-
damental frequency to the DC voltage is given from (14.121) as
4
v1 1.273
π
Since the current amplitude of the fundamental frequency is the same as in case of
Class-A amplifier, the maximum output power of the ideal Class-F power amplifier
is 1.273 times the power corresponding to Class-A amplifier configuration, which is
about 1-dB higher. The efficiency of the Class-F power amplifier can be calculated
in the following way:
Suppose the amplifier is biased as in Class-B. Therefore, from (14.120) and
(14.122) the DC power PDC is given by
vdp Idp
PDC VDC × IDC ×
2 π
Whereas, the maximum RF power PRF max from (14.120) and (14.122) is given by
2vdp Idp vdp Idp

PRF max √ × √ ×
π 2 2 2 2 π
Thus, maximum efficiency in Class-F amplifier:
ηmax PRF max /PDC 100%
Here, the device considered is an ideal device with zero Knee voltage. This highly
efficient operation of Class-F power amplifier can be obtained only when resonant
circuit tuned to an infinite number of odd and even harmonics of the fundamental
frequency is added into the load network.
Another version of Class-F power amplifier mode of operation is inverse Class-F
mode. In case of inverse Class-F mode of operation, the device output RF voltage
waveform is half sinusoidal and the current waveform is square shaped. Design of
a Class-F and inverse Class-F amplifier involves the design of matching network at
the fundamental frequency, and harmonic tuned load networks.
In conventional Class-F or inverse Class-F power amplifiers, the fundamental
and harmonic load-network impedances are optimized by short-circuit termination
and open-circuit peaking corresponding to finite number of harmonics to control the
voltage and current waveforms at the device output to obtain maximum efficiency
and output power. Practically, few harmonics such as up to third- or fifth-order
harmonics are considered to reduce the complexity for improvement of efficiency
and RF output power. To achieve maximum possible output power and efficiency
considering proper termination of up to a finite number of harmonics, the relative
amplitude should be chosen such that the amplitude of the fundamental component
becomes maximum keeping the composite voltage and current waveform within the
capability of the device. In this way, without overstressing the device, higher RF
output power and efficiency can be achieved from Class-F power amplifier with a
finite number of harmonic tuning. Figure 14.48 shows the dependence of RF output
power and efficiency of Class-F power amplifier on the amplitude of third harmonic
voltage component when up to third harmonic is considered for proper termination.
It shows that in absence of any harmonics in voltage waveform, the efficiency is
nothing but the classical value of 78.5% corresponding to the Class-B amplifier.
With the increase of the amplitude of the third harmonic component, the efficiency
increases up to a value of about 90.7%. In this condition, the RF output power is about
0.63 dB more compared to maximum power available from Class-A configuration.
With further increase of amplitude of the third harmonic component, the efficiency
as well as RF output power decreases.
Table 14.11 shows the normalized amplitudes of fundamental and harmonic com-
ponents for different cases when up to 3rd, 5th, 7th, 9th, 11th, 13th, 15th and infinite
number of harmonics are considered. These values are determined numerically by
optimizing the amplitude of harmonic components to maximize the signal amplitude
of fundamental components keeping the normalize waveform envelope within the
allowable limit of unity.
1.0 100
Increase of RF Power (dB)

90.7%
0.8
90
Efficiency (%)
0.63 dB
0.6
78.5% 80
0.4
70
0.2
0.0 60
0.0 0.1 0.2 0.3 0.4
Normalized amplitude of 3rd harmonic
Fig. 14.48 RF output power and efficiency dependence on amplitude of third harmonic component
Table 14.11 Normalized amplitude of fundamental and odd harmonics

Harm v1 v3 v5 v7 v9 v11 v13 v15
1 1 – – – – – – –
3 1.155 0.194 – – – – – –
5 1.207 0.283 0.075 – – – – –
7 1.229 0.336 0.132 0.046 – – – –
9 1.240 0.364 0.170 0.081 0.035 – – –
11 1.245 0.381 0.191 0.105 0.055 0.026 – –
13 1.250 0.390 0.208 0.125 0.076 0.045 0.026 –
15 1.252 0.398 0.216 0.133 0.084 0.053 0.032 0.018
Infinite 4/π 4/3π 4/5π 4/7π 4/9π 4/11π 4/13π 4/115π
Figure 14.49 shows the device voltage waveforms corresponding to fundamental

frequency as well as composite waveform when up to different harmonic components
are considered for waveform shaping to maximize RF output power and efficiency.
The normalized amplitude of the fundamental frequency component becomes 1.155,
1.207, 1.229, 1.240, 1.245, 1.250 and 1.252 when the harmonics are considered
up to 3rd , 5th , 7th , 9th , 11th , 13th and 15th , respectively. Thus, with the increase of
the considered number of harmonics, the amplitude of the fundamental frequency
component increases though the maximum value of the composite waveform is within
the allowable limit of unity. It is also to be noted that, with the increase of a number
of considered harmonics, the composite voltage waveform becomes more and more
flatten, i.e. similar to a square waveform.
The improvement in efficiency and RF output power level with the increase of
number of harmonics considered are shown in Figs. 14.50 and 14.51, respectively.
From the plots, it is evident that theoretical efficiency of about 95% and increase of
RF output power level of about 0.8 dB is possible when only up to the 5th harmonic
components are considered. Thus, practically without going into the realization of
(a) (b)
Amp Amp
1.5 1.5
1.155
1.000
1.0 1.0
0.5 0.5
0.0 0.0
0 0
(c) (d)
Amp Amp
1.5 1.5
1.207 1.229
1.0 1.0
0.5 0.5
0.0 0.0
0 0
(e) Amp (f) Amp

1.5
1.5 1.245
1.240
1.0 1.0
0.5 0.5
0.0 0.0
0 0
(g) (h)
Amp Amp
1.5 1.5
1.250 1.252
1.0 1.0
0.5 0.5
0.0 0.0
0 0
Fig. 14.49 Device voltage waveform (composite and fundamental component) considering, a no
harmonic component tuned, harmonic components tuned up to b 3rd , c 5th , d 7th and e 9th , f 11th ,
g 13th . h 15th
100.0
100 97.4 97.8 98.1 98.4
96.5
94.8
95
90.7
Efficiency (%) 90
85
80 78.5
75
70
1 3 5 7 9 11 13 15 infinite
Harmonic considered up to
Fig. 14.50 Efficiency of Class-F power amplifier
1.20 1.05
0.94 0.95 0.97 0.98
Rela ve RF output (dB)
1.00 0.89
0.82
0.80
0.63
0.60
0.40
0.20
0.00
0.00
-0.20
1 3 5 7 9 11 13 15 infinite
Harmonic considered up to
Fig. 14.51 RF output power of Class-F power amplifier
very complex output networks, higher efficiency of the order of 90% with at least
0.5 dB more output power compared to Class-A is expected from Class-F microwave
power amplifiers.
14.8 FETs Output Power Capability
The maximum linear RF output power from a Class-A FET amplifier is given by
(14.81):

IF VdgB − VgsO − VK
Prf (max)
8

Neglecting VgsO and VK compared to the gate-to-drain breakdown voltage VdgB
the maximum RF output power is given by
IF × VdgB
Prf (max) (14.123)
8
Assuming uniform channel doping of density ND , the maximum drain current
(IF ) can be written using (6.3) as
IF q ND dvn max Wg
Here, d is channel depth, Wg is gate width and vn max is the maximum velocity of
the channel carriers. The breakdown voltage is given by
εEdgB
2
VdgB
2qND
Here, EdgB is the breakdown field and ε is the dielectric constant of the FET
channel material. The maximum RF output power (14.123) can be expressed as
dεWg
Prf (max) 2
vnmax EdgB (14.124)
16
Suppose effective gate length of the FET is lgeff . The maximum frequency of
operation fmax can be expressed as
vnmax
fmax (14.125)
2π lgeff
Combining (14.124) and (14.125):

2 2
Prf (max) ε vnmax EdgB
× fmax (14.126a)
Wg 32π lgeff /a

The term, Prf (max)/Wg is the RF power density of a FET. Suppose, prf (max)
Prf (max)/Wg . Then
2 2
ε vnmax EdgB
prf (max) × fmax (14.126b)
32π lgeff /a
Here, (lgeff /a) is the aspect ratio of FET and practically is about 10–5. Considering
the factor (lgeff /a) is a constant, the product of maximum RF power density and maxi-
mum frequency of operation depends only on the material properties (vnmax × EdgB )2 .
Thus, the product (power × frequency) is a constant for a particular material tech-
nology and is figure of merit for power FET technology. Power amplifiers based
on GaN technology provides better (power × frequency) product due to its greater
gate-to-drain breakdown voltage and higher saturated electron velocity compared to
others [30–32].
14.8 FETs Output Power Capability 461
Table 14.12 Properties of GaN and GaAs materials

Material Dielectric Breakdown field (V/cm) Electron saturation velocity (cm/s)
constant (εr )
GaAs 13.1 0.65 × 106 1.3 × 107
GaN 9.0 3.5 × 106 2.5 × 107
Example 14.27 Compare the power frequency product between GaAs and GaN FETs
technologies.
Solution
Material properties of GaAs and GaN required for computing the power frequency
products of power FET technologies are shown in Table 14.12.
Putting parameter values in (14.126b), the ratio of power frequency products of
GaN and GaAs is

prf (max) × fmax GaN 9.0 2.5 × 3.5 2
73.7
prf (max) × fmax GaAs 13.1 1.3 × 0.65
This shows that power density of GaN HEMTs are about 74 times that of RF
GaAs FETs with both the devices having same maximum frequency of operation.
Example 14.28 Two GaAs FETs are designed with different gate lengths, one for
fmax 1 GHz and another for fmax 10 GHz maintaining their aspect ratio as 8.
Calculate RF power density of both the FETs.
Solution
The power density of a FET is given from (14.126b) as follows:
2 2
ε vnmax EdgB 1
Power density prf (max)
32π lgeff /a fmax
From Table 14.12 for GaAs FETs:

Dielectric constant, ε 13.1 × εo 13.1 × 8.854 × 10−12 F/m
vnmax 1.3 × 107 cm/s 1.3 × 105 m/s

EdGB 0.65 × 106 V/cm 0.65 × 108 V/cm
Aspect Ratio lgeff /a 8 (as given)

2
13.1 × 8.854 × 10−12 1.3 × 105 × 0.65 × 108 1
Power density W/m
32π 10 fmax (Hz)
1
10.3 W/mm
fmax (in GHz)
Thus, for GaAs FET with fmax 1 GHz power density is 10.3 W/mm and for
GaAs FET with fmax 10 GHz, power density is 1.03 W/mm.
Example 14.29 Two GaN FETs are designed with different gate lengths, one for
fmax 10 GHz and another for fmax 100 GHz maintaining their aspect ratio as 8.
Calculate RF power density of both the FETs.
Solution
The power density of a FET is given by (14.126b):
2 2
ε vnmax EdgB 1
Power density prf (max)
32π lgeff /a fmax
From Table 14.12, for GaN FETs:

Dielectric constant, ε 9 × εo 9 × 8.854 × 10−12 F/m
vnmax 2.5 × 107 cm/s 2.5 × 105 m/s

EdgB 3.5 × 106 V/cm 3.5 × 108 V/cm
Aspect ratio lgeff /a 8 (as given)

2
5 8
9 × 8.854 × 10−12 2.5 × 10 × 3.5 × 10 1
Power density W/m
32π 10 fmax (Hz)
1
758.6 W/mm
fmax (in GHz)
Thus, for GaN FET with fmax 10 GHz, power density is 75.86 W/mm and for
GaN FET with fmax 100 GHz power density is 7.58 W/mm.
The RF output power per unit gate width of a FET which is designed for a particular
maximum frequency fmax is practically determined by its material technology. For a
particular fmax , with the increase of device gate width, output current increases, thus
increases the output RF power.
A practical power amplifier designed to operate at about 4 GHz has fmax
20 GHz. Thus, from Example 14.27, power density of the GaAs power FET intended
to be operated at about 4 GHz is around 0.5 W/mm. Considering the gate-drain
breakdown voltage 20 V, maximum current can be determined from (14.123) as
IF 8 × Prf (max)/VdgB 8 × 0.5/20 200 mA/mm. Neglecting VgsO and VK
compared to VdgB , the load resistance for maximum output power in Class-A opera-
tion is, RL VdgB /IF (20/200) k 100 .
Thus, a GaAs power FET designed for fmax 20 GHz with gate width of 1.0 mm
can provide RF output power of about maximum 0.5 W when it is operated in linear
Class-A operation with a load resistance of 100 . A typical power device cell can
have about 1.5 mm gate width, thus the maximum RF output power from a power
device cell will be about 0.75 W with the required load resistance of 67 . Practically,
several such cells are connected in parallel to increase the effective gate width as
shown in Fig. 14.52a. Suppose 10 such cells are connected in parallel to design
power FET with 15 mm of effective gate width. The achieved maximum RF power
is about 7.5 Watts, and the load impedance is about 10 times less, i.e. about 6.7 .
Thus, with the increase of FET gate width to increase RF output power capability,
the required value of output load resistance decreases. So the conversion ratio to
transfer 50 to the required load resistance will be larger, making matching more
difficult over a wide band of frequency. Output powers of the individual chips are
combined using chips with pre-matched ports before parallel combination, or without
prior port impedance matching. To avoid impedance level problems, in general, chip-
level FETs are individually matched and transferred to the reference impedance (Z o
50 ) before combining power. In wide bandwidth applications, it is preferable
to use lesser gate width FETs and using of proper power combining technique, as
shown in Fig. 14.52b, instead of just paralleling the FETs to achieve more RF output
power. In the case of GaN HEMTs, higher RF output power density allows much
smaller size devices with the same output power. Therefore, for the same output
power capability, output impedance of GaN device is much higher compared to
GaAs devices. Higher output impedance of GaN power devices enables design of
broadband output matching network for realization of broadband power amplifiers.
14.8.1 Microwave Power Combing Techniques
There are many applications where higher RF power is required compared to what
is available from a single transistor. To meet the higher power requirement it is
required to combine the output power of several transistors. One technique to get
higher output power is to connect more and more number of transistors in parallel as
shown in Fig. 14.52a. This leads to lower terminal impedances mainly at the output
(b) Transistors
(a) Transistors
Input Output
power power
divider combiner
Fig. 14.52 Microwave power combining techniques, a parallel connection of transistors, b true
power combining with proper impedance transformation
Transistors
2-way 2-Way
power power
divider combiner
2-way 2-way
RF RF
power power
IN OUT
divider combiner
2-way 2-Way
power power
divider combiner
Fig. 14.53 Schematic block diagram of corporate-type power combining technique for combining
four transistors
of the combined amplifier which makes the impedance matching more difficult over
a broadband of frequency. This problem can be addressed by partially matching indi-
vidual transistors to transfer terminal impedances to higher values before combining
by parallel connection. Another problem is the phase difference among the signals
combining at its output due to the path differences, which leads to reduced gain
and reduced combined output power. This problem becomes more severe with the
increase of operating frequency and numbers of transistors to be connected in par-
allel. The phase imbalance problem can be reduced by placing the transistors close
to each other. However, this leads to the problem of heat dissipation and increase
the channel temperature of the device which reduces device reliability. To avoid the
problems of lowering port-impedances and phase imbalance in the simple parallel
combining of transistors, true power combining techniques using input power divider
and output power combiner are used as shown in Fig. 14.52b.
There are various types of power combining techniques such as serial power com-
bining, N-way power combining and corporate power combining. Among all these
techniques, corporate power combining also called tree method power combining
is mostly used. Schematic block diagram of a corporate-type power combining for
four transistor chips is shown in Fig. 14.53. In this case, the number of devices
changes with 2n , where n is an integer, thus it is also called binary power combining.
In the ideal case, where all the power chips are identical and perfectly matched and
there is no loss of the power divider and combiner networks, combined output power
will be 2n times of one chip but the gain will be equal to the gain of one device.
14.8.2 In-Phase Power Combiners
The Wilkinson splitter is the most widely used two-way in-phase power
divider/combiner used for microwave power combining. Schematic diagram of
Wilkinson power splitter is shown in Fig. 14.54. The power splitter consists of two
o/4, Z1
(a) o/4, Z1 (b) 2 Zo
2 Zo 1
1 A Riso/2
1 Riso
Riso/2 2
3
o/4, Z1 3 Zo
o/4, Z1
(c) o/4, Z1 (d) o/4, Z1

2 Zo 2 Zo
2Zo 2Zo 1
Riso/2 Riso/2
1
open short
Fig. 14.54 Wilkinson power divider, a schematic diagram, b shown as symmetric structure, c
equivalent circuit under even-mode signal (v1 v2 vs ), d equivalent circuit for odd-mode signal
(v1 −v2 vs )
branch lines of length λo /4 and a resistor in between the branches to provide isola-
tion between the output ports. The Wilkinson splitter has symmetric structure and
thus ideally it provides zero amplitude and phase difference between two outputs
irrespective of operating frequency. The characteristic impedance, Z1 of the lines
and value of the isolation resistor Riso can be determined by even- and odd-mode
analysis as follows:
With reference to Fig. 14.54b, under even mode excitation of the output ports, i.e.
when v1 v2 vs , electrical open can be considered along the line of symmetry as
shown by the dotted line “A”. Under the even mode excitation, the equivalent circuit
can be considered as shown in Fig. 14.54c. For impedance matching at the input port,
we can write:
2Zo Z12 /Zo
Thus,
√
Z1 2Zo (14.127a)
Similarly, under odd-mode excitation among the output ports, i.e. when v1
−v2 vs , electrical short can be considered along the line of symmetry as shown.
Thus, under the odd-mode excitation, the equivalent circuit will be as shown in
Fig. 14.54d. For impedance matching at the output port, we can write:
Riso /2 Zo
Fig. 14.55 Schematic block diagram of corporate-type power combining technique for combining
8 transistors
Thus,
Riso 2Zo (14.127b)
Four and other 2n ways corporate-type power combiner/divider can be realized

by just connecting more and more 2-way splitters in cascade configuration as shown
in Fig. 14.55.
Schematic diagram of a special type 4-way power combiner/divider suitable for
high power combing with low loss is shown in Fig. 14.56 [33, 34]. Using even-
and odd-mode analysis, branch impedances and values of isolation resistors can be
determined as follows:
Under the even-mode signal excitation, i.e. for v1 v2 v3 v4 vs , all the
isolation resistors (Riso1 and Riso2 ) of Fig. 14.56a can be considered as open. Thus,
for input impedance matching, we can write:
1 Z2
Zo × 2 1
2 Z2 /2Zo
This leads to
Z1 Z2 (14.128a)
This implies that the required characteristic impedances of the branches are inde-
pendent of the reference impedance Zo . Therefore, in case of planner circuit, wider
(low impedance)
√ transmission lines can be used instead of narrow transmission line
of impedance 2Zo to minimize conductor loss of the power combiner.
For even–odd-mode signal excitation with v1 v2 −v3 −v4 vs , the line
of symmetry “A” can be considered as short but the line of symmetry “B” can be
considered as open. In this condition, the equivalent circuit can be represented as
shown in Fig. 14.56 b. From this
2 Zo
o/4, Z2
Riso2/2 1
o/4, Z1 B
(a)
Riso2/2 2
Riso1/2 3
o/4, Z2
Zo 1 Zo
A
4 Zo
o/4, Z2
Riso1/2
Riso2/2 3
o/4, Z1 B
Riso2/2 4
5
o/4, Z2
Zo
2 Zo
o/4, Z2
Riso2/2
o/4, Z1 open B
Riso2/2
(b) Riso1/2 o/4, Z2
3
Zo
1 Zo
short A
2 Zo
o/4, Z2
(c) Riso2/2
short B
Fig. 14.56 a 4-way power divider, b equivalent circuit under even mode signal (v1 v2 v3
v4 vs ), d c equivalent circuit for odd-mode signal (v1 −v2 vs )
Riso1 1 Z2
× 2
2 2 Zo
This leads to
Riso1 Z22 /Zo (14.128b)
For other combinations of even–odd-mode signal excitation with v1 −v2

v3 −v4 vs , the line of symmetry “B” can be considered as short. At this
condition, the equivalent circuit can be represented as shown in Fig. 14.56c. From
this
(a) o /4, 50/ 2Ω o /4, 50/ 2Ω
RF RF
100 Ω 100 Ω
IN OUT
o /4, 50/ 2Ω o /4, 50/ 2Ω

Power
transistors
(b) < o/4 < o/4
50Ω 50Ω
o /4 o /4
RF < o/4 < o/4 RF

50Ω 50Ω
IN OUT
< o/4 < o/4
o /4 o /4
50Ω 50 Ω
< o/4 < o/4

Power
transistors
Fig. 14.57 Schematic diagram of combining, a two power transistors, b four power transistors
using 4-way splitters with low impedance (wider transmission line) output combiner used
Riso2
Zo
2
This leads to
Riso2 2Zo (14.128c)
Thus, the design parameters for this special 4-way power splitter are Z1
Z2 , Riso1 Z22 /Zo , Riso2 2Zo .
For a special case of Z1 Z2 Zo 50 , Riso1 Zo 50 and Riso1
2Zo 100 .
Schematic diagram of a power amplifier combining two transistors using two-
way Wilkinson power splitters is shown in Fig. 14.57a. Though the Wilkinson power
splitters are excellent in terms of amplitude and phase balance of the two paths, this
combiner suffers from the requirement of a very good port VSWRs of the transistors
and higher power handling requirement of the output isolation resistor.
Any amplitude and/or phase imbalance between the two transistors lead to power
dissipation through the isolation resistor. Thus, high power RF resistance is required
to realize the power module using Wilkinson power splitter. In practical applications,
particularly for chip level combining matched transistors are used without using
isolation resistors as shown in Fig. 14.57b. Here, for example, the special 4-way
power splitter is shown without using isolation resistors. Here impedances of all the
branch line impedances are kept the same as reference impedance 50 with shorter
length of the two branches as shown to minimize the splitter losses and to realize the
amplifier module very compact. Loss of the input power divider and output power
combiner reduces the gain of the combined power module, whereas only the loss of
the output power combiner is responsible to reduce the output power of the power
module. Therefore, loss of the output power combiner is the most critical factor to
realize highly efficient power module. Another in-phase combiner is the rat-race
power combiner. Its frequency bandwidth is very narrow thus it is not widely used
as Wilkinson power splitter.
14.8.3 Balanced Power Combining
In case of in-phase power combining, VSWR of all the transistors should be very
good. The VSWR of a combined power amplifier module is never better than the
VSWR of the transistors. In general, the output port of power amplifiers are matched
for maximum output power, which is different from conjugate matching thus, output
VSWR is poor. To make a power amplifier module having better VSWR of both the
ports using devices with poor port VSWRs, balanced power combining technique
is used. To realize balanced power amplifier, quadrature power splitters are used.
Schematic diagram of a balanced power amplifier with signal levels using 3-dB 90°
power splitters is shown in Fig. 14.58. Two amplifiers (AMPa and AMPb ) having
transmission coefficients S21a k1 θ1 and S21b k2 θ2 and input reflection coef-
ficients S11a ρ1 φ1 and S11b ρ2 φ2 , respectively, are used. When port-1 of the
power amplifier module is excited by signal vs through the reference resistance Ro
and all the other terminals are terminated with the same reference resistance Ro , the
signal reflected to the input port is given by
vs
vr1 (ρ1 (0◦ + φ1 ) + ρ2 (−180◦ + φ2 )) (14.129a)
4
vs
(ρ1 φ1 − ρ2 φ2 ) (14.129b)
4
Therefore, the input reflection coefficient of the combined amplifier is given by
2vr1 (ρ1 φ1 − ρ2 φ2 )
Γr1 (14.129c)
vs 2
Thus, in terms of S-parameters, the input reflection coefficient can be written as

Fig. 14.58 Balanced power amplifier configuration with signal levels
1
S11 (AMP) Γr1 (S11a − S11b ) (14.129d)
2
For the batch-matched amplifiers AMPa and AMPb , only with ρ1 ρ2 , and φ1
φ2 , i.e. with S11a S11b , the input refection coefficient of the balanced amplifier,
Γr1 0, i.e. S11 (AMP) 0. Thus, the balanced amplifiers provide ideally no
reflection of RF power to its input port even if individual amplifiers have poor return
losses. In a similar way, it can be shown that the output reflection coefficient of
balanced amplifiers is ideally zero for matched pair of amplifiers.
The signal level at the isolated port of the input power divider (port-4) is given by
vs
viso1 (ρ1 (−90◦ + φ1 ) + ρ2 (−90◦ + φ2 )) (14.130a)
4
−jvs
(S11a + S11b ) (14.130b)
4
For the batch-matched amplifiers with ρ1 ρ2 ρ, and φ1 φ2 φ, i.e. with
S11a S11b the signal level at the isolated port is
−jvs −jvs −jvs
viso1 (ρ φ) S11a S11b (14.131)
2 2 2
The reflected power will be dissipated through the isolation resistor without going
to the input port.
The signal level at port-3 of the combined amplifier is given by
vs
vt4 (k1 θ1 − 90◦ + k2 θ2 − 90◦ ) (14.132a)
4
Fig. 14.59 Schematic diagram of balanced power amplifier using Wilkinson power splitter
Therefore, transmission coefficient from port-1 to port-3, i.e. gain of the balanced
amplifier is given by
vt4 1
g vs (k1 (θ1 − 90◦ ) + k2 (θ2 − 90◦ )) (14.132b)
2
2
−j
(S21a + S21b ) (14.132c)
2
For the matched pair amplifiers with k1 k2 k and θ1 θ2 θ , the gain can
be written as
g k S21a S21b (14.132d)
Thus, gain of combined amplifiers with same gain coefficients is the same as the
gain of the individual amplifiers.
The signal level at the port-2 of the combined amplifier is
vs
viso2 (k1 θ1 + k2 (θ2 − 180◦ )) (14.133a)
4
vs
(S21a − S21b ) (14.133b)
4
For the matched batch amplifiers with k1 k2 k and θ1 θ2 θ , the signal
level at the port-2 is
viso2 0 (14.133c)
Thus, for batch matched amplifiers there will be no power at the output isolated
port.
There are various types of quadrature power splitters. The Wilkinson power splitter
with λo /4 more length of one arm than the length of other arm works as quadrature-
phase power splitter. Schematic drawing of a balanced amplifier using Wilkinson
power splitter is shown in Fig. 14.59.
Among the different types of quadrature couplers, branch line coupler as shown
in Fig. 14.60a is mostly used for high power applications. Suppose the lengths of
all the branch lines are quarter wavelength, i.e. λo /4 corresponding to the operating
(a) o/4, Z1
1 2
A
o/4, Zo o/4, Zo
4 o/4, Z1 3
(b) o/4, Z1 (c) o/4, Z1

1 2 Zo 1 2 Zo
Zo o/8
Zo o/8 o/8 Zo Zo o/8 Zo
open short
Fig. 14.60 a Branch line power divider, b equivalent circuit under even-mode signal (v1 v2
vs ), c equivalent circuit for odd-mode signal (v1 −v2 vs )
frequency. The characteristic impedance of the shunt lines are considered as Zo . Using
even- and odd-mode analysis, branch impedances of the coupler can be determined
as follows:
Under even-mode signal excitation at port-2 and port-3, along the line of sym-
metry, “A” can be considered as open and the equivalent circuit can be considered
as shown in Fig. 14.60b. Similarly, under odd-mode signal excitation, the equivalent
circuit can be considered as shown in Fig. 14.60c. With reference to Fig. 14.60b,
using (7.80) the input impedance of the open circuited λo /8 length transmission line
is given by
Zin(λo /8) −jZo
Thus, for input impedance matching, we can write
Z12
Zo − jZo
−jZo Zo
This leads to
Zo2 2Z12
or,
√
Z1 Zo / 2 (14.134a)
Similarly, with reference to Fig. 14.60c, using (7.79) the input impedance of the
short-circuit λo /8 length transmission line is given by
Fig. 14.61 Photograph of a branch line 3 dB 90° power divider/combiner
Zin(λo /8) +jZo
Thus, for input impedance matching we can write
Z12
Zo +jZo
+jZo Zo
This also leads to
Zo2 2Z12 (14.134b)
or
√
Z1 Zo / 2 (14.134b)
Therefore, characteristic
√ impedances of the series quarter wave length transmis-
sion liners are Zo / 2.
Photograph of a branch line coupler using microstrip transmission line on a 20 mil-
thick alumina substrate is shown in Fig. 14.61. Here, meander transmission lines are
used to realize the coupler over a very small area at lower microwave frequency band.
Schematic diagram of a balanced amplifier using branch line couplers is shown in
Fig. 14.62. Branch line couplers are widely used for realization of high power ampli-
fiers due to the feasibility of using wider planner transmission line and possibility
for mounting high-power loads to serve as isolation resistors.
There are also various types of quadrature power splitters such as broadside and
edge-coupled transmission line couplers. Among them, Lange couplers are widely
used due to its very wide frequency band performance. The balanced amplifiers are
not only used for achieving high RF output power, these configurations are also used
for low-power amplifiers to realize very broadband amplifiers.
Fig. 14.62 Schematic diagram of balanced power amplifier using branch line couplers
Example 14.30 Amplifier-A is having 6 dB gain and 3.8 dB NF. Amplifier-B is

having 20 dB gain and 4 dB NF. What should be the sequence of the cascaded
two-stage amplifier for lowest NF. What will be the value of the lowest NF for this
arrangement?
Solution
Gain of the amplifier-A and amplifier-B are
GA 6 dB 4 (in factor)
GB 20 dB 100 (in factor)
Noise figure of amplifier-A and amplifier-B are
NFA 3.8 dB 2.4 (in factor)

NFB 4 dB 2.5 (in factor)
Case-1: Amplifier-A as first amplifier: Overall noise figure NFAB
NFAB NFA + (NFB − 1)/GA

2.4 + (2.5 − 1)/4 2.775
4.43 dB
Case-2: Amplifier-B as first amplifier: Overall noise Figure NFBA
NFBA NFB + (NFA − 1)/GB

2.5 + (2.4 − 1)/100 2.524
4.024 dB
Thus, cascaded lowest noise figure is 4.024 dB where amplifier-B is used as first
amplifier in cascaded two-stage amplifier.
Example 14.31 For a Class-A FET power amplifier with 10 V drain supply and 2 A
drain current bias providing RF load current of 1 A amplitude. What is the DC to RF
efficiency for a load resistance of 5 ?
Solution
DC power consumption of the Class-A power amplifier:
PDC 10 V × 2 A 20 W
RF power delivered to the load:

2
PRF 1/2 × 1 × 5 2.5 W
Thus, efficiency:
2.5
η 12.5%
20
Example 14.32 For a Class-B FET power amplifier with 15 V drain supply providing
12 V peak signal to 15 load. What is DC to RF efficiency of the amplifier?
Solution
In Class-B operation, the efficiency:
π Vdrf
η ×
4 VDD
π 12
× 62.8 %
4 15
Example 14.33 A high-gain MESFET packaged device of case-to-channel thermal
resistance 4.5 °C/Watt provides 8 W RF output taking 20 W DC power. What will
be the channel temperature of the device if the case temperature of the device is
maintained at 55 °C?
Solution
Case-to-channel thermal resistance:
Rth 4.5 ◦ C/W
Dissipated power within the device:
Pdissi (20 − 8) W 12 W
Increase of temperature:
T 12 W × 4.5 ◦ C/W 54 ◦ C
Thus, channel temperature:
Tch (55 + 54) ◦ C 109 ◦ C
14.9 Temperature Compensation of Microwave Amplifiers
Linear gain, saturated output power and efficiency of GaAs- and GaN-based FETs
depend strongly on its temperature [35–39]. At fixed drain and current and voltage
bias, transconductance as well as unity current gain cut-off frequency decreases
with the increase of temperature. One of the most important parameters determining
the gain of a microwave amplifier is unity current gain cut-off frequency, which
decreases steeply, irrespective of bias condition, with the increase of temperature.
Therefore, at a fixed voltage and current bias, the performance parameters such as
gain, output power and efficiency of the amplifiers decrease with the increase of
temperature. Practically, the linear gain of these amplifiers decreases approximately
by the amount 0.015 dB/°C with the increase of device temperature.
There are many techniques to achieve temperature compensated RF performance
of the FET-based amplifiers. Active-controlled voltage/current bias is used to vary the
drain current and/or voltage to compensate gain and output power of the amplifiers.
However, care must be taken to avoid any instability due to the change of bias with
the temperature. Another technique is to use temperature-dependent resistance as
feedback element to compensate linear gain variation over temperature. This type
of compensation is more effective and suitable for MMIC based amplifier modules
[36, 37]. In most of the practical applications, where amplifiers are operated in linear
region, a separate voltage variable amplifier/attenuator with temperature-dependent
bias is used to compensate linear gain of the overall system.
References
1. Rollett JM (1962) Stability and power gain invariants of linear two-ports. IRE Trans Circuit
Theory 29–32
2. Owens PJ, Woods D (1970) Reappraisal of the unconditional stability criteria for active 2-port
networks in terms of S parameters. Electron Lett 6:315
3. Woods D (1976) Reappraisal of the unconditional stability criteria for active 2-port networks
in terms of S-parameters. IEEE Trans Circuits Syst 23(2):73–81
4. Jackson RW (2006) Rollett proviso in the stability of linear microwave circuits—a tutorial.
5. Rothe H, Dahlke W (1956) Theory of Noisy Fourports. Proc Inst Radio Eng 44:811–815
6. Haus H, Atkinson W, Branch G, Davenport W, Fonger W, Harris W, Harrison S, McLeod W,
Stodola E, Talpey T (1960) Representation of noise in linear twoports. Proc IRE 48:69–74
References 477
7. Fukui H (1966) Available power gain, noise figure, and noise measure of two-ports and their
graphical representations. IEEE Trans Circuit Theory CT-13:137–142
8. Tucker R (1975) Low-noise design of microwave transistor amplifiers. IEEE Trans Microw
Theory Tech 23(8):697–700
9. Hillbrand H, Russer P (1976) An efficient method for computer aided noise analysis of linear
amplifier networks. IEEE Trans Circuits Syst 23(4):235–238
10. Garg VK, Singh RV, Jain VK, Bera SC (1999) New amplifier design eliminates chip capacitors.
IETE Tech Rev 16(6):197–201
11. Deal W, Mei X, Radisic V, Liu P, Uyeda J, Barsky M, Gaier T, Fung A, Lai R (2007) Demonstra-
tion of a S-MMIC LNA with 16-dB gain at 340 GHz. In: 2007 IEEE compound semiconductor
integrated circuits symposium, Oct 2007, pp 1–4
12. Samoska L, Gaier T, Fung A, Mei XB, Yoshida W, Lee J, Uyeda J, Liu PH, Deal WR, Radisic
V, Lai R (2008) Submillimeter-wave InP MMIC amplifiers from 300–345 GHz. IEEE Microw
Wirel Compon Lett 18(1):61–63
13. Bera SC, Shah LB, Raval DU, Pandey S, Kumar V, Singh S, Das DK (2012) Design and
development of V&W band amplifier modules. In: Proceedings of international conference on
microwaves, antenna, propagation & remote sensing, Jodhpur, India, Dec 2012
14. Raab FH (1997) Class-F power amplifiers with maximally flat waveforms. Proc IEEE Conf
Microw Theory Technol 45:2007–2012
15. Pukala D, Raab LF (2001) Maximum efficiency and output of Class-F power amplifiers. IEEE
Trans Microw Theory Tech 49(6):1162–1166
16. Kuroda K, Ishikawa R, Honjo K (2008) High efficiency GaN-HEMT Class-F amplifier operat-
ing at 5.7 GHz. In: Proceedings of 38th European microwave conference, Amsterdam, Nether-
land, Oct 2008, pp 440–443
17. Ando A, Takayama Y, Yoshida T, Ishikawa R, Honjo K (2008) A high-efficiency class-F GaN
HEMT power amplifier with a diode pre-distortion linearizer. In: Proceedings of Asia-Pacific
microwave conference, Macau, China, Dec 2008, pp 1–4
18. Khan FN, Mohammadi FA, Yagoub MCE (2008) A GaNHEMT Class-F amplifier for
UMTS/WCDMA applications. In: Proceedings of IEEE international RF and microwave con-
ference, Kuala Lumpur, Malaysia, Dec 2008, pp 478–482
19. Yoo C-U, Kim M-S, Yang Y (2009) High-efficiency Class-F amplifier design using defected
ground structure. In: Proceedings of international conference on new trends in information and
service science, Beijing, China, July 2009, pp 700–703
20. Wu D, Korolkiewicz E, Lu Q, Liu L (2010) To design and model a Class-F amplifier and
investigate the effect of losses on the efficiency of DC to AC power conversion. In: Proceed-
ings of 7th international symposium on communication systems networks and digital signal
processing, New Castle upon Tyne, UK, July 2010, pp 119–122
21. Kuroda K, Ishikawa R, Honjo K (2010) Parasitic compensation design technique for a C-Band
GaN HEMT Class-F amplifier. IEEE Trans Microw Theory Tech 58(11):2741–2750
22. Kim J, Jo G, Oh J, Kim Y, Lee K, Jong J (2011) Modeling and design methodology of
high-efficiency Class-F and Class-F−1 power amplifiers. IEEE Trans Microw Theory Tech
59(1):153–165
23. Grebennikov A (2011) Load network design technique for Class F and inverse Class F power
amplifiers. High Freq Electron 10(5):58–76
24. Tuffy N, Guan L, Zhu A, Brazil TJ (2012) A simplified broadband design methodology for
linearized high-efficiency continuous class-F power amplifiers. IEEE Trans Microw Theory
Tech 60(6):1952–1963
25. Carrubba V et al (2012) The continuous inverse class-F mode with resistive second-harmonic
impedance. IEEE Trans Microw Theory Tech 60(6):1928–1936
26. Kenle C, Dimitrios P (2012) Design of broadband highly efficient harmonic-tuned power
amplifier using in-band continuous Class-F1/F mode transferring. IEEE Trans Microw Theory
Tech 60(12):4107–4116
27. Lu Z, Chen W (2013) Resistive second-harmonic impedance continuous class-F power ampli-
fier with over one octave bandwidth for cognitive radios. IEEE J Emerg Sel Top Circuits Syst
3(4):489–497
28. Arnous MT et al (2014) Highly efficient and wideband harmonically tuned GaN-HEMT power
amplifier. In: Proceedings of 20th International Conference on Microwave, Radar, and Wireless
Communication (MIKON), June 2014, pp 1–4
29. Yefymovych A, Krizhanovski V, Giofre R, Colantonio P (2014) Load network design technique
for Class-F power amplifier. In: Proceedings of 20th international conference on microwaves,
radar and wireless communications, Gdansk, Poland, June 2014, pp 1–3
30. Kawano A et al (2005) High-efficiency and wide-band single-ended 200 W GaN HEMT power
amplifier for 2.1 GHz W-CDMA base station application. In: Asia-Pacific Microwave Confer-
ence
31. Kimball DF et al (2006) High-efficiency envelope-tracking W-CDMA base-station amplifier
using GaN HFETs. IEEE Trans Microw Theory Tech 54(11):3848–3856
32. Wu YF, Wood SM, Smith RP, Sheppard S, Allen ST, Parikh P, Milligan J (2006) An internally-
matched GaN HEMT amplifier with 550-watt peak power at 3.5 GHz. In: International electron
devices meeting, Dec 2006, pp 1–3
33. Bera SC, Singh RV, Gagr VK, Arora NS, Nair SS (2005) PBG effect on a modified Wilkinson
power divider. In: Proceedings of Indian conference on microwave and remote sensing-2005,
pp 303–307
34. Bera SC, Singh RV, Garg VK (2008) Modified Wilkinson power divider with harmonic sup-
pression characteristic. Microw J (2008)
35. Anholt RE, Swirhun SE (1992) Experimental investigation of the temperature dependence of
GaAs FET equivalent circuits. IEEE Trans Electron Devices 39(9)
36. Yamauchi K et al (2001) X-band MMIC power amplifier with an on-chip temperature com-
pensation circuit. IEEE Trans Microw Theory Tech 49(12):2501–2506
37. Huang CW, Chang SJ, Wu W, Wu CL, Chang CS (2005) A Ku-Band four-stage temperature
compensated pHEMT MMIC power amplifier. Microw Opt Technol Lett 44(5):480–485
38. Torregrosa-Penalva G, Asensio-Lopez A, Ortega-Gonzalez FJ, Bianco-del-Campo A (2003)
PAE improvement and compensation of small-signal gain drift due to temperature on power
amplifiers through active biasing. Microw Opt Technol Lett 38(5):389–392
39. Bera SC (2006) Remarks on ‘PAE improvement and compensation of small-signal gain drift
due to temperature on power amplifiers through active biasing’. Microw Opt Technol Lett
48(4):827–828
Chapter 15
Microwave Limiters
Abstract This chapter presents design, analysis and characteristics of various

microwave limiter circuits. Both open- and closed-loop-type limiters based on p-
i-n diodes, Schottky barrier diodes and transistors are discussed. Complete design,
analysis and application of a closed-loop-limiter system with temperature compen-
sation mechanism is presented for microwave systems.
15.1 Introduction
To protect sensitive devices from unintentional or intentional exposure to high level

of RF power, limiters are used in receivers, solid-state power amplifiers, detectors,
etc. Basically, there are two types of limiters; open-loop limiter and closed-loop or
feedback-type limiter. In case of open-loop limiter, the same device detects the RF
power level and acts to limit the power level. Here, limiting device works due to its
self-bias effect, whereas a closed-loop limiter circuit uses separate detector circuit to
detect the RF power level and the detected signal is applied to other devices/circuits to
limit the power level in closed-loop configuration. Diodes being two terminal devices
are very stable in operation and are commonly used as limiting devices. Schottky
and p-i-n diodes are the most suitable limiting device for microwave applications
[1–11]. Transistor-based microwave amplifiers are also used as limiter under suit-
able bias condition. The amplifiers are used as limiter by operating them in gain
compression/output power saturation region.
Limiter may be considered as an attenuator which provides attenuation beyond
a predefined RF power level. The predefined power level is called threshold power
level. Beyond the threshold power level, amount of attenuation will be governed by
self-biasing of the limiting device which is determined by the RF power level. Ideally,
the amount of self-biasing will be such that there will be 1-dB attenuation for 1-dB
increase of power level. Therefore, beyond the threshold power level, ideally, there
will be no change of output power with the increase of input power level. However,
below the predetermined threshold power level, output power changes linearly with
the input power level. That is, output power level changes by 1-dB with change of

480 15 Microwave Limiters
1-dB input power level. In case of zero-bias limiter circuit, the threshold power level
is determined by the construction of the device itself. To achieve adjustable threshold
power level, external control bias may be applied to the limiting device.
The absence of minority carrier storage effect in Schottky barrier diodes makes
them efficient to rectify microwave and mmwave signal. Therefore, Schottky diodes
s are superior than p-i-n diodes as microwave limiters and suitable for realization
in MMIC technology. Though Si- and GaAs-based Schottky diode limiters are suit-
able for low power level applications, SiC-based Schottky diodes can be used as
high power limiter [8–10]. To increase the overall limiting power range, multistage
limiters with different threshold power levels are also used. Small-signal microwave
amplifiers using MESFETs with appropriate values of gate and drain bias resistors
have capability for overdrive protection. Thus, small-signal amplifiers are also used
as limiter in microwave systems to protect other devices and circuits in the system.
15.2 Limiter Characteristics
Sensitive microwave/RF devices such as BJTs, FETs and HEMTs are used in
receivers, channel amplifiers and solid-state power amplifiers as well as in many
measurement instruments. These devices are very sensitive and susceptible to dam-
age from input signals having amplitudes of more than the safe limit of these devices.
The safe limits are of the order of few mW for low-power devices which are used in
receivers, channel amplifiers and small-signal stages of solid-state power amplifiers,
whereas the safe limit of power level may be of the order of Watts for high-power
devices which are used in the medium- and high-power stages of power amplifiers.
The unwanted higher power level may come from signals which are within the oper-
ating frequency band or may be out of the normal operating frequency band of the
device/circuit. Practically, a limiter should work over a very wide band of frequency
extending beyond the system’s operating frequency bands in both the sides to protect
sensitive devices used in communication subsystems and other instruments.
A limiter should provide very low insertion loss over the operating frequency
band when the signal levels are within the system’s operating dynamic range. This
requirement is more critical when it is used at the front end of a receiver to avoid its
impact on system noise figure. It should provide very high attenuation to the incom-
ing signals over any frequency band which exceeds the predefined threshold level of
the respective devices and units. The limiter should be very fast to provide protec-
tion within about nanoseconds of arrival of the high-level signal. Another important
requirement of a limiter circuit is that it must survive exposure to high-power signal
level providing sufficient attenuation and it must return to its low insertion loss state
within nanoseconds after the disappearance of high-level signal power.
Important performance parameters of a limiter are operating frequency range,
insertion loss, flat and spike leakage and limiting power range. Operating frequency
range defines the range over which the limiter must protect the succeeding devices
or systems. In general, the operating frequency range of a limiter is broader than the
15.2 Limiter Characteristics 481
5
Threshold Plateau
Linear point
Output Power (dB)
region
0
Practical
limiters
-5 Insertion
Ideal
loss Flat limiter
leakage
-10
-10 0 10 20 30
Input Power (dB)
Fig. 15.1 Typical characteristic of limiter
operating frequency range of the circuits or subsystems required to be protected from

high level of RF power levels. Insertion loss of a limiter is defined as the insertion
loss over its linear region, i.e. small-signal operating region over its entire dynamic
range as shown in Fig. 15.1.
When high input power is applied to a limiter, most of the power is absorbed
or reflected by the limiter and only remaining small portion of the power can pass
through it. The small portion of the applied power which passes through the limiter
is called flat leakage or only leakage power. Generally, the leakage power is specified
as an absolute level, i.e. in dBm or Watts. The input limiting power range is defined
from the threshold power level to the maximum power level up to which the limiter
provides attenuation to keep its output power level within specified limit. Ideally,
over this region the output power of the limiter remains constant. However, in most
of the limiter, over the input dynamic range, the output increases gradually with its
input as shown in Fig. 15.1 as plateau. The region between the initiation of limiting
and the region of maximum isolation is termed the plateau region. For input signals
above the maximum input power level, the output will begin to increase again at a
greater slope. The output power variation over the input limiting power range may
also initially decrease and then increase with greater slope as shown in the figure.
This type of limiter characteristic is observed for fast response devices due to their
efficient rectification of RF signal thereby providing more attenuation over the initial
limiting power range compared to the higher level of signal power. Another limitation
is the presence of spike leakage as shown in Fig. 15.2 [4–6]. For a sudden increase
of input power level, limiter takes some time to settle the output at its leakage power
level. This time is called recovery time. Spike leakage is a transient phenomenon and
sometimes is a risk factor to sensitive electronic devices. This becomes critical for
some sensitive devices such as microwave FETs and HEMTs when comparatively
slow limiting diodes such as wide intrinsic region p-i-n diodes are used as limiters. In
these cases, fast response devices such as Schottky barrier diodes as limiting devices
are preferable in place of p-i-n diodes.
Fig. 15.2 Limiter input

characteristic
Spike leakage
Flat leakage
Power
output
Time
Ideally, both ports of a limiter should have good impedance matching over its entire
input power range. To achieve this, diodes are needed to be mounted in a balanced
configuration as it is used in case of diode-based attenuator circuits. Another aspect
of a limiter is to precisely protect sensitive devices such as pHEMT, HFET, etc. To
protect such sensitive devices from overdrive power situation, provision of threshold
power level adjustment is required. This adjustability is required to protect the system
without affecting the system’s normal performance over its entire operating dynamic
range.
15.3 P-I-N Diode Limiters
Nonlinear devices are required to achieve limiter function for protecting RF and
microwave devices and circuits from high-level signals while maintaining very low
loss, below the limiting threshold power level. In the RF and lower microwave fre-
quency range, p-i-n diodes are popularly used as limiting power device. Generally
p-i-n diodes have slow response time. In the microwave frequency range, special
types of p-i-n diodes with relatively short carrier lifetime, called limiter diodes, are
constructed with lower device width to achieve comparatively fast response time.
Intrinsic layer thickness of the limiter diodes is between 2 to 7 μm. These diodes
with relatively short carrier lifetime, efficiently rectify RF power when it exceeds a
specified level and provide required attenuation by self-biasing.
Generally, p-i-n diodes are used to realize limiter due to their capability of handling
high power levels and providing reasonable leakage level. Shunt-connected p-i-n
diode limiters are very popular because of their very high impedance at OFF condition
(low signal power level) and very low impedance when forward biased under high-
level signal power. In the presence of small-level signals, the shunt diode does not
conduct, and therefore represents high impedance and thus provides very low loss
to the RF signal. When RF signal of high power level is applied, p-i-n diode goes
into the forward-bias condition and provides attenuation. A schematic circuit of a
15.3 P-I-N Diode Limiters 483
DC Rs
Sensitive
Input return Devices
path /circuits Ri Ci
Fig. 15.3 Limiter circuit diagram of single shunt-connected p-i-n diode and its equivalent circuit
shunt-connected p-i-n diode limiter is shown in Fig. 15.3. The RF choke shown in
the figure provides a DC return path for the self-bias current.
The diode is shunt mounted across the transmission line, and is provided with a
DC bias return by an RF choke, as shown in Fig. 15.3. For RF input signal level which
is below the predetermined threshold level, the diode acts as an ordinary unbiased
p-i-n diode. Thus, the diode acts as parallel combination of high-value resistor with
low-value capacitor and provides very low loss to the RF signal. However, when the
incident signal level exceeds the threshold power level, the diode’s I-layer is flooded
with carriers during the positive half cycle of the input RF signal. Due to the large
carrier lifetime, most of these carriers remain in the I-region through the negative
half cycle. Thus, rectified DC current begins to flow through the diode and the diode
biases itself to a low value of RF resistance. The diode’s RF resistance falls to a
very low value under the influence of this self-generated bias current, called the self-
biasing effect. Reflecting the large signal back to its source, the limiter circuit then
acts as a reflective switch, and thus limits the output RF power. Additionally, some
part of the signal is also absorbed by the diode resistance. When the large amplitude
input signal has disappeared, recombination of the carriers in the diode’s I-region
leads to decrease or diminishing of the circulating bias current. Due to the absence
of self-bias current, the RF resistance of the p-i-n diode becomes very high, allowing
small signals to pass without attenuation.
The S-parameter matrix of shunt-connected diode limiter is the same as derived
in case of shunt-connected attenuator and is given by (9.5)
⎡ ⎤
−Z o 2Z T
S11 S12
⎣ ⎦
2Z T +Z o 2Z T +Z o
(15.1)
S21 S22 2Z T −Z o
2Z T +Z o 2Z T +Z o
where Z T is the equivalent impedance of the p-i-n diode and Z o is the system reference
impedance. Therefore, fraction of power that is reflected from the diode to source is
given by

−Z o 2

pr |S11 |
2 (15.2)
2Z T + Z o
And the fractional power transmitted to the output is given by

2Z T 2
|S |
pt 21 2 (15.3)
2Z T + Z o
Fractional power absorbed, i.e. dissipated within the diode impedance is given by

pd 1 − |S11 |2 + |S21 |2 1 − ( pr + pt ) (15.4)
Therefore, in the limiter circuit, some of the incident power is reflected back to
the source, some is passed to the load, i.e. at the output load and some is absorbed
and dissipated in the diode resistor as heat. The part of the incident power dissipated
in the shunt device is a function of its impedance Z T and is governed by (15.4).
From the equivalent circuit of the p-i-n diode as shown in Fig. 15.3, its impedance
is given by
Z T (Ri ||Ci ) + Rs (15.5)
Neglecting the parasitic components, i.e. series resistor Rs and shunt capacitor C j ,
the p-i-n diode impedance can be considered as pure resistance. Different fractional
power components are shown in Fig. 15.4. From the figure, it is clear that at a diode
resistance value of 25 in a 50 system, the fractional dissipated power reaches
a maximum of 50% of the incident power. This 25 diode resistance corresponds
to the limiter attenuation (|S21 |) of 6 dB and return loss (|S11 |) of 6 dB. Thus, a
slow limiter which turns ON with some delay, may pass through this maximum RF
power dissipation of 3-dB point after sufficient exposure of high RF power level.
Accordingly, power rating of the limiting diode needs to be selected.
Typical characteristics of a shunt-connected p-i-n diode limiter circuit are shown
in Fig. 15.5. All the plots are at steady-state condition and power levels are in dB
scale with reference to the threshold input and output power levels. It may be noted
that over the limiting region, the loss of the limiter increases and the return loss
decreases. For the p-i-n diode limiter, the output power increases again with the
increase of input signal levels above the maximum input power limiting level. The
increase of output power beyond the maximum limiting point is due to the inability
to provide further attenuation by the limiter circuit with the increase of RF power
level.
Figure 15.6 shows different power components of the limiter circuit at steady-
state condition over the input RF power range. It shows that the returned power as
well as absorbed power increases with the increase of input power to the limiter. It
is clear that though the fractional dissipated power becomes maximum at the 6 dB
attenuation point; however, in steady-state condition, the power dissipated in the
diode increases with the increase of input power level.
From the above characteristic, it is also clear that the return loss of the single shunt-
connected p-i-n diode limiter decreases with the decrease of diode resistance, that is,
with the increase of input RF power level. This is undesirable for any RF/microwave
Transmitted
Reflected
1.0
Fractional power Absorbed
0.5
0.0
1 10 100 1000
Diode Resistance (Ω)
Fig. 15.4 Fractional power limiter with variation of diode resistance
15
Output power
0 |S11|
(dB)
-15
|S21|
-30
-15 -5 5 15 25 35
Input power (dB)
Fig. 15.5 Characteristic of shunt-connected p-i-n diode limiter with input power variation
circuits which affect the matching of cascaded circuits and systems with it. To achieve
better input and output return losses, more number of p-i-n diodes are used with sepa-
ration of λ/4 transmission line length which cancel the reflected power. Schematic cir-
cuit diagram of a limiter with two shunt-connected p-i-n diodes is shown in Fig. 15.7.
40
Returned power
20 Absorbed power
Output power
(dB)
-20
-40
-15 -5 5 15 25 35
Input power (dB)
Fig. 15.6 Characteristic of shunt-connected p-i-n diode limiter with input power variation
λ /4
RF Sensitive
Input Devices
choke
/circuits
Fig. 15.7 Limiter circuit diagram with two shunt-connected p-i-n diodes
Example 15.1 Calculate maximum possible dissipated power within a p-i-n diode
in a shunt-connected p-i-n diode limiter circuit operating at frequency of 4 GHz.
Consider, the parasitic series resistance and capacitance of the diode are 5 and
0.01 pF, respectively.
Solution
Figure 15.4 shows that maximum fraction of input power is dissipated through the
diode when the diode resistance becomes 25 . At the operating frequency of 4 GHz,
the capacitive reactance is
1
− j39789
( j2π × 4 × 0.00001)
Thus, reactive part of the diode can be ignored compared to its resistive part.
During its limiting action, when the diode passes to its net RF resistance of 25 i.e.
including its parasitic resistance (5 ), it will experience maximum power dissipa-
tion.
Fraction of input power that is reflected from the diode to the source is given by
(15.2)

−Z o 2
pr |S11 |
2
2Z T + Z o

−50 2
0.25
50 + 50
Therefore, reflected power 10 W × 0.25 2.5 W

And the fractional power transmitted to the output is given by (15.3)

2Z T 2
pt |S21 |
2
2Z T + Z o

50 2
0.25
50 + 50
Therefore, transmitted power 10 W × 0.25 2.5 W

Thus, the maximum power dissipated within the p-i-n diode
input power − (reflected power + transmitted power)

10 W − (2.5 W + 2.5 W)
5W
15.4 Schottky Diode Limiters
At lower frequencies, p-i-n diodes are better choice for limiter circuits. Because of
the absence of minority carrier storage effect, Schottky barrier diodes can rectify
microwave and mmwave signals more effectively. Therefore, for better performance
and suitability of realization in MMIC technology, Schottky diodes are preferable at
microwave and mmwave frequency range as limiter [12].
Though GaAs-based Schottky diode limiters are suitable for low-power operation,
SiC-based Schottky barrier diodes are capable to operate over higher power level.
Practically, multistage limiters of different threshold power levels using Schottky
and p-i-n diodes are used for high-power-limiting applications.
Under small-signal operating condition, RF resistance of Schottky barrier diodes
can be written as
∂ Id ηkT 1
Rd (Id ) 1/ (15.6)
∂ Vd q (Id − Io )
where I o is reverse saturation current, I d is forward current, η is ideality factor, k is

Boltzmann constant and q is electron charge.
The expression (15.6) of Schottky diodes RF resistance is valid only for small-
signal levels. Operation of limiter in its limiting power region cannot be considered
No Vb
1.5 Pin (dBm) 6 3 0 RF
L3 RL
L1 L
Cb
1.0
Id (mA)
VS
0.5 L2
0.0
-0.50 -0.25 0.00 0.25 0.50 0.75 1.00
Vd (V)
Fig. 15.8 I–V characteristic of Schottky diode in presence of RF signal
as small-signal operation. I–V characteristic of Schottky diodes deviate from the

exponential relationship at large signal operating condition, as shown in Fig. 15.8.
Typical I–V characteristic of a Schottky diode in presence of different RF power
levels at microwave frequency is shown in Fig. 15.8. It is clear that with the increase
of RF power level, the I–V curves deviate more and more from their exponential
characteristics. In presence of RF signal, the deviation from the exponential charac-
teristic is due to the self-biasing effect caused by rectification of RF signal. The load
line L 1 is for low supply voltage and low-bias resistance, whereas the load line L 2
is for high supply voltage and high-bias resistance. From the plot, it is clear that RF
resistance of the diode decreases with the increase of RF power level for the load
line L 1 , whereas for the load line L 2 , RF resistance of the diode increases with the
increase of RF power level. For the shunt-connected diode-based limiter circuits as
shown in Figs. 15.3 and 15.7, RF resistance of the diode should decrease with the
increase of RF power level. Thus, for these shunt diode limiter circuits, the diode
should be biased through a very low series resistance with zero or very low voltage
bias, i.e. bias condition similar to load line L 1 .
To achieve good impedance matching of both the ports of a limiter over its entire
input power range, diodes with balanced configuration are used as shown in Fig. 15.9.
Here, for the limiter circuit, a 3-dB 90° coupler is used. The output port of the limiter
circuit is the isolated port of the coupler. Two Schottky barrier diodes are connected
to coupled and direct port of the hybrid. Other remaining port is used as input of the
limiter. The control voltage Vc is applied through a very low value of bias resistor Rb
(about 0–10 ) connected to the bias point of the RF circuit using high-impedance
short-circuited λ/4 transmission lines.
The Schottky diode’s RF equivalent impedance, as shown in Fig. 15.9, is a combi-
nation of RF resistance (Rd ), parasitic capacitance (Cd ) and parasitic series resistance
15.4 Schottky Diode Limiters 489
Vc
Rb
λ /4 high RF λ/4 high

impedance line short impedance line
RF RF
3-dB 90o Coupler
IN OUT
Rs
D1 D2 Rd Cd
Fig. 15.9 Limiter circuit diagram based on balanced diode configuration
Rs . RF resistance (Rd ) of the diode is determined by bias current and the RF power
level at the diodes. Depending on its value, some portion of the RF power will be
reflected back from the diode and rest of the power will be dissipated within the
diode.
At the output port of the circuit, the reflected power from both the diodes will
be combined. Considering same characteristic for both the diodes (batch-matched
diodes), there will be no reflected power at the input port of the circuit. Thus, both
the ports of the limiter circuit will remain in matched condition at any value of RF
impedance of the diodes. Thus, there will be no reflected power to the input port.
A limiter circuit is realized on a 25-mil-thick alumina (εr 9.9) substrate. Pho-
tograph of the realized circuit is shown in Fig. 15.10. For the 3-dB power divider,
Lange coupler is used at the centre frequency of 11.5 GHz. Beam-lead Schottky bar-
rier diodes are used for its low parasitic capacitances. Accordingly the circuit layout
is made. Beam-lead Schottky diode of medium-barrier height is used for the circuit
for its lower values of the parasitic resistor (<5 ) and capacitor (<0.05 pF).
Adjustable threshold power level is required for protecting very sensitive devices
like HBTs and pHEMTs without affecting their performance over nominal operating
range. To achieve this, adjustable bias voltage to the diodes is required. Behaviour of
the limiter circuit for different supply voltages is shown in Fig. 15.11. The attenuation
of the limiter increases with the increase of input power levels when the supply
voltage is lower than certain value (here, it is 0.4 V). For higher supply voltages,
the attenuation of the limiter decreases with the increase of RF power levels. The
attenuation of the limiter should increase with the increase of RF power level for
Radial stub for RF ground
RF IN
RF Out
diodes
Fig. 15.10 Photograph of the limiter based on beam-lead diodes
+5.0
Normalized S21 (dB)
+2.5 0.42V
0.41V
0.0 0.40V
0.39V
-2.5
0.38V
-5.0
-20 -15 -10 -5 -2 0
Pin (dBm)
Fig. 15.11 Supply voltage dependency of limiter characteristics
limiting function. Thus, the adjustable supply voltage of the limiter diodes should be
of lower value.
Figure 15.12 shows I–O characteristic (output power variation with input power)
of the limiter for different values of bias resistor (Rb ) at zero control voltage (Vc ).
Over the small-signal condition, the insertion loss of the limiter is nearly 0.9 dB
for all the bias resistors. Over the limiting region of the limiter, the slope of the
output power with the change of input power level depends on the value of bias
resistor Rb . For higher values of bias resistors (e.g. 15 and 20 ), I–O characteristics
15.4 Schottky Diode Limiters 491
+5.0
+2.5 20Ω
15Ω
Pout (dBm)
10 Ω
0.0
5Ω
-2.5
0Ω
-5.0
-5 0 5 10 15 20
Pin (dBm)
Fig. 15.12 Bias resistor dependency of Schottky diode limiter characteristic
have positive slope, whereas for zero and low values of bias resistor, initially the
output power decreases then increases with the increase of input power level. This
behaviour is due to the more self-biasing effect of the Schottky diodes at lower values
of bias resistor compared to higher resistor values. From the plots, it is clear that the
rate of change of attenuation is a function of input RF power level and bias resistor
(Rb ). From the bias load lines L 1 and L 3 of Fig. 15.8 this can be explained. With
increase of RF power level, RF resistance of the diodes drop sharply in case of L 3 ,
corresponding to lower value of bias resistor, than in case of L 1 , corresponding to
higher value of bias resistor. From the plot, it is also clear that there is an optimum
bias resistor to keep output power nearly constant over the limiting power level. In
this case, it is about 10 .
For different values of control voltages (Vc ), the input–output characteristic of
the limiter at bias resistor of 10 is shown in Fig. 15.13. The plot shows that
by changing the control voltage, Vc , from −0.2 to +0.2 V, the threshold power
level changes maintaining its linearity below the threshold power level. However,
for higher positive control voltage, as shown in the figure for Vc +0.3 V, the I–O
characteristic shows nonlinearity and also increase of insertion loss. The nonlinearity
below the threshold power level, i.e. above the normal operating power region is
undesirable for any communication system.
With the control voltage from Vc −0.2 to +0.2 V, the adjusted threshold power
levels are from −10 to +5 dBm. The limiting power range is about 15 dB for each
control voltage. It is also shown that beyond the range of limiting power level, output
power level increases with the increase of input power. This is due to the decrease
of diode impedance below 50 .
10
5
-0.2V
0.0V
Pout (dBm)
0
-0.1V
+0.1V
-5
+0.2V
-10 +0.3V
-15
-20
-15 -10 -5 0 5 10 15 20
Pin (dBm)
Fig. 15.13 Threshold power level adjustment by bias voltage
(a) (b)
POUT
Limiting characteristic
PDC
PIN POUT
PDISSI
PIN
Fig. 15.14 Output power saturation of microwave amplifier
15.5 Amplifier-Based Limiter
Microwave amplifiers maintain linear characteristic over their small-signal operat-

ing region. However, with the increase in input power level beyond linear operat-
ing region, gain of the amplifiers starts to decrease and ultimately output power
saturates. Suppose PIN , POUT , PDC and PDISSI are the input RF power, output RF
power, DC power and power dissipated within the amplifier, respectively, as shown
in Fig. 15.14a. The relationship among these powers can be written as
PDC + PIN POUT + PDISSI (15.7)

15.5 Amplifier-Based Limiter 493
VG VDD
RG RD
Output RF
matching OUT
Input ID network
RF
matching
IN IG
network
Fig. 15.15 FET amplifier with high value gate and drain resistors
or
PDISSI PDC − PIN (G − 1) (15.8)
Here, G is the gain of amplifier. Suppose, DC power (PDC ) supplied to an amplifier

is fixed and the input RF power level to the amplifier goes on increasing. The equa-
tion (15.8) suggest that beyond certain input power (PIN ) level, the power dissipated
(PDISSI ) within the amplifier becomes negative if we assume gain of the amplifier (G)
remains constant. However, dissipated power cannot be negative (implies generation
of power), thus the gain of the amplifier must decrease with the increase of input
RF power level beyond certain limit. Thus, due to the limited DC power supplied to
an amplifier, which is governed by capability of the amplifying device, gain of an
amplifier decreases and acts as limiter as shown in Fig. 15.14b.
Under high RF power overdrive condition, excessive gate current may flow due to
the rectification effect at the gate-to-channel Schottky barrier junction of GaAs FETs
which reduces device life. GaAs FET-based amplifiers are designed with high value
of gate resistor (RG ) to limit the gate current (IG ) under high power level overdrive
condition and used as limiter circuit as shown in Fig. 15.15. In some cases, moderate
value of drain resistor (R D ) is also used to control the drain current (I D ) as well as
voltage across the drain-to-source terminal under overdrive condition. GaAs FETs
are used for low power amplifiers operating at microwave frequency bands to develop
receivers, solid-state power amplifiers (SSPAs) and other subsystems. In addition to
function as amplifier in its nominal operating power level, the low power amplifying
stages can also function as limiter and protect the following high-power-sensitive
devices during overdrive.
15.6 Closed-Loop Limiter
Very precise restriction of RF power level is required for highly sensitive microwave
circuits and systems such as pHEMT-based receiver and power amplifiers to protect
them from RF overdrive without affecting their RF performance over nominal oper-
ating power level. In this case, open-loop limiters based on p-i-n diode or Schottky
diode cannot serve the purpose for their soft-limiting characteristics. A hard limiter,
i.e. a limiter with very low input power difference from its highest linear operating RF
power level to threshold power level is required to protect such sensitive subsystems.
To set the threshold power level very precisely, the hard limiter should also have a
provision of adjustable threshold power level. Schematic circuit diagram of such a
closed-loop limiter configuration is shown in Fig. 15.16. An RF power sampler is
used to sample the RF power level for detection. The detected voltage is amplified
and used as control signal to control the attenuation of a voltage-controlled attenuator
for controlling the signal level (Fig. 15.17).
Sensitive
Subsystem RF power
Sampler
RF Voltage RF
IN Controlled OUT
Attenuator
DC Amplifier
DETECTOR VC
Fig. 15.16 Schematic diagram of closed-loop overdrive level control
Output
Power
Cold
Ambient
Hot
Input power
Fig. 15.17 Temperature behaviour of OLC system

15.6 Closed-Loop Limiter 495
The closed-loop overdrive level control scheme is capable of protecting all the
sensitive subsystems which are connected after the power level control point [13, 14].
In Fig. 15.6, all the devices connected after the voltage-controlled attenuator circuit
is protected under overdrive condition. Thus, any level of high power output can be
controlled using this scheme just by changing the coupling factor of the coupler.
Mainly, Schottky barrier diode is used as RF power detector at microwave frequency
range. To adjust the limiting power level precisely, i.e. to adjust the threshold power
level of the limiter, a control voltage (Vc ) is applied to the detector diode. A DC
amplifier is used to amplify the detected voltage and fed to a voltage-controlled
attenuator. The voltage-controlled attenuator may be realized using p-i-n diodes
or FETs. The limiting range of the limiter is determined by the attenuation range
of the voltage-controlled attenuator. Single- or multistage analog voltage-variable
attenuators are used depending on the required range of overdrive level protection.
15.6.1 Temperature Behaviour of OLC System
Attenuation of voltage-variable attenuators based on p-i-n diodes or FETs changes

with change of the device’s temperature. Gain of the sensitive subsystems used within
a closed-loop system is also temperature dependent. Therefore, input power level of
the detector changes with the change of temperature. Moreover, detected DC voltage
not only depends on the RF power level but it is also very sensitive to the temperature
of the detector diode itself. The gain requirement of the DC amplifier is very high, in
case of closed-loop hard limiter. Thus, a small change of the detector diode voltage
significantly changes the control voltage to the voltage-controlled attenuator. Typical
I–O characteristic of an OLC system with the change of operating temperature is
shown in Fig. 15.7.
The change of input power level due to the attenuation variation of the voltage-
controlled attenuator, gain variation of the sensitive subsystem and DC amplifier are
inherently compensated by the inherent property of the closed-loop negative feed-
back system. The negative feedback system automatically changes the output of the
detector to keep its input power level constant by controlling the attenuation of the
voltage-controlled attenuator. Thus, it is important to achieve constant/unchanged
I–O characteristic of the detector circuit in terms of its detected voltage versus its
input power level over the operating temperature range. The unchanged I–O charac-
teristic of the detector over the operating temperature range, automatically adjusts the
attenuation of the voltage-controlled attenuator by detecting change of the detector’s
input power level by any means and generating the required control voltage.
15.6.2 Temperature Compensation of OLC System
To compensate the variation of limiting power level over the change of operating
temperature, a temperature-controlled bias voltage (Vc ) is applied to the detector
diode as shown in Fig. 15.18. The detector diode is biased using a temperature-
dependent control voltage Vc , and a bias resistor Rc , through a bias network that
consists of an RF decoupling network which in turn consists of a high-impedance
λo /4 transmission line and a capacitor Cb .
The detected voltage is filtered by a low-pass filter (LPF) and is fed to one input
of the differential DC amplifier. The control voltage Vc is connected to the other
input of the differential amplifier. The output voltage of the differential amplifier
controls the voltage-controlled attenuator (VCA). The threshold power level of the
differential amplifier is determined by the diode characteristic and controlled volt-
age Vc . When the RF power level exceeds the threshold power level, the differential
amplifier provides controlled bias voltage to the VCA. This controlled bias voltage
controls the attenuation of the attenuator to limit the RF power to a predetermined
limiting power level. To compensate the threshold power variation with the operat-
ing temperature, a temperature-dependent voltage drop across a p–n junction diode
D1 and associated resistive networks and supply voltages are used to generate the
temperature-dependent control voltage Vc . Here, the diode D1 is used as temperature
sensor. The op-amp-based DC amplifier is used to achieve the required sensitivity
of the controlled voltage over the operating temperature range. Both positive and
Low power High power

Amplifier Amplifier
Coupler
RF RF
VCA
IN OUT
Temp. controlled
Diode detector voltage generator
Rc +V
Dd
Cb
R1
DC amp
Vc
Differential DC Amplifier LPF D1
Ra Rf
Cf R4
Rb
R3 R2
-V
Fig. 15.18 Temperature-compensated OLC system for high power amplifiers

15.6 Closed-Loop Limiter 497
negative supply voltage sources are used to bias the p–n junction diode to get the
control voltage with both polarities in case it is required to compensate the limiting
power level over temperature.
15.7 Applications
Limiters are used to protect sensitive devices used in receivers, solid-state power
amplifiers and other microwave subsystems from intentional or unintentional expo-
sure to high power levels. Such a scheme is shown in Fig. 15.19. Here, a limiter is
used before a predistortion linearizer cascaded with a travelling wave tube ampli-
fier (TWTA). In this case, the limiter circuit not only protects the TWTA from RF
overdrive condition but it also improves the linearity of the system beyond and near
saturation of the TWTA by preventing the output power drop beyond its saturation.
Limiters protect the subsystems from severe EMI conditions. Limiters are also
used in narrow sub-band channels to avoid communication disruption of other sub-
TWTA
RF RF
IN Limiter Linearizer OUT
Fig. 15.19 Schematic diagram of an application of limiter to protect linearized TWTA
High power
amplifier Circulator
RF Voltage 1 2 RF
DR
IN Controlled OUT
Filter
Attenuator
3
Reflected
power
Detector
RF power
DC Sampler
Amplifier
VC
Fig. 15.20 Application of closed-loop OLC to protect DR filter from high power dissipation under
OOB carrier
(a)
(b)
|S21| (dB) |S21| (dB)
input
input
output output
Frequency Frequency
Fig. 15.21 Input and output of the filter a without OLC and b with OLC
bands within the same channel. In some applications, both Schottky and p-i-n diode
based limiters are used to provide better protection than a p-i-n diode-only limiter
due to its lower limiting threshold [15]. However, it has more insertion loss compared
to either only p-i-n diode or only Schottky diode-based limiters.
15.7.1 OLC for Out-of-Band Carrier Mitigation in Filters
Power handling capability of dielectric-resonator-based filter is limited by power dis-

sipation when it is exposed to out-of-band (OOB) carriers. It becomes more severe
in case of narrowband high-power filters when realized based on dielectric resonator
(DR). Dielectric-resonator-based high-power filters are preferred option for satellite
transponders due to its compact size and less mass along with stable RF performance
over temperature. However, average power handling is limited due to the excessive
power dissipation at OOB frequency, particularly in the frequency range where the
group delay is maximum. A closed-loop limiter system is shown in Fig. 15.20 to mit-
igate the OOB carrier heating problem. The DR filter is used in a satellite transponder
as output filter and connected at the output of high power amplifier such as TWTA.
Poor return loss of the DR filter over its OOB frequency region is used to detect
and limit the RF power level. A circulator is used between the TWTA and the filter
to get the reflected power from the DR filter at port-3. The out-of-band frequency
return loss of the filter is poorer than its in-band frequency return loss. Thus, suitably
setting the threshold limiting power level by proving control bias to the detector, the
presence of OOB carrier beyond certain level can be detected. The higher in-band
return loss of the filter compared to its out-of band return loss ensures very less
in-band power return compared to OOB carrier power. Thus, in absence of power
corresponds to OOB frequency, closed-loop system will not provide any attenuation
to the signal. A typical response of filter with and without closed-loop limiter circuit
is shown in Fig. 15.21. It shows that power level at the input of the filter for out-of-
References 499
band frequency range becomes lower than the in-band frequency power level due to
the OLC function.
References
1. Leenov D (1965) The silicon PIN diode as a microwave radar protector at megawatt levels.
IEEE Trans Electron Dev 12:53–61
2. Smith DG, Heston DD, Heston J, Heimer B, Decker K (1979) Designing high-power limiter
circuit with GaAs PIN diodes. In: IEEE MTT-S international microwave symposium digest,
pp 329–331, Apr 1979
3. Alpha Application Note 80300 (1985) Characteristics of semiconductor limiter diodes. Speci-
fications, application notes, packages, Alpha Industries Inc., Semiconductor Division, pp 3–45
4. Tan RJ, Ward AL, Garver RV, Brisker H (1988) PIN diode limiter spike leakage, recovery time
and damage. In: IEEE international microwave symposium digest, vol 1, pp 275–278, May
1988
5. Tan RJ, Ward AL, Kaul R (1989) Transient response of PIN limiter diodes. In: IEEE interna-
tional microwave symposium digest, vol 3, pp 1303–1306, June 1989
6. Ward AL, Tan RJ, Kaul R (1994) Spike leakage of thin Si PIN limiters. IEEE Trans Microw
Theory Tech 42(10):1879–1885
7. Smith DG, Heston DD, Allen DL (1999) Designing high-power limiter circuits with GaAs PIN
diodes. In: IEEE international microwave symposium digest, vol 1, pp 329–332, June 1999
8. Erikson J, Rorsman N, Zirath H (2001) Microwave silicon carbide Schottky diodes. Electron
Lett 37:250–252
9. Erikson J, Rorsman N, Zirath H (2003) 4H-silicon carbide Schottky barrier diodes for
microwave applications. IEEE Trans Microw Theory Tech 51:796–804
10. Sudow M, Andersson K, Billstrom N, Grahn J, Hjelmgren H, Nilsson J, Nilsson PA, Stahl
J, Zirath H, Rorsman N (2006) An SiC MESFET-based MMIC process. IEEE Trans Microw
Theory Tech 54:4072–4078
11. Coaker BM, Dowthwaite (2007) Planar limiters and receiver protectors. In: Proceedings of the
4th European radar conference, pp 405–408, Oct 2007
12. Bera SC, Basak K, Jain VK, Singh RV, Garg VK (2010) Schottky diode-based microwave
limiter with adjustable threshold power level. Microw Optical Technol Lett 52(7):1671–1673,
July 2010
13. Miyazaki IK, Ohata T, Ono T, Kamikokura A, Hayashi R, Seino K, Hirose H, Kurebayashi H
(1998) L-band 50 W SSPA incorporating overdrive limitation circuit and power saving circuit
for MTSAT. In: 17th international communications satellite systems conference and exhibit,
Yokohama, Japan, 23–27 Feb 1998
15. Lim CL (2012) Reduce losses in RF Schottky-PIN limiters circuits. Electron Des
Chapter 16
Microwave Linearizers
Abstract This chapter presents design and analysis of different types of lineariz-
ers for microwave power amplifiers. Working principles of feedback-, feedforward-
and predistortion-type linearizers are discussed. In addition to series and shunt diode-
based predistortion linearizers, complete design details of a diode-based predistortion
linearizer in vector modulator configuration with temperature-compensated perfor-
mance suitable for space use is presented. Different circuit topologies for digital
linearizer and analog broadband linearizers using equalizers are also presented.
16.1 Introduction
Linearity of amplifiers plays a major role in design of communication systems.

Throughout transmit and receive chain linear amplification of modulated signals
is required to accurately decode the signals to extract information. Any amplitude
and/or phase distortions of the modulated signal reduce the ability to decode these
signals properly. Distortion cannot be dissociated from a signal because it modi-
fies the signal. Distortion also cannot be detected when the signal source is shut
down. Amplifier’s nonlinearity becomes more severe when more bandwidth effi-
cient higher order modulations such as QPSK, OQPSK, and DQPSK schemes are
used. Even higher order modulation codes such as 8APSK, 16APSK, 16QAM, etc.
exhibit greatly improved bandwidth efficiency. However, these are more susceptible
to amplifier’s nonlinearity due to their large non-constant envelope. Signal levels in
receiving sections are very low, thus amplifiers in the receiving section operate in
linear region. However, problem arises when linear amplification is required in the
transmitting section where power amplifiers are used. To operate the power ampli-
fiers in linear region it is required to operate them in power back-off condition which
reduces the amplifier’s DC to RF efficiency as discussed in Chap. 14. Most of the
DC power is consumed in the transmitting power amplifiers. Thus, it is very impor-
tant to operate the transmitting power amplifiers near to their saturation point to get
highest possible efficiency. In other way, it is required to use more expensive higher
power amplifier and operate it at back-off condition to achieve required transmitted

502 16 Microwave Linearizers
power with required linearity. In addition to the requirement of higher DC power

and higher overall system cost, another drawback of the system is the thermal man-
agement and reliability of the final power device. With the reduction of efficiency,
power dissipation within the amplifying device will be more which demands more
efficient cooling system to operate the device within its allowable temperature limit
otherwise the device will operate at higher temperature which will lead to reduction
of its reliability.
The efficiency of a power amplifier can be improved by operating a Class-A
amplifier towards hard saturation, i.e. clipping the signal waveform. This is done
by biasing the device at lower quiescent current level such as in case of Class-AB,
B and C configuration. When the devices are biased at lower quiescent current, the
amplifiers are operated in reduced conduction angle. Class-A amplifiers operating
towards hard saturation and the amplifiers with reduced conduction angle that are
operating in their nonlinear region generates distortion on modulated signals. There-
fore, some linearization technique is required which improves the overall linearity
of the system even though the amplifiers in the transmitter section operate in their
nonlinear region for achieving higher DC to RF efficiency [1–15].
16.2 Types of Distortion and Amplifier Nonlinearity
Signal distortion is linear or nonlinear transformation of the signal. Linear distortion

is the change in signal form due to the variation of gain and time delay over the
frequency range of interest. Linear distortion causes different gains and time delays
for different frequencies present in the signal which causes change in signal form.
Linear distortion in a communication system mainly arises in dynamic circuits such as
filters and can result in severe signal spectrum shaping in case of analog transmission
chain and inter-symbol interference in digital transmission chain. To minimize the
linear distortions, amplitude tilt equalizer and group delay equalizers are used in the
transmission chain.
Nonlinear distortion produces modifications of gain and signal phase with the
change of signal strength which results in the generation of new frequency compo-
nents that were not present in the original signal. The generated new components
behave as random noise to the desired signal. In addition to the original components,
nonlinear distortion produces many other spectral components which is known as
spectral regrowth. Thus, linear distortion only modifies the gain and phase of each
spectral component, whereas nonlinear distortions modify the spectrum qualitatively
by generating new spectral components which act as random noise to the original
signal. The spectral regrowth happens not only within the usable bandwidth of the
signal but it also falls outside the allocated bandwidth which may violate the regu-
lations about spectrum transmission outside the allocated bandwidth which defines
the maximum allowable adjacent channel interference levels. Another effect of the
nonlinearity is the shifting of bias point, i.e. quiescent point of the operating device
from its original bias point at no RF condition. This effect modifies the DC power
16.2 Types of Distortion and Amplifier Nonlinearity 503
AM/AM AM/PM
Gain, Pout (dB) Phase (deg.)
Pout SSPA
Gain
TWTA
Pin (dB) Pin (dB)
Fig. 16.1 Amplitude and phase characteristic of power amplifier
consumption depending on the signal power strength. Any linear amplifier is not
perfectly linear; there are always nonlinear effects which can be modelled as RF
signal level dependent equivalent circuit elements. Also, it can be remembered that
any amplifier can be considered as linear amplifier up to certain signal level, but the
same amplifier will have nonlinear behavior when the signal level crosses certain
limit.
Nonlinearities of power amplifiers are characterized by output power or gain
against the variation of input power level, called AM/AM conversion, and output
phase variation against the input power variation called AM/PM conversion. First
is the input power dependent output power or gain and second is the input power
dependent output signal phase. Typical amplitude and phase characteristics of solid-
state power amplifier (SSPA) and travelling wave tube amplifier (TWTA) are shown
in Fig. 16.1. The characteristic of TWTAs is more nonlinear compared to the charac-
teristic of SSPAs. Practically, gain compression (decrease of gain) of TWTA is about
6 dB at its maximum output power, whereas it is about 2 dB (recommended maximum
operating gain compression) in case of solid-state power amplifiers. Phase decrease
of TWTAs is approximately 45°, whereas it is approximately 20° for SSPAs. Another
important point may be noted that, in general, with the increase in input power level,
the phase of the output signal decreases for TWTAs, whereas it increases in case of
SSPA as shown in Fig. 16.1.
A very common solution to minimize distortion due to power amplifier nonlinear-
ity is backing off their operating region away from the saturation point. The output
power is much smaller than its saturated output power when the power amplifier
operates in the linear region. This solution not only increases the cost of the system
and reduces the dynamic range of an amplifier but also drastically reduces its DC
to RF efficiency which is undesirable. Power efficiency is a critical consideration in
case of portable- and space-based systems where batteries and solar cells are used.
More heat dissipation may also reduce the reliability of the device used in the system.
More elegant solution is the use of linearization techniques to improve overall

linearity of the system and operating the power amplifier nearer to its saturation
to achieve better DC to RF efficiency. Use of linearizer reduces in-band distortion
and as well as adjacent channel interference by reducing the overall nonlinearity of
the system. Thus, it is desirable to design a moderately linear PA and employ some
linearization technique for further improvement of linearity. The linearization system
maximizes the spectral efficiency operating the power amplifier in its near-saturated
region. Thus, linearization allows a high power amplifier (HPA), for a given level of
distortion, to operate at a higher power level with greater efficiency. At microwave
frequencies, the use of a linearizer has an even greater advantage since available HPA
power level is more limited and cost is greater than at lower frequencies.
16.3 Types of Linearizers
Different types of linearization techniques are used for microwave power amplifiers
in communication systems. These techniques are broadly classified into three groups:
(1) feedback, (2) feedforward and (3) predistortion. The feedback and predistortion
linearization techniques can be implemented at intermediate frequency (IF), RF or at
baseband level, whereas the feedforward linearization technique is realized directly at
RF level. Depending on frequency and bandwidth, different linearizers can be realized
in digital or in analog domain. Some linearization techniques are combination of two
or more different techniques. For example, adaptive baseband predistortion is the
combination of both the feedback and predistortion techniques.
Among all the linearization techniques, predistortion technique is the most com-
monly used technique where a nonlinear circuit module is inserted between input
signal and power amplifier to improve overall linearity of the system. Following
sections will discuss the different linearization techniques.
16.3.1 Feedback Linearizers
To carry more traffic over a given spectrum, it is required to linearize high power
amplifiers used in a transmitter section. This leads to increased use of spectrally
efficient modulation schemes. Feedback linearization technique is a widespread and
successful application to meet this requirement. A generic feedback scheme for
linearization of power amplifiers is shown in Fig. 16.2. Here, input signal vi is
compared with feedback signal to generate error signal. The error signal passes
through the loop compensation network of gain G(s) and the power amplifier of
gain G A . Here, the distortion generated within the amplifier is modelled as signal
vd . The output of the amplifier is sampled and feedback for compensation is given
through compensation network. The feedback factor including the coupling network
16.3 Types of Linearizers 505
vd
vi vo
G(s) GA
Power
Amplifier
H(s)
Fig. 16.2 A generic feedback scheme for linearization of power amplifier
is considered as H (s). Due to the feedback action, the error will be minimized by
forcing the output vo to track the input signal.
Here, the output signal can be written as
GA × G 1
vo vi + vd (16.1a)
1 + GA × G × H 1 + GA × G × H
Here, G A G H is the loop gain. In case of large loop gain, where G A G H 1, the
output signal can be written as
1 1
vo × vi + × vd (16.1b)
H G AG H
In case of large loop gain, the second term of the output signal corresponds to the
distortion and hence is minimized. The amount of reduction of the distortion signal is
by the factor of 1/(1 + G A G H ). Increasing the loop gain, the effect of the distortion
can be made very small. However, making the loop gain too large can result in the
instability of the feedback loop. Another advantage of the feedback linearization
technique is that it tracks the input signal (vi ) more accurately by the feedback signal
regardless of the forward gain. It is clear from (16.1b) that the overall gain for the
input signal is primarily determined by the feedback factor 1/H . Therefore, realizing
H using only stable passive components such as resistors will stabilize the overall
gain of the feedback system significantly.
16.3.2 Feedforward Linearizers
Feedforward linearization technique is realized directly at RF level. In the scheme, at

first the distortion generated in the main power amplifier is extracted by comparing
the output of the power amplifier with its delayed input. Then, this extracted distortion
is amplified using an auxiliary power amplifier and destructively combined with the
Splitter Coupler-1 Combiner

RF
RF Ga a
OUT
IN
Delay-1
Main Power
Amplifier
Auxiliary
Amplifier
b Gb
Delay-2 Coupler-2
Fig. 16.3 A generic feedforward scheme for linearization of power amplifier
delayed output of the main power amplifier using a power combiner. Thus, ideally,
the output of the system will be distortion free.
Schematic diagram of a typical feedforward scheme for linearization of microwave
power amplifiers is shown in Fig. 16.3. The working principle of the scheme is
demonstrated in the diagram for a two-tone input signal. Two time delay circuits,
one each in the main amplifier path and another in the auxiliary amplifier path are
used to match delay for achieving proper extraction and subtraction of the distortion.
Any deviation from the required amplitude and phase matching will degrade the
extraction of distortion component and subsequently distortion cancellation after
combination at the output of the system. Amplitude and phase matching becomes
critical due to the drift of amplifier characteristics with temperature.
Main drawback of the feedforward linearization technique is the requirement of
auxiliary amplifier with output power comparable to the main amplifier to overcome
the loss through the output coupler/combiner. The efficiency of the feedforward sys-
tem is reduced by the power consumption of the auxiliary amplifier which must be
operated in its linear region. Another reason for reduction of efficiency of the system
is the loss of the output coupler (coupler-1), combiner and delay circuit (delay-1)
which are connected at the output of the main power amplifier. However, feedfor-
ward linearization technique provides broad bandwidth performance with reasonable
improvement in linearity. Also, this technique has the advantage of inherent stability.
16.3.3 Predistortion Linearizers
Among the various types of linearization techniques, predistortion (PD) linearization

is the most popular for satellite communication. The popularity of predistortion lin-
earizers is due to its low power consumption, simplicity and ability to linearize over
16.3 Types of Linearizers 507
wide bandwidth of power amplifiers operating at near saturation condition. Predis-

tortion linearizers provide relatively wide-band performance and are able to function
as stand-alone unit, which makes it preferable for microwave systems. Predistor-
tion linearizer creates inverse nonlinearity of the transmitting amplifiers such as of
TWTAs or SSPAs in order to compensate for the distortion. Working principle of the
predistortion linearization scheme is shown in Fig. 16.4. A predistortion linearizer
creates amplitude and phase distortions of the signal opposite to the distortion of the
HPA which is to be linearized. At first, the signal to be amplified by an amplifier is
passed through the linearizer to introduce opposite distortion to the signal. When the
signal distorted by the linearizer passes through the amplifier, it negates the distortion
and produces a near-linear transfer characteristic as shown in Fig. 16.4. Amplitude
and phase nonlinearities are generally considered as unwanted. In case of predis-
torter, it is required to generate the amplitude and phase nonlinearity intentionally
with the same amount in opposite direction with that of the power amplifier. Thus,
in this technique, a nonlinear transfer function is generated which is complimentary
to the amplitude and phase nonlinearities of the power amplifier to be linearized.
This is used to predistort the incoming signal such that when the distorted signal
passes through the nonlinear amplifier, it is further restored in the opposite way to
get its original waveform back. This is done by increasing the gain of the linearizer
by the same amount as the amplifier gain decreases with the increase of input power
level. Likewise, the phase shift introduced by the linearizer is equal and opposite to
that of the amplifier and results in the phase improvement as shown in Fig. 16.4. In
general, there is no feedback in the predistortion linearization technique, thus it is
an open-loop system. Therefore, the distortion characteristics (amplitude and phase
variation with the power level) of the power amplifier to be cancelled must be known
in advance.
16.4 Implementation of Predistortion Linearizers
Predistortion linearizers can be implemented at RF as well as intermediate frequency

(IF) level. The advantage of IF predistortion is that the same design can be used for
different carrier frequencies (RF) by simply varying the local oscillator (LO) fre-
quencies of up- and downconverters. RF predistortion is done immediately before
the RF power amplifier. However, in case of IF predistortion, there will be upconverter
and amplifying stages in between the RF power amplifier and the predistortion lin-
earizer. Thus, characteristics of the IF predistortion linearizer needs to be designed by
considering the nonlinearity as well as gain variation of the in-between subsystems.
A predistorter should have its amplitude and phase characteristics opposite to
that of the power amplifier. Therefore, the device for designing the predistortion
linearizer must be biased in such a way that it should operate in its nonlinear region
at the operating power level. Practically, Schottky barrier diodes, varactor diodes
and transistors (FETs and BJTs) are used to generate nonlinear characteristics for
designing predistortion linearizers [1–15]. These devices are biased in such a way
RF Predistortion Power RF
IN Linearizer Amplifier OUT
Pin Pin Pin
gain phase
Fig. 16.4 Predistortion scheme for linearization of power amplifier
that some of its parameters vary with the change of its input RF power level. In case
of Schottky barrier diodes, RF resistance variation with its input RF power level
is used to generate proper amplitude and phase nonlinearities, whereas capacitance
variation of a varactor diode with its input RF power level is used for designing a
varactor diode-based linearizer. Gain compression and phase shift of a low power
transistor-based amplifier can also be used to linearize performance of a high power
RF amplifier. However, amplifier in a transistor-based linearizer needs to operate at
very high level of compression to achieve required amplitude and phase nonlinearities
that may be a concern for high-reliability operation over an extended lifetime. Among
all types of predistortion linearizers, diode-based linearizers are most suitable for its
compactness, low power consumption and high-reliability operation. For microwave
as well as mmwave frequency range linearizers, Schottky diode is most suitable for
its high-frequency operation to generate amplitude and phase distortions due to its
self-biasing effect.
16.4.1 Schottky Diode Predistortion Linearizers
Diode-based predistortion linearizers are most suitable for many applications includ-
ing for space use due to its compactness, less power consumption and higher reli-
ability [8–15]. Schottky diodes are preferable for designing microwave/mmwave
linearizers for its superior detection capability over these frequency ranges. Depen-
dency of RF resistance of the Schottky diodes on RF power level is used to generate
amplitude and phase nonlinearities required for predistortion linearizers.
16.4 Implementation of Predistortion Linearizers 509
Vb
No RL
1.5 Pin (dBm) 6 3 0 RF
Cb L
Load line for L1
1.0 series diode VS
Id (mA)
linearizer
Load line for

0.5 L2 parallel diode
linearizer
0.0
-0.50 -0.25 0.00 0.25 0.50 0.75 1.00
Vd (V)
Fig. 16.5 I–V characteristic of Schottky diodes in presence of RF power
A typical I–V characteristic of Schottky diodes in presence of different RF power

levels including schematic circuit diagram is shown in Fig. 16.5. The exponential I–V
characteristic deviates more and more from its exponential nature with the increase
of RF signal level [11–15]. RF resistance of Schottky diodes is given by inverse slope
of the I–V curve at particular operating condition and is given by
∂ Id
Rd (Id ) 1/ (16.2)
∂ Vd
Under large signal RF operation, DC voltage and current of the diode are not
only the function of RF signal level but are also dependent on bias load line as
shown in Fig. 16.5. The two types of bias load lines, L 1 and L 2 , as shown in the
diagram correspond to nearly constant-voltage-bias and constant-current-bias con-
ditions, respectively. It is clear from the figure that for the bias load line L 1 , RF
resistance of the diode decreases with the increase of RF signal level, whereas in
case of the bias load line L 2 , RF resistance of the diode increases with the increase of
RF signal level. Thus, RF resistance of the diode may increase or decrease with the
increase of RF power level depending on the bias load line. Based on this property of
the RF resistance of Schottky diodes, two types (1) series diode and (2) shunt diode
predistortion linearizer circuits are designed.
Among all types of predistortion linearizers, series and shunt diode linearizers are
simple and least expensive to implement. However, both types of linearizers offer
modest improvement in linearity suitable for solid-state power amplifiers. Thus, these
types of linearizers are good candidates for mobile applications such as for small
handheld radios or cell phones due to their simplicity in circuitry.
Fig. 16.6 Schematic of a VS

series diode linearizer circuit Ri
Rb Rs
Ci
L
Cb1 Cb2
16.4.1.1 Series Diode Predistortion Linearizer
Schematic circuit diagram of a series diode linearizer is shown in Fig. 16.6. Here,
diode is biased through a very small bias resistance Rb to operate the diode at nearly
constant-voltage-bias condition to achieve decrease of RF resistance of the diode with
the increase of RF power level. Two RF chokes are used to provide DC current path
without affecting the RF performance. Two capacitors Cb1 and Cb2 are used for DC
blocking. The Schottky diode can be represented by its equivalent circuit as shown in
the diagram. Here, diode equivalent capacitance Ci and the series resistance Rs can
be considered as constant over the operating RF power level, whereas RF resistance
Ri of the diode varies with the RF power level. The equivalent impedance Z d of the
Z d Rs + Ri ||Ci (16.3)
Neglecting the effect of bias networks and blocking capacitors on the RF per-
formance, the S-parameters of the series diode linearizer can be written as from
(7.162c)
⎡ ⎤
Zd 2Z o
S11 S12
⎣ ⎦
2Z o +Z d 2Z o +Z d
(16.4)
S21 S22 2Z o Zd
2Z o +Z d 2Z o +Z d
Therefore, the amplitude and phase of the series diode linearizer can be written
as

(2Z o )2 + (4π f Z o Ci Ri )2
|S21 | (16.5a)
(Ri + Rs + 2Z o )2 + [2π (Rs + 2Z o ) f Ci Ri ]2
4π f Z o Ci Ri 2π (Rs + 2Z o ) f Ci Ri
S21 tan−1 − tan−1 (16.5b)
2Z o Ri + Rs + 2Z o
|S21|, |S11| (dB) S21 (deg)

Ci (pF) Ci (pF)
|S11| 0 50
0.2 0 0.3
0.2
0.1
0.3 -10
-10 25
0.2
|S21|
0.1
-20
-20 0
1000
1000 100
100 10
10 1000
1000 100
100 10
10
Ri (Ω) Ri (Ω)
Fig. 16.7 Amplitude and phase characteristics of response of series diode linearizer
The amplitude and phase variation of the series diode predistortion linearizer with
the variation of diode RF resistance are plotted in Fig. 16.7 for operating frequency of
2 GHz using (16.5a) and (16.5b), respectively, for different values of capacitance of
the diode. It shows that the amplitude of the linearizer increases with the increase of
diode resistance as required to linearize amplitude characteristic of a power amplifier.
The phase of the linearizer decreases with the decrease of diode resistance which is
required to linearize phase characteristic of a power amplifier, particularly for SSPAs.
Here, dependency of shape as well as amount of amplitude and phase shift on the
capacitance value Ci is also shown. It suggests that different combinations of ampli-
tude and phase characteristics can be achieved to linearize different power amplifiers
by selecting diodes of different capacitances. Practically, an external capacitor can be
used in parallel to the diode to achieve adjustable amplitude and phase characteristics
of the linearizers.
The series diode linearizer, biased nearly constant-voltage-bias condition, gen-
erates amplitude expansion and phase decrease characteristic with the increase of
RF power level. Though the series diode linearizer is very simple with less power
consumption, it has drawback of poor port return losses as shown in Fig. 16.7. Prac-
tically, it is desirable to independently adjust the phase and amplitude characteristic
of a linearizer to compensate the amplifier’s amplitude and phase nonlinearity of dif-
ferent combinations. However, the series diode linearizers are not capable to provide
this flexibility.
16.4.1.2 Shunt Diode Predistortion Linearizer
Schematic circuit diagram of a shunt diode linearizer is shown in Fig. 16.8 [9, 10].
Here, diode is biased through a high-value bias resistance Rb to operate the diode
at nearly constant-current-bias condition to achieve increase of RF resistance of the
Schottky diode with the increase of RF power level. RF choke L is used to provide
the bias to the diode without affecting RF performance.
Fig. 16.8 Schematic circuit VS

diagram of a shunt diode
predistortion linearizer Rb
L
Cb1 Cb2
Ri
Rs
Ci
Neglecting the effect of bias networks and blocking capacitors on the RF per-
formance, the S-parameters of the shunt diode linearizer can be written as from
(7.163d)
⎡ ⎤
Z d ||Z o −Z o 2Z d ||Z o
S11 S12 ⎢ Z d ||Z o +Z o Z d ||Z o +Z o ⎥
⎣ ⎦ (16.6a)
S21 S22 2Z d ||Z o Z d ||Z o −Z o
Z d ||Z o +Z o Z d ||Z o +Z o
⎡ ⎤
−Z o 2Z d
⎣ ⎦
2Z d +Z o 2Z d +Z o
(16.6b)
2Z d −Z o
2Z d +Z o 2Z d +Z o
Therefore, the amplitude and phase of the shunt diode linearizer can be written as

(2Ri + 2Rs )2 + (4π f Ci Ri Rs )2
|S21 | (16.7a)
(2Ri + 2Rs + Z o )2 + [2π (2Rs + Z o ) f Ci Ri ]2
4π f Ci Ri Rs 2π (2Rs + Z o ) f Ci Ri
S21 tan−1 − tan−1 (16.7b)
2(Ri + Rs ) 2(Ri + Rs ) + Z o
With the increase of RF resistance of the diode, the amplitude and phase variation
of the shunt diode predistortion linearizer is shown in Fig. 16.9 based on (16.7a) and
(16.7b) considering the frequency of operation 2 GHz. It shows that the amplitude
as well as phase of the linearizer increases with the increase of RF resistance of the
diode, i.e. with the increase of RF power level. It also shows that amount of phase
expansion increases with the increase of capacitance of the diode.
The shunt diode predistortion linearizers are used in case of amplifiers where
gain as well as phase expansions are required such as travelling wave tube amplifiers
(TWTAs). These are not suitable for SSPAs where decrease of phase with the increase
|S21|(dB) Ci (pF)
0 0.1 S21 (deg) Ci (pF)
1.0 50
1.0
-5
25
0.1
-10 0
10 100 1000 10 100 1000
Ri (Ω) Ri (Ω)
Fig. 16.9 Amplitude and phase characteristics of shunt diode linearizer
of RF power level is required. Moreover, shunt diode linearizers suffer for their
poor input and output return losses and provide no flexibility for achieving various
combinations of amplitude and phase characteristics.
16.4.2 Linearizer with Vector Modulator Configuration
The series and shunt diode linearizers are the simplest and least expensive to imple-
ment. However, both types of linearizers offer only modest improvement in linearity.
These types of linearizers are good candidates for mobile applications such as small
handheld radios or cell phones due to simplicity in circuitry. For applications, where
efficiency and linearity are very critical like TWTAs in space communication, very
accurate amplitude and phase compensations are required. To achieve this, flexible
linearizer is needed to linearize different power amplifiers encountered in practice
with the capability of generating adjustable amplitude and phase characteristics.
Linearizer with vector modulator configuration can provide various combinations of
adjustable amplitude and phase nonlinearities which are required for practical power
amplifier linearization. This configuration is also having very good input and output
return losses over the entire operating RF power level which is required for cascading
the linearizer with power amplifier at its output and other circuit at its input.
Schematic block level diagram of a predistortion linearizer with vector modulator
configuration is shown in Fig. 16.10 [11–16]. The input RF signal to the predistortion
linearizer at first is divided into two parts and passes through two arms consists of
different circuits. One of the arms consists of phase shifter and attenuator/amplifier
operating in their linear region and the other arm consists of an attenuator or amplifier
operating in nonlinear region to generate distortion. The signals coming from both
the arms are combined by an RF combiner to get final output. Conceptual working
principle of the predistortion linearizer in vector modulator configuration is shown
in Fig. 16.11. Here, gl is gain of the linear arm; gnl and g’nl are the gains of the
nonlinear arm at small signal and large signal conditions respectively. Whereas, GL
Phase Linear
Shifter Amp./Attn.
Power Combiner
Power Divider
RF RF
IN OUT
Distortion
Generator
Fig. 16.10 Block schematic of predistortion linearizer with vector modulator configuration
Fig. 16.11 Conceptual working principle of predistortion linearizers based on vector modulator
configuration using phasor diagram
and GN are the resultant gain of the linearizer at small-signal and large-signal condi-
tions respectively. With the increase of RF power level, the gain/loss of the nonlinear
circuit may increase or decrease depending on its characteristic. For example, gener-
ally gain of transistor based amplifiers decreases with the increase of RF power level,
whereas loss of Schottky diode based circuit may increase or decrease with increase
of RF power level depending on its bias condition. In Fig. 16.11a, it is shown that the
expansion of gai (ΔG) and decrease of phase (Δφ) are achieved with the increase
of gain of the nonlinear arm with the increase of RF power. Figure 16.11b shows the
expansion of gain (ΔG) and increase of phase (Δφ) of the output signal due to the
increase of gain of the nonlinear arm with the increase of RF power level, whereas
Fig. 16.11c shows the expansion of gain and decrease of phase of the output signal
due to the gain compression of the nonlinear circuit with the increase of RF power
level.
Rip
Rbp Rsp
VSD, ISD
VO
Cip Pin diode
λ o/4 Ca λo/4
3dB in-phase power combiner

3-dB 90o power divider Cb Cb
D3 D4
linear circuit
RF RF
IN OUT
nonlinear circuit
D1 D2
Cb Cb
λ o/4 Ca λ o/4
Ris
Rss
VO
Rbs
VPD, IPD
Cis Schottky diode
Fig. 16.12 Circuit schematic of diode-based predistortion linearizers in vector modulator config-
uration
16.4.2.1 Diode Linearizer with Vector Modulator Configuration
Schematic RF circuit diagram of a linearizer based on Schottky and p-i-n diodes in

vector modulator configuration is shown in Fig. 16.12. The input RF signal to the
linearizer is divided into two equal and quadrature phase parts using a 3-dB 90° power
divider. The one part passes through the linear circuit and another through nonlinear
circuit and then combined using an in-phase power combiner. The nonlinear arm
consists of a 3-dB 90° hybrid coupler in which two Schottky barrier diodes (D1 and
D2 ) are connected to its direct and coupled ports, whereas the linear arm consists of
another 3-dB 90° hybrid coupler in which two p-i-n diodes (D3 and D4 ) are connected
to its direct and coupled ports. Isolated ports of both the couplers act as output of the
nonlinear and linear arms. Considering both the diode pairs are matched, inherent
property of the 3-dB 90° hybrid structure, always ensures the perfectly matched
input and output ports of both the arms for all values of the diode resistance under
any bias condition and RF power level. A simple equivalent circuit of p-i-n and
Schottky barrier diodes consists of a parallel combination of RF resistance Ri and
capacitance C i with a series-connected parasitic resistance Rs . Therefore, amplitude
ρ and phase φ of the linear and nonlinear attenuator circuits with 3-dB 90° coupler
and two numbers of matched diodes can be expressed as
1/2
ρ (16.8a)

φ − tan (16.8b)
(Ri + Rs + Z o )(Ri + Rs − Z o ) − (2π f Ci Ri )2 Z o2 − Rs2
where ρ ρsd and φ φsd for Ri = Ris and C i = C is in case of the nonlinear
attenuator with Schottky diodes.
And, ρ ρ pd and φ φ pd for Ri Rip and C i C ip in case of linear attenuator
with p-i-n diodes.
The amplitude |S 21 | and insertion phase S21 of the linearizer is given by
2 2 1/2
1 1 1
|S21 | L Z ρsd + ρ pd + ρsd ρ pd cos θ (16.9a)
2 2 2

−1 ρ pd sin θ
S21L Z tan (16.9b)
ρsd + ρ pd cos θ
Here, θ is the phase angle in between the two signals combined at the output port
of the linearizer and is given by
θ φsd − φ pd − 90 (16.10)
Figure 16.13 shows the plot of amplitude and phase of the linearizer with the vari-
ation of Schottky diode’s resistance (Rsd ) for different settings of the p-i-n diode’s
RF resistance (Rpd ) taking the typical values of C pd 0.01 pF and C sd 0.1 pF at
the frequency of f 11.5 GHz. The plot shows that amplitude and phase expansion
can be achieved when Schottky diode’s resistance increases beyond 50 with the
increase of RF power level. Thus, Schottky diodes of the nonlinear arm used to be
biased through a high resistance with small-signal current corresponding to its RF
resistance more than or equal to 50 when p-i-n diodes are biased corresponding
to their RF resistance more than 50 . It is also clear that depending upon the set-
ting of p-i-n diode’s resistance and initial biasing of the Schottky diodes, different
combinations of amplitude and phase can be obtained. It can be shown that with
different combinations of DC biasing of Schottky and p-i-n diodes, various possible
combinations of amplitude and phase variations with respect to the RF power level
can be obtained. In this regard, the linearizer with vector modulator configuration is
very much versatile in nature.
Photograph of a Ku-band linearizer is shown in Fig. 16.14. Here, beam-lead
medium-barrier Schottky diodes with a 3-dB 90° Lange coupler is used for nonlinear
attenuator and beam-lead p-i-n diodes with another 3-dB 90° Lange coupler is used
for linear attenuator of the linearizer. Input power divider is also 3-dB 90° Lange
coupler and output power combiner is a Wilkinson power combiner. The entire circuit
is realized on a single 0.75 in. × 0. 5 in. alumina substrate of thickness 25 mm. For
different DC bias conditions, measured amplitude and phase characteristics of the
linearizer are shown in Fig. 16.15. It shows that by the selection of only the DC bias
currents of the Schottky diodes and p-i-n diodes used in the nonlinear arm and linear
Rpd ( ) = 100 150 200

0 150
|S21| (dB) -5 S21 100
S21 (deg.)
-10 |S21| 50
-15 0
10 100 1000
Rsd ( )
Fig. 16.13 Amplitude and phase characteristics of linearizer of Fig. 16.12
RF IN RF Out
S-Band Linearizer Ku-Band Linearizer
Fig. 16.14 Photograph of diode-based linearizer in vector modulator configuration
arm, respectively, various combinations of amplitude and phase characteristics over

the RF power level can be obtained. Thus, the diode-based linearizer with its vector
modulator configuration is very versatile to generate required I–O characteristic to
linearize different power amplifiers with different nonlinear characteristics.
16.4.3 Temperature Behaviour of Diode Linearizer
Among the various types of linearizers, the vector modulator configuration based
predistortion linearizers using diodes are most suitable for generating various com-
binations of amplitude and phase characteristics with the RF power level. Thus,
this configuration is very useful for linearization of different power amplifiers with
various nonlinear I–O characteristics.
Temperature-dependent forward-biased I–V characteristic of Schottky barrier and
p-i-n diodes can be written as
|S21| (dB) S21(deg) |S21| (dB) S21 (deg)

10 50 12
Isd = 0.6 mA 100
8 40 10 Isd = 0.85 mA
Ipd = 1.3 mA 80
8 Ipd = 1.5 mA
6 30
60
6
4 20 40
4
2 10 2 20
0 0 0 0
-20 -15 -10 -5 0 5 -20 -15 -10 -5 0 5
Pin (dBm) Pin (dBm)
|S21| (dB) S21(deg) |S21| (dB) S21(deg)

5 40 8 100
4 Isd = 0.93 mA Isd = 0423 mA
30 80
Ipd = 1.0 mA 6 Ipd = 0.4 mA
3 60
20 4
2 40
10 2
1 20
0 0 0 0
-20 -15 -10 -5 0 5 -20 -15 -10 -5 0 5
Pin (dBm) Pin (dBm)
Fig. 16.15 Various combinations of measured amplitude and phase of the linearizer of Fig. 16.14
for different bias conditions

q(Vd −E b )

q Vd ν
Id (Vd (T ), T ) Io exp aT exp ηkT
(16.11)
ηkT
Here, Io aT ν exp(−q E b /ηkT ), the reverse-bias saturation current of the diode

which is negligible compared to its forward current Id . k is Boltzmann constant, q is
the electron charge, and a is a temperature independent constant. E b is the bandgap
potential of the semiconductor in volts for p-i-n diodes and barrier potential φb in
case of Schottky barrier diode. η is the ideality factor of the diode.
The temperature- and bias current-dependent RF resistance of p-i-n diodes is
given by
T n−m
Rd (Id (T ), T ) z p (16.12a)
Id
Here, z, m, p, ν, n all are temperature-invariant constants. For p-i-n diodes, the

approximate values of the parameter are m 1, p 1, ν 3, n 2.
The small-signal RF resistance of Schottky diodes is determined by
1 ηkT
(16.12b)
∂ Id /∂ V q Id
Thus, (16.12b) is the same as the expression of p-i-n diode’s RF resistance given by
(16.12a) with z ηk/q, m 1, p 1, ν 2, n 2. From (16.12a), it is clear that
under constant-current-bias condition (constant Id ), with the increase of operating
temperature, the RF resistance of both the diodes increases linearly. Thus, amplitude
and phase characteristics of the linearizer change with the operating temperature of
the diodes [16].
To achieve temperature-insensitive amplitude and phase characteristics, RF resis-
tance of the diodes should be temperature independent, i.e. ∂ Rd /∂ T 0. Differenti-
ating (16.12a) with respect to T and putting the temperature independency condition
∂ Id (n − m)Id
(16.13)
∂T Tp
Suppose, the diodes are biased by a voltage source VO through a resistor Rb . The
current of the diode can be written as
Id (VO − Vd )/Rb (16.14)
Differentiating (16.14) with respect to T and combining with (16.13), we can

write the condition as
∂ Vd /∂ T −(n − m)(VO − Vd )/T p (16.15)
Combining (16.11) with (16.12), then differentiating it, we can write (16.16).
∂ Vd E b − Vd νp + m − n ηk
− − . (16.16)
∂T T p q
Combining (16.15) and (16.16), we can write

p n−m− p νp − n + m ηkT
VO Eb + Vb + (16.17a)
n−m n−m n−m q
Thus, for p-i-n diodes, putting m 1, p 1, ν 3, n 2, we can write

2ηkT
VO E b + (16.17b)
q
And for Schottky diodes, putting m 1, p 1, ν 2, n 2, we can write

ηkT
V O φb + (16.17c)
q
This (given by 16.17b and 16.17c) is the optimum voltage (open-circuit voltage)
required to achieve temperature-insensitive RF resistance of p-i-n diode and small-
signal Rf resistance of Schottky barrier diode. The bias resistor Rb is determined by
Fig. 16.16 Schematic circuit diagram of the diode linearizer with temperature compensation net-
works
(16.14), where diode current (Id ) is at ambient temperature. The p-i-n diodes operate
in linear region over the full operating RF power level. Thus, it is valid for p-i-n
diodes, whereas Schottky barrier diodes operate in nonlinear region, thus rectify the
signal, and thus the diode voltage and current changes with the RF power level.
Practically, the required optimum open-circuit voltage is different from (16.17c),
and increases with the increase of RF power level. Therefore, bias resistor for the
p-i-n diodes is determined by (16.14), whereas temperature-dependent bias resistor is
used for Schottky barrier diodes for achieving temperature-insensitive RF resistance.
Figure 16.16 shows the schematic circuit diagram of the linearizer with bias networks
for temperature-dependent voltage/current to the Schottky diodes. Voltage source Vs
(+5 V) in combination with resistors R1 and R2 are used to get VO Vs R2 /(R1 + R2 )
and Rb R1 ||R2 + R3 for p-i-n diodes. Resistors R p and Rq are used to set opti-
mum voltage VO Vs Rq /(R p + Rq ) for Schottky diodes. Resistor Rr in parallel
combination with a negative temperature coefficient thermistor Rth is used to get
temperature-dependent bias resistor for Schottky diodes.
16.4.4 Broadband Linearizers
Predistortion linearizers are cascaded with a power amplifier such as solid-state power
amplifiers (SSPAs) and travelling wave tube amplifiers (TWTAs) to improve nonlin-
ear performance of a communication system. The predistortion linearizer generates

inverse amplitude and phase nonlinearity with respect to the power amplifier’s ampli-
tude and phase nonlinearity. Diode-based linearizers use Schottky barrier diodes as
nonlinear attenuator and p-i-n diodes as linear attenuator to generate the amplitude
and phase nonlinearity. Radio frequency (RF) impedance of the diodes determined
by RF resistance (Ri ) in combination with a parallel equivalent capacitor (Ci ) deter-
mines the nonlinear amplitude and phase characteristic of the linearizer. Therefore,
gain expansion and phase expansion of the diode-based linearizers vary with the
change of operating frequency. The frequency dependencies of the linearizer’s per-
formances are not only due to the frequency dependency of diodes impedance. It is
also due to the frequency-dependent characteristics of other used circuit elements
such as attenuators and the input and output couplers of the distortion generation
bridge. Moreover, nonlinear responses of the power amplifiers to be linearized also
vary over the operating frequency range that needs compensation. Schematic circuit
of a broadband diode linearizer using amplitude tilt equalizers is shown in Fig. 16.17.
The amplitude tilt equalizers provide RF amplitude slope over operating frequency
range. The equalizer circuit may be based on active or passive components. The
advantage of active equalizer is its ability to generate slope adjustability by adjusting
its bias current/voltage. In case of active equalizer, depending on bias voltage/current
it can generate positive, negative or zero amplitude slope. Due to the amplitude
slope of the RF signal, linearizer can provide different types of amplitude, amplitude
expansion and phase expansion over the broad operating frequency range required
for linearization of broadband amplifiers. Here, equalizers are used outside the dis-
tortion generation bridge, one at the input and another at the output of the bridge.
The equalizers can also be put within the distortion generation bridge as shown in
Fig. 16.18, to develop broadband linearizers. The equalizer used within the distortion
generator bridge may be placed before the nonlinear circuit as shown in Fig. 16.18a
or after the nonlinear circuit as shown in Fig. 16.18b.
In case of linear attenuator with p-i-n diodes, suitable optimum open-circuit bias
voltage can serve the purpose of gain compensation over the operating temperature
range. Temperature-controlled biasing circuitry is made by exploiting the property
of diode’s RF resistance over temperature. It is known that RF resistance of the
diode may increase or decrease or remain zero depending on the supply voltage or
open-circuit voltage of the bias circuit of the diodes. Therefore, by providing a bias
circuit with optimum open-circuit voltage and current, any temperature coefficient
of the attenuators and equalizers can be obtained for achieving temperature-invariant
amplitude, amplitude expansion and phase expansion of the linearizer. Here, another
linear attenuator at the output of the distortion generation unit is used to compensate
the overall gain/loss variation of the linearizer over the operating temperature range.
The component values of the bias networks correspond to temperature-
compensated performance and can be determined experimentally. RF circuit of the
diode-based linearizers to be temperature compensated is located inside a chamber
in which the temperature can be maintained at a required level in the range varying
from lowest (say −20 °C) to highest (say +80 °C). The amplitude, amplitude expan-
sion and phase expansion of the linearizer are monitored using a vector network
Non-linear Phase RF
RF Attenuator Shifter
OUT
IN Linear
Equa-1 Equa-2
Attn-2
Linear
Attn-1
Temp. dependent
resistive network
VS
Distortion Generation Bridge
Fig. 16.17 Schematic block diagram of a broadband diode linearizer with temperature compensa-
tion networks
(a)
Non-linear Phase RF
RF Equa-1 Attenuator Shifter
OUT
IN Linear
Attn-2
Linear
Equa-2
Attn-1
(b)
Non-linear Phase RF
RF Equa-1
Attenuator Shifter
OUT
IN Linear
Attn-2
Linear
Equa-2
Attn-1
Fig. 16.18 Schematic block diagram of a broadband diode linearizer with equalizer within the
distortion generation bridge, equalizer a at input b at output of the nonlinear attenuator
analyzer (VNA). The voltage and current supply to the nonlinear attenuator, linear
attenuators and equalizers at different operating temperatures can be obtained. The
recorded voltage and current data of the linear attenuator are plotted as shown in
Fig. 16.19. From the best-fitted straight load line data, the open-circuit voltage and
equivalent open-circuit resistance can be obtained as shown in the plot. Similarly,
Linear attenuator Current (mA)

2.0 Slope = 1/RO
1.0 +80OC
VO
+30 OC
-20 OC
0.0
0.0 0.5 1.0 1.5
Linear Attenuator Voltage (Volt)
Fig. 16.19 Experimental determination of optimum bias resistance and open-circuit voltage for
the linear attenuator
Slope = 1/R+80
Nonlinear attenuator Current
2.0
+80 OC
Slope = 1/R+30
+30 OC
Slope = 1/R -20
1.0
VS
-20 OC
0.0
0.0 1.0 2.0 3.0 4.0 5.0
Nonlinear Attenuator Voltage (Volt)
Fig. 16.20 Experimental determination of optimum bias resistance and open-circuit voltage for
the nonlinear attenuator
recorded voltage and current data for the nonlinear attenuator are plotted as shown
in Fig. 16.20. Three load lines are drawn corresponding to three different tempera-
tures. The value of the temperature-dependent resistive network can be obtained to
achieve the resistances R+80 , R+30 and R−20 at the temperatures of +80, +30, −20 °C,
respectively.
16.5 Digital Predistortion Linearizers
Using analog circuit technique, it is very difficult to generate exact opposite ampli-
tude and phase nonlinearities to that of a power amplifier to be linearized. Digital
predistortion linearizers with advanced digital processing techniques are capable
Coupler Voltage Voltage

RF Variable Variable RF
IN Attn/Amp. Phase Shifter OUT
Control signals
Envelop
Detector DAC DAC
ADC Digital Processor
Fig. 16.21 Schematic block diagram of a digital predistortion linearizer variable amplifier and
phase shifter
of generating nearly ideal nonlinear amplitude and phase characteristics required to

linearize different types of power amplifiers.
Schematic diagram of a digital predistortion linearizer is shown in Fig. 16.21.
Here, input RF signal is sampled and information about its envelope is extracted
using an envelope detector circuit. The detected signal is then digitized using an
analog-to-digital converter (ADC). The digitized signal is then processed using a
high-speed digital processor and converted to analog control signals using digital-to-
analog converters (DACs) for controlling a voltage-variable attenuator/amplifier and
a phase shifter. The variable attenuator/amplifier and the phase shifter are connected
in the signal path which controls the amplitude and phase of the signal to generate
proper amplitude and phase nonlinearities to linearize the power amplifier.
Schematic diagram of another digital predistortion linearizer is shown in
Fig. 16.22. The scheme is similar to as shown in Fig. 16.21. Here, the distortion
generator circuit is implemented using two voltage-variable amplifier/attenuators,
(I-VVA) and (Q-VVA), in vector modulator (I–Q) configuration instead of cascaded
amplifier and phase shifter. By controlling the amplitude of I-VVA and Q-VVA
signals, both amplitude and phase of the signal can be controlled according to the
envelope of the input RF signal.
The circuit elements used in the RF signal path, i.e. VVA, phase shifter, power
divider and combiner, power coupler and detectors of the digital linearizers need
to work at the operating RF frequency over its bandwidth. However, the digital
processing unit, ADCs and DACs shall process envelope of the modulated signal.
Thus, bandwidth of the RF signal will be limited by the processing speed of ADC,
DAC, and signal processing units.
References 525
3 dB in phase
power combiner
Q-VVA
3 dB 90o RF
power divider OUT
RF I-VVA
IN
Envelop DAC DAC

Detector
ADC Digital Processor
Fig. 16.22 Schematic block diagram of a digital linearizer using variable amplifiers/attenuators in
I & Q signals
References
1. Inada R, Ogawa H, Kitazume S, Desantis P (1986) A compact 4-GHz linearizer for space use.
2. Imai N, Nojima T, Murase T (1989) Novel linearizer using balanced circulators and its appli-
cation to multilevel digital radio systems. IEEE Trans Microw Theory Tech 37(8):1237–1243
3. Yamauchi K, Mori K, Nakayama M, Itoh Y, Mitsui Y, Ishida O (1996) A novel series diode
linearizer for mobile radio power amplifiers. In: IEEE MTT-S international microwave sym-
posium digest, pp 831–834
4. Zhang W-M, Yuen C (1998) A broadband linearizer for Ka-band satellite communication. In:
IEEE MTT-S international microwave symposium digest, pp 1203–1206
5. Haskins C, Winslow T, Raman S (2000) FET diode linearizer optimization for amplifier pre-
distortion in digital radios. IEEE MGWL 10(1):21–23
6. Yi J, Yang Y, Park M, Kang W, Kim B (2000) Analog predistortion linearizer for high-power
RF amplifiers. IEEE Trans Microw Theory Tech 48(12):2709–2713
7. Yang Y, Woo YY, Kim B (2002) New predistortion linearizer using low-frequency even-order
intermodulation components. IEEE Trans Microw Theory Tech 50(2):446–452
8. Katz A, Sudarsanam R, Aubert C (1985) A reflective diode linearizer for spacecraft applications.
In: IEEE MTT-S international microwave symposium digest, pp 661–664, June 1985
9. Yamauchi K, Mori K, Nakayama M, Mitsui Y, Takagi T (1997) A microwave miniaturized
linearizer using a parallel diode. IEEE MTT-S digest, pp 1199–1202, June 1997
linearizer using a parallel diode with a bias feed resistance. IEEE Trans Microw Theory Tech
45(12):2431–2435
11. Bera SC, Bhardhwaj PS, Singh RV, Garg VK (2003) A diode linearizer for microwave power
amplifiers. Microw J 46(11):102–113
12. Bera SC, Singh RV, Garg VK (2004) A compact Ku-band linearizer for space application. In:
Proceedings of Asia Pacific microwave conference, pp 37–38
13. Bera SC, Singh RV, Garg VK (2008) Diode-based predistortion lineariser for power amplifiers.
IEE Electron Lett 44(2):125–126
14. Garg VK, Bera SC (2009) Predistortion linearizer with novel temperature compensation tech-
nique for satellite transponder. In: 60th international science congress, Korea
15. Bera SC, Jain A, Raval DU, Shah LB (2010) L-band predistortion linearizer for satellite
transponder. In: Proceedings of international conference on microwaves, antenna, propaga-
tion & remote sensing, Jodhpur, India
16. Bera SC, Kumar V, Singh S, Das DK (2013) Temperature behavior and compensation of diode-
based predistortion linearizer. IEEE Microwave Wirel Compon Lett 23(4):211–213
Chapter 17
Microwave Frequency Multipliers
Abstract This chapter presents the working principle, design and analysis of var-
ious microwave frequency multipliers based on diodes and transistors. Different
frequency multiplier circuits based on Varactor diode, Schottky barrier diode, step
recovery diodes (SRDs) and FETs are presented. The chapter also presents diode
and transistor-based balanced configuration of frequency multiplier circuits.
17.1 Introduction
A frequency multiplier generates an output signal whose frequency is a multi-

ple of its input frequency. The basic realization principle for frequency multipli-
cation is the generation of multiple higher frequencies by distorting (amplitude
or phase) a sinusoidal signal and then filtering out the desired higher frequency
component [1, 2]. The distortion of a sinusoidal waveform is carried out using
non-linear characteristic of devices such as diodes, transistors, etc. Various tech-
niques exist for the realization of frequency multipliers. At radio frequencies, these
techniques generally use a non-linear device to generate the desired multiplication
of standard/reference frequency. The non-linear elements such as varactor diodes,
Schottky barrier diodes (SBDs), step recovery diodes (SRDs) and bipolar junction
transistors (BJTs) or field effect transistors (FETs) are used as non-linear devices
[1–14].
Frequency multipliers are used to realize RF source, e.g. local oscillator (LO) of
higher frequency from low-frequency standard. With the use of higher frequency for
wireless communications, to take advantage of available spectrum, the required LO
frequencies are also increasing. The performances of microwave solid-state oscilla-
tors degrade with the increase of frequency. To overcome this problem, highly stable
oscillators operating at a lower frequency are used, followed by frequency multi-
pliers with low conversion loss for realization of microwave frequency generator.
In case of multiplier-based local oscillator, a crystal oscillator followed by multipli-
ers, filters and amplifiers is used to generate high frequency source with required
output power. Other approaches for realizing local oscillator at higher frequencies

528 17 Microwave Frequency Multipliers
are the use of transistor-based oscillator stabilized by a dielectric resonator or use

of phase-locked loop (PLL) using crystal source, frequency divider, phase detector
and voltage-controlled oscillator. Among all these approaches, multiplier-based local
oscillators are simple in circuitry and lower in cost.
17.2 Principle of Multiplier Operation
A pure sinusoidal waveform is distorted without affecting its zero crossing point in
the time axis, i.e. maintaining its periodicity when it passes through a non-linear
(amplitude and/or phase) system. Thus, the distorted output signal consists of only
the fundamental and other components of frequencies exactly integer multiple of the
fundamental frequency, called harmonics. In this process, any component of frequen-
cies other than the exact multiple of the fundamental frequency cannot be generated
maintaining the zero crossing point unaltered. Therefore, any non-sinusoidal peri-
odic waveform (distorted sine wave) contains energy at fundamental and harmonics
of the fundamental frequency. The design of multiplier is to design a non-linear cir-
cuit with filtering that produces a waveform with significant signal strength at the
desired harmonic. Conversion gain G c is used to characterize microwave frequency
multiplier’s conversion efficiency and is defined as the ratio of the output power
delivered to the load to the available power from the input source. The goal of the
circuit design is to maximize the conversion gain for a given device and input/output
frequencies.
Generalized block diagram with associated input and output spectrum of a fre-
quency multiplier is shown in Fig. 17.1. Here, devices such as diodes and transistors
are biased to operate in strongly non-linear region to distort the input sinusoidal
waveform of frequency f o for the generation of higher order harmonics in which
one desired higher order harmonic passes through the bandpass filter, suppressing
other frequency components [1, 2]. The input and output matching networks provide
required terminations at each harmonic frequency to get output sinusoidal waveform
of frequency n f o with maximum possible amplitude, i.e. with maximum conversion
gain G c . The use of LPF and BPF provide high impedance to all the unwanted har-
monic components. However, it is desirable to allow the currents of the undesired
harmonics to flow so that the inter-modulation products of those harmonics will con-
tribute to the desired output harmonic. Therefore, to maximize conversion efficiency
it is desirable to short the currents of the undesired components. Another performance
requirement is the suppression of fundamental as well as harmonic components other
than the wanted frequency component, i.e. n f o at the output of the multiplier. The
worst case suppression S, as shown in Fig. 17.1, should be as high as possible. The
importance of the suppression obviously depends on the particular application. In
general, the input matching network should provide proper matching to the funda-
mental frequency so that most of the input power reaches the non-linear device and
the output matching network should provide proper matching to the desired multi-
plied frequency so that most of the output power transferred to the load. Except the
17.2 Principle of Multiplier Operation 529
LPF& BPF &

Non-linear
Ro input output
device with Ro
matching matching ( )
bias network
network network
( ) ( )
Fig. 17.1 Generalized block schematic of a frequency multiplier
fundamental and desires multiplied frequencies, all other frequencies are referred
to as idle frequencies. The matching networks should provide proper termination to
the idle frequencies to transfer maximum output power to the load corresponds to
desired multiplied frequency and suppress the undesired frequency components. In
principle, reactive loads at idle frequencies are desirable so that they do not dissipate
active power.
Any periodic non-sinusoidal wave can be expressed by Fourier series as
∞
∞

vo (ωt) a0 + an cos nωt + bn sin nωt (17.1)
n1 n1
Here, ω is the angular frequency of the RF signal a0 , an and bn are constants.

Here, a0 represents the average DC value of the waveform, and it is not important for
multiplier application, thus may be ignored. The other constants can be determined
by performing the following integrations over any 2π radians:
2π
1
an vo (θ ) cos nθ dθ (17.2a)
π
0
2π
1
bn vo (θ ) sin nθ dθ (17.2b)
π
0
Fig. 17.2 Harmonic components for various waveforms
Here, ωt θ . These integrals can be easily calculated and some may be com-
puted by inspection. As example, a very narrow pulse has about the same amplitude
for all the harmonics, i.e. comblike spectrum. Similarly, any waveform of ‘half-wave
symmetry’ like symmetrical clipped sinusoidal waveform has the only odd harmonic
components. Some other waveforms with their harmonic components are shown in
Fig. 17.2. It is to be noted that for waveforms with sharp edges (e.g. square wave)
have higher harmonics with larger amplitudes due to 1/n dependency of amplitudes
compared to other waveforms (e.g. rectified sine wave) of 1/n 2 dependency. It is
also to be noted that some waveforms contain only even or only odd harmonic com-
ponents, which reduces the requirement of filtering out the undesired harmonic to
get desired frequency component. The harmonic components not only depend upon
the shape of the waveform, but also depend on the duty cycle. Figure 17.3 shows the
waveform of different pulse and their spectral components. This type of half and full
wave current waveforms are generated when a sinusoidal signal applied to diode-
based circuits. It shows that with the decrease in width of the waveform, level of
the higher harmonic components increases. Suppose, a square waveform have a duty
cycle T1 /T2 , as shown in Fig. 17.4. The coefficients an and bn of the waveform can
be derived as
θ1
1
an A cos nθ dθ (17.3a)
π
0
Here, θ ωt and θ1 ωT1 . Therefore,

A
an sin(nθ1 ) (17.3b)
nπ
17.2 Principle of Multiplier Operation 531
a
1.00
cba b
0.50 c
0 0.00
0 0.5π 1π 1.5π 2π
Spectrum (dBc)
- 20
- 40
- 60
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Harmonic numbers
a
1.00
b
c ba 0.50 c
0.00
0 0 0.5π 1π 1.5π 2π
Spectrum (dBc)
-20
-40
-60
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Harmonic numbers
Fig. 17.3 Waveform and spectrum of different waveforms
Similarly,
θ1
1
bn A sin nθ dθ (17.4a)
π
0
A
[1 − cos(nθ1 )] (17.4b)
nπ
Thus, the resultant amplitude of the nth harmonic Cn can be written as

Cn an2 + bn2 (17.5)
Figure 17.4 shows the normalized harmonic amplitudes of square waves with dif-
ferent duty cycles corresponds to θ1 0 to π . The plot is very useful to decide duty
Amplitude
t
Amplitude
(normalized) Harmonic content
0.7 n=1
0.6
n=2
0.5
n=3
0.4 n=4
n=5
0.3 n=6
0.2
0.1
0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50
Fractional Pulse width ( )
Fig. 17.4 Harmonic amplitudes of square waves with different duty cycles
cycle of a square waveform for particular multiplier. It shows that square wave of
very small duty cycle contain all harmonics of nearly equal amplitudes, i.e. comblike
spectrum whereas, square wave of 50% duty cycle has only the odd harmonic compo-
nents. It also suggests that to design a frequency doubler using a square-shaped pulse,
though most second harmonic energy will be generated for duty cycle of 25%; it is
preferable to choose duty cycle of 33% for suppressing the third harmonic without
any extra filter.
Other aspects of a frequency multiplier are the DC power consumption, bandwidth
of operation, and reliability of the system. Bandwidth is important to provide variable
output frequency with the use of variable fundamental frequency oscillator at input of
the multiplier. Reliability is important since the device operates in highly non-linear
region. Precaution in terms of bias condition as well as input RF level should be
taken to prevent operation of the device goes beyond its safe operating limit.
In case of ideal multiplier considering noise less multiplication and no phase noise
generation of its own, the process of frequency multiplication inevitably increase the
phase noise. The reason for the increase of phase noise is that a frequency multiplier
is in fact also a phase multiplier. Thus, it multiplies the frequency as well as phase
deviations of the input signal. All frequency multipliers will increase the phase noise
by the same factor ‘n’ that they multiply, in dB this is 20 log(n). However, generation
of microwave frequency signal by multiplying a very stable low-frequency reference
signal can have better phase noise than producing it directly in the microwave fre-
quency range.
17.3 Diode Multiplier 533
17.3 Diode Multiplier
Diode-based frequency multipliers are very popular in microwave frequency range

for their simple structure and higher reliability. Uses of diode-based multipliers
extend to millimetre and submillimetre wave systems [3–9]. Diode-based multi-
pliers are the only convenient technique of realizing local oscillators for frequencies
above 200 GHz. DC power consumption of diode multipliers is very less or zero;
obviously, it does not provide any conversion gain. Therefore, cascading an amplifier
with the diode multiplier is required to achieve the required RF power level of the
multiplied frequency output. To obtain higher order frequency multiplication, it is
required to cascade several multipliers which will increase complexity as well as
conversion efficiency. At microwave and millimetre wave frequency regions, mainly
varactor, Schottky and step recovery diodes are used as non-linear device for multi-
plier. In case of Schottky diode, strong non-linearity of the conduction current is the
mechanism for frequency multiplication, called resistive multiplication. Mainly non-
linearity of capacitance is the mechanism for frequency multiplication of varactor
and step recovery diodes, known as reactive multiplication.
Ideally, resistive multipliers have infinite bandwidth. However, the presence of
associated junction capacitance and reactive parasitic elements of the diodes limit
the high-frequency operation. Moreover, input and output matching networks further
limit the bandwidth. In addition to that, power loss within the resistive non-linear
component increases the conversion loss. The conversion efficiency of a non-linear
resistance multiplier is proportional to 1/n 2 , where n is the harmonic number. Thus,
resistive multipliers are not suitable in case of applications where conversion loss is
critical. Resistive multipliers are efficiently used in balanced configurations. Singly
balanced or doubly balanced configurations have intrinsic property of fundamental
and odd harmonic frequency rejection. This reduces the need for filtering out the odd
harmonic frequency components.
Ideally, reactive multipliers have zero conversion losses. This requires proper
reactive loading at all frequencies other than the input (fundamental) frequency and
the output (multiplied) frequency. However, strong frequency dependency of the
reactive impedances limits the operating bandwidth of the multiplier, i.e. reactive
multipliers have narrowband performance.
17.3.1 Schottky Diode Frequency Multiplier
Schottky diodes for multiplier can be used as non-linear resistor, i.e. varistor or non-
linear capacitor, i.e. varactor. There are tradeoffs in choosing a resistive vs. reactive
mode of operation. Schottky diodes are used as non-linear resistor for multiplier
design to provide high-frequency multiplication with very high bandwidth, with
the penalty of higher conversion loss, whereas frequency multipliers using Schot-
tky diode as a varactor have high conversion efficiency with the penalty of narrow
I
current waveform
V
0
voltage waveform
Fig. 17.5 Input signal and diode current waveform
Li1 Lo1
Ci Co
Li2 Lo2
Fig. 17.6 A simple schematic circuit diagram of a frequency multiplier
bandwidth and need high Q circuit, thus very sensitive to small change of structural
dimension.
Resistive frequency multipliers use non-linearity of I-V characteristic of a Schot-
tky barrier diode to distort a sinusoidal waveform for generating harmonic compo-
nents as shown in Fig. 17.5. Level of the harmonic components increases with the
increase in amount of distortion. However, resistive frequency multipliers have higher
conversion loss; thus, the level of maximum attainable power level of the harmonic
components is less. A simplified schematic circuit diagram of a resistive Schottky
diode frequency multiplier is shown in Fig. 17.6. Here, input inductors (L i1 , L i2 ) and
capacitor (Ci ) match the diode impedance corresponding to frequency fo to the source
impedance for transferring maximum fundamental signal to the diode. Whereas, out-
put inductors (L o1, L o2 ) and capacitor (Co ) match the diode impedance corresponds
to desired nth harmonic to the load impedance for transferring maximum harmonic
power to the load. Input inductor–capacitor combination also resonates at f o and out-
put inductors–capacitor combination resonates at n f o . Thus, the resonators provide
short circuits to the unwanted harmonics and decouple the input from the output. The
resonators also put the diode in parallel with the input at the fundamental frequency
and in parallel with the output at the desired harmonic.
A Schottky diode frequency doubler using microstrip distributed components suit-
able for microwave frequency range is shown in Fig. 17.7. Here, the input side of
the diode is terminated to ground through a λo /4 short-circuited stub. It will provide
Fig. 17.7 Schottky diode frequency doubler realization using microstrip line distributed circuit
elements
high impedance (ideally open) circuit at the input side of the diode thus it will allow
the fundamental frequency to reach the diode. The short-circuited λo /4 transmission
line will act as short-circuited half-wavelength transmission line corresponding to
second harmonic, thus it will prevent the output power generated in the diode trav-
elling towards the input. Similarly, the use of λo /4 open-circuited transmission line
at the output side of the diode creates an RF short at the output for the fundamental
frequency. Thus, it causes the input signal reaching the output of the device reflected
back to the diode.
High impedance transmission line elements at the input and output of the diode are
used as inductor to resonate the diode junction capacitance. Finally, λo /4, i.e. quarter
wavelength impedance transformer is used at the input of the circuit to transform 50-
source impedance to the required diode impedance for transferring maximum signal
to the diode. Similarly, λo /8 transformer which is quarter wavelength transmission
line at 2 f o transform the 50 load impedance to the diode output impedance, to
transfer maximum second harmonic power to the load.
17.3.2 Varactor Diode Frequency Multiplier
Non-linear capacitive reactance of varactor diode distorts sinusoidal signal to gener-

ate its harmonics. A typical non-linear capacitance–voltage characteristic of varactor
diodes is shown in Fig. 17.8. Considering zero series resistance, reactive multipliers
have high conversion efficiency, when matching networks are designed using only
reactive (loss less) elements. Here, frequency multiplication is based on the non-
linear reactance, thus the multipliers have tuned input and output matching circuits
and often tuned idler circuits are used for achieving higher conversion gain. Due to
the reactive nature of the matching networks, it exhibits narrow bandwidth and high
sensitivity to operating conditions, like bias voltage/current and component values.
Practically, varactor diode-based frequency multipliers generate very little phase as
well as amplitude noise. It only contributes to thermal noise due to the series resis-
tance of the varactor and loss resistances associated with the matching networks.
Capacitance
Diode voltage
0
Fig. 17.8 Typical non-linear capacitance characteristic of a varactor diode
Diode bias
Tuned Rb Tuned Tuned
to fo to 2fo to 3fo
Li1 Lo1
Cb Ca Cb
Ci Co
La
Li2 Lo2
Fig. 17.9 A simple schematic circuit diagram of a frequency tripler using lumped components
To generate desired higher order harmonic components with sufficient level, it

is required to provide short circuit to other harmonic components. The short circuit
networks are known as idlers. Idlers are usually realized using a series resonating
network using lumped components or using shunt short or open transmission line of
proper length for a particular harmonic component. Without idlers, varactor multi-
plier does not generate harmonic components efficiently beyond the second order.
An idler corresponds to second harmonic allowed second harmonic current through
the diode which mixed with the fundamental and can generate third order harmonic.
In the same way, providing idlers at second and third harmonics it is possible to
generate fourth harmonic efficiently.
Metal–semiconductor junction diode, i.e. Schottky diode and p–n junction diode
both are used as varactor diode for frequency multiplier. However, Schottky barrier
varactor diode-based multiplier can generate frequencies to several hundred of GHz.
Figure 17.9 shows a simple schematic circuit diagram of a multiplier using varactor
diode as the non-linear device. The bias resistor Rb is used to bias the diode, which
also changes the DC voltage across the diode with the change of RF power level, thus
increases non-linearity. This is an example of tripler (×3) circuit. The series resonator
circuit consists of Ca and L a is used to provide short to the second harmonic. The input
network consists of L i1 , L i2 and Ci is for matching the source impedance to the input
Diode bias
Rb
2fo
Cb idler
BPF O/P
I/P matching O/P matching
LPF for fo for 3fo
I/P Ci
via hole
ground
Fig. 17.10 Varactor diode frequency tripler realization using microstrip line distributed circuit
elements
of the device corresponds to fundamental frequency f o . It will also provide rejection

to the third as well as other higher order harmonics at the input side. Whereas, the
output network consists of L o1 , L o2 and Co is for matching the load impedance to
the output of the device corresponding to third harmonic component, 3 f o . It will
also provide rejection to fundamental as well as other higher order harmonics at the
output.
Figure 17.10 shows a tripler circuit layout using varactor diode and microstrip
distributed elements. The diode is grounded using via holes. The bias network is
designed with λo /4 line with short circuit using capacitor Cb . At the device end it
provides very high impedance (ideally open) for fundamental frequency, f o but very
low impedance (ideally short) for second harmonic, 2 f o . Thus, the bias network
will provide required short circuit to the second harmonic to generate higher order
harmonics in addition to provide the bias voltage/current to the diode. The input and
output matching networks are designed using λo /4 microstrip impedance transformer
for f o and 3 f o , respectively. The required low-pass filter for input and bandpass filter
for output are also designed using microstrip circuit elements as shown.
17.3.3 SRD Frequency Multiplier
Step recovery diodes are based on p-i-n configuration and behave as an ideal non-
linear capacitor thus used for realization of higher order multipliers. Step recovery
diodes are used as multipliers for all orders of multiplication, starting from 2 through
20 for both narrow and wide bandwidths. Ideally, it acts as a two-state capacitor with
zero switching time between the states. Under forward charge storage condition, it
Fig. 17.11 Typical Capacitance

non-linear capacitance
characteristic of SRD
C is low &constant
Diode voltage
0
acts as a large (ideally infinite) capacitance, thus behave as low impedance whereas
under reverse bias condition it acts as a small capacitance (the depletion capacitance)
thus behave as high impedance as shown in Fig. 17.11. The change from its low
impedance state to high impedance state takes place very quickly, in a time interval
called the diode transition time (TT ). Thus, when an RF signal applied to the diode,
on the one half-cycle, it stores charge and appears as low impedance and on the
second half-cycle, the diode conducts until the stored charge is removed and then
switches off very rapidly at a speed governed by the transition time, TT . This change
produces a narrow pulse of voltage from the applied RF signal. This is equivalent
to a number of frequencies which are multiples of the applied input frequency, i.e.
harmonics of the fundamental frequency. In general, it is desirable that the minority
carrier lifetime (τ ) be greater than 10 times the period of the input frequency, while
the transition time TT should be less than the period of the output frequency. Thus, the
highest order harmonic is limited by the narrowness of the pulse that is determined
by the transition time, TT , of the SRD.
SRD multipliers are in the category of switching reactance multiplier thus it does
not have efficiency limitation (1/n 2 ) as in the case of resistive multipliers. In case
of high-order frequency multiplication, the SRD multipliers provide much higher
efficiency even compared to varactor multiplier.
A multiplier using step recovery diode consists of impulse generator, input and
output matching networks, and input and output filters as shown in Fig. 17.12. The
impulse generator converts the energy in each input cycle into a narrow large ampli-
tude voltage pulse as shown in Fig. 17.13. It consists of a step recovery diode (SRD),
a drive inductor L 3 and capacitor C3 . The capacitor C3 is used to provide short at
the required output frequency. Thus, the RF output signal cannot get back into the
input section of the frequency multiplier circuit. Another function of the capacitor
C3 is to act as part of the input matching network. Here, bias resistor R B is used
for self-biasing the diode. Thus, it will provide reverse bias to the diode due to the
rectification of the RF signal.
The input matching circuit provides a match to the source impedance at the funda-
mental frequency to the input of the impulse generator circuit. The output matching
circuit is basically a resonator which converts the generated impulse into a damped
ringing waveform at the desired output frequency. When the output of the multiplier
is terminated directly to the load impedance Ro , it works as a broadband multiplier
and its output waveform looks like a comb spectrum. But when the output of the mul-
Input Input Output

Filter Matching Impulse Generator Matching
BPF
L1 L2 L3 L4
SRD
C1 C2 C3 C4
Fig. 17.12 Schematic of a series diode linearizer circuit
Input
voltage
Time
Input
Current
Time
Output
voltage
Time
Fig. 17.13 Typical impulse response of SRD-based impulse generator
tiplier is terminated in a resonant network, i.e. a bandpass filter, it works as a narrow

band multiplier, and the output level will be the same energy but will be concentrated
around the resonant frequency, n f o . The output filter reflects the unwanted harmonics
back to the SRD where they mix to increase the power level at the desired harmonic
frequency. The BPF ultimately pass the desired frequency to the load impedance
rejecting the undesired frequency components. A typical spectrum at the output of
the impulse generator circuit and at the output of a 15th-order frequency multiplier
circuit is shown in Fig. 17.14.
(a) (b)
Amplitude Amplitude
15th
Frequency Frequency
Fig. 17.14 Typical spectrum at the output of a impulse generator and b BPF
17.4 Transistor Multiplier
Frequency multipliers using transistors have better conversion efficiency compared

to the diode-based frequency multipliers [10–14]. The diode multipliers always have
conversion loss but FET and BJT multipliers provide conversion gain with good DC
to RF efficiency. Transistor-based circuits generate harmonics of an input signal by
distorting it when the level of the input signal and bias conditions are such that it
drives the transistor in its non-linear region.
Origin of non-linearities of BJT and FETs are usually the non-linearities in the
output current characteristics with the input voltage. Practically, output current clip-
ping is the main contributor for generating harmonics of the input signal. This non-
linearity is due to the non-linear characteristic of common emitter current gain and
transconductance gain in case of BJTs and FETs, respectively. Figure 17.15 shows a
typical output current characteristic and transconductance gain of FETs. Other than
the non-linearities of the output current characteristic, non-linear input and output
capacitances as well as feedback capacitance also contribute to the generation of har-
monic components. For example, non-linearities of gate–source, drain–source and
gate–drain capacitances of FETs under large signal operating condition contribute
to the generation of harmonic components. Practically, the presence of these non-
linearities enhances the harmonics level where the main contributor is the output
current clipping caused by proper biasing and input signal level.
A simplified model of transistors can be considered to derive the harmonic con-
tents of a transistor-based multiplier. It can be considered that the harmonics are
generated due to only clipping of the output current waveform. Thus, the output
current waveform can be described as piecewise linear model as shown in Fig. 17.16
as dotted line in case of FETs. The corresponding transconductance gain is constant
with finite positive value over −Vgso to VF and is zero beyond these values as shown
in the figure. This simple piecewise linear model can be used assuming the load line
does not reach the ohmic and breakdown regions. Thus, it can be assumed that the
operating voltage is always higher than the knee voltage and smaller than the break-
down voltage by proper drain voltage biasing and by suitable choice of the output
load impedance.
17.4 Transistor Multiplier 541
(a) Id (b)
IF
Vd Vgs
Vgs -Vgso 0 VF VK VdgB -Vgso 0
Fig. 17.15 Typical a drain current and b transconductance gain of FETs and their piecewise linear
approximate plots
Fig. 17.16 Output current waveform with asymmetric clipping
Output current waveform may clip only in one side by the pinch-off of the channel
in case of FETs and by cut-off region of the BJT. In this case, transistors are biased for
Class A–C regions. One-sided clipping may also happen by gate–channel junction
forward conduction in case of FETs and operating BJTs at saturation region biasing
in Class-A. The disadvantage of this operating condition for frequency multiplier is
the high DC power dissipation and possibility of device damage for excessive gate
junction current for FETs. Both sided symmetric and asymmetric clipping may also
happen depending on the biasing and input voltage/current swings.
A generalized output current waveform of a transistor-based multiplier with both
side clipping is shown in Fig. 17.16. Here, I Q is the quiescent bias current of the
transistor, I F is the maximum peak current and I p is the amplitude of corresponding
sinusoidal output current waveform. The output current waveform is clipped with
amount of 2θ1 and 2θ2 as shown in the diagram. Thus, the sinusoidal waveform can
be expressed as
i o (θ ) I Q + I p sin ωt (17.6)
Here, ω is the angular frequency of the input RF signal and ωt θ . The clipping
of the output current waveform can be expressed as
π
3π
io − θ1 I F and i o − θ2 0 (17.7)
2 2
From (17.6) and (17.7), I Q and I p can be expressed in terms of I F as

cos θ2 1
IQ I F and I p IF (17.8a)
cos θ1 + cos θ2 cos θ1 + cos θ2
The bias point of the transistor can be derived from (17.8a). For FET multiplier,
gate bias voltage can be derived from:
IQ
Vgs −Vgso + Vgso + VF (17.8b)

IF
cos θ2
−Vgso + Vgso + VF (17.8c)

cos θ1 + cos θ2
This periodic output current waveform can be expressed with Fourier series as
∞
∞

i o (t) a0 + an cos nωt + bn sin nωt (17.9)
n1 n1
Here, ω is the angular frequency of the input RF signal; a0 , an and bn are constants
and can be determined by
2π
1
a0 i o (θ )dθ (17.10a)
2π
0
2π
1
an i o (θ ) cos nθ dθ (17.10b)
π
0
2π
1
bn i o (θ ) sin nθ dθ (17.10c)
π
0
Here, θ ωt. As shown in the figure, the clippings of the current waveform are
over 2θ1 and 2θ2 . Here, a0 is DC component of the output current waveform which
depends on the initial biasing of the transistor and clipping of the waveform. The
constants an and bn will tell us about the amplitude of the fundamental and harmonics
at the output of the transistor. The resultant output amplitude Ion corresponds to nth
harmonic can be expressed as

Ion an2 + bn2 (17.11)
The output current waveform can be expressed as
i o (θ ) I Q + I p sin θ, over 0 ≤ θ ≤ π/2 − θ1

I F , over π/2 − θ1 ≤ θ ≤ π/2 + θ1
I Q + I p sin θ, over π/2 + θ1 ≤ θ ≤ 3π/2 − θ2
0, over 3π/2 − θ2 ≤ θ ≤ 3π/2 + θ2
I Q + I p sin θ, over 3π/2 + θ2 ≤ θ ≤ 2π (17.12)
Therefore, the DC component of the waveform can be determined as follows:

2π
1
a0 i o (θ)dθ
2π
0
⎡ π π 3π
−θ1
2 +θ1
2 2 −θ2
1 ⎢

⎣ I Q + I p sin θ dθ + I F dθ + I Q + I p sin θ dθ
2π
π π
0 2 −θ1 2 +θ1
⎤
2π

⎥
+ I Q + I p sin θ dθ ⎥
⎦
3π
2 +θ2
or,
π − θ1 − θ2 sin θ2 − sin θ1 θ1
a0 IQ + I p + IF (17.13)
π π π
Similarly, coefficients an can be determined as
2π
1
an i o (θ ) cos nθ dθ
π
0
⎡ π π
−θ1 2 2 +θ1
1⎢
⎣ I Q + I p sin θ cos nθ dθ + I F cos nθ dθ

π
π
0 2 −θ1
−θ2
3π
2

+ I Q + I p sin θ cos nθ dθ
π
+θ1
⎤
2
2π

⎥
+ I Q + I p sin θ cos nθ dθ ⎦ (17.14)
3π
2 +θ2
or,
π π
IQ 3π 3π
an sin n − θ1 + sin n − θ2 − sin n + θ1 − sin n + θ2
nπ 2 2 2 2
π
Ip 3π
+ cos (n − 1) − θ1 + cos (n − 1) − θ2
2π (n − 1) 2 2
π
− cos (n − 1) + θ1 + cos(n − 1)(2π ) − 1
2
3π
− cos (n − 1) + θ2
2
π
Ip 3π
+ cos (n + 1) + θ1 − cos (n + 1) − θ2
2π (n + 1) 2 2
π
− cos (n + 1) − θ1 − cos(n + 1)(2π ) + 1
2
3π
+ cos (n + 1) + θ2
2
IF π π
+ sin n + θ1 − sin n − θ1 (17.15)
nπ 2 2
Similarly, the coefficients bn can be determined as
2π
1
bn i o (θ ) sin nθ dθ
π
0
⎡ π π
−θ1
2 2 +θ1
1⎢
⎣ I Q + I p sin θ sin nθ dθ + I F sin nθ dθ

π
π
0 2 −θ1
−θ2
3π
2

+ I Q + I p sin θ sin nθ dθ
π
2 +θ1
2π

+ I Q + I p sin θ sin nθ dθ (17.16)
3π
2 +θ2
or,
π π
IQ 3π 3π
bn cos n + θ1 − cos n − θ1 + cos n + θ2 − cos n − θ2
nπ 2 2 2 2
π
Ip 3π
+ sin (n − 1) − θ1 + sin (n − 1) − θ2
2π (n − 1) 2 2
π
3π
− sin (n − 1) + θ1 + sin(n − 1)(2π ) − sin (n − 1) + θ2
2 2
Ip π π
+ sin (n + 1) + θ1 − sin (n + 1) − θ1
2π (n + 1) 2 2

3π 3π
− sin (n + 1) − θ2 − sin(n + 1)(2π ) + sin (n + 1) + θ2
2 2
IF π π
+ cos n − θ1 − cos n + θ1 (17.17)
nπ 2 2
In case of one-sided clipping by the pinch-off of the channel, amplitude of fun-
damental and up to fifth harmonics as well as DC current components are shown
in Fig. 17.17 with the variation of the conduction angle using (17.13), (17.15) and
(17.17). Thus, here θ1 0 and the conduction angle of the current waveform are
2π − 2θ2 . It shows that, for the best conversion gain, Class-C bias must be selected.
For example, for optimum doubler operation, an operating angle of approximately
0.66π must be selected that yields a second harmonic current amplitude about 0.275
times of the maximum output current, I F . For a frequency tripler, an operating angle
of approximately 0.45π must be selected for a third harmonic current amplitude
equal to about 0.185 times of the maximum output current. Similarly, for fourth-
and fifth-order harmonics amplitude will be maximum of values 0.138 and 0.110
times of the maximum output current for the conduction angle of 0.34π and 0.26π ,
respectively.
The output load impedance must be such that the voltage swing corresponding to
the required harmonic component becomes maximum without crossing the voltage
Fig. 17.17 Normalized amplitudes of harmonics with current conduction angle (2π − 2θ2 ), for
one-sided clipping (θ1 0)
breakdown limit. All output harmonics must be shorted except the desired output
harmonic. Thus, the optimum output load is the resistance that maximizes the voltage
swing within the hard non-linear limits imposed by breakdown and knee voltage for
FETs. Thus, In case of FET multiplier, the load impedance of nth harmonic is given
by
Vdsn Vdg B + 2Vgs − VF − Vk
R Ln (17.18)
Idsn 2Idsn
Here, Vgs is given by (17.8c). Symmetric clipping of the output current waveform,
one side by pinch-off and other side by saturation of the FET channel, happen when
FET is biased for Class-A operation. Thus, here θ1 θ2 and the conduction angle of
the current waveform is 2π −2θ2 . Using (17.13), (17.15) and (17.17), the normalized
amplitude of the harmonic components for the symmetric clipped waveform are
shown in Fig. 17.18. There is no existence of even order harmonic components due to
the symmetric nature of the current waveform. However, relatively larger maximum
current amplitude of all the odd order harmonics are obtained compared to the one
side clipped current waveform. For example, maximum amplitudes of the third and
fifth harmonic current components are 0.212 and 0.127 times, respectively, of the
maximum current, I F . Thus, more output power, i.e. higher conversion gain can be
obtained for the odd harmonic components due to the symmetric current clipping
compared to the one side clipped waveform but at the expense of a more critical
reliability due to the forward gate current conduction and higher dissipated power.
The direct generation of harmonics of higher order can be obtained by biasing the
device at different conduction angles which are determined by the gate biasing of
the transistors. For even order harmonic generation the device needs to be biased in
Class-AB or C, in order to get rectified sinusoidal output current rich in even order
harmonics. For odd order harmonics generation, the optimum bias condition is to
generate an output current waveform with both sides clipping. Thus, Class-A biasing
about centre of the output current is required and the input voltage/current should
have high enough amplitude to clip the output current waveform in both sides. The
magnitude of harmonic components decreases with the increase of its order thus it is
required to increase the load impedance with the increase of harmonic order to get
output voltage of higher amplitude.
By varying the bias voltage and the amplitude of the input voltage, all possible
combinations of upper and lower current clippings can be obtained. The normalized
amplitude of different harmonics in this general case of non-symmetric clipping can
be obtained using Eqs. (17.13), (17.15) and (17.17). The results are plotted as contour
plots of constant second harmonic current amplitude, as shown in Fig. 17.19. The
plot is symmetric with respect to the bisector 2θ1 + 2θ2 2π . The points on the
bisector represent square waves with duty cycle from 0 (2θ1 0, 2θ2 2π ) to 1
(2θ1 2π, 2θ2 0). The point at middle on this line (2θ1 2θ2 π ) corresponds
to symmetric square wave; thus, there is no second harmonic component. The plot
is anti-symmetric with respect to the bisector line 2θ1 2θ2 , which represents
symmetric clipping of the output current waveform from a Class-A bias condition;
Fig. 17.18 Normalized amplitudes of harmonics with current conduction angle (2π − 2θ2 ) for
symmetric clipping (θ1 θ2 )
thus, there is no second harmonic component along this bisector. It is clear from
the constant second harmonic contour plot that the normalized maximum current
component is about 0.318 which is higher than the one side clipped current waveform
of value 0.275. From the contour, the maximum normalized amplitude (0.318) of the
second harmonic component corresponds to asymmetric clipping around a square
waveform with duty cycle equal to 1/4 (or symmetrically 3/4).
The constant amplitudes of the third harmonic current component for different val-
ues of clipping angles are shown in Fig. 17.20. In case of third harmonic component
also, the plot is symmetric with respect to the bisector 2θ1 + 2θ2 2π . The points on
the bisector represent square waves with duty cycle from 0 (2θ1 0, 2θ2 2π ) to 1
(2θ1 2π, 2θ2 0). The point at middle on this line (2θ1 2θ2 π ) corresponds
to symmetric square wave; the amplitude of the third harmonic component becomes
highest of normalized value 0.212. It is interesting to see that another two combina-
tions of clippings are corresponded to the maximum third harmonic waveform exists.
The maximum normalized amplitude (0.212) of the third harmonic component cor-
responds to asymmetric clipping around a square waveform with duty cycle equal
to 1/6 (or symmetrically 5/6). The plot is asymmetric with respect to the bisector
line 2θ1 2θ2 , which represents symmetric clipping of the output current waveform
from a Class-A bias condition.
Fig. 17.19 Constant amplitude contour of second harmonic over clipping angles
Fig. 17.20 Constant amplitude contours of third harmonic over clipping angles
17.5 Realization of FET Multipliers 549
17.5 Realization of FET Multipliers
A generalized schematic circuit diagram of a FET multiplier is shown in Fig. 17.21.

Ideally, the input and output matching networks provide the required impedance
to each harmonic component independently. The transistor is biased at sufficiently
negative bias to operate it in Class-AB, B or C mode depending on the requirement
of output current clipping as discussed in the previous section. To achieve higher
conversion gain, in most of the cases the requirement of gate voltage will be smaller
than the pinch-off voltage, Vgso , and the FET will remain OFF during most of the
signal cycle. Thus, in most of the cases, the duty cycle will be smaller than 50%. The
drain bias is applied depending on its breakdown voltage and required output voltage
swing. The duty cycle of the output current waveform is adjusted by adjusting the gate
bias voltage. The input matching network provides complex conjugate match at the
input port of the device corresponds to fundamental frequency for maximum power
transfer into the device if permissible from the stability point of view. Similarly, the
output port is also complex conjugate matched at the desired harmonic frequency to
transfer maximum power to the load for the desired harmonic. The input matching
circuit also consists of resonators, also called idler circuit, which provides short circuit
to the desired output as well as other harmonic frequencies. The output matching
circuit also consists of resonators which provide fundamental and other harmonic
frequency components except for the desired frequency. It is desirable to realize the
idler circuits at input and output ports using reactive elements for minimizing power
dissipation, i.e. for maximizing conversion gain. Also, reactive terminations reflect
back the signal into the active device; this contribution can be exploited for increasing
the conversion gain by constructive interference.
Microstrip circuit layout of a microwave frequency doubler using FET is shown
in Fig. 17.22. Here, negative gate bias is fed to the device through a short-circuited
λo /4 transmission line. On the other hand, drain bias is given to the device through a
VG VDD
RG RD
Output
matching
Input network
matching
network
Fig. 17.21 Generalized circuit diagram of FET multiplier

Fig. 17.22 Circuit layout of an FET doubler
short-circuited λo /8 high impedance transmission line which is quarter wavelength

long corresponds to the desired second harmonic frequency. The gate bias network
also provides short to the second harmonic frequency at the input of the device. The
input of the device is complex conjugate matched to the source impedance by using
series and stunt matching network. Two idler circuits are connected at the output
of the device correspond to undesired fundamental and third harmonic components
using open stub transmission lines. A high impedance transmission line is used as
impedance transformer to match the output port for the desired second harmonic
component.
Practical way to design a frequency multiplier without having its non-linear model
is the use of non-linear harmonic load pull test method. In this case, the active device
is inserted into a circuit where the input port is tuned for the fundamental frequency
and the output port is tuned for the maximum possible harmonic components for
achieving the maximum amplitude of the desired harmonic component. After that,
the active device is removed and both the input and output matching networks are
measured using a vector network analyzer to determine the desired impedances. Then
the desired impedances can be synthesized using linear circuit analysis method.
17.6 Balanced Frequency Multipliers
Balanced frequency multipliers are preferred configuration at microwave and mil-

limetre wave frequency for simplifying idler circuits and providing higher efficiency,
i.e. more conversion gain at the desired frequency. Another property of the balanced
frequency multipliers is the inherent rejection of certain unwanted harmonic fre-
quency components. Balanced frequency-multiplier configuration can be realized
with diodes as well as transistors.
17.6 Balanced Frequency Multipliers 551
17.6.1 Balanced Frequency Multipliers Using Diodes
A simple schematic circuit diagram of a diode-based balanced frequency multiplier

is shown in Fig. 17.23. Here, two diodes with antiparallel connection are series
connected to the transmission line. A low-pass filter (LPF) at the input side of the
diodes is used to allow the fundamental frequency component to the diodes, and to
prevent leaking the produced third order harmonic into the input side (i.e. source).
The third harmonic produced by the diodes will be passed through the bandpass filter
to reach the output load. The bandpass filter (BPF) will reject the fundamental as
well as other unwanted harmonic components and prevents to reach the output load
resistance.
The non-linear diode current can be written as
i d a0 + a1 vd + a2 vd2 + a3 vd3 + a4 vd4 + a5 vd5 + · · · (17.19)
Here, vd is the voltage across the diode from anode to cathode and i d is the diode
current also with the direction from anode to cathode. Therefore, the currents through
the diode as shown in Fig. 17.23 can be expressed as
i d1 a0 + a1 vd1 + a2 vd1
2 3
+ a3 vd1 4
+ a4 vd1 5
+ a5 vd1 + ··· (17.20a)
i d2 a0 + a1 vd2 + 2
a2 vd2 + 3
a3 vd2 + 4
a4 vd2 + 5
a5 vd2 + ··· (17.20b)
Here, the magnitude of the diode voltages are same but of opposite direction.
Thus, vd2 −vd1 vd , (say).
Therefore, the load current can be written as
i o i d1 − i d2 2a1 vd + 2a3 vd3 + 2a3 vd5 + · · · (17.21)
id1
LPF & BPF &

I/P M.N. O/P M.N.
io
id2
Fig. 17.23 Simple schematic circuit of a diode balanced frequency multiplier

LPF & LPF &

I/P M.N. O/P M.N.
Fig. 17.24 Schematic circuit of a shunt connected antiparallel diode balanced frequency multiplier
This shows that the output current of the antiparallel diode contains only odd
harmonics. The even harmonic components are not generated due to the balanced
configuration of the multiplier. This is also clear from the output current waveform
which has half-wave symmetry, thus contains only odd harmonic components. The
bandpass filter at the output allows only the required odd harmonic component and
will reject all other harmonic components. The output matching network transfer
the output impedance of the diodes to the load resistance corresponds to the desired
harmonic component.
Another scheme to design a balanced multiplier is shown in Fig. 17.24. Antipar-
allel diodes are shunt connected with the transmission line. From the same argument,
it can be shown that this configuration also will generate only odd harmonic com-
ponents. The input low-pass filter (LPF) allows the fundamental frequency signal to
the diodes. It also rejects the higher order harmonics. The input matching network
matches the source impedance to the input impedance of the diodes for maximum
signal transfer to the diodes. The bandpass filter (BPF) only passes the desired har-
monic and rejects the other harmonics.
17.6.2 Balanced Frequency Multiplier Using Transistors
Balanced frequency multipliers using transistors generally provide broadband per-

formance due to the improved input and output impedance matching over the broad
frequency range. The performance of balanced multipliers also becomes very stable
since the devices are terminated with 50 over the very wide frequency range. Bal-
anced frequency multipliers provide 3-dB more harmonic power than an equivalent
single-device multiplier circuit.
Schematic circuit diagram of a balanced frequency multiplier using 3-dB 90°
coupler is shown in Fig. 17.25. Here, resistors Rg , Rs and drain supply voltage VD
used for self-biasing the transistors. Values of the resistors are chosen to operate
the individual FETs in Class-B or C mode. The input coupler divides the signal
17.6 Balanced Frequency Multipliers 553
Fig. 17.25 A transistor (FET) based balanced multiplier
into two equal amplitudes with 90° phase shift before applied to the transistors. The
output coupler will introduce another 90° phase shift in between the two signals
before combining at the output port of load impedance Ro . Therefore, fundamental
frequency component and all the odd harmonics generated by the transistors will
have 180° phase shift in between the signals coming from the two paths at the output
load and cancelled each other. These odd harmonic powers will reach to the isolation
resistor Ro2 and dissipated. In this configuration, only the even order harmonics
coming from the two paths will reach the load termination with relative 0o phase
shift and added in power. Using proper bandpass filter at the output, required even
order harmonic frequency component is selected. Balanced multiplier circuit also can
be realized using a 180° rat-race hybrid coupler at the input to drive the transistors in
anti-phase and combining the outputs from both the transistors by using an in-phase
combiner or simply by parallel connection.
References
1. Gilmore R (1987) Concepts in the design of frequency multipliers. Microw J 129–139

2. Thomas DG, Branner GR (1996) Optimization of active microwave frequency multiplier
performance utilizing harmonic terminating impedances. IEEE Trans Microw Theory Tech
44(12):2617–2624
3. Cronin NJ, Law VJ (1985) Planar millimeter-wave diode mixer. IEEE Trans Microw Theory
Tech 33(9):827–830
4. Benson K, Frerking MA (1985) Theoretical efficiency for triplers using real varister diodes at
submillimeter wavelengths. In: IEEE MTT-S international microwave symposium digest, pp
315–318
5. East J, Kollberg E, Frerldng M (1993) Performance limitations of varactor multipliers. In:
Presented at the fourth international conference on space terahertz technology, Los Angles,
California, March 1993
6. Louhi J, Raisanen A (1994) On the modeling and optimization of varactor frequency multipli-
ers at submillimeter wavelengths. In: Presented at the fifth international conference on space
terahertz technology, Ann Arbor, Michigan, May 1994
7. Erickson NR, Tuovinen J, Rizzi BJ, Crowe TW (1994) A balanced doubler using a planar diode
array for 270 GHz. In: 5th International symposium on space terahertz technology, pp 409–413
8. East J (1996) High frequency limitations of diode frequency multipliers. In: Seventh interna-
tional symposium on space terahertz technology, Charlottesville, pp 116–124, March 1996
9. Zhao M, Fan Y, Wu D, Zhan J (2008) The investigation of W-band microstrip integrated high
order frequency multiplier based on the nonlinear model of avalanche diode. Prog Electromag-
netics Res 85:439–453
10. Gopinath A, Rankin JB (1982) Single-gate MESFET frequency doublers. IEEE Trans Microw
Theory Tech MTT-30(6):869–875
11. Rauscher C (1983) High-frequency doubler operation of GaAs field-effect transistors. IEEE
Trans Microw Theory Tech MTT-31(6):462–473
12. Borg M, Branner GR (1991) Novel MIC bipolar frequency doublers having high gain wide
bandwidth and good spectral performance. IEEE Trans Microw Theory Tech 39(12):1936–1946
13. Chen SW et al (1993) A high performance 94-GHz MMIC doubler. IEEE Microw Guided
Wave Lett 3(6):167–169
14. Augusto JS, Rosario MJ, Vaz JC, Freire JC (1993) Optimal design of MESFET frequency
multipliers. In: Proceedings of the 23rd European microwave conference, Madrid, pp 402–405,
September 1993
Chapter 18
Microwave Frequency Mixers
Abstract This chapter presents the working principle, design and analysis of various
types of microwave frequency mixers based on diodes and transistors for commu-
nication systems. Noise performance analysis and importance of image reject filters
in the design of mixers are discussed. Design details and analysis of single-ended,
single-balanced, double-balanced, image reject and subharmonic mixers are also
presented along with examples.
18.1 Introduction
Mixers are used in transmitters and receivers to translate signals in frequency domain.
It mixes two input signals to provide an output signal of frequency either sum and/or
difference frequency of the input signals. It is fundamentally a multiplier having three
ports: a radio frequency (RF) port, a local oscillator (LO) port and an intermediate
frequency (IF) port. Practically, the LO signal is CW, whereas the RF and IF signals
are modulated carriers. In the transmitter section, mixers are used for modulation
and upconversion to ease wireless transmission. In the receiver section, mixers
are used for downconversion and demodulation. In principle, any nonlinear device
can be used to design a mixer circuit. Mainly Schottky barrier diodes, GaAs FETs
and BJTs are used as nonlinear device for realization of microwave mixers [1–15].
It is also possible to use CMOS technology to realize microwave and millimetre
wave mixers. However, for high-performance applications at higher microwave and
mmwave frequency range, Schottky diode mixers are preferable [1, 2].
18.2 Working Principle of Frequency Mixers
Frequency mixers produce output signal of frequency equal to the sum or difference of
the frequencies of the two input signals, as shown in Fig. 18.1. Specifically, in case
of downconversion, the frequency (fOUT ) of the output signal, called intermediate

556 18 Microwave Frequency Mixers
Fig. 18.1 Conceptual block diagram of the frequency mixer
frequency (fIF ), is the difference of frequencies of the two input signals namely
input frequency (fIN ), called RF frequency (fRF ), and nth harmonic of local oscillator
frequency (nfLO ) and is given by
fOUT fIF |nfLO − fRF | (18.1)
Whereas in case of upconversion, frequency of the output signal (fOUT ), called RF

frequency (fRF ), is the sum of frequencies of the input signal (fIN ) called IF frequency
(fIF ) and nth harmonic of local oscillator frequency (nfLO ) and is given by
fOUT fRF nfLO + fIF (18.2)
Mixers of desired output with fundamental frequency component of the local oscil-
lator, i.e. for n 1, are known as fundamental mode mixers. Whereas, mixers having
desired output with harmonics of the local oscillator, i.e. for n 2 or 3 or 4 . . ., are
known as subharmonic mixers (SHM). In case of the subharmonic mixing, the LO
frequency is internally multiplied, thus producing mixing components from the RF
frequency and an integer multiple of the LO frequency. Practically, it is easier to
design a lower frequency LO with the required higher output power than the design
of higher frequency LO with same output power. In SHM, local oscillator of lower
frequency is required. The reduction of LO frequency can simplify the LO design,
and this makes the subharmonic mixing technique very attractive.
An important requirement for mixers is linearity in its RF performance. Thus, to
maintain the input–output linear relationship for the RF signal, the device of the mixer
needs to operate in nonlinear condition to produce frequency translation. For the
mixer to operate in the linear mode, LO power should be sufficient to drive the device
by a suitable sinusoidal signal in its saturated condition to generate nonlinearity and
the RF signal level should be small enough compared to the LO level to maintain
linearity.
A mixer consists of at least one device forced to operate in its nonlinear region to
generate the frequency translation. The nonlinearity of a device can be expressed by
the following nonlinear input–output relationship:
18.2 Working Principle of Frequency Mixers 557
vo a0 + a1 vi + a2 vi2 + a3 vi3 + . . . (18.3)
Here, a0 , a1 , a2 , a3 are the constants. Suppose the input signal (vi ) with two
different frequencies ωIN 2π fIN and ωLO 2π fLO is applied to the device. Thus,
vi vIN cos ωIN t + vLO cos ωLO t
Here, vIN and vLO are amplitude of the signals. Considering up to third-order
terms, the output will be given by
vo a0 + a1 (vIN cos ωIN t + vLO cos ωLO t) + a2 (vIN cos ωIN t + vLO cos ωLO t)2
+ a3 (vIN cos ωIN t + vLO cos ωLO t)3
2 2
a2 vIN a2 vLO
a0 + +
2 2
3 2
3a3 vIN 3a3 vIN vLO
+ a1 vIN + + cos ωIN t
4 2
3 2
3a3 vLO 3a3 vIN vLO
+ a1 vLO + + cos ωLO t
4 2
2 2
a2 vIN a2 vLO
+ cos 2ωIN t + cos 2ωLO t
2 2
+ a2 vIN vLO cos(ωIN + ωLO )t ⇒ desired for upconversion
+ a2 vIN vLO cos(ωIN − ωLO )t ⇒ desired for downconversion
3
a3 vIN vLO a3 vLO
− cos ω3IN t + cos 3ωLO t
4 4
2
3a3 vIN vLO
+ cos(2ωIN + ωLO )t
4
2
3a3 vIN vLO
+ cos(2ωLO + ωIN )t ⇒ desired 2 × subharmonic mixing
4
2
3a3 vIN vLO
+ cos(2ωIN − ωLO )t
4
2
3a3 vIN vLO
+ cos(2ωLO − ωIN )t ⇒ desired 2 × subharmonic mixing (18.4)
4
From (18.4), it is clear that a large number of spectral components are produced
by the device nonlinearity. Output frequencies are shown in Fig. 18.2 up to higher
order mixing considering higher LO signal level compared to the RF signal level.
For example, the squared term in (18.3) produces the sum and difference frequen-
cies, (ωIN + ωLO ) and (ωIN − ωLO ), respectively, which are desired for fundamental
frequency mixing as well as undesired double frequency terms, 2ωIN and 2ωLO .
The cubic term in (18.3) produces frequency components (2ωLO ± ωIN ), which are
desired for 2× subharmonic mixing in addition to the undesired frequency compo-
Fig. 18.2 Output spectrum of mixer
nents 3ωIN , 3ωLO , (2ωIN − ωLO ) and (2ωIN + ωLO ). Therefore, square term of the
nonlinearity produces fundamental mixing whereas higher order terms can generate
subharmonic mixing. It is also clear that in addition to the desired frequency compo-
nent, other undesired frequency components are also generated. In the linear region
operation corresponding to the RF signal level, the components corresponding to the
harmonic of RF frequency, e.g. 2ωIN , 3ωIN , (2ωIN ± ωLO ), etc., can be neglected due
to their negligible level compared to the desired component. Other undesired com-
ponents need to be eliminated by proper filtering and circuit topology. For example,
in case of fundamental frequency mixing, a bandpass filter may be used to get only
(ωIN − ωLO ) as shown in Fig. 18.2. Most practical mixers use filters to reduce the
levels of the unwanted spurious outputs. Another commonly used technique, for the
reduction of some unwanted spurious, is the use of balanced mixer topology.
18.2.1 Image Frequency in Mixer
For the design of frequency mixer, special care shall be taken for the frequency band
which is the image of the desired RF band with respect to the LO frequency [13]. The
image frequency is twice the IF frequency away from the RF and will be converted
directly to the same IF frequency as the RF as shown in Fig. 18.3. Practically, it is
required to translate/detect a particular band of frequency. Thus, noise and unwanted
signals present at this frequency band can severely degrade the system performance
due to the interference with the desired signal. One way to get rid of from the image
frequency is the use of a preselector filter preceding the mixer input to reject the
image frequency band allowing the required frequency to the mixer input. Another
way is the use of mixer topology with image reject property.
In the case of downconverting fundamental mixers, IF frequency is given by
Fig. 18.3 Image spectrum for fundamental and 2× subharmonic mixing
fIF fRF − fLO (18.5a)
Thus the image frequency is given by
fIMAGE fRF − 2fIF (18.5b)

fIMAGE |2fLO − fRF | (18.5c)
Similarly for 2× subharmonic mixer, the image frequency will be given by
fIMAGE |4fLO − fRF | (18.5d)
Practically, desired RF frequency is a band of frequency around the centre fre-

quency. Correspondingly, the image frequency will also be a band of frequency. The
passband of the preselect filter should have proper passband performance without
introducing any distortion in the desired RF frequency band and very high rejection
over the entire image frequency band.
Example 18.1 The receive frequency band for mobile satellite communication sys-
tem is 6450–6485 MHz. For processing the spectrum, it is translated to IF band of
50–85 MHz using a local oscillator of frequency 6400 MHz. Find out the image
frequency band.
Solution
The image frequency band can be determined from (18.5b) as
fIMAGE fRF − 2fIF

(6450 to 6485) MHz − 2 × (50 to 85) MHz
(6350 to 6315) MHz
18.2.2 Conversion Loss of Frequency Mixer
The conversion gain of mixers is an important parameter. It is defined as the ratio

of output power (POUT ) of the desired output frequency to the input power (PIN ) of
the input frequency as shown in Fig. 18.1. Basic scheme of an idealistic mixer is
shown in Fig. 18.4. At the RF (input) and LO sides, a balun is necessary in order to
convert unbalanced inputs into balanced signals. Conversely, no adapter is needed at
the IF output, which is already unbalanced. In low-frequency mixers, the baluns may
be implemented with small transformers. However, at higher frequencies, microstrip
networks may be the preferred option for realization of baluns. The basic assumptions
are that the LO signal vLO (t) is large enough so that the device (e.g. Schottky diodes
or FETs) acts as an ideal commutator under its control, and the input RF signal vIN (t)
is very small compared to it. Thus, a mixer is almost always operated with the LO
input saturated at the nominal operating condition, whereas the input (RF) power
is usually well below the saturation level to get linear RF performance. During the
positive half-cycle of vLO (t), one switch will be ON and other switch will remain
OFF. During the other half-cycle, the roles are interchanged. Therefore, the output
signal vOUT (t) switches between +vIN (t) and −vIN (t) depending on the sign of vLO (t).
This is equivalent to multiplying vIN (t) by a square wave of amplitude ±1 that takes
the sign of vLO (t). Considering matched RF and IF ports and resistively termination
of all the intermodulation (IM) products and no parasitic resistive or reactive losses,
the output voltage can be represented mathematically by multiplying the input signal
voltage with the Fourier series for the square wave function as
Fig. 18.4 Idealistic mixer configuration

⎡ ⎤
4
vOUT (t) vIN sin(ωIN t)⎣ sin(nωLO t)⎦ (18.6)
n1,3,5,...
π n
The terms within the square bracket are the Fourier series representation of square
wave function (from Fig. 17.2, putting A 2). Thus, inside the mixer, the local
oscillator signal can be considered as the sum of the fundamental and all its odd
harmonic components with amplitudes (4/nπ ), where n 1, 3, 5, . . . ∞. Using
trigonometry identity,
2vIN
vOUT (t) [sin(ωLO − ωIN )t − sin(ωLO + ωIN )t]
π
2vIN
+ [sin(3ωLO − ωIN )t − sin(3ωLO + ωIN )t]
3π
2vIN
+ [sin(5ωLO − ωIN )t − sin(5ωLO + ωIN )t]
5π
+ ... (18.7)
Output of the mixer consists of two sidebands corresponding to the fundamental

and all odd harmonics of the LO frequency. The signal amplitudes of each sideband
of all the products are
2vIN
, n 1, 3, 5, . . . , ∞
nπ
Therefore, output signal power of each sideband is given by
2
1 4vIN
POUTn , n 1, 3, 5, . . . , ∞ (18.8)
Ro n2 π 2
Thus, the total output power is the sum of both the sidebands of all the harmonics
power:
2
1 4vIN 2
vIN
POUT (Total) 2 × (18.9)
R n2 π 2
n1,3,...∞ o
Ro
This is nothing but equal to the input RF power. Thus, the total power is conserved.
The conversion loss is the reciprocal of the conversion gain and can be written as
PIN
Conversion Loss (18.10)
POUT
Here PIN is the power of the RF input, and POUT is the power of the selected
frequency at the output. For the fundamental frequency mixer, one of the sideband
will be selected. The conversion loss of the fundamental mixer corresponds to one
sideband is
2
vIN 2
1 4vIN π2
Conversion Loss / (18.11)
Ro Ro π 2 4
π 2
Conversion Loss(in dB) 10 log dB 3.92 dB (18.12)
4
This is the minimum loss of the single-sideband corresponds to output frequency
of (ωLO − ωIN ) or (ωLO + ωIN ).
The above analysis shows that when matched loads are presented to each IM
product, and the RF, IF, and image signals are matched terminated, then the theoretical
minimum conversion loss of a fundamental sideband is 3.92 dB. The theoretical
conversion loss is 3 dB when all the IM products are reactively terminated and the
IF and RF ports are matched terminated and also the image signals are properly
terminated. In this case, the input RF power is equally divided in between the image
band and required IF band. Thus, it can be concluded that the theoretical limit for
the conversion loss is 0 dB in case all the IM products are reactively terminated and
the image signals are short or open circuited. However, practically conversion loss
has three major components: (a) mismatch loss at RF and IF ports, (b) loss due to
the finite value of diode spreading and junction resistance, and (c) loss due to the
generation of sideband and IM products.
18.2.3 Noise Performance of Frequency Mixer
Noise figure of a mixer is also defined as the signal-to-noise power ratio (SNR) at
the input to the SNR at the output. All practical mixers fold the RF spectrum around
the local oscillator frequency [1–7]. Thus, it creates an output which contains the
summation of the spectrum on both sides since fIF |fRF − fLO |. Thus, a mixer
(down converter) converts both the desired and the image bands to the desired output
frequency (IF) band. This implies folding the noise present at the image frequency on
top of the desired output frequency as shown in Fig. 18.5. Therefore, the overall noise
at the translated output frequency band is (a) the noise at desired input frequency band
converted to desired output frequency band, (b) the noise at input image frequency
band converted to desired output frequency band and (c) the noise added by the mixer
circuit itself.
In case of heterodyne receivers, only one side frequency band is used thus, in this
case, one of these sides acts as spurious, i.e. noise and the other side as desired signal.
Therefore, image reject filtering (in practice a bandpass filter) as shown in Fig. 18.6
or image cancelling mixer circuit topology is used to remove one of these responses.
However, in case of direct-conversion receivers, both the sidebands, i.e. above and
below fIF fLO are converted and utilized for the intended signal. Based on the
applications and circuitry, noise figures (NF) of the mixer can be categorized as (a)
single-sideband noise figure without image rejection mechanism (b) single-sideband
noise figure with image rejection mechanism and (c) double-sideband noise figure.
Image noise
Signal noise
Output
Signal
Input
Signal Ideal
noise-less
mixer
Fig. 18.5 Contribution of image frequency noise to IF band
Image reject
filter
Fig. 18.6 Block diagram of mixer with image reject filter
In case of single-sideband NF without image frequency band rejection mechanism,

it is considered that there is no signal at the image frequency. Thus, the image
frequency band contributes to noise power. Suppose, over the desired frequency
band, the input signal power and gain of the mixer are SIN and GIN . Then the output
signal power over the desired frequency band is
SOUT SIN × GIN (18.13)
The total output noise power over the desired single-sideband is
NOUT NIN × GIN + NIM × GIM + NMIXIN × GIN + NMIXIM × GIM (18.14)
Here, NIN and NIM are the input noise in the desired frequency band and image
frequency band, respectively. NMIXIN , NMIXIM are the noise power generated by the
mixer at the desired and image frequency band, respectively, referred to the input of
the mixer. And GIM is the gain of the mixer at the image frequency band.
Therefore, ratio of the output signal power to the output noise power is
SOUT SIN × GIN

(18.15)
NOUT NIN × GIN + NIM × GIM + NMIXIN × GIN + NMIXIM × GIM
Assuming GIN GIM , NMIXIN NMIXIM , NIN NIM , the output signal-to-noise
ratio can be written as
SOUT SIN

(18.16)
NOUT 2 × NIN + NMIXIN
Therefore, the single-sideband noise factor without image frequency band rejec-
tion FSSB1 can be written as

SIN SOUT NMIXIN
FSSB1 / 2 1+ (18.17)
NIN NOUT NIN
In case of single-sideband NF, with image frequency band rejection mechanism

(e.g. using image reject filter as in Fig. 18.6), it can be assumed that there is no signal
power at the image frequency and no contribution of the image band noise at the
output of the mixer. Then, the output signal power over the desired frequency band
is
NOUT NIN × GIN + NMIXIN × GIN + NMIXIM × GIM (18.18)
SOUT SIN × GIN
(18.19)
NOUT NIN × GIN + NMIXIN × GIN + NMIXIM × GIM
Assuming GIN GIM , NMIXIN NMIXIM , the output signal-to-noise ratio can be
written as
SOUT SIN
(18.20)
NOUT NIN + 2 × NMIXIN
Therefore, the single-sideband noise factor with image frequency band rejection
FSSB2 can be written as
SIN SOUT 2 × NMIXIN
FSSB2 / 1+ (18.21)
NIN NOUT NIN
Therefore, in case of SSB signal, the use of image reject filter improves the noise
figure by about 3-dB when contribution of noise by the mixer itself is less.
In case of double-sideband NF, it is considered that the image band contains both
noise and an image signal which is identical to the desired input signal. Here, the
image frequency band contributes to noise NIM as well as signal power SIM . Thus,
the output signal power is
SOUT SIN × GIN + SIM × GIN (18.22)
The total output noise power over the desired single-sideband is
NOUT NIN × GIN + ×GIM + NMIXIN × GIN + NMIXIM × GIM (18.23)
SOUT SIN × GIN + SIM × GIN
(18.24)
NOUT NIN × GIN + NIM × GIM + NMIXIN × GIN + NMIXIM × GIM
Assuming GIN GIM , NMIXIN NMIXIM , NIN NIM , the output signal-to-noise
ratio can be written as
SOUT SIN

(18.25)
NOUT NIN + NMIXIN
Therefore, the double-sideband noise factor without image frequency band rejec-
tion FDSB can be written as
SIN SOUT NMIXIN
FDSB / 1+ (18.26)
NIN NOUT NIN
It is clear from (18.17) and (18.26) that the single-sideband noise factor FSSB1
without any image frequency mechanism is exactly double of the double-sideband
noise factor FDSB , i.e. 3 dB higher. However, the difference of noise figure in between
single-sideband noise factor with and without image frequency mechanism is not
exactly 3-dB, actually the difference approached 3-dB as the mixer noise contribution
of mixer itself is very low.
18.3 Single-Ended Mixers
Schottky barrier diodes are used as nonlinear device for the microwave as well as
mmwave mixers because of its low forward threshold voltage and fast switching
capability [1, 2]. Mostly unbiased diodes are used for mixers. However, sometimes
small forward current bias is given to the diodes to achieve lower conversion loss.
This is necessary when limited LO drive level is available. The diode is forward
biased to have a quiescent operating point close to the region of maximum nonlinear
characteristic, as shown in Fig. 18.7a, to generate maximum possible distortion by
lower levels of the LO drive. The equivalent circuit of a Schottky diode is shown
in Fig. 18.7b. Here, RS is the parasitic resistance and Rd and Cd are the nonlinear
junction resistance and capacitance of the Schottky diode, respectively. Under posi-
tive half-cycle of LO signal of sufficient level, the diode will be forward biased and
(a) (b) (c)

Current Current
Rs
Maximum
nonlinear Rd Cd
region
Voltage Voltage
Fig. 18.7 I-V characteristic of (a) Schottky diode and (b) its equivalent circuit (c) idealistic com-
mutator characteristic
thus the signal will experience very small impedance and during its other half-cycle
diode resistance will be very high thus it will experience very high diode impedance.
An idealized diode can be considered as ideal commutator which switches transcon-
ductance instantaneously as shown in Fig. 18.7c and yields the theoretically optimal
diode mixer performance.
Single-ended mixers consist of single device and are unbalanced, thus, in general,
there are no inherent isolations in between LO and RF ports. These mixers do not have
any intrinsic property to eliminate any undesired frequency components generated
by the mixing process. Diplexer may be used to combine LO and RF signal to provide
the isolation before feeding to the mixer device. However, it is difficult to do this
when the LO and RF frequencies are very close.
Basic block schematic of a single-ended mixer using single diode is shown in
Fig. 18.8. Here, the LO, RF and IF ports are electrically same, and they are only
separated by filters to prove port to port isolations. The required bandwidths of the
filters must be sufficiently apart to provide high isolation. An LO signal of large
amplitude compared to the RF signal level is combined at the anode of the diode
after passing through respective bandpass filters. For idealistic operation, stronger
LO signal needs to be applied so that the LO signal level has the capability to switch
the transconductance of the diode in between high and low values. Level of the RF
signal is low enough compared to the LO level to achieve linear RF performance.
Practically no Schottky diode behaves as an ideal commutator. It has finite turn ON
transition as shown in Fig. 18.7a. In addition to this, though the RF signal level is
small compared to LO level, it will also modulate the diode transconductance to
some extent, leading to nonlinearity in RF signals.
Suppose, RF and LO signal level vIN and vLO , with frequencies ωIN 2π fIN and
ωLO 2π fLO , respectively. Then the diode voltage
vd vIN cos ωIN t + vLO cos ωLO t

18.3 Single-Ended Mixers 567
RF IF
input output
The diode current can be written as

vd
id Io exp −1
Vt

vd 1 vd 2 1 vd 3 1 vd 4
Io + + + + ... (18.27)
Vt 2! Vt 3! Vt 4! Vt
Here, Io is the reverse saturation current and Vt is the diode thermal voltage.
From (18.27), it is clear that exponential characteristic of the diode produces large
number of spectral components due to the presence of square, cubic and other higher
order terms. Thus, practical mixer circuit will produce all possible harmonic mixing
components. Since, the single-ended mixer topology does not have any inherent
spurious suppression capability, its output contain all harmonic components. Proper
filtering at the output of the mixer is used to filter out the desired frequency band. This
is the reason for unpopularity of single-ended mixer topology for modern mixers.
However, in case of higher mmwave frequency design, single-ended mixer circuits
are realized for their simplicity. Microstrip layout of a typical single-ended diode
mixer is shown in Fig. 18.9.
Transistors are used for active as well as passive mixers [9, 14, 15]. Transistors are
operated as switch for passive mixers. Drain–source resistance of a FET behaves as
voltage-variable RF resistance. A simple schematic circuit diagram of a FET passive
mixer is shown in Fig. 18.10. Here, the FET is operated at zero drain and source bias
condition. LO signal acts as controlling the FET switch. The RF signals are applied at
the drain terminal after passing through a bandpass filter. The IF signals are taken from
the source terminal. Capacitor Co at the drain terminal is used to provide a short circuit
to the RF frequency. At IF port, an LPF is used to filter out the undesired frequency
components which are generated due to the mixing of LO and RF components. Large
enough LO signal is required to turn the FET OFF on the negative cycle in case of
zero gate bias. Thus, it will drive the gate–source junction into forward bias on its
Fig. 18.9 Microstrip circuit layout of a simple single-ended diode mixer
FET acts
as switch
RF IN IF OUT
ID
Rg
Co
VGO
LO IN VG VGO
Fig. 18.10 Simple schematic circuit of passive mixer using FET as switch
positive half-cycle, which degrades the device reliability. To operate the FET as a
switch by the LO signal without reliability problem, proper gate biasing is required.
Therefore, it is necessary to set the gate bias voltage appropriately to operate the
FET in switching mode keeping forward gate current within the allowable limit and
without crossing the reverse gate breakdown voltage.
In case of active FET mixers, the nonlinear transconductance of the transistor is
used for mixing the RF signals with LO signal. The LO signal is used to vary the
transconductance of the transistor. Active FET mixers provide conversion gain rather
than loss and also have lower noise figures than passive mixers. A simple circuit dia-
gram of an active FET mixer is shown in Fig. 18.11. Gate and drain terminals of
the FET are biased through λRF /4 and λIF /4 transmission lines, respectively. Output
capacitor Co is used to ensure the short circuit for RF and LO signal at the drain ter-
18.3 Single-Ended Mixers 569
VG VD
λ
λ Cb
LO IN
IF OUT
RF IN Cb Co
Fig. 18.11 A simple schematic circuit of a FET active mixer
Dual Gate
FET IF OUT
LO IN
Co
RF IN
minal of the FET to keep nearly constant drain bias with and without LO signal which
maximizes time-varying transconductance for achieving maximum conversion gain.
The difficulty of the circuit of Fig. 18.11 is the isolation in between RF and LO ports.
Here, some form of diplexing of the RF and LO signal is required to separate the RF
and LO inputs which are incident on the single gate terminal. To avoid this problem,
dual-gate FET is used to realize mixer circuit as shown in Fig. 18.12. Dual-gate FET
consists of two gate terminals to control its channel properties. Transconductance of
the FET can be controlled independently by applying signal in both the gate termi-
nals. Here, RF and LO signals are applied to the two separate gate terminals. Thus,
the dual-gate FET structure provides isolation in between RF and LO for its inherent
characteristics. Very compact active mixer circuit is designed using the dual-gate
FET structure with the disadvantage of lower linearity compared to passive mixers.
18.4 Balanced Mixers
The mixer topologies so far discussed consist of a single device and are unbalanced.
Thus, it is very difficult to achieve isolation in between LO and RF ports. It is not
possible to design single-ended frequency mixers with isolation among RF, LO and
IF ports, resulting in undesired signal interactions and one signal reaches to another
port. Moreover, such circuitry does not have any intrinsic property to eliminate any
undesired frequency components generated by the mixing process. Diplexer may be
used to combine LO and RF signal to provide the isolation before feeding to the
mixer device. However, it is difficult to do this when the LO and RF frequencies are
very close. To provide isolations among LO, RF and IF ports and also to eliminate
some unwanted mixing products from the output of a mixer, it is required to balance
the signals. Balancing is out-phasing the signal to create reactive cancellation. The
reactive cancellation is accomplished with a channelization technique using phase-
correlating power dividers. The cancellation of undesired mixing products at the
output of a mixer is determined by its balancing. At the RF and LO ports, a net-
work called balun is necessary in order to convert the unbalanced inputs into the
balanced signals. However, no such type of balun is required at the IF output, which
we need unbalanced. Single-balanced and double-balanced mixers are popularly
used which are classified depending on their configuration. Single-balanced mixers
strongly attenuate either the input signal or the LO signal at the output port, but not
both the signals simultaneously. Thus, less numbers of the above mixing products
go to its output compared to single-ended mixers. A double-balanced mixer exhibits
superior isolation among its LO, RF and IF ports. Here, only the sum and difference
frequencies of the input and LO frequencies go to the output, while attenuating both
the LO and RF signals, and provide significant rejection to three-quarters of the pos-
sible mixer products at the output. Thus, the balancing of the mixers minimizes the
requirement of filtering out the undesired mixing products.
The balanced mixer uses more number of devices (diodes, transistors) thus, the
requirement of LO power increases [8, 10, 11, 14, 15]. This increases the linearity of
a balanced mixer due to the sharing of the power among the more number of devices
working in the nonlinear region. A higher barrier height diode or more numbers of
diodes in one arm may be used to improve linearity further. This enables higher drive
level without overcoming the maximum limit of individual devices.
18.4.1 Single-Balanced Mixers
Single-balanced mixer uses two devices. It provides isolation in between one pair
of signals. Here, either the LO signal or the RF signal is balanced. Any one of
the signals added destructively at the IF port of the mixer and is rejected inherently.
The level of rejection depends on the amplitude and phase balance of the balun,
18.4 Balanced Mixers 571
4 1
LO input RF Input
Balun BPF
5
LPF
3 2
IF Output
Fig. 18.13 Schematic circuit of a single-balanced diode mixer using balun
and the matching between the two devices. Depending on the configuration, single-
balanced mixer rejects certain unwanted mixing products at the IF port.
Schematic circuit diagram of a single-balanced mixer shown in Fig. 18.13 consists
of one balun two matched diodes. The balun converts the unbalanced LO signal to a
balanced signal before feeding to the diodes. A balanced balun has characteristic to
maintain phase angles of ±90° at port-1, ±90° at port-2 with respect to ground and 0°
at port-5. Other important parameters of a balun are insertion loss and output-to-input
impedance ratios. The single-balanced mixers have better LO-to-IF and LO-to-RF
isolations due to the balanced characteristic of the balun and matched diode pair.
Matched diode pair means the DC and RF characteristics of both the diodes are
same. Due to the matched property of the diodes, a virtual ground is created between
the diodes, which makes a null for the LO signal, thus providing isolation to the RF
and IF ports. Thus, single balance mixers cancel the LO power which thus prevents
LO leakage to the RF and IF ports. Thus, single-balanced mixers provide superior
LO-to-IF and LO-to-RF isolation compared to single-ended mixers. The two filters
BPF for RF port and LPF for IF port are used to provide isolations in between RF
and IF ports. Thus, these filters are used not only to pass the required frequencies;
they also provide rejections to unwanted mixing components.
Schematic circuit diagram of a single-balanced mixer using 180° hybrid is shown
in Fig. 18.14. Microstrip layout diagram of a rat-race 180° hybrid is shown in the
figure. Here, LO signal is applied at port-4 and RF signal is applied at port-3 of
the hybrid. Thus, the LO is applied 180° out of phase and RF is applied in phase
to the two diodes. A low-pass filter at the output passes only the IF frequency and
rejects other frequency components, also it provides a short circuit to the LO and RF
frequencies to isolate IF from LO and RF. An λo /4 open-circuit stub may also be
used as shown to provide a short circuit to the RF signal at the IF port.
Suppose, the amplitudes and frequency of the LO and RF signals are AL , AR and
ωL , ωR , respectively. Thus, the resultant voltages vd 1 and vd 2 at port-1 and port-2,
respectively, can be expressed as (eliminating common phase shifts):
AL AR
vd 1 √ cos ωL t + √ cos ωR t (18.28a)
2 2
Fig. 18.14 Schematic circuit of a single-balanced diode mixer using 180° hybrid
Fig. 18.15 Circuit schematic of a single-balanced mixer with 90° hybrid
AL AR
vd 2 − √ cos ωL t + √ cos ωR t (18.28b)
2 2
The diode I-V (id − vd ) characteristic from (18.27) can be expressed as
id a1 vd + a2 vd2 + a3 vd3 + a4 vd4 + . . . (18.28c)
Thus, the diode currents id 1 and id 2 can be written as

AL AR
id 1 id (vd 1 ) a1 √ cos ωL t + √ cos ωR t
2 2
2
AL AR
+ a2 √ cos ωL t + √ cos ωR t
2 2
3
AL AR
+ a3 √ cos ωL t + √ cos ωR t
2 2
4
AL AR
+ a4 √ cos ωL t + √ cos ωR t + ···
2 2
Whereas

AL AR
id 2 id (−vd 2 ) −a1 − √ cos ωL t + √ cos ωR t
2 2
2
AL AR
+ a2 − √ cos ωL t + √ cos ωR t
2 2
3
AL AR
− a3 − √ cos ωL t + √ cos ωR t
2 2
4
AL AR
+ a4 − √ cos ωL t + √ cos ωR t + · · ·
2 2
Therefore, the IF current iIF id 1 − id 2 can be written as

2
AL AR AL AR
iIF a1 √ cos ωL t + √ cos ωR t + a2 √ cos ωL t + √ cos ωR t
2 2 2 2
3 4
AL AR AL AR
+ a3 √ cos ωL t + √ cos ωR t + a4 √ cos ωL t + √ cos ωR t + · · ·
2 2 2 2
2
AL AR AL AR
+ a1 − √ cos ωL t + √ cos ωR t − a2 − √ cos ωL t + √ cos ωR t
2 2 2 2
3 4
AL AR AL AR
+ a3 − √ cos ωL t + √ cos ωR t − a4 − √ cos ωL t + √ cos ωR t + · · ·
2 2 2 2
√
2a1 AR cos ωR t + 2a2 AL AR cos ωL t cos ωR t
√ √
+ 1.5 2a3 A2L AR cos2 ωL t cos ωR t + 0.2 2a3 A3R cos3 ωR t
+ 2a4 A3L AR cos3 ωL t cos ωR t + 2a4 AL A3R cos ωL t cos3 ωR t
+ ··· (18.28d)
This shows that there is no fundamental component of LO frequency at the

output. Thus, due to the balance configuration, there is no LO leakage to the IF
port. Also, there are no even-order harmonic components of the RF frequency. Thus,
the requirement of the rejection of undesired components at the diode output is lesser
than the requirement in case of single-ended mixers. Moreover, considering very less
RF level signal compared to the level of the LO, higher order RF components will
have less contribution to the overall harmonic power level.
In the 3-dB 90° hybrid version single-balanced mixer, the RF and LO signals are
applied to the mutually isolated two ports as shown in Fig. 18.15. Here, RF and LO
signals are applied to port-1 and port-2 which are mutually isolated over its designed
frequency band. However, due to the 90° phase difference of the hybrid, there will
be no cancellation of the LO and RF as well as intermodulation components at the
IF port. To provide a short circuit to the LO signal at the IF port, a λo /4 open-circuit
LO IN RF IN
D4 D1
IF
Balun1 Balun2
D3 D2
Fig. 18.16 Schematic circuit of a double-balanced diode mixer
stub is used. In case of RF frequency near to the LO frequency, this stub will also
provide a low impedance to the RF signal. Thus, IF will also be isolated from the
RF and LO signals. The low-pass filter at the IF port is used to pass only the desired
IF frequency component to the output eliminating the generated undesired mixing
components.
18.4.2 Double-Balanced Mixers
Single-balanced mixers can provide inherent isolation among either LO to IF or RF

to IF but not both simultaneously. Double-balanced mixers can be considered as
they consist of two single-balanced mixers with different configurations [8, 10, 11,
16–18]. Schematic configuration of a double-balanced mixer is shown in Fig. 18.16.
The configuration consists of two baluns with four devices (here diodes). Here,
LO port isolation from the RF and IF ports is achieved in the same way as it is
achieved in single-balanced mixers due to the virtual ground in between the matched
diode pairs. The IF signal is taken from the RF balun instead of from the LO balun
to achieve better isolation in between LO and IF ports. The isolation between the
RF and the IF ports depends on the balancing of the balun. Theoretically, double-
balanced mixers generate only one-fourth of the possible intermixing products, which
consist of only odd harmonics of LO and odd harmonics of RF frequencies. The
other mixing products suppressed due to the balanced configurations. The degree of
suppression depends on the amount of balance of the baluns, as well as matching
among the diodes. However, practically, the suppression of even-even products will
be more than the even-odd and odd-even products. Double-balanced mixers consist
of four diodes, thus LO drive will be more than the single-ended as well as single-
balanced mixers. Therefore, the upper limit of the RF level will be higher for the
same amount of IM3 level, i.e. its 1-dB compression point will be higher than the
other two configurations.
mixer-1
RF
LPF LSB
IN
o o
3-dB 90 3-dB 90
Hybrid LO IN Hybrid
LPF USB
mixer-2
Fig. 18.17 Circuit schematic of an image rejection mixer
18.4.3 Image Rejection Mixers
Image rejection mixers (IRM) are used as a converter which isolated lower sideband
frequency spectrum from upper sideband spectrum [13]. Schematic diagram of an
IRM using two 3-dB 90o couplers is shown in Fig. 18.17. Suppose, the LO and IF
frequencies are ωLO and ωIF , respectively. The two sideband RF frequencies can be
represented as
vRF vL cos(ωLO − ωIF )t + vU cos(ωLO + ωIF )t (18.29)
Here, vL and vU are the amplitudes of lower and upper RF sidebands. Suppose
the LO signal is vLO cos ωLO t.
The inputs of the mixer-1 and mixer-2 respectively can be written as
vL vU
vim1 √ cos(ωLO − ωIF )t + √ cos(ωLO + ωIF )t (18.30a)
2 2
and
vL vU
vim2 √ sin(ωLO − ωIF )t + √ sin(ωLO + ωIF )t (18.30b)
2 2
Considering only the second-order term of the mixers with multiplication factor
km , the output of the mixer-1 can be written as

vL vU
vom1 km √ cos(ωLO − ωIF )t + √ cos(ωLO + ωIF )t vLO cos ωLO t
2 2
km vL vLO km vU vLO
√ cos(2ωLO − ωIF )t + √ cos(2ωLO + ωIF )t
2 2 2 2
km vL vLO km vU vLO
+ √ cos ωIF t + √ cos ωIF t
2 2 2 2
The first two terms will be eliminated by the LPF connected at the output of the
mixer. Thus, the LPF output will be
km vL vLO km vU vLO
vof 1 √ cos ωIF t + √ cos ωIF t (18.31a)
2 2 2 2
In the same way, the output of the second mixer after LPF will be
km vL vLO km vU vLO
vof 2 − √ sin ωIF t + √ sin ωIF t (18.31b)
2 2 2 2
Thus, one of the outputs of the output hybrid will be

km vL vLO km vL vLO
vo1 cos ωIF t + cos ωIF t
4 4
km vL vLO
cos ωIF t (18.32a)
2
From (18.32a), it is clear that the output vo1 contains the IF component which
corresponds to only lower sideband of the RF signal.
Another output of the second hybrid will be
km vU vLO km vU vLO
vo2 sin ωIF t + sin ωIF t
4 4
km vU vLO
sin ωIF t (18.32b)
2
From (18.32b), it is clear that the output vo2 contains the IF component which
corresponds to only upper sideband of the RF signal.
Thus, the image rejection mixer separates lower and upper sideband RF frequency
components into its two output ports. Depending upon applications, the IF frequency
band corresponding to either the lower sideband (ωLO − ωIF ) or the upper sideband
(ωLO + ωIF ) can be selected for use.
18.5 Subharmonic Mixers
The desired output of a subharmonic mixer (SHM) is a mixing component of RF

frequency and an integer multiple (2, 3, …) of LO frequency. In SHM, the LO
frequency is internally multiplied and mixes with the RF frequency to produce the
desired output frequency [6, 7, 12, 19–21]. Thus, the requirement of frequency of the
LO is lower. The reduction of LO frequency simplifies the LO design, particularly at
high frequencies where it is difficult to design an LO with the required output power.
Therefore, the subharmonic mixer is advantageous and is more attractive at higher
microwave and mmwave frequency range. With reference to Fig. 18.1, the output
frequency of SHM is given by
18.5 Subharmonic Mixers 577
Filter &
matching Filter &
matching
Filter &
matching
Fig. 18.18 Scheme of subharmonic mixer using an antiparallel diode pair
fOUT |fIN ± nfLO | (18.33)
Here, n is the order of the subharmonic mixer. For 4× subharmonic mixer, n 4,

which requires LO with one-fourth the frequency that would be required with a
fundamental frequency mixer.
As the subharmonic mixers are more advantageous at higher microwave and
mmwave frequency range, mostly Schottky barrier diodes are used as nonlinear
device to realize the circuits as it is difficult to realize transistor-based mixer circuits
at such high frequencies. Schematic circuit of a subharmonic mixer using antiparallel
diode pair is shown in Fig. 18.18. Here, sum of LO and RF signals is applied to the
diodes and can be written as
vd vRF cos ωRF t + vLO cos ωLO t (18.34a)
Considering the individual diode current represented by
id (vd ) a1 vd + a2 vd2 + a3 vd3 + a4 vd4 + · · · (18.34b)
The total current can be expressed as
iT id 1 (vd ) − id 2 (−vd )
a1 (vRF cos ωRF t + vLO cos ωLO t) + a1 (vRF cos ωRF t + vLO cos ωLO t)
+ a2 (vRF cos ωRF t + vLO cos ωLO t)2 − a2 (vRF cos ωRF t + vLO cos ωLO t)2
+ a3 (vRF cos ωRF t + vLO cos ωLO t)3 + a3 (vRF cos ωRF t + vLO cos ωLO t)3
+ a4 (vRF cos ωRF t + vLO cos ωLO t)4 − a4 (vRF cos ωRF t + vLO cos ωLO t)4
+ a5 (vRF cos ωRF t + vLO cos ωLO t)5 + a5 (vRF cos ωRF t + vLO cos ωLO t)5
+ ···
2a1 (vRF cos ωRF t + vLO cos ωLO t) + 2a3 (vRF cos ωRF t + vLO cos ωLO t)3
+ 2a5 (vRF cos ωRF t + vLO cos ωLO t)5 + · · · (18.35)
This shows that there are no even-order products; it only generates odd-order
frequency mixing components. Generally, the amplitude of LO signal is much higher
than the amplitude of the RF signal. Therefore, using the condition vLO vRF , the
higher order terms of RF components can be neglected compared to the fundamental
component of RF signal level and higher order LO signal levels. Thus, the total
current can be written as
iT 2a1 vRF cos ωRF t + 2a1 vLO cos ωLO t + 6a3 vRF vLO
2
cos ωRF t cos2 ωLO t
3
+ 2a3 vLO cos3 ωLO t + 20a5 vRF
2 3
vLO cos2 ωRF t cos3 ωLO t
4
+ 10a5 vRF vLO cos ωRF t cos4 ωLO t + 2vLO
5
cos5 ωLO t + · · · (18.36)
The output current contains (2ωLO − ωRF ), (2ωLO + ωRF ), (4ωLO − ωRF ),
(4ωLO + ωRF ) and other mixing products including fundamental RF and LO fre-
quency components. This shows that the antiparallel diode pair is capable to generate
mixing products which are required for subharmonic mixer. Though the circuit struc-
ture inherently eliminates the even-order intermodulation products, the fundamental
RF and LO signals will pass to the output. Therefore, an output filter is required
to pass only the desired subharmonic mixing products eliminating other undesired
mixing products.
Example 18.2 A system consists of a low-noise amplifier (LNA) cascaded with a
mixer without any image rejection filter in between them. The gain of the LNA at its
desired and image frequency bands are 30 and 32 dB, respectively. The noise figures
of the LNA at its desired and image frequency bands are 2.0 and 3.0 dB, respectively.
Calculate the noise figure of the system at its desired frequency band considering
ideal noise-less mixer of 0 dB insertion loss.
Solution
The system with low-noise amplifier (LNA) and mixer without any image frequency
band rejection filter is shown in Fig. 18.19.
This is the case of single-sideband signal. Thus, there is no signal at the image
frequency band. Without image frequency band rejection mechanism, the noise of
the image frequency band contributes to total output noise at the desired frequency
band.
Suppose, over the desired frequency band the input signal power, gain and noise
figure of the LNA are SIN , GIN and NFIN , respectively. Also suppose, over the image
frequency band, the gain and noise figure of the LNA are GIM and NFIM , respectively.
Then, the signal power at the output of LNA over the desired frequency band is
SOUT SIN × GIN (18.37)
The mixer is lossless. Thus, signal power at the output of the mixer is the same
SOUT_MIXER SIN × GIN (18.38)

18.5 Subharmonic Mixers 579
Image noise
Signal noise
Signal Signal
Signal Ideal
noise-less
mixer
Fig. 18.19 Contribution of image frequency noise to IF band in the absence of image frequency
band reject filter
Suppose, noise power input to the LNA at the desired and image frequency bands
are same and is denoted as NIN .
Thus, the noise power at the output of the LNA.
Over the desired frequency band is NLNA_OUT _IN NIN × GIN × NFIN
Over the image frequency band is NLNA_OUT _IM NIN × GIM × NFIM
Thus, the total output noise power over the desired single-sideband at the output
of the mixer is
NOUT_MIXER NLNAOUTIN + NLNAOUTIM

NIN × GIN × NFIN + NIN × GIM × NFIM (18.39)
SOUT_MIXER SIN × GIN
(18.40)
NOUT_MIXER NIN × GIN × NFIN + NIN × GIM × NFIM
Thus, the single-sideband noise factor without image frequency band rejection
can be written as
SIN SOUT_MIXER GIN × NFIN + GIM × NFIM
NFMIXER / (18.41)
NIN NOUT_MIXER GIN

GIM × NFIM
NFIN 1 + (18.42)
GIN × NFIN
Here,
NFIN 2 dB 1.585
NFIM 3 dB 1.995
GIN 30 dB 1000
GIM 32 dB 1585.893
Thus from (18.42),

1585.893 × 1.995
NFMIXER 1.585 1 +
1000 × 1.585
4.747 6.764 dB
Thus, the absence of image reject filter degrades the system noise figure by
(6.764 − 2) dB 4.764 dB!!
References
1. Mark R (1967) Barber, noise figure and conversion loss of the Schottky barrier mixer diode.
IEEE Trans Microw Theor Tech 15(11):629–635
2. Gewartowski JW (1971) Noise figure for a mixer diode. IEEE Trans Microw Theor Tech
29(5):481
3. Held DN, Kerr AR (1978) Conversion loss and noise of microwave and millimeter-wave mixers:
Part 1—theory. IEEE Trans Microw Theory Tech 26(2):49–55
4. Held DN, Kerr AR (1978) Conversion loss and noise of microwave and millimeterwave mixers:
Part 2—experiment. IEEE Trans Microw Theor Tech 26(2):55–61
5. Kerr AR (1979) Shot-noise in resistive-diode mixer and the attenuator noise model. IEEE Trans
Microw Theory Tech 27(2):135–140
6. Kerr AR (1979) Noise and loss in balanced and subharmonically pumped mixers: Part I—the-
ory. IEEE Trans Microw Theor Tech 27(12):938–943
7. Kerr AR (1979) Noise and loss in balanced and subharmonically pumped mixers: Part II—ap-
plication. IEEE Trans Microw Theor Tech 27(12):944–950
8. Ogawa HA, Masayoshi A, Kozo M (1980) K-band integrated double-balanced mixer. IEEE
Trans Microw Theory Tech 28(3):180–185
9. Maas SA (1987) A GaAs MESFET mixer with very low intermodulation. IEEE Trans Microw
Theor Tech 35(4):425
10. Basu S, Maas SA (1993) Design and performance of a planar star mixer. IEEE Trans Microw
Theor Tech 41(11):2028–2030
11. Maas SA, Chang KW (1993) A broadband, planar, doubly balanced monolithic Ka-band diode
mixer. IEEE Trans Microw Theor Tech 41:2330–2335
12. Madjar A (1997) A novel general approach for the optimum design of microwave and millimeter
wave subharmonic mixers. IEEE Trans Microw Theor Tech 44(11):1997–2000
13. Marki FA (1977) Miniature image reject mixers and their use in low noise front ends in con-
junction with GaAs FET amplifiers. In: 11th Asilomar Conference on Circuits, Systems and
Computers, pp 159–162, Nov 1977
14. Fong KL, Meyer RG (1999) Monolithic RF active mixer design. IEEE Trans Circ Syst-II:
Analog Digital Signal Process 46(3):231
15. Rahman NA, Majlis BY (2004) A GaAs PHEMT single-ended mixers for 28 GHz applications.
In: The 4th annual seminar of national science fellowship
16. Kuo C, Kuo C, Kuo S, Maas HW (2008) Novel miniature and broadband millimeter-wave
monolithic star mixers. IEEE Trans Microw Theor Tech 56(4):793–802
References 581
17. Lai Y-A, Chen C-N, Su C-C, Hung S-H, Wu C-L, Wang Y-H (2011) A compact double-balanced
star mixer with novel dual 180° hybrid. Progr Electromag Res C 24:147–159
18. Guo J, Xu Z, Qian C, Dou W-B (2011) Design of a microstrip balanced mixer for satellite
communication. Progr Electromagn Res 115:289–301
19. Kallfass I, Massler H, Leuther A (2007) A 210 GHz, Subharmonically pumped active FET mixer
MMIC for radar imaging applications. In: 2007 IEEE CSIC symposium digest, pp 71–74, Oct
2007
20. Gunnarsson S (2008) Analysis and design of a novel X4 subharmonicaliy pumped resistive
HEMT mixer. IEEE Trans Microw Theor Tech 56(4):809–816
21. Lee Y-C, Liu C-H, Hung S-H, Su C-C, Wang Y-H (2012) A 20–31 GHz high image rejection
ratio subharmonic mixer. Progr Electromagn Res C 27:197–207
Chapter 19
Microwave Communication Systems
Abstract This chapter presents terrestrial mobile communication and satellite

communication systems with emphasis on design and analysis of various subsys-
tems used in receiver and transmitter sections. Homodyne and heterodyne receivers
and different transmitter architectures for wireless mobile communication systems
are described. Satellite communication systems defining important parameters like
EIRP, G/T and SFD are also presented. Detailed design, analysis and realization
of various microwave subsystems for satellite transponders like receivers, channel
amplifiers, linearizers, TWTAs, SSPAs, microwave power module (MPM) and mul-
tiport power amplifier (MPA) are also described.
19.1 Introduction
Wireless communication technology is the fastest growing technology in the world.

The successful development of wireless technologies has greatly improved people’s
ability to remain connected socially as well as for business. The phenomenal success
of wireless mobile communications is reflected by the rapid pace of technology inno-
vation. From the first-generation (1G) analog mobile communication system in 1980
to the 3G system which was first launched in 2001 through the second-generation
(2G) mobile communication system in 1991, the wireless mobile network has trans-
formed from a pure telephony system to a rich multimedia content transport system.
The fourth-generation (4G) wireless system can support data rates of up to 1 Gbps
for low mobility and up to 100 Mbps for high mobility. The target of a wireless
communication system is to connect and achieve seamless communications between
anybody, anything, wherever they are, whenever they need and by whatever elec-
tronic means they wish with affordable cost by consuming less power. Upcoming
wireless communication systems are aiming to fulfil these requirements by configur-
ing next-generation terrestrial wireless and satellite communication systems. As the
lower frequency bands become congested, the communication systems need to shift
towards higher and higher frequency bands to meet the capacity and data rate targets.
Higher microwave and millimetre wave frequency bands are much less congested

584 19 Microwave Communication Systems
and can potentially provide multi-gigahertz spectrum. This is the reason why the
satellite communication systems are already using higher microwave and mmwave
frequency bands for high-throughput applications. It is also envisaged that the next-
generation (5G) mobile communication system needs to use higher microwave and
mmwave frequency bands to meet the requirement of spectrum [1]. This chapter will
discuss terrestrial mobile communication and satellite communication systems with
associated microwave circuits and subsystem.
19.2 Mobile Communication Systems
Mobile communication system is one of the greatest innovations in the history of

technology. The rapid adoption of smartphones by a wide range of users and the
explosive growth of data traffic of these devices have been phenomenal. A simplified
block diagram of a generic transceiver for wireless communication system is shown
in Fig. 19.1. It consists of transmit and receive RF section and a baseband processing
section. The transmitter modulates the carrier signal with particular scheme and
transmits through an antenna. The receiver recovers the information from the signal
received by the antenna.
A bandpass signal v(t) can be considered as a sinusoidal waveform of frequency
fc , with varying amplitude A(t) and/or phase angle ϕ(t) with time, which can be
written as

v(t) A(t) cos 2π fc t + ϕ(t) (19.1)
Suppose
vi (t) A(t) cos ϕ(t) (19.2a)
Fig. 19.1 Block diagram of generic wireless communication system

19.2 Mobile Communication Systems 585
vq (t) A(t) cos(ϕ(t) + 90◦ ) (19.2b)
Then, (19.1) can be expressed as
v(t) vi (t) cos(2π fc t) + vq (t) cos(2π fc t + 90◦ ) (19.3)
This shows that a bandpass signal can be expressed completely by its in-phase
and quadrature-phase components. The components vi (t) and vq (t), respectively, are
the in-phase (I) and quadrature-phase (Q) components of the signal. In a generic
transmitter, the baseband signal processor converts the information signal into an in-
phase (I) and quadrature-phase (Q) component in the digital domain. The properties
of the I–Q signals determine the type of modulation. Thus, the actual data symbols to
be transmitted are coded to the I and Q signals in baseband circuitry. The I–Q signals
are finally amplified by the power amplifier and transmitted through an antenna. In
the generic receiver, the received signal from the antenna is amplified by a low-noise
amplifier and finally the transmitted symbols are decoded by recovering the original
I and Q signals from the received RF signal. The amplitude and phase information
of the signal can be extracted from the I–Q signals as follows:

A(t) vi (t)2 + vq (t)2

vq (t)
ϕ(t) tan−1
vi (t)
19.2.1 Receiver Architecture for Mobile Communication
The basic function of a wireless communication receiver is to distinguish the desired

signal from all other signals and to amplify the signal to a sufficient level for reli-
able detection by its baseband signal processing section. The simplest topology of
its RF section is a bandpass filter tuned to the same frequency with the signal. The
frequency response of a band select filter receiver in the presence of adjacent chan-
nels/interferences is shown in Fig. 19.2. The filter should be a tunable or there should
be a bank of selectable bandpass filters with different centre frequencies to select
one channel at a time among desired multiple channels. This scheme is suitable
only if desired channels are separated sufficiently, and there should not be any inter-
ference signal near to the desired signals to design a practical filter to select only
one channel in the presence of other channels. However, in practice, many narrow
frequency bands spaced very closely to utilize the total allocated frequency bands
with minimum wastage. Thus, it becomes very difficult to design a practical filter
to select only the desired channel suppressing all other closely spaced channels and
interferences. In some situations, the signal strength of the neighbouring channels
becomes much higher than the selected channel due to the path loss and/or channel
fading. In a practical scenario, a channel filter with very sharp cut-off response is
Fig. 19.2 Band-select filter receiver response
required to select a narrow bandwidth signal at higher carrier frequency. This needs a
filter with very high-quality factor which is practically impossible at least in an inte-
grated form. To overcome this problem practically, heterodyne as well as homodyne
receiver architectures are used for wireless communication receivers.
In heterodyne receiver architecture, the requirement of an extremely narrowband
channel select filter centred around a higher carrier frequency is eliminated by per-
forming the channel selection at a lower frequency. Block diagram of simple hetero-
dyne receiver architecture is shown in Fig. 19.3. In this case, the channel select filter
operates at a lower frequency with unchanged pass bandwidth leading to practical
realizable filter with achievable quality factor. Another advantage of this architecture
is that the different channels can be selected using fixed channel select filter sim-
ply by changing the frequency of the local oscillator. The frequency translation is
achieved by a process called heterodyning. In heterodyning, the original signal vi (t)
at frequency fo1 is translated to a new intermediate frequency (IF) fo1 − fLO by mul-
tiplying it in the time domain with another signal vLO (t) at frequency fLO . Cascading
several mixing stages with an appropriate filter, the signal spectrum can be translated
gradually to any other frequency band. Such receivers are called superheterodyne
receivers. Though the heterodyne receiver architecture solves the problem of channel
selection, it suffers from the requirement of image reject filter and associated inser-
tion loss of it. There are some other receiver architectures to overcome the image
rejection problem such as image reject mixer, direct conversion, low IF and wideband
IF. Among them, image reject mixers are popularly used and already discussed in
the previous chapter.
The direct-conversion receiver is also known as zero-IF receiver and homodyne
receiver. Block diagram of zero-IF receiver architecture is shown in Fig. 19.4. Here,
the centre frequency of the incoming signal and the LO signal are same, and thus
the mixer translates the input spectrum directly around DC. The low-pass filters at
outputs of the mixers attenuate all unwanted signals outside the desired frequency
band and divide the signal into I and Q components for further processing by
Fig. 19.3 Block diagram of heterodyne receiver
Fig. 19.4 Block diagram of a homodyne receiver
digital processor for information extraction. The homodyne receivers have several
advantages over the conventional heterodyne architecture. These architectures do
not require any highly selective image reject filters; this leads to savings in size.
Here, simple low-pass filters are designed at lower frequencies as an integrated
unit to select the desired channel. The homodyne architecture is attractive for the
possibility of realizing integrating receiver using low-frequency amplifiers and
digital-to-analog converter (DAC) followed by digital signal processing unit just
after the low-pass filters. However, there are some practical issues to implement the
homodyne receivers since the desired signal translated around DC. The architecture
is highly sensitive to all sources of DC power, such as LO self-mixing, re-receiving
of LO signal after reflection from other circuitry (antenna), DC offsets in the
circuitry, down-mixing of signals due to even-order distortion in LNA and mixer,
etc. This architecture also sufferers from 1/f noise of active devices working in the
low-frequency regions. Thus, to achieve full benefit from the homodyne architecture,
both the LNA and the mixers should be highly linear to avoid even-order distortion.
The low IF architecture receiver is the combination of good properties of the image
reject receiver and the zero-IF architecture. In this architecture, the intermediate
frequency (IF) is non-zero, but it is much lower than the conventional image reject
receiver. The low IF frequency leads to the use of low-frequency/low-power circuit
configuration which is more suitable for integration than the structures of the image
reject receiver. It also helps to circumvent the DC offset and 1/f noise problems of
the zero-IF architecture.
19.2.2 Transmitter Architecture for Mobile Communication
The function of a wireless communication transmitter is to modulate a carrier

frequency by information signal with particular scheme and to transmit through
an antenna. Contrary to the receiver, the transmitter circuitry processes signals of
well-known strength and spectral contents. The challenges in transmitter design are
mainly related to the spectral purity of the final power-amplified modulated signal
entering the antenna, particularly in a system where many channels are closely spaced
in frequency domain. Practically, the final power amplifier which amplifies the
modulated signals generally operates in nonlinear region to achieve better DC-to-RF
efficiency. Thus, the most severe source of spectral deterioration is due to nonlinear-
ity of the power amplifier that generates spectral re-growth which disturbs in-band
as well as adjacent channel signals. The power amplifier is generally designed using
III–V group semiconductors for achieving high RF power level with better efficiency
and it is integrated with linearizer to minimize the overall nonlinear effect.
Block diagram of a simple architecture direct-conversion transmitter is shown in
Fig. 19.5. The direct-conversion transmitter consists of an I/Q modulator, bandpass
filter and power amplifier. The simple architecture suffers severely from sensitivity to
the imbalance in between the mixers and to the phase and amplitude errors between
the LO signals. These error mechanisms will lead to increased LO leakage and
non-optimal suppression of the image frequency. Another shortcoming of the direct-
conversion transmitter is the sensitivity to LO pulling, where the strong output signal
of the power amplifier disturbs the operation of local oscillator due to their same
frequency. This effect becomes worse if power amplifier and local oscillator are
integrated into the same chip. One way to minimize LO pulling is to offset the local
oscillator from the output frequency. Two local oscillators with offset frequency
may be used to produce final frequency using a mixer. In this case, it is possible
to keep frequency of the LOs far away from the output frequency. Another way to
reduce the LO pulling is the use of heterodyne transmitter architecture as shown in
Fig. 19.6. Here, the I/Q modulation is performed at a lower LO frequency and the
final frequency translation is done by a second mixer stage to translate the spectrum
to output frequency.
Fig. 19.5 Block diagram of a direct-conversion transmitter
Fig. 19.6 Block diagram of a heterodyne transmitter
19.2.3 Transceiver for Mobile Communication
Though the functionalities of the transmitter and receiver for base station and for
the mobile units are same, the practical realizations are different due to the different
driving forces behind their development. The integration of the mobile transmitter and
receiver units is extremely desirable to extend battery life, reducing weight and size
which make the handheld mobile unit more desirable and attractive to the customer.
This can be done by increasing the level of integration. Reliability of the integrated
complete unit is also directly influenced by the integration of multiple functions into
a single chip and the mass production of ICs.
The RF section of a transmitter usually consists of an oscillator, a modulator,
an upconverter, filters and power amplifier as shown in Fig. 19.7. It also consists
of a phase-locked oscillator or synthesizer. The signals to be transmitted modulate
Fig. 19.7 Transceiver block diagram for mobile communication systems
the oscillator frequency through a suitable modulation scheme such as amplitude

modulation, frequency modulation, phase modulation or other digital modulation
schemes such as PSK, FSK, etc. In WCDMA system, QPSK modulation scheme
is used as shown in the Fig. 19.7. Here, the data stream from baseband processor
is split into odd and even streams and are separately modulated before combining
into a QPSK signal. The signals are then amplified by amplifier and upconverted to
a higher desired carrier frequency by an upconverter. The upconverted RF signals
are amplified by a power amplifier and transmitted by the antenna. The function of
the receiver is to extract the useful information from the received RF signals from
antenna. Receiver section in an RF transceiver consists of a low-noise amplifier,
downconverter, IF amplifier, bandpass filters and demodulator. The demodulated
signal processed in the baseband processor for extraction of the information.
Enormous growth of data traffic volume is one of the main drivers behind the next-
generation (5G) mobile communication systems [1]. Mobile wireless communication
with ultra-wide bandwidth will be the key requirement for the next-generation mobile
system to meet the ever-increasing needs for higher capacity. Presently, wireless
communication demands tens of billions of Internet of things (IoT) and machine-
to-machine (M2M) communications. The next-generation wireless communication
system needs enormous spectrum to fulfil the requirements for smart city, smart
home, smart grid, smart remotely controlled and self-driven vehicle, e-health which
is available only at higher microwave and millimetre wave frequency bands. In addi-
tion to the mmwave frequency bands, the fundamental technologies that enable next-
generation mobile communications include massive multiple-input-multiple-output
(MIMO) system and small-cell configuration. The mmwave technologies enable to
utilize the wideband spectrum; massive MIMO extends the range and increases spec-
tral efficiency in these frequencies by employing a large number of antennas, and
small-cell technologies provide the means to deploy a wide area mobile network with
a large number of small cells, scalable backhaul and proper interference mitigation
techniques. All these technologies together enable next-generation mobile commu-
nication systems with much larger capacity, much higher data rate and much denser
deployment than 4G systems. It is envisaged that the next-generation mobile com-
munication systems (5G) will work in higher microwave and mmwave frequency
bands.
The main functions of the RF section of the transceiver chain are amplification
of signal, conversion of frequencies and analog beamforming. The signal amplifi-
cation is done by power amplifiers in transmit chain and by LNA in receive chain.
The frequency conversions are done by the mixers using suitable local oscillators as
shown in Fig. 19.8. At higher microwave and mmwave frequencies, patch antenna
arrays are most viable option for implementing antenna arrays with large number of
elements in a small size to achieve sufficient range. A simplified block schematic of
a mobile transceiver with hybrid beamforming architecture using 64 patch elements
is shown in Fig. 19.8. Maintaining most of the capability to dynamically form beams
and adapt the MIMO processing schemes according to the channel conditions, it is
required to minimize the number of RF chains for reduction of cost and complexity.
Here, this is shown by using antenna sub-arrays, analog beamforming and digital
MIMO processing. Analog beamforming can be implemented using RF, IF or LO
beamforming technology. Here, it is shown in RF section. The 64 antenna elements
are grouped into 16 numbers of 4 × 1 sub-arrays, and thus only 16 numbers of RF
chains are required for 64 elements. Each RF signal is phase shifted and combined
in a group of four before frequency conversion. This reduces the number of con-
verter and MIMO streams. This scheme needs to handle only four MIMO streams
with four converters (each in transmit and receive section) to drive the entire 64-
element antenna array which reduces the complexity in the digital processing unit.
For RF signal phase shifting, reflective-type phase shifter, loaded line phase shifter
or switched delay phase shifters can be implemented at microwave and mmwave
frequency range.
RF section of the mobile transceiver mainly consists of low-noise amplifier (LNA),
power amplifier (PA), up- and downconverters, IF amplifiers, phase shifters, modula-
tors, switches and filters as shown in Fig. 19.8. There are several viable semiconduc-
tor device technologies to realize RF transceiver circuits working at microwave and
mmwave frequency bands. Gallium arsenide (GaAs) and indium phosphide (InP)
and high-electron-mobility transistor (HEMT) and hetero-junction bipolar transistor
(HBT) devices based on GaAs and SiGe can be used for solid-state power amplifiers
to transmit few Watts of power. GaAs and InP p-HEMTs are among the best choices
for RF Tx/Rx switches and LNAs. SiGe HBTs and RF CMOS have also been used for
low-cost transceiver circuits in low-to-medium RF power amplifications. The mobile
transceiver can be realized in multi-chip modules and surface mount packages using
MMIC chips. Thus, the whole transceiver can be integrated on a single printed circuit
board.
Fig. 19.8 Block diagram of mobile transceiver with MIMO system

Another critical technology required for the 5G mobile communication is the

high-speed analog-to-digital converter (ADC). To support multi-Gbps data rate, it is
required to use multiple channels of ADC with a sampling rate of few hundred MHz
to few GHz.
19.3 Satellite Communication Systems
Microwave frequencies are used in satellite communication due to its higher pene-
tration capability through the atmosphere, higher percentage bandwidth and higher
directive coverage. A microwave communication system mainly consists of a
receiver, a transmitter and an antenna system. Block diagram of a satellite com-
munication system is shown in Fig. 19.9. Like any other communication systems,
the space and the ground segments consist of a transmitter, a receiver and an antenna
system. A simple transponder for satellite communication is a repeater system. It
amplifies the signal received by a receive antenna with minimum addition of noise
and then translates the signal frequency band to other bands. The frequency-translated
signal further amplified to high power level by transmitter and then transmitted to
ground by a transmit antenna. For example, Fig. 19.9 shows a C × S-band transpon-
der. It receives the signal in C-band and transmits it in S-band. Frequency translation
provides high level of isolation to the sensitive receive sections from the high level of
transmitted RF power. This ensures the stability of the transponder avoiding reinjec-
tion of transmitted output signal to the receive section by using filtering. A satellite
transponder has three most important performance parameters: saturation flux den-
sity (SFD), receive antenna gain (GR ) to system temperature (TR ) ratio (GR /TR ) and
satellite-saturated effective isotropic radiated power (EIRP). Saturation flux density
is the received power flux density at the satellite which saturates the transponder,
i.e. to saturate final high power amplifier (HPA) used in the transmitter. Practically,
the satellite SFD is not an independent parameter; its minimum value is limited by
the ratio GR /TR . Suppose the equivalent noise temperature of a satellite receiver
(including losses in between receiver and receive antenna output) is Te , and the total
equivalent system noise power Pn can be written as
Pn k(Te + To )B (19.1a)
Here, B is the noise bandwidth of the system and To is the antenna noise temper-
ature which is about 290 K and k is Boltzmann constant. Thus, the overall system
noise temperature TR is (Te + To ). The received power PR at the output of the satel-
lite receive antenna should be more than the total noise power Pn . Considering the
received power flux density at the satellite is PFD, the received power at the output
of the receive antenna can be written as
PR PFD × Ae
Fig. 19.9 A simplified block diagram of a satellite communication system
Here, Ae is the effective aperture of the receive antenna, and it is related to the
gain of the antenna GR by
GR 4π Ae /λ2R
Here, λR is the wavelength corresponding to the receive signal frequency. Thus,

the received power is given by
PFD × GR × λ2R
PR (19.2)
4π
For reliable communication, the received power must be more than the total noise
power. The amount will be determined by the required signal-to-noise power ratio
needed for particular data rate, used modulation code, etc. for a communication.
Here, to derive the expression for minimum SFD level, we will impose the condition
PR > Pn (19.3)
Combining (19.1), (19.2) and (19.3),
PFD × GR × λ2R
> kTR B (19.4)
4π
or
4π TR
PFD > ×k ×B×
λR
2 GR
Expressing all the parameters in dB (19.4), it can be written as

19.3 Satellite Communication Systems 595

4π GR
[PFD]dBW/m2 > + [k]dBW/Hz−K + [B]dBHz − (19.5)
λ2R dB/m2 TR dB/K
This shows that for a transponder operating at particular frequency and bandwidth,
the minimum operating PFD limit is determined by its receive antenna gain-to-system
noise temperature ratio, i.e. (GR /TR ). Depending upon the requirement of uplink
carrier-to-noise power ratio, which is determined by the communication data rate,
modulation codes, etc.; the operating PFD limit is selected which is higher than the
value given by (19.5).
Example 19.1 A satellite transponder operates at receive frequency 5.9 GHz of band-
width 35 MHz. The gain of the receive antenna is 27 dBi and the receiver noise figure
(including loss in between receive antenna and receiver) is 3.5 dB. Calculate the min-
imum received power flux density limit of the transponder.
Solution
Receive frequency 5.9 GHz
Thus, λR (0.3/5.9) m 0.05085 m
Receiver noise figure
NF 3.5 dB 2.2387(in factor)
Receiver noise temperature
Te (NF − 1)To (2.2387 − 1)290 K 359 K
Thus, the system noise temperature
TR Te + To (290 + 359) K 649 K 28 dBK
Thus,

GR
(27 − 28) dB/K −1 dB/K
TR dB/K

4π 4π
10 log 36.87 dB/m2
λR dB
2
λ2R
[k]dBW −228.6 dBW/Hz − K
[B]dBHz 10 log(35 × 106) dB − Hz 75.4 dBHz

4π GR
[PFD]dBW/m2 > + [k]dBW/Hz−K + [B]dBHz −
λ2R dB/m2 TR dB/K

GR
> 36.87 − 228.6 + 75.4−
TR dB/K

GR
> −116−
TR dB/K
> −115 dBW/m2
Thus, the minimum PFD is −115 dBW/m2 .

The saturated power flux density (SFD) also should be higher than the value given
by (19.5). Otherwise, the transponder will be saturated with the noise power itself.
Saturated effective isotropic radiated power (saturated EIRP) of a satellite
transponder is determined by the final high power amplifiers output power capa-
bility (Psat ), output loss (LOUT ) and satellite’s transmit antenna gain (GT ). The output
loss is the RF loss of all the elements in between output of the power amplifier and
input of the antenna. Therefore, saturated RF power at the input of the antenna (PT sat )
is given by
PT sat Psat − LOUT (19.6a)
Thus, the saturated EIRP is given by
EIRPsat PT sat × GT (19.6b)
The overall gain of a transponder is determined by its saturated EIRP and SFD.
From (19.2), under saturated flux density condition, the received power at the input
of the receiver is given by
SFD × GR × λ2R
PRsat (19.7)
4π
Therefore, the overall gain (GTransponder ) of the transponder is given by
PT sat EIRPsat
GTransponder (19.8)
PRsat GT × PRsat

GTransponder dB [EIRPsat ]dBW − [GT ]dBi − [PRsat ]dBW (19.9)
Example 19.2 A satellite transponder operates at receive frequency 5.9 GHz and
transmit frequency 2.6 GHz of bandwidth 35 MHz. The gain of receive and transmit
antennae are 27 and 42 dBi, respectively. Calculate the gain of the transponder if
SFD is −95 dBW/m2 and saturated EIRP is 65 dBW.
Solution
Receive frequency 5.9 GHz
Thus, λR (0.3/5.9) m 0.05085 m
Receive antenna gain 27 dBi
From (19.6b), the saturated transmit power
EIRPsat
PT sat
GT
[PT sat ]dBW [EIRPsat ]dBW − [GT ]dBi
65 dBW − 42 dBi
23 dBW 53 dBm
From (19.7),
SFD × GR × λ2R
PRsat
4π

λ2R
[PRsat ]dBW [SFD]dBW/m2 + [GR ]dBi +
4π dBm2
(−95 + 27 − 36.87) dBW −104.87 dBW
−74.87 dBm
From (19.9),

GTransponder dB [EIRPsat ]dBW − [GT ]dBi − [PRsat ]dBW
65 dBW − 42 dBi + 104.87 dBW
127.87 dB ∼
128 dB
Thus, the gain of the transponder is 128 dB.

Block diagram of a typical satellite communication transponder with power lev-
els at different stages based on the previous examples is shown in Fig. 19.10. The
transponder consists of a preselect filter (PSF) at the input of the transponder to select
the desired frequency band. The PSF restricts the noise and interferences outside the
desired bandwidth entering into the receiver. The signals are received by the receiver,
amplified by a low-noise amplifier (LNA) and then frequency translated by mixer
and further amplified by IF amplifiers. A bandpass filter (BPF) is used just after the
mixer to suppress the spurious products generated by the mixer. The output signals
from the receiver are divided into several channels by channelization filters and then
amplified to high power level using different transmitter chains. The channelized
high power signals then combined in frequency domain by using an output mul-
tiplexer. The multiplexed signal passes through a harmonic reject filter to provide
sufficient rejection to the harmonics generated by the power amplifiers operated in
their nonlinear region. Total gain of a transponder of about 130 dB is distributed
among the receiver and transmitter. The gains of receive and transmit sections are
about 50 and 84 dB, respectively, as shown in Fig. 19.10. The loss of the preselect
filter and feeder cable in between antenna output and the receiver input needs to be
taken into account to determine the overall system noise temperature. Similarly, loss
of the output multiplexer, HRF and interconnecting cable up to antenna input are to
be taken into account to determine final output power transmitted by the antenna to
achieve required saturated EIRP.
Fig. 19.10 Block diagram of a satellite communication system with power levels
Example 19.12 Calculate insertion loss and return losses of a 2-port network of
S-parameter matrix:

0.05 − 80◦ 0.84 − 25◦

[S]
0.84 − 25◦ 0.05 − 80◦
Calculate insertion loss and return losses when such two networks connected in
cascade.
Solution
Insertion loss (IL) of the network is
ILdB −20 × log(|S21 |)

−20 × log(0.84)
1.51 dB
Return loss (RL) of the network is
RLdB −20 × log(|S11 |)

−20 × log(0.05)
26.02 dB
ABCD parameters of the network can be calculated using (7.124)

(1 + S11 )(1 − S22 ) + S12 S21
A
2S21
(1 + 0.05 − 80◦ )(1 − 0.05 − 80◦ ) + 0.84 − 25◦ × 0.84 − 25◦

2 × 0.84 − 25◦
(1 + 0.009 − 0.049i)(1 − 0.009 + 0.049i) + (0.761 − 0.355i) × (0.761 − 0.355i)

2 × (0.761 − 0.355i)
0.924 4.662◦
(1 + S11 )(1 + S22 ) − S12 S21

B Zo
2S21
(1 + 0.05 − 80◦ )(1 + 0.05 − 80◦ ) − 0.84 − 25◦ × 0.84 − 25◦
50 ×
2 × 0.84 − 25◦
(1 + 0.009 − 0.049i)(1 + 0.009 − 0.049i) + (0.761 − 0.355i) × (0.761 − 0.355i)
50 ×
2 × (0.761 − 0.355i)
21.252 63.159◦
(1 + S11 )(1 − S22 ) − S12 S21

C Yo
2S21
1 (1 + 0.05 − 80◦ )(1 − 0.05 − 80◦ ) − 0.84 − 25◦ × 0.84 − 25◦
×
50 2 × 0.84 − 25◦
1 (1 + 0.009 − 0.049i)(1 − 0.009 + 0.049i) − (0.761 − 0.355i) × (0.761 − 0.355i)
×
50 2 × (0.761 − 0.355i)
0.010 75.463◦
(1 − S11 )(1 + S22 ) + S12 S21

D
2S21
(1 − 0.05 − 80◦ )(1 + 0.05 − 80◦ ) + 0.84 − 25◦ × 0.84 − 25◦

2 × 0.84 − 25◦
(1 − 0.009 + 0.049i)(1 + 0.009 − 0.049i) + (0.761 − 0.355i) × (0.761 − 0.355i)

2 × (0.761 − 0.355i)
0.924 4.662◦
Thus, ABCD parameters in matrix form of the 2-port network are

0.924 4.662◦ 21.252 63.159◦

[ABCD]
0.010 75.463◦ 0.924 4.662◦
[ABCD] parameters of such two 2-port networks connected in cascade will be

given by
[Ac Bc Cc Dc ] [ABCD] × [ABCD]

0.924 4.662◦ 21.252 63.159◦ 0.924 4.662◦ 21.252 63.159◦

×
0.010 75.463◦ 0.924 4.662◦ 0.010 75.463◦ 0.924 4.662◦

0.740 21.976◦ 39.283 67.821◦

0.018 80.125◦ 0.740 21.976◦
[S] parameters of the cascaded 2-port network can be derived using (7.125)
Ac + Bc Yo − Cc Zo − Dc
Cascaded S11
Ac + Bc Yo + Cc Zo + Dc
0.740 21.976◦ + 0.786 67.821◦ − 0.910 80.125◦ − 0.740 21.976◦

0.740 21.976◦ + 0.786 67.821◦ + 0.910 80.125◦ + 0.740 21.976◦
0.220 − 50.338◦

2.841 50.049◦
0.077 − 100.387◦
Thus, return loss (RL) of the cascaded network is
Cascaded RL (in dB) −20 × log(0.077)

22.22 dB
2
Cascaded S21
2

0.740 21.976◦ + 0.786 67.821◦ + 0.910 80.125◦ + 0.740 21.976◦
2

2.841 50.049◦
0.704 − 50.049◦
Thus, insertion loss (IL) of the cascaded network is
Cascaded IL (in dB) −20 × log(0.704)dB

3.05 dB !! ( about (1.51 + 1.51) dB)
This example shows that the insertion loss of the cascaded network (3.049 dB) is
equal to sum of the insertion losses of the individual networks.
Example 19.13 Calculate insertion loss and return losses of a 2-port network of
S-parameter matrix:

0.30 − 90◦ 0.84 − 26◦

[S]
0.84 − 26◦ 0.30 − 90◦
cascade.
Solution
Insertion loss (IL) of the network is
IL (in dB) −20 × log(|S21 |)

−20 × log(0.84)
1.51 dB
Return loss (RL) of the network is
RL (in dB) −20 × log(|S11 |)

−20 × log(0.30)
10.46 dB
ABCD parameters of the network can be calculated using (7.124)

(1 + S11 )(1 − S22 ) + S12 S21
A
2S21
(1 + 0.3 − 90◦ )(1 − 0.3 − 90◦ ) + 0.84 − 26◦ × 0.84 − 26◦

2 × 0.84 − 26◦
(1 + 0 − 0.3i)(1 − 0 + 0.3i) + (0.755 − 0.368i) × (0.755 − 0.368i)

2 × (0.755 − 0.368i)
◦
0.966 5.961

(1 + S11 )(1 + S22 ) − S12 S21

B Zo
2S21
(1 + 0.3 − 90◦ )(1 + 0.3 − 90◦ ) − 0.84 − 26◦ × 0.84 − 26◦
50 ×
2 × 0.84 − 26◦
(1 + 0 − 0.3i)(1 + 0 − 0.3i) + (0.755 − 0.368i) × (0.755 − 0.368i)
50 ×
2 × (0.755 − 0.368i)
◦
14.215 20.717

(1 + S11 )(1 − S22 ) − S12 S21

C Yo
2S21
1 (1 + 0.3 − 90◦ )(1 − 0.3 − 90◦ ) − 0.84 − 26◦ × 0.84 − 26◦
×
50 2 × 0.84 − 26◦
1 (1 + 0 − 0.3i)(1 − 0 + 0.3i) − (0.755 − 0.368i) × (0.755 − 0.368i)
×
50 2 × (0.755 − 0.368i)
◦
0.015 93.638

(1 − S11 )(1 + S22 ) + S12 S21

D
2S21
(1 − 0.3 − 90◦ )(1 + 0.3 − 90◦ ) + 0.84 − 26◦ × 0.84 − 26◦

2 × 0.84 − 26◦
(1 − 0 + 0.3i)(1 + 0 − 0.3i) + (0.755 − 0.368i) × (0.755 − 0.368i)

2 × (0.755 − 0.368i)
◦
0.966 5.961

Thus, ABCD parameters in matrix form of the 2-port network are

0.966 5.961◦ 14.215 20.717◦

[ABCD]
0.015 93.638◦ 0.966 5.961◦
[ABCD] parameters of such two 2-port networks connected in cascade will be

given by
[Ac Bc Cc Dc ] [ABCD] × [ABCD]

0.966 5.961◦ 14.215 20.717◦

0.015 93.638◦ 0.966 5.961◦

0.966 5.961◦ 14.215 20.717◦

×
0.015 93.638◦ 0.966 5.961◦

0.911 25.027◦ 27.459 26.677◦

0.029 99.598◦ 0.911 25.027◦
Ac + Bc Yo − Cc Zo − Dc
Cascaded S11
0.911 25.027◦ + 0.549 26.677◦ − 1.437 99.598◦ − 0.911 25.027◦

0.911 25.027◦ + 0.549 26.677◦ + 1.437 99.598◦ − 0.911 25.027◦
1.380 − 58.039◦

3.090 52.000◦
0.447 − 110.039◦
Cascaded RLdB −20 × log(0.447)

7.00 dB
2
Cascaded S21
2

0.911 25.027◦ + 0.549 26.677◦ + 1.437 99.598◦ − 0.911 25.027◦
2

3.090 52.000◦
0.647 − 52.000◦
Thus, insertion loss (RL) of the cascaded network is
Cascaded IL (in dB) −20 × log(0.647)

3.78 dB !! (different from (1.51 + 1.51) dB)
This example shows that the insertion loss of the cascaded network (3.78 dB) is
more than the sum of the insertion losses of the individual networks. This is due to
the poor port return losses (10.46 dB) of the individual networks.
Example 19.14 Calculate insertion loss and return losses of the two 2-port networks
P and Q of S-parameter matrixes:

0.6 − 91◦ 0.8 − 1◦

[S]P
0.8 − 1◦ 0.6 − 91◦
and

0.5 26◦ 0.9 − 63◦

[S]Q
0.9 − 63◦ 0.5 26◦
cascade.
Solution
Insertion loss (IL) of the network P is
IL (in dB) −20 × log(|S21 |)

−20 × log(0.8)
1.938 dB
Return loss (RL) of the network P is
RL (in dB) −20 × log(|S11 |)

−20 × log(0.6)
4.44 dB
Insertion loss (IL) of the network Q is
IL (in dB) −20 × log(|S21 |)

−20 × log(0.9)
0.92 dB
Return loss (RL) of the network Q is
RL (in dB) −20 × log(|S11 |)

−20 × log(0.5)
6.02 dB
ABCD parameters of the network P can be calculated using (7.124)

(1 + S11 )(1 − S22 ) + S12 S21
A
2S21
(1 + 0.6 − 91◦ )(1 − 0.6 − 91◦ ) + 0.8 − 1◦ × 0.8 − 1◦

2 × 0.8 − 1◦
(1 − 0.01 − 0.6i)(1 + 0.01 + 0.6i) + (0.8 − 0.014i) × (0.8 − 0.014i)

2 × (0.8 − 0.014i)
1.250 0◦
(1 + S11 )(1 + S22 ) − S12 S21

B Zo
2S21
(1 + 0.6 − 91◦ )(1 + 0.6 − 91◦ ) − 0.8 − 1◦ × 0.8 − 1◦
50 ×
2 × 0.8 − 1◦
(1 − 0.01 − 0.6i)(1 − 0.01 − 0.6i) − (0.8 − 0.014i) × (0.8 − 0.014i)
50 ×
2 × (0.8 − 0.014i)
36.409 − 90◦
(1 + S11 )(1 − S22 ) − S12 S21

C Yo
2S21
1 (1 + 0.6 − 91◦ )(1 − 0.6 − 91◦ ) − 0.8 − 1◦ × 0.8 − 1◦
×
50 2 × 0.8 − 1◦
1 (1 − 0.01 − 0.6i)(1 + 0.01 + 0.6i) − (0.8 − 0.014i) × (0.8 − 0.014i)
×
50 2 × (0.8 − 0.014i)
0.015 90◦
(1 − S11 )(1 + S22 ) + S12 S21

D
2S21
(1 − 0.6 − 91◦ )(1 + 0.6 − 91◦ ) + 0.8 − 1◦ × 0.8 − 1◦

2 × 0.8 − 1◦
(1 + 0.01 + 0.6i)(1 − 0.01 − 0.6i) + (0.8 − 0.014i) × (0.8 − 0.014i)

2 × (0.8 − 0.014i)
1.250 0◦
Thus, ABCD parameters in matrix form of the 2-port P network are

1.250 0◦ 36.409 − 90◦

[ABCD]P
0.015 90◦ 1.250 0◦
In similar way, ABCD parameters in matrix form of the 2-port Q network are

0.516 − 3.535◦ 78.865 90.039◦

[ABCD]Q
0.009 92.513◦ 0.516 − 3.535◦
Therefore, [ABCD] parameter matrix of the cascaded networks P and Q can be

written as
[ABCD]PQ [ABCD]P × [ABCD]Q

0.984 − 1.448◦ 79.817 90.880◦

0.020 90.060◦ 0.575 175.949◦
A + BYo − CZo − D
Cascaded S11
A + BYo + CZo + D
0.984 − 1.448◦ + 1.596 90.880◦ − 0.981 90.060◦ − 0.575 175.949◦

0.984 − 1.448◦ + 1.596 90.880◦ + 0.981 90.060◦ + 0.575 175.949◦
1.658 22.379◦

2.540 81.295◦
0.653 − 58.916◦
Cascaded RL (in dB) −20 × log(0.653)

3.70 dB
2
Cascaded S21
A + BYo + CZo + D
2

0.984 − 1.448◦ + 1.596 90.880◦ + 0.981 90.060◦ + 0.575 175.949◦
2

2.540 81.295◦
0.787 − 81.295◦
Thus, insertion loss (RL) of the cascaded network is
Cascaded IL (in dB) −20 × log(0.787)

2.08 dB ! ! (different from (1.938 + 0.92) dB)
This example shows that the insertion loss (2.08 dB) of the cascaded network is
less than the sum of the insertion losses of the individual networks. This is due to
the poor port return losses of the individual networks.
Examples 19.12, 19.13 and 19.14 show that in case of good return losses of the
individual networks, insertion loss and gain of the combined (cascaded) network
are the algebraic sum of the individual networks’ gain (loss), whereas, in case of
poor return losses of the individual networks, combined (cascaded) gain (loss) may
increase or decrease.
19.4 Receiver
Function of on-board receiver of a communication satellite is to amplify the received

signal linearly with minimum possible addition of noise and translating the frequency
band of the received signal to the required downlink frequency band using a local
oscillator. Most important parameters of a communication receiver are operating
frequency, bandwidth, gain, noise figure, linearity, spurious levels and frequency
translation error [2–4]. Block diagram of a typical receiver with operating gain of
about 50 dB and overall noise figure of 1.7 dB is shown in Fig. 19.11. Front end
of the receiver consists of a low-noise amplifier (LNA) to achieve overall low-noise
temperature of the system. The input matching network of an LNA corresponds to
its optimum noise figure which is different from complex conjugate matching; thus,
input VSWR of an LNA is poor. An isolator is used at input of the LNA to avoid
mismatch of the receive antenna output to the receiver. A high-gain LNA (28 dB in
this example) is used to minimize the noise contributions due to the losses of the
following elements such as filters and mixer. Here, a three-stage low-noise amplifier
using pHEMT device is used for total gain of 28 dB and noise figure of 1.7 dB. In
general, double-balanced mixer is used to downconvert the receive frequency band
to transmit IF frequency band using a local oscillator. The frequency of the local
oscillator is the frequency difference in between receive and transmit frequencies.
A bandpass filter (BPF) at the input of the mixer is used to pass only the required
frequency band and rejecting the unwanted out-of-band frequencies including the
rejection of image frequency band. The BPF at the output of the mixer is used to
reject various mixing products generated by the mixer. An IF amplifier is used to
provide rest of the gain required for the receiver. The IF amplifier is designed to
extract maximum power gain from the device, which operates in its linear region
with moderate noise figure. Thus, if stability criterion allows, the IF amplifiers are
designed with complex conjugate matching at its input as well as output for achieving
maximum power gain from the used device. The amplifiers may be realized in hybrid
microwave integrated circuit (HMIC) or in monolithic microwave integrated circuit
(MMIC). In case of HMIC implementation, discrete components including active
devices are mounted on printed substrate to realize the complete circuit, whereas,
in MMIC implementation, all passive and active components including matching
elements are built in a single substrate. Due to the absence of packaging of individual
components and interconnecting elements in MMIC realization, the frequency of
operation and achievable bandwidth are more. Photograph of a 3-stage microwave
amplifier, realized using discrete components, is shown in Fig. 19.12. Here, packaged
pHEMT, chip resistors and chip capacitors are used as discrete components which are
mounted in an alumina substrate on which matching transmission microstrip elements
are printed. Photograph of a double-conversion receiver is shown in Fig. 19.13 which
is realized in HMIC.
19.4 Receiver 607
Fig. 19.11 Block diagram of a single conversion communication receiver
pHEMT
Fig. 19.12 Photograph of a microwave amplifier realized in HMIC
Example 19.3 Calculate overall noise figure (NF) and noise temperature of a receiver
of block diagram as shown in Fig. 19.11. Consider loss of the isolator 0.2 dB.
Solution
Isolator loss: 0.2 dB
LNA gain: 28 dB, LNA NF: 1.7 dB
Filters and mixer combined loss: (1 + 8 + 1) dB 10 dB
IF amplifier gain: 32 dB, IF amplifier NF: 3 dB
Simplified block diagram of the receiver with gain and noise figures of individual
modules is shown in Fig. 19.14. Using the following Friis formula for calculation of
receivers overall noise figure (NFRX ),
NF2 − 1 NF3 − 1 NF4 − 1
NFRX NF1 + + +
G1 G1 × G2 G1 × G2 × G3
0.479 9.0 0.995
1.047 + + +
0.955 0.955 × 631 0.955 × 631 × 0.1
1.5803
Fig. 19.13 Photograph of a communication receiver
Fig. 19.14 Block diagram of the receiver for noise calculation
[NFRX ]dB 1.987 dB
Noise temperature TRx of the receiver is
TRX (NFRX − 1)To

(1.5803 − 1)290 K 168.3 K
Example 19.4 Calculate overall third-order intermodulation product (IM3) of a

receiver (shown in the block diagram in Fig. 19.11) at input power level of −76 dBm.
The Po1dB (output power at 1-dB gain compression point) of LNA and IF amplifiers
are 5 and +7 dBm, respectively. The PoIP3 (output third-order intercept point) of the
mixer is 10 dBm.
19.4 Receiver 609
Solution
Po1dB of LNA: +5 dBm
Po1dB of IF amplifier: +7 dBm
PoIP3 of mixer: +10 dBm
Considering the I–O characteristic of LNA, mixer and IF amplifier can be
expressed by a power series up to third order, the third-order intercept output power
level PoIP3 can be written in terms of Po1dB as
PoIP3 Po1dB + 10.63
Using this equation, PoIP3 of the LNA and IF amplifier are +15.63 and +17.63 dBm,
respectively.
Output power levels of each element at the input carrier power level of −76 dBm
are shown in Fig. 19.15, considering all the elements are operating with constant
gain. The third-order intermodulation power level PoIM 3 in dBc with respect to the
output carrier level is given by (14.15b)

PoIM 3 Po2f1 −f2 − Pof1 2 Pof1 − PoIP3
Here, Pof1 Pout . Therefore, the third-order intermodulation power level of LNA,
mixer and IF amplifiers at the output of respective elements are
PoIM 3 (LNA) 2(−48.2 − 15.6) dBc −127.7 dBc ⇒ −175.9 dBm

PoIM 3 (Mixer) 2(−57.2 − 10) dBc −134.4 dBc ⇒ −191.6 dBm
PoIM 3 (IF Amp) 2(−26.2 − 17.6) dBc −87.7 dBc ⇒ −113.86 dBm
Considering the linear gain of the IM 3 power levels by the following stages, the
overall IM 3 level at the output of the receiver will be
PoIM 3 (−175.9 + 22) dBm + (−191.6 + 31) dBm + (−113.86) dBm

−113.86 dBm
Fig. 19.15 Block diagram of the receiver for IM3 calculation

The IM 3 level in dBc with respect to the carrier level at the output is (−113.86 +
26.2) dBc −87.7 dBc.
This shows that for this receiver configuration, the contribution of LNA and mixer
on the overall IM 3 is negligible. It is fully governed by the nonlinearity of the final
stage, i.e. IF amplifier.
Therefore, if the receiver operates at higher input power levels say by 10 or 20 dB
more than overall IM 3 level of the receiver will be more by 20 and 40 dB, respectively.
Thus, for the receiver input power level of −66 and −56 dBm, the IM 3 level will be
−67.7 and −47.7 dBc, respectively.
19.4.1 Local Oscillator
The role of a local oscillator in satellite transponder is to provide a stable reference

RF frequency with sufficient output power level to drive mixer circuit in its nonlinear
region for frequency translation. The frequency of the local oscillator is the difference
between the centre frequency of the uplink band and the centre frequency of the
downlink frequency band. Frequency stability, spectral purity in terms of phase noise,
and spurious products and output power level are the most important parameters of
a local oscillator. Generally, in a satellite application, local oscillators of the desired
frequency are derived from a reference low-frequency source. The stability of local
oscillator is determined by the stability of the reference frequency source.
Mostly, crystal-based oscillators are used as reference source for the local oscil-
lators. Crystal cut in the form of a plate determines the fundamental frequency of
oscillation. Frequency stability of a reference crystal oscillator is mainly influenced
by the change of temperature and time. To minimize the influence of change of tem-
perature, temperature-compensated crystal oscillator (TCXO) is used. Here, TCXO
encases the oscillator circuit and temperature compensating networks in a closed
container. Another option is the use of oven-controlled crystal oscillator (OCXO) as
reference oscillator. Here, crystal oscillator and temperature-sensitive elements are
kept in a thermally insulated container along with a heater. The heater maintained the
inside temperature at oscillators’ minimum sensitive region. The stability of a local
oscillator is specified as long-term stability over the lifetime and short-term stability
over the specified operating temperature range. In general, for satellite transponders,
temperature-controlled crystal oscillator (TCXO) of short-term stability of ±1 ppm
(parts per million, i.e. error of 1 Hz in 1 MHz) and long-term stability of ±10 ppm
for 15 years is used. In some applications, where more stable frequency is required,
oven-controlled crystal oscillator (OCXO) of one order that has better stability com-
pared to TCXO is used. Two types of local oscillators are realized. One is based on
frequency multiplier and another is using phase-locked loop (PLL).
19.4 Receiver 611
(a)
(b)
Fig. 19.16 a Block diagram of a multiplier-based local oscillator. b Simple block diagram of a
PLL-based local oscillator
19.4.1.1 Multiplier-Based Local Oscillator
Block diagram of a multiplier-based LO is shown in Fig. 19.16a. Here, reference

crystal oscillator frequency of 132 MHz and required LO output frequency of 3.3 GHz
is considered. Two stages of X5 multiplication (overall X25) are done to generate
3.3 GHz frequency from the reference frequency of 130 MHz. Bandpass filters at
different stages are used to suppress unwanted frequency components to a sufficient
level (at least −60 dBc) with respect to the desired frequency component. Any
unwanted frequency component of an LO acts as discrete frequency component and
may severely affect the communication.
Amplifiers (AMP1 and AMP2) are used before each multiplier circuit to provide
required power level to the multiplying device to operate it highly nonlinear con-
dition. The final amplifier (AMP3) is used to increase the power level of the final
frequency component to achieve the desired output level required for operating mixer
in the receiver in its nonlinear region. In general, the final amplifier stage operates
in saturation region to provide nearly constant output power level over the operat-
ing environmental (temperature, bias voltage variation, etc.) condition to ensure the
operation of mixer in its fixed conversion gain/loss condition.
19.4.1.2 PLL-Based Local Oscillator
Simplified block diagram of a phase-locked loop (PLL)-based local oscillator is

shown in Fig. 19.16b. It consists of a phase detector, loop filter, frequency divider
and voltage-controlled oscillator (VCO). Generally, varactor diode is used to achieve
voltage-dependent frequency of the VCO. Output frequency of the VCO is divided by
a frequency divider and fed back to one input of the phase detector circuits. The phase
detector compares the output frequency from the divider to the reference frequency
of the reference oscillator. A phase error signal is generated by the phase detector
and creates a signal whose magnitude is proportional to the phase error. This phase
error signal is then low-pass filtered by the loop filter and fed to the control input
of the VCO. The control signal controls the output frequency of the VCO. At the
locked condition of the PLL, the two inputs to the phase detector are in-phase and
the output frequency is equal to the reference oscillator frequency multiplied by the
divider ratio, N.
19.5 Satellite Transmitter
The transmitter section amplified the channelized signal to the required RF power
level before transmitting through a transmit antenna. As shown in Fig. 19.10, this
section consists of a driver amplifier (DA) to boost the signal to proper drive level
required for high power amplifier (HPA) which ultimately provides the required
transmit power level. In a communication system, major contributions of nonlin-
earities are due to the final power amplifier that affects the overall communication
performance severely [5–7]. A linearizer also used at input of the HPA to minimize
the effect of nonlinearity of the HPA on transmitted signal. The driver amplifier (DA)
is also called channel amplifier (CAMP) for its function to amplify the channelize
frequency band. The CAMPs consist of several control systems for on-board con-
trolling the transponder gain by ground command or automatically by sensing its RF
power level.
The channel amplifier and linearizer are low-power systems, and thus these are
realized very compactly using solid-state technology. However, depending on down-
link frequency and required output power level, two types of high power amplifiers
are used: Solid-state power amplifier (SSPA) and travelling wave tube amplifier
(TWTA). Use of TWTAs is the only option in case of requirement of higher RF
over high microwave frequency range. However, due to the advancement of solid-
state device technology, it can provide required RF power level at least over lower
microwave frequency range. Thus, SSPAs are preferable for its compact size and
better linearity compared to TWTAs.
19.5.1 Driver Amplifier (DA)
Function of an on-board driver amplifier (DA) for a communication satellite is to

amplify the channelized signal linearly, i.e. without much distortion, for providing
sufficient drive level to the high power amplifier [8]. It has also provision for on-board
gain setting of the transponder by ground command or automatically by sensing its
RF power level. A driver amplifier for satellite transponder has two operating modes:
fixed gain mode (FGM) and automatic level control (ALC) mode. Most important
parameters of a driver amplifier are its operating frequency, bandwidth, gain, linearity,
output power, gain setting range in FGM and dynamic range in ALC mode. In a
19.5 Satellite Transmitter 613
Fig. 19.17 Block diagram of a satellite channel amplifier
satellite transponder, a driver amplifier amplifies the signal after channelization, and
thus it is also called channel amplifier (CAMP).
Block diagram of a channel amplifier is shown in Fig. 19.17. The CAMP consists
of RF circuits and bias and control circuits. The RF lineup consists of several amplifier
stages (A1–A5) to achieve required gain. Digital attenuators (DAT1 and DAT2) are
used for commandable gain setting of the CAMP by issuing digital command. The
analog attenuators (AAT1 and AAT2) are used to provide attenuation automatically
by detecting the power level using RF power detector (DET). The detected voltage
at the output of the detector amplified by a differential DC amplifier and applied to
the control terminal of the analog attenuators through a commandable analog switch.
In ALC mode of operation, the output of the DC amplifier will be connected to the
analog attenuators. Another input of the differential DC amplifier is connected to a
temperature-controlled voltage to keep the output RF power level constant in ALC
mode over the operating temperature range. In FGM operation, the control terminals
of the analog attenuators will be connected to temperature-controlled voltage to keep
the constant gain over the operating temperature range. The digital control circuit
processes the command data and generates appropriate control data to select the
mode of operation (in between ALC and FGM) and to provide commandable gain
setting in both the operation modes. An amplitude tilt active equalizer (EQ) is used to
achieve broadband frequency response. The amplifiers and both types of attenuators
are distributed in the lineup to achieve required noise figure and linearity of the
CAMP over its entire gain (dynamic range) setting conditions for both the operation
modes.
Photograph of a Ku-band CAMP of about 60 dB gain is shown in Fig. 19.18. The
amplifier and attenuator modules are realized using MMIC technology. The full RF
Fig. 19.18 Photograph of a Ku-band CAMP
Fig. 19.19 Frequency

response of the CAMP with
and without equalizer
circuit is packaged in three compartments with narrow slits (0.7 mm × 2 mm) through
which the circuits are interconnected using gold ribbons. The slit acts as waveguide of
cut-off frequency below 10 GHz, and thus provides high isolations for the operating
frequency in between two adjacent compartments, which prevents waveguide mode
of propagation in operating frequency band, and thus ensures overall stability of the
CAMP.
Frequency response of the CAMP with and without equalizer is shown in
Fig. 19.19. It shows that the use of equalizer improves the gain flatness from 6
to 1.5 dB over the frequency range of 10.5–13 GHz. In fixed gain mode (FGM)
operation, the digital attenuators DAT1 and DAT2 are used to control the gain of the
CAMP to set the saturation flux density (SFD) of the transponder and also to oper-
ate the transponder at required power back-off condition. The typical gain setting is
about 30-dB in steps of 1-dB. I–O characteristics of the channel amplifier in FGM
operation for different gain setting conditions are shown in Fig. 19.20. The output
of the CAMP saturated under higher input power level is due to the saturation of
the final amplifier stage. Under nominal operating condition, the channel amplifier
always operates in its linear I–O characteristic region.
Fig. 19.20 I-O characteristic of CAMP in FGM operation for different gain setting condition
Fig. 19.21 I-O characteristic of CAMP in ALC mode operation for different attenuation in DAT2
In ALC mode of operation, the output power will remain constant irrespective
of its input power level with the specified ALC dynamic range. Typical ALC dynamic
range is about 30 dB. Here, the RF power level is detected before the digital attenuator
DAT2 to control the output power level in ALC mode for operating the HPA back-off
condition if required. Final amplifier stage is used after the detector and DAT2 to
enable the use of low-power device for the final amplifier stage meeting the output
power requirement. I–O characteristics of the CAMP in its ALC mode of operation
are shown in Figs. 19.21 and 19.22 for setting of the digital attenuators DAT2 and
DAT-1, respectively. In ALC mode of operation, the digital attenuator DAT2 is used
to provide adjustable (variable) constant output power level as shown in Fig. 19.21.
Typically, 15 dB attenuation range in steps of 0.5 dB is kept for this purpose. The
digital attenuator DAT1 is used to slide the ALC range keeping the same ALC
dynamic range as shown in Fig. 19.22. This provides the flexibility of SFD range
setting of the transponder in ALC mode.
Fig. 19.22 I-O characteristic of CAMP in ALC mode operation for different attenuation in DAT1
19.5.2 Travelling Wave Tube Amplifier (TWTA)
Travelling wave tube amplifier (TWTA) is one of the most economically costliest
subsystems which used most critical but highly matured technology for realization
[9]. In most of the satellite transponders, TWTAs are used as high power amplifier
(HPA) for its ability to provide higher output power with higher DC-to-RF efficiency
at higher frequency of operation over broader frequency band compared to solid-
state power amplifiers (SSPAs). A TWTA consists of travelling wave tube (TWT)
and electronic power conditioner (EPC). The EPC provides required DC voltage and
currents to the TWT and provides various controls and protection mechanisms of
the TWT. In a satellite transponder, TWTAs amplify signal taking from the output
of a channel amplifier and provides required output power. In most of the cases,
a predistortion linearizer is used in between the channel amplifier and TWTA to
minimize the effect of nonlinearity of a TWTA on communication system.
In travelling wave tube amplifier, amplification of microwave signal takes place
due to the continued interaction between the wave and the high-energy electron beam
travelling along the signal. Functional diagram of a TWT is shown in Fig. 19.23.
Structurally, a TWT can be divided into three sections: electron gun, slow-wave
structure and collector. The electron is generated by heating a cathode which travel
towards anode due to high electric field generated by applying a very high potential
difference in between anode and cathode. The electron beam after passing through
the helix is collected at the collector. To fulfil the requirement of continued interac-
tion of waves with the electron beam for long time, the microwave signal is passed
through a slow-wave structure through which the electron beam flows and the elec-
tron beam is focused applying a longitudinal static magnetic field using permanent
magnets as shown in Fig. 19.23. The helical slow-wave structure slowed down the
microwave signal with its phase velocity of about cp/2π r, where p is the pitch of the
helix of radius r. Due to the propagation of electromagnetic wave along the helix, a
longitudinal electric field will be generated. This time-varying electric field results
in velocity modulation in the electron beam passing through the helix. This velocity
modulation will result in bunching of electrons in regular intervals of one wavelength
Fig. 19.23 Functional diagram of TWT
Fig. 19.24 Typical I–O, gain and phase characteristic of TWTA
of the applied signal. Thus, with the condition of equal phase velocity of microwave
signal and electron beam, a continuous interaction takes place between the beams and
the waves in the helix and bunches grow as the beam moves ahead. This continued
interaction results in the amplification of microwave signal flowing through the helix
by picking power from the electron beam. The amplified microwave signal is then
coupled out of the helix at the output port. The collected electrons at the collector
dissipate its rest of the energy at the collector.
Most important RF performance parameters of a TWTA are operating frequency,
bandwidth, saturated output power, DC-to-RF efficiency and linearity. Capability
to provide higher output power of about 250 W at S-band and 150 W at Ka-band
with DC-to-RF efficiency about 65–70% for space grade TWTAs makes them very
attractive for using as HPA in high-power satellite transponders. However, it has
comparatively poor linearity, size and more mass compared to SSPAs. I–O char-
acteristic, gain and phase dependency on RF power level are shown in Fig. 19.24.
Typical values of gain compression and total phase shift are about 6.6 dB and 45°,
respectively. These nonlinearities lead to distortion in amplified signal. One way to
minimize the effect of these nonlinearities on amplified signal is to operate the TWTA
at power back-off condition. However, efficiency decreases with the increase of back-
off of TWTA. Practically, a linearizer is used with TWTA to minimize the effect of
nonlinearity of TWTAs and thus minimizes the requirement of TWTA back-off.
19.5.3 Solid-State Power Amplifier (SSPA)
Solid-state power amplifiers (SSPAs) are used as high power amplifiers where output
power requirement is less. SSPAs are advantageous for its better linearity, lower
mass and smaller footprint area compared to TWTAs. Wherever SSPAs are capable
to provide the required RF power level, it is preferable to use SSPAs. For example, at
L-, S- and C-bands, SSPAs as well as TWTAs are used depending on the output power
requirement. However, beyond C-band there is no option other than using TWTAs
for space communication due to the non-availability of SSPAs with required power
level. With the present technology, space qualified SSPA of output power up to 60 W
is realized using GaAs-based MESFET/HFET power devices. With the advancement
of GaN technology, presently devices are available to realize space qualified SSPAs
of output power level about 250 W at L- and S-bands and 120 W at C-band.
Block diagram of a solid-state power amplifier of output RF power 53 dBm
(200 W) is shown in Fig. 19.25. It consists of RF section, electronic power condi-
tioner (EPC) and bias and temperature control circuits. The RF section consists of
medium power amplifier and high power amplifier stages to provide output power
level of 200 W with overall gain of 53 dB. Important parameters of an SSPA are
operating frequency, bandwidth, output power level, DC-to-RF efficiency and lin-
earity. To achieve higher efficiency, in general, the final two amplifier stages operate
in Class-AB mode with harmonic tuning. However, the medium power amplifier
stages operate in Class-A mode to achieve overall better linearity without sacrificing
overall efficiency. The efficiency of a final stage amplifier at its maximum allowable
operation power level is comparable to the efficiency of a TWTA which is about
70%. However, overall efficiency of an SSPA is significantly reduced due to the use
of other amplifying stages to achieve gain comparable with a TWTA. For example,
the efficiency of a C-band 40 W GaAs-based SSPA (including EPC) is about 45%.
Another important aspect of design of a power amplifier is the protection of power
devices under intentional or unintentional RF overdrive condition. Lineup of the
power amplifier is so selected that full range of overdrive cannot pass to final ampli-
fier stage. Another option is the use of open-loop limiter or closed-loop overdrive
protection circuit to protect the devices from overdrive condition [10].
The bias and control circuits provide required bias voltages/currents to all the
devices which are temperature-controlled to achieve temperature-compensated gain
as well as output power level. Practically, it is required to increase the drain voltage
of the final amplifying FET device with the increase of temperature to keep the
output power constant over the operating temperature range. Photograph of an SSPA
realized in hybrid microwave integrated circuit (HMIC) is shown in Fig. 19.26.
Near the saturation point, SSPA behaves differently from TWTA. Typical I–O,
gain and phase characteristics of an SSPA are shown in Fig. 19.27. In case of TWTA,
the output power decreases when it operated beyond its saturated point; however, for
an SSPA, it remains nearly constant. SSPAs are not permitted to operate beyond its
2-dB gain compression point to ensure its reliability for space use. Excessive gate
current may flow when a power FET operates beyond its 2-dB gain compression
Fig. 19.25 Block diagram of a solid-state HPA
Fig. 19.26 Photograph of a solid-state power amplifier
point. Excessive flow of gate current for a power FET over a prolonged duration may
lead to failure of the device. However, low-power FETs may be operated at more gain
compression without compromising its life. Typical total phase shift for an SSPA is
about 20° up to its 2-dB gain compression point. Another important point is to be
noted that in case of SSPAs the phase of the output signal increases with the increase
of power level as shown in Fig. 19.27, whereas it decreases in case of TWTAs as
seen in Fig. 19.24.
Thus, it is clear that SSPAs are in general more linear than TWTAs when both
are operated at their respective rated output power conditions. In practice, an SSPA
Fig. 19.27 Typical I–O characteristics of an SSPA
without linearizer has about equally good overall linearity as linearity of a linearized
TWTA. Thus, there is very less scope for improvement of linearity of SSPAs using
a linearizer. More precisely, using linearizer, there is large scope of phase as well
amplitude nonlinearity improvement for TWTAs. However, there is only scope of
phase nonlinearity improvement for an SSPA.
19.6 Linearizer
Among the various types of linearizers, predistortion (PD) linearizers are mostly
used for satellite communication due to their simplicity, low power consumption,
and ability to linearize over wide bandwidth of a power amplifier operating at near
saturation condition [11–15]. Predistortion linearizer creates inverse nonlinearity of
the transmitting amplifiers such as of TWTAs or SSPAs in order to compensate for
the distortion. It is able to function as standalone unit and can be cascaded in between
CAMP and HPA with proper input and output power level matching.
Block diagram of a broadband predistortion linearizer is shown in Fig. 19.28. Heart
of the linearizer is the distortion generator module. In general, it is realized using
Schottky and p-i-n diodes in a vector modulator configuration as discussed in Chap.
16 to generate required amplitude and phase nonlinearities. Two amplitude tilt active
equalizers (EQ1 and EQ2) are used to make the linearizer broadband. An analog
attenuator (AAT1) is used to compensate overall gain variation of the equalizer over
its operating temperature range. The amplifier modules (A1, A2 and A3) are used to
match the linearizer’s input and output power levels with the output power level of
CAMP and input power level of HPA in addition to compensate loss of the distortion
generator, equalizers and attenuator. Photograph of a linearizer operating in Ku-band
is shown in Fig. 19.29. The linearizer is realized using MMIC amplifier modules and
other circuits realized on alumina substrate using HMIC technology. The distortion
generator unit is realized as vector modulator with one arm nonlinear circuit using
Schottky barrier diodes and the other arm with a linear circuit using p-i-n diodes. The
19.6 Linearizer 621
Fig. 19.28 Block diagram of a broadband predistortion linearizer
Fig. 19.29 Photograph of a Ku-band linearizer
amplitude tilt active equalizers and analog attenuator are realized using p-i-n diodes
as voltage/current variable resistors. Temperature compensation of the unit is done
using optimum load-line bias technique of Schottky barrier and p-i-n diodes as well as
providing temperature-controlled voltage/current to the analog attenuator (AAT1).
Typical nonlinear amplitude and phase performances of a Ku-band linearizer cas-
caded with a TWTA are shown in Fig. 19.30. It shows that gain compression of a non-
linearized TWTA improves from 6.5 to ±0.7 dB when it is cascaded and optimized
with linearizer. Similarly, nonlinearized TWTA’s total phase shift improves from
45° to ±4°. Thus, amplitude and phase nonlinearity of linearized TWTA (LTWTA)
improves significantly and comparable, even better than linearity of a nonlinearized
SSPA. Two-tone third-order intermodulation levels (IM 3 ) of TWTA and LTWTA are
shown in Fig. 19.31. It shows that there is no significant improvement of IM 3 level
near saturation region of the TWTA. However, there is a significant IM 3 improve-
ment at around 7 dB input back-off (IBO) of each carrier which corresponds to 4 dB
IBO of the TWTA with respect to its saturation point.
Fig. 19.30 Typical I–O, gain and phase characteristic of TWTA and Linearized TWTA
Fig. 19.31 Typical 3rd order IM levels of TWTA and LTWTA
19.7 Microwave Power Module (MPM)
Microwave power module (MPM) is the technology where both solid-state and vac-
uum tube technologies are combined to realize very compact with low mass transmit-
ter module to provide highest possible output power with highest efficiency [16, 17].
The solid-state technology is advantageous for its miniaturized configuration using
MMIC technology with excellent linearity. However, its output power giving capa-
bility and efficiency is less compared to TWTA. Presently, space qualified MMICs
with medium output power up to 10 W are available. On the other hand, TWTAs are
capable of providing highest output power level with high DC-to-RF efficiency. But
its linearity is poor and also size is more in case of high-gain TWTAs. In MPM, short-
length TWT is used as final power amplifier with lower gain which needs input power
about 1–10 W which is achievable from solid-state MMIC configuration. Thus, rest
19.7 Microwave Power Module (MPM) 623
Fig. 19.32 Block diagram of microwave power module (MPM)
of the transmitter gain with medium-level output power is achieved using solid-state
technology in MMICs. To improve the linearity of the total system, a predistortion
linearizer is realized in solid-state technology cascaded at suitable position as shown
in Fig. 19.32. To minimize the overall mass and footprint area, a single electronic
power conditioner is used for all these microwave subsystem and a single mechani-
cal assembly including all the RF units and the EPC. To realize a microwave power
module, challenge is the thermal management due to the very compact assembly of
the unit. EMI/EMC-related issues are also critical due to the proximity of the small
signal and high power units in a single package.
19.8 Multiport Amplifier (MPA)
Multiport amplifiers (MPAs), also known as matrix amplifiers, have multiple inputs
and multiple outputs with several high power amplifiers connected in parallel. These
are used to provide flexibility in terms of power allocation, since combined output
power from parallel-connected several amplifiers is shared between the output ports.
Here, RF power can be allocated among the output ports as per requirements of
a communication system. Therefore, the combined power of all the high power
amplifiers is available for any output port provided that the other ports do not require
any power at the same time. The multiport amplifier configurations are very useful
in case of multibeam satellite communication systems, where beam-to-beam service
requirements such as number of users/data rate changes dynamically over time. In
case of multiport amplifier configuration, all the power amplifiers provide same RF
output power irrespective of different powers required by the different output ports.
Thus, operating conditions (input/output back-off) of all the amplifiers remain the
same though RF powers taken from different output ports are different.
Fig. 19.33 Block diagram of a 4 × 4 MPA with signal flow paths (bold lines) when power fed to
only Port-I 1
A multiport amplifier (MPA) consists of an array of power amplifiers (PAs) in

parallel and a pair of complementary Butler matrix networks that consist of 90o
hybrid networks [18]. Schematic diagram of a 4 × 4 MPA is shown in Fig. 19.33.
The 4 × 4 MPA consists of four high power amplifiers connected in parallel using
4 × 4 input and 4 × 4 output networks. The signal at each input in the MPA is divided
into four signals (in general n signals) with particular phase relationships. These
signals are amplified separately in each power amplifier and are recombined in the
output network. Thus, signal at each input is amplified by all the power amplifiers,
however, assembled at the corresponding output ports. Figure 19.33 shows the flow
of signal when it is applied to the input port-1. Thus, the signal is applied in port-1
amplified by all the amplifiers and then recombined at the output port-1. In a similar
way, applied signal at input ports 2, 3 and 4 is amplified by all the amplifiers and
recombined at the respective output ports 2, 3 and 4. In ideal condition, signal applied
to the port-1 becomes available at the output port-1 only. Thus, no part of the signal
applied at the input port-1 will be available at the output ports 2, 3 and 4. To achieve
this ideal performance from an MPA, it is important to equalize the amplitude and to
synchronize the transmitted phase among the signals. Practically, infinite isolation
is not possible due to various nonideal electrical characteristics of each signal path.
The finite isolation leads to signal loss as well as interference among signals coming
from different input ports. Another drawback is the unwanted multicarrier operation
of the power amplifiers. This multicarrier operation reached as all the input signals
are amplified at each power amplifier even when a single carrier is fed at each input.
The multiport amplifier is adapted to applications that require flexibility in terms of
power allocation at its different output ports and is really advantageous if the input
signals are already multicarrier.
Design of compact input and output Butler matrix networks with minimum inser-
tion loss and proper amplitude and phase matching is crucial. The input network
operates at lower RF power level, and thus loss of the input network can be easily
19.8 Multiport Amplifier (MPA) 625
(a) (b)
Fig. 19.34 4 × 4 Butler matrix using 3-dB 90° hybrid couplers a with crossover, b planner structure
without crossover
Fig. 19.35 8 × 8 Butler matrix using 3-dB 90° hybrid couplers
compensated by providing more gain at the driver amplifier stages. It is always prefer-
able to realize input network in planar transmission line configuration very compactly
with the compromise of insertion loss. Figure 19.34 shows planner configuration of
a 4 × 4 Butler matrix with and without crossover connection and Fig. 19.35 shows
planner configuration of an 8 × 8 Butler matrix suitable for realization in microstrip
line configuration.
Any loss of the output butler network decreases total available output power.
Generally, waveguide-based output network is designed to achieve low insertion
Fig. 19.36 Block diagram of the system of Example 19.5
loss, though it is bulky. Followings are the various examples related to microwave
communication systems and subsystems.
Example 19.5 Derive the overall noise figure of a communication system which
consists of a receiver and transmitter as shown in Fig. 19.36. Calculate the overall
system noise figure for gain and noise figure of the receiver: 50 and 2 dB, respectively,
and gain and noise figure of the transmitter: 80 and 20 dB, respectively.
Solution
Gain of the receiver GRX 50 dB 105
Noise figure of the receiver NFRX 2 dB 1.585
Gain of the transmitter GTX 80 dB 108
Noise figure of the transmitter NFTX 20 dB 100
Suppose the available noise power at the input of the receiver is Pni ; this is the
thermal noise over the noise bandwidth B of the receiver and is given by
Pni kTo B
Thus, the noise power output, PnoRX , of the receiver can be written as
PnoRX PniTX (Pni ) × (GRX NFRX )

Pni × GRX + (NFRX − 1)Pni × GRX
The first term is the output noise due to the amplification of the input available
thermal noise, and the second part is the noise added by the receiver (Fig. 19.36).
In the similar way, the total noise power at the output of the transmitter can be
written as
PnoTX PniTX × GTX + (NFTX − 1)Pni × GTX

PnoTX Pni × GRX × NFRX × GTX + (NFTX − 1)Pni × GTX (19.10a)
Suppose NFRXTX is the overall noise figure of the system. The overall gain of the
system is GRX × GTX . Thus, the noise power at the output of the transmitter can be
written as
PnoTX (Pni ) × (GRX × GTX × NFRXTX ) (19.10b)

Comparing (19.10a) and (19.10b), the overall system noise figure can be written
as
(NFTX − 1)
NFRXTX NFRX + (19.11)
GRX
This is known the Friis’s formula, already derived in Chap. 14. This formula shows
that in case of sufficiently high receiver gain, the overall noise factor is dominated
by the noise factor of the receiver.
Putting the values, the overall noise factor of the system
(100 − 1)
NFRXTX 1.585 + 1.586
100,000
2.003 dB
This shows that though the noise figure of the transmitter is poor, its effect on the
overall system is negligible due to the high gain of the receiver.
Example 19.6 Derive overall power-added efficiency of a communication system

which consists of a receiver and transmitter as shown in Fig. 19.37. Calculate the
overall power-added efficiency of the system for gain- and power-added efficiency
of the receiver: 50 dB and 1%, respectively, and gain- and power-added efficiency
of the transmitter: 80 dB and 50%, respectively.
Solution
Gain of the receiver GRX 50 dB 105
Power-added efficiency of the receiver ηRX 1% 0.01
Gain of the transmitter T RX 80 dB 108
Power-added efficiency of the transmitter ηTX 50% 0.5
Suppose
DC power to the receiver PDCRX
DC power to the transmitter PDCTX
Input RF power of the receiver PINRX
Input RF power of the transmitter PINTX PORX
Output RF power of the transmitter POTX
From the definition of power-added efficiency,

PORX − PINRX
ηRX (19.12a)
PDCRX
POTX − PINTX
ηTX (19.12b)
PDCTX
The overall power-added efficiency of the system ηRXTX can be written as

POTX − PINRX
ηRXTX (19.12c)
PDCRX + PDCTX
Putting GRX PORX /PINRX , GTX POTX /PINTX and using (19.12a), (19.12b),
the overall efficiency of the system can be written as
1 GRX − 1 1 (GTX − 1)GRX 1
× + × (19.13)
ηRXTX GRX GTX − 1 ηRX GRX GTX − 1 ηTX
Putting GRX 105 , ηRX 0.01, TRX 108 and ηTX 0.5
8
1 105 − 1 1 10 − 1 105 1
13 × + ×
ηRXTX 10 − 1 0.01 1013 − 1 0.5
or
ηRXTX 0.5 50%
This shows that though the power-added efficiency of the receiver is poor, its
effect on the overall system is negligible due to the high gain of the transmitter and
receiver.
Considering GRX 1 and GTX 1, the overall power-added efficiency of the
system (19.13) can be written as
1 GRX 1 GTX GRX 1
× + × (19.14a)
ηRXTX GRX GTX ηRX GRX GTX ηTX
1 1 1 1
× + (19.14b)
ηRXTX GTX ηRX ηTX
or
ηRXTX ηTX (19.14c)
Example 19.7 Derive worst-case overall two-tone third-order intermodulation level

of a communication system which consists of a receiver and transmitter as shown
in Fig. 19.38. Calculate the overall two-tone third-order intermodulation level of the
system for IM 3 of the receiver and transmitter, which are (a) 20 and 10 dBc (b) 10
and 10 dBc, respectively.
Solution
Suppose
Output RF power of the receiver PORX PINTX
Output RF power of the transmitter POTX
Therefore, level of IM3RX at the output of the receiver is
PORX × 10−(IM3RX /10) (19.15a)
Here, consider that the third-order intermodulation signal generated by the

receiver will be linearly amplified by the transmitter. Thus, at the output of the
transmitter, the third-order intermodulation signal level will be
10−(IM3RX /10) × PORX × GTX + POTX × 10−(IM3TX /10) (19.15b)

POTX × 10−(IM3RX /10) + 10−(IM3TX /10) (19.15c)
Here, the first term is the contribution by the receiver and the second term is the
contribution by the transmitter.
Thus, the overall IM 3 level

POTX × 10−(IM3RX /10) + 10−(IM3TX /10)
IM3RXTX
POTX
10−(IM3RX /10) + 10−(IM3TX /10)

[IM3RXTX ]dB 10 log 10−(IM3RX /10) + 10−(IM3TX /10) (19.15d)
(a) Putting the value of IM3RX 20 dBc and IM3TX 10 dBc

[IM3RXTX ]dB 10 log 10−(20/10) + 10−(10/10)
10 log(0.01 + 0.1)
10 log(0.01 + 0.1) 9.59 dB
(b) Putting the value of IM3RX 10 dBc and IM3TX 10 dBc

[IM3RXTX ]dB 10 log 10−(10/10) + 10−(10/10)
10 log(0.1 + 0.1)
−6.99 dB
This example shows that effect of nonlinearity of receivers and transmitters

equally affects the overall linearity of the system.
Example 19.8 Derive expressions for output powers of a transmitter of I-O charac-
teristic governed by vo a1 vi + a3 vi3 for a two-tone carrier input, vi A cos ω1 t +
B cos ω2 t. Considering the transmitter is matched at its input and output ports with
Ro 50 , and a1 10, a3 −0.04, calculate the output power levels corre-
sponding to the fundamental and third-order harmonics for the following cases:
(a) For total input power level of 0 dBW of equal power levels of two carriers.
(b) For total input power level of 0 dBW of unequal power levels of two carriers by
6 dB.
(c) For total input power level of 0 dBW of unequal power levels of two carriers by
10 dB.
Also, plot gains of both the carriers over the total input power level of −20 to
0 dBW for the case-b and case-c.
Solution
I–O characteristic of the transmitter is given by
vo a1 vi + a3 vi3 (19.16)
The two carriers’ input excitation is
vi A cos ω1 t + B cos ω2 t (19.17)
Therefore, the output is given by
vo a1 (A cos ω1 t + B cos ω2 t) + a3 (A cos ω1 t + B cos ω2 t)3 (19.18)

3 3
vo a3 A2 B + a3 AB2
2 2
3 3
+ a1 A + a3 A3 + a3 AB2 cos ω1 t
4 2

3 3
+ a1 B + a3 B3 + a3 A2 B cos ω2 t
4 2

1 1
+ a3 A3 cos 3ω1 t + a3 B3 cos 3ω2 t
4 4

3
+ a3 A2 B [cos(2ω1 − ω2 )t + cos(2ω1 + ω2 )t]
4

3
+ a3 AB2 [cos(2ω2 − ω1 )t + cos(2ω2 + ω1 )t] (19.19)
4
Amplitude of the output voltage corresponding (vo1 ) to carrier ω1 is given by

3 3
voω1 a1 A + a3 A3 + a3 AB2
4 2
Thus, the output power corresponding to carrier ω1 is given by
2
1 voω1 2 3 3
Poω1 √ a1 A + a3 A + a3 AB /2Ro
3 2
(19.20a)
Ro 2 4 2
Similarly, the output power corresponding to carrier ω2 is given by

2
1 voω2 2 3 3
Poω2 √ a1 B + a3 B3 + a3 A2 B /2Ro (19.20b)
Ro 2 4 2
The output power corresponding to each third-order intermodulation product

(2ω1 ± ω2 ) is given by
2
3
Po(2ω1 ±ω2 ) a3 A2 B /2Ro (19.21a)
4
Similarly, the output power corresponding to each third-order intermodulation

product (2ω1 ± ω2 ) is given by
2
3
Po(2ω2 ±ω1 ) a3 AB2 /2Ro (19.21b)
4
The input power corresponding to carrier ω1 is given by

√ 2
Piω1 A/ 2 /Ro A2 /2Ro (19.22a)
The input power corresponding to carrier ω2 is given by

√ 2
Piω2 B/ 2 /Ro B2 /2Ro (19.22b)
Therefore, total input power is given by

PiTotal A2 + B2 /2Ro (19.23)
(a) For equal power levels of both the carriers with PiTotal 0 dBW 1 W. Thus,
from (19.23),
A B 7.071 V
Fig. 19.39 Power levels under two-tone excitation of equal input power levels of Example 19.8
Here, A B 7.071 V and a1 10, a3 −0.04 and Ro 50 .

From (19.22a) and (19.22b),
Piω1 Piω2 A2 /2Ro 7.0712 /100 0.5 W −3.01 dBW
From (19.20a) and (19.20b),

2
3 3
Poω1 Poω2 a1 A + a3 A3 + a3 AB2 /2Ro
4 2
2
3 3
10 × 7.071 − × 0.04 × 7.0713 − × 0.04 × 7.071 × 7.0712 /100
4 2
15.13 W 11.80 dBW
From (19.21a), (19.21b),

2
3
Po(2ω1 ±ω2 ) Po(2ω2 ±ω1 ) a3 A2 B /2Ro
4
2
3
× 0.04 × 7.0713 /100
4
1.12 W 051 dBW
The input and output power levels of the two equal power carriers are shown in
Fig. 19.39. Under the two-tone carrier excitation of equal power levels, the output
power levels for both the fundamental frequency components are same and also the
power levels of both the third-order IMD components are same. The levels of the
third-order IMD components are
Po(2ω1 ±ω2 ) − Poω1 Po(2ω2 ±ω1 ) − Poω2

(0.51) dBW − 11.80 dBW −11.29 dBc
(b) For unequal power levels by 6 dB with PiTotal 0 dBW 1 W.
From (19.22a) and (19.22b),

Piω1 − Piω2 10 log A2 /2Ro − 10 log B2 /2Ro 6 dB (19.24a)
And from (19.23),

PiTotal 10 log A2 + B2 /2Ro 0 dBW (19.24b)
From (19.24a) and (19.24b),
A 8.94 V, B 4.48 V
The input powers corresponding to carriers ω1 and ω2 are given by,
Piω1 A2 /2Ro 8.942 /100 0.799 W −0.973 dBW

Piω2 B2 /2Ro 4.482 /100 0.201 W −6.973 dBW
From (19.20a),
2
3 3
Poω1 a1 A + a3 A3 + a3 AB2 /2Ro
4 2
2
3 3
10 × 8.94 − × 0.048.94 − × 0.04 × 8.94 × 4.48 /100
3 2
4 2
32.71 W 15.15 dBW
From (19.20b),
2
3 3
Poω2 a1 B + a3 B3 + a3 A2 B /2Ro
4 2
2
3 3
10 × 4.48 − × 0.04 × 4.48 − × 0.04 × 8.94 × 4.48 /100
3 2
4 2
4.25 W 6.29 dBW
From (19.21a),
2
3
Po(2ω1 ±ω2 ) a3 A2 B /2Ro
4
Fig. 19.40 Power levels under two-tone excitation of unequal input levels by 6 dB of Example
19.8
2
3
× 0.04 × 8.942 × 4.48 /100
4
1.15 W 0.62 dBW
From (19.21b),
2
3
Po(2ω2 ±ω1 ) a3 AB2 /2Ro
4
2
3
× 0.04 × 8.94 × 4.482 /100
4
0.29 W −5.38 dBW
The input and output power levels of the two unequal power level by 6 dB are
shown in Fig. 19.40. The difference of output power levels among the fundamental
carriers increases to 8.86 dB from the difference of 6 dB at input. This is due to
the power transfer to the various harmonics and intermodulation components. This
phenomenon is known as ‘power robbing’ when multiple carriers amplified by an
amplifier operating in its nonlinear (gain compression) region. Though both the carri-
ers pass through the same amplifier, they experience different levels of amplification.
Gain responses of both the carriers are shown in Fig. 19.41 over the input power
level of −20 to 0 dBW. At lower power levels where the amplifier operates in linear
region, the gains of both the carriers are same. But over the nonlinear region of the
Fig. 19.41 Gain of the carriers with unequal input power levels by 6 dB of Example 19.8
amplifier, the weaker signal amplifies less compared to the stronger signal as shown
in Fig. 19.41.
(c) For unequal power levels by 10 dB with PiTotal 0 dBW 1 W.
From (19.22a) and (19.22b),

Piω1 − Piω2 10 log A2 /2Ro − 10 log B2 /2Ro 10 dB (19.24a)
And from (19.23),

PiTotal 10 log A2 + B2 /2Ro 0 dBW (19.24b)
From (19.24a) and (19.24b),
A 9.535 V, B 3.015 V
The input powers corresponding to carriers ω1 and ω2 are given by
Piω1 A2 /2Ro 9.5352 /100 0.909 W −0.414 dBW

Piω2 B2 /2Ro 3.0152 /100 0.091 W −10.414 dBW
From (19.20a),
2
3 3
Poω1 a1 A + a3 A + a3 AB /2Ro
3 2
4 2
2
3 3
10 × 9.535 − × 0.04 × 9.5353 − × 0.04 × 9.535 × 3.0152 /100
4 2
41.142 W 16.14 dBW
From (19.20b),
2
3 3
Poω2 a1 B + a3 B + a3 A B /2Ro
3 2
4 2
2
3 3
10 × 3.015 − × 0.04 × 3.015 − × 0.04 × 9.535 × 3.015 /100
3 2
4 2
1.660 W 2.20 dBW
From (19.21a),
2
3
Po(2ω1 ±ω2 ) a3 A2 B /2Ro
4
2
3
× 0.04 × 9.5352 × 3.015 /100
4
0.68 W −1.70 dBW
From (19.21b),
2
3
Po(2ω2 ±ω1 ) a3 AB2 /2Ro
4
2
3
× 0.04 × 9.535 × 3.0152 /100
4
0.07 W −11.70 dBW
The input and output power levels of the two carriers with unequal power levels
by 10 dB are shown in Fig. 19.42. Here, it can be noted that due to the power
robbing phenomenon, the difference of output power levels of the fundamental carrier
increases to 13.94 dB compared to the difference of 10 dB at input. Gain responses of
both the carriers are shown in Fig. 19.43 over the input power level of −20 to 0 dBW.
It can be noted that the gain of the weaker carrier becomes 3.94 dB lower compared
to the gain of the stronger carrier at 0 dBW total input power level. Thus, weak signal
becomes further weak when it passes through a nonlinear (gain compression region)
amplifier in the presence of stronger signal.
Example 19.9 Write the expressions for output powers of a transmitter of IO
characteristic governed by vo a1 vi + a3 vi3 for a two-tone carrier input, vi
A cos ω1 t + B cos ω2 t. Considering the transmitter is matched at its input and out-
put ports with Ro 50 , and a1 4, a3 0.02, calculate the output power
levels corresponding to fundamental components and third-order harmonics for the
following cases:
(a) For total input power level of 0 dBW of equal power levels of two carriers.
(b) For total input power level of 0 dBW of unequal power levels of two carriers by
10 dB.
19.8
Fig. 19.43 Gain of the carriers with unequal input levels by 10 dB of Example 19.8
Also, plot gains of both the carriers over the total input power level of −20 to
0 dBW for the case-b.
Solution
I–O characteristic of the transmitter is given by
vo a1 vi + a3 vi3
Under the two-carrier input excitation of vi A cos ω1 t + B cos ω2 t, the output

power corresponding to carrier ω1 is given by (19.20a)
2
1 voω1 2 3 3
Poω1 √ a1 A + a3 A + a3 AB /2Ro
3 2
Ro 2 4 2
Similarly, the output power corresponding to carrier ω2 is given by (19.20b)

2
1 voω2 2 3 3
Poω2 √ a1 B + a3 B3 + a3 A2 B /2Ro
Ro 2 4 2

(2ω1 ± ω2 ) is given by (19.21a)
2
3
Po(2ω1 ±ω2 ) a3 A2 B /2Ro
4

(2ω1 ± ω2 ) is given by (19.21b)
2
3
Po(2ω2 ±ω1 ) a3 AB2 /2Ro
4
The input power corresponding to carrier ω1 is given by (19.22a)

√ 2
Piω1 A/ 2 /Ro A2 /2Ro
The input power corresponding to carrier ω2 is given by (19.22b)

√ 2
Piω2 B/ 2 /Ro B2 /2Ro
Therefore, total input power is given by (19.23)

PiTotal A2 + B2 /2Ro
(a) For equal power levels of both the carriers with PiTotal 0 dBW 1 W. From
(19.23),
A B 7.071 V
Here, A B 7.071 V and a1 4, a3 +0.02 and Ro 50 .

From (19.22a) and (19.22b),
Piω1 Piω2 A2 /2Ro 7.0712 /100 0.5 W −3.01 dBW

Fig. 19.44 Power levels under two-tone excitation of equal input levels of Example 19.9
From (19.20a) and (19.20b),

2
3 3
Poω1 Poω2 a1 A + a3 A3 + a3 AB2 /2Ro
4 2
2
3 3
4 × 7.071 + × 0.02 × 7.0713 + × 0.02 × 7.071 × 7.0712 /100
4 2
19.53 W 12.91 dBW
From (19.21a) and (19.21b),

2
3
Po(2ω1 ±ω2 ) Po(2ω2 ±ω1 ) a3 A2 B /2Ro
4
2
3
× 0.02 × 7.0713 /100
4
0.28 W −5.51 dBW
The input and output power levels of the two equal power carriers are shown in
Fig. 19.44. Under the two-tone carrier excitation of equal power levels, the output
power levels for both the fundamental frequency components are same and also the
power levels of both the third-order IMD components are same. The levels of the
third-order IMD components are
Po(2ω1 ±ω2 ) − Poω1 Po(2ω2 ±ω1 ) − Poω2

−5.51 dBW − 12.91 dBW −18.42 dBc
(b) For unequal power levels by 10 dB with PiTotal 0 dBW 1 W.
Thus, from (19.22a) and (19.22b),

Piω1 − Piω2 10 log A2 /2Ro − 10 log B2 /2Ro 10 dB
And from (19.23),

PiTotal 10 log A2 + B2 /2Ro 0 dBW
From (19.24a) and (19.24b),
A 9.535 V, B 3.015 V
The input powers corresponding to carriers ω1 and ω2 are given by
Piω1 A2 /2Ro 9.5352 /100 0.909 W −0.414 dBW

Piω2 B2 /2Ro 3.0152 /100 0.091 W −10.414 dBW
From (19.20a),
2
3 3
Poω1 a1 A + a3 A3 + a3 AB2 /2Ro
4 2
2
3 3
4 × 9.535 + × 0.02 × 9.535 + × 0.02 × 9.535 × 3.015 /100
3 2
4 2
28.88 W 14.61 dBW
From (19.20b),
2
3 3
Poω2 a1 B + a3 B + a3 A B /2Ro
3 2
4 2
2
3 3
4 × 3.015 + × 0.02 × 3.0153 + × 0.02 × 9.5352 × 3.015 /100
4 2
4.28 W 6.32 dBW
From (19.21a),
2
3
Po(2ω1 ±ω2 ) a3 A2 B /2Ro
4
2
3
× 2.02 × 9.5352 × 3.015 /100
4
0.17 W −7.72 dBW
19.9
From (19.21b),
2
3
Po(2ω2 ±ω1 ) a3 AB2 /2Ro
4
2
3
× 0.02 × 9.535 × 3.0152 /100
4
0.02 W −17.72 dBW
The input and output power levels of the two unequal power levels by 10 dB are
shown in Fig. 19.45. Here, the difference of output power levels of the fundamental
carrier deceases to 8.29 dB from the difference of 10 dB at input. Gain responses of
both the carriers are shown in Fig. 19.46 over the input power level of −20 to 0 dBW.
It can be noted that the gain of the weaker carrier becomes 1.71 dB higher compared
to the gain of the stronger carrier at 0 dBW total input power level.
Example 19.10 A transceiver consists of a receiver, ALC driver amplifier and high
power amplifier as shown in Fig. 19.47. The ALC driver amplifier controls the gain
of the transponder from 110 to 140 dB depending on its input power level to operate
the transponder at saturation condition over its 30 dB dynamic range.
Calculate the noise and carrier powers at the transponder output when the
transponder operates in saturated condition for (a) minimum and (b) maximum gain
considering the following parameters:
System noise temperature (Ts ) 650 K
Transponder noise bandwidth (BN ) 100 MHz
Saturated gain of the HPA 50 dB
Saturated output power of HPA 140 W
Fig. 19.46 Gain of the carriers with unequal input levels by 10 dB of Example 19.9
Fig. 19.47 Gain of the carriers with unequal input levels by 6 dB
Plot carrier and noise power at the output of the transponder over its gain of
110–140 dB considering the transponder are operating at saturation condition.
Solution
The system noise temperature (Ts ): 650 K
Input noise power (PN in ) over the frequency band of 100 MHz:
PN in kTs BN
1.38 × 10−23 × 650 × 100 × 106 W
−90.47 dBm (19.25)
(a) At minimum gain of the transponder:

Total input noise power PN in −90.47 dBm
Gain of the transponder: 110 dB
Thus,
Total output noise power PN out (−90.47 + 110) dBm

19.53 dBm
0.09 W
Saturated output power of the transponder: 140 W
Considering saturated output power of the HPA consists of amplified noise and
carrier power only (neglecting intermodulation powers), the carrier power can be
written as
output saturated power − output noise power

140 W − 0.09 W
139.91 W
51.46 dBm
(b) At maximum gain of the transponder:

Total input noise power PN in −90.47 dBm
Gain of the transponder: 140 dB
Thus,
Total output noise power PN out (−90.47 + 140) dBm

49.53 dBm
89.70 W
Saturated output power of the transponder: 140 W
Considering saturated output power of the HPA consists of amplified noise and
carrier power only (neglecting contribution of intermodulation powers) and assuming
saturated power of HPA is same for single and multicarrier conditions, the carrier
power can be written as
output saturated power − output noise power

140 W − 89.70 W
50.30 W
47.02 dBm
Plot of the output carrier and noise power of the transponder over its entire gain
range of 110–140 dB is shown in Fig. 19.48.
Example 19.11 Show the use of 3-dB 90° hybrid coupler to separate RHCP and
LHCP signals from an antenna.
Solution
Electric field of a right-hand circular polarized (RHCP) signal can be expressed as
E( r , t) Ea sin(ωa t − ka z)êx + Ea cos(ωa t − ka z)êy (19.26)

Fig. 19.48 Carrier and noise powers versus gain at output of the transponder
Fig. 19.49 Scheme for the separation of LHCP and RHCP signals using 3-dB 90° hybrid
And the electric field of another left-hand circular polarized (LHCP) signal can
be expressed as
E( r , t) −Eb sin(ωb t − kb z)êx + Eb cos(ωb t − kb z)êy (19.27)
Two input ports of a 3-dB 90° hybrid coupler are connected to the two ports of an
antenna as shown in Fig. 19.49. Here, horizontal and vertical feeds are connected to
the two inputs (1 and 2) of the hybrid. At the port-1 (say, z 0), the signal voltage
can be written as
v1 (t) va sin(ωa t) − vb sin(ωb t) (19.28)
Similarly, at the port-2 (say, z 0), the signal voltage can be written as
v2 (t) va cos(ωa t) + vb cos(ωb t) (19.29)

Thus, the signal voltage at port-3 can be written as

va vb
v3 (t) √ sin(ωa t − π) − √ sin(ωb t − π )
2 2
va π vb π
+ √ cos ωa t − + √ cos ωb t −
2 2 2 2
va vb va vb
− √ sin(ωa t) + √ sin(ωb t) + √ sin(ωa t) + √ sin(ωb t)
2 2 2 2
vb vb
√ sin(ωb t) + √ sin(ωb t)
2 2
√
2vb sin(ωb t) (19.30)
This shows that the output port-3 of the hybrid contains only the frequency com-
ponent corresponding to LHCP.
Thus, the signal voltage at the port-4 can be written as
va π vb π
v3 (t) √ sin ωa t − − √ sin ωb t −
2 2 2 2
va vb
+ √ cos(ωa t − π ) + √ cos(ωb t − π )
2 2
va vb va vb
− √ cos(ωa t) + √ cos(ωb t) − √ cos(ωa t) − √ cos(ωb t)
2 2 2 2
va va
− √ cos(ωa t) − √ cos(ωa t)
2 2
√
2va sin(ωa t) (19.31)
This shows that the output port-4 of the hybrid contains only the frequency com-
ponent corresponding to RHCP.
In a similar way, it can be shown that two signals of different frequency bands
can be transmitted one in LHCP and other in RHCP through a single antenna using
a 3-dB 90° hybrid.
References
1. Wang C-X et al (2014) Cellular architecture and key technologies for 5G wireless communi-
cation networks. IEEE Commun Mag, Feb 2014
2. Pandian JD, Baker L, Cortes G, Goldsmith PF, Deshpande AA, Ganesan R, Hagen J, Locke L,
Wadefalk N, Weinreb S (2006) Low-noise 6-8 GHz receiver. IEEE Microw Mag 7:74–84
3. Garg VK, Singh RV, Jain VK, Bera SC (1999) New amplifier design eliminates chip capacitors.
IETE Tech Rev 16(6):197–201
4. Bera SC, Shah LB, Raval DU, Pandey S, Kumar V, Singh S, Das DK (2012) Design and
development of V&W band amplifier modules. In: Proceedings of international conference on
microwaves, antenna, propagation & remote sensing, 2012, Jodhpur, India
5. Buch SD, Bera SC (2013) Transponder nonlinearity characterization & mitigation techniques:
present scenario & future trends. In: International workshop on sensor network and wireless
communication, ADIT, Oct 2013
6. Yadav SP, Bera SC (2014) Nonlinearity effect of high power amplifiers in communication sys-
tems. In: International conference on advances in communication and computing technologies
(ICACACT), Aug 2014
7. Yadav SP, Bera SC (2014) Nonlinearity effects of power amplifiers in wireless communication
systems. In: Proceedings of IEEE international conference on electronics, communication and
computational engineering (ICECCE 2014), Hosur, India, pp 1011–1016, Nov 2014
8. Bera SC, Singh RV, Garg VK (2006) Design and temperature compensation of a Ku-band
channel amplifier with ALC for a satellite transponder. Microw J 49(4):68–82
9. Hu Y, Feng J (2016) The development and new trends of microwave vacuum electronic devices.
In: IEEE international conference on emerging technologies (ICET), pp 1–5, Oct 2016
linearizer using a parallel diode with a bias feed resistance. IEEE Trans Microw Theory Tech
45(12):2431–2435
12. Bera SC, Bhardhwaj PS, Singh RV, Garg VK (2003) A diode linearizer for microwave power
amplifiers. Microw J 46(11):102–113
13. Bera SC, Singh RV, Garg VK (2004) A compact Ku-band linearizer for space application. In:
Proceedings of Asia Pacific microwave conference, 2004, pp 37–38
14. Bera SC, Singh RV, Garg VK (2008) Diode-based predistortion lineariser for power amplifiers.
IEE Electron Lett 44(2):125–126
15. Bera SC, Kumar V, Singh S, Das DK (2013) Temperature behavior and compensation of diode-
based predistortion linearizer. IEEE Microw Wirel Compon Lett 23(4):211–213
16. Abrams RH, Parker RK (1993) Introduction to the MPM: what it is and where it might fit.
IEEE MTT-S Int Microw Symp Dig 1:107–110
17. Kowalczyk R, Zubyk A, Meadows C et al (2016) High efficiency E-band MPM for communi-
cations application. In: 17th IEEE international vacuum electronics conference, pp 513–514,
2016
18. Bera SC, Singh RV, Garg VK (2008) Modified Wilkinson power divider with harmonic sup-
pression characteristic. Microw J, Nov 2008
Chapter 20
Multiple Choice Questions with Answers
20.1 Multiple Choice Questions
(1) Electron mobility of the following undoped elements in descending order is

(A) GaAs, Ge, Si
(B) GaAs, Si, Ge
(C) Si, Ge, GaAs
(D) Ge, Si, GaAs
(2) Common-base current gain of a p-n-p bipolar transistor is 0.99. The common-
emitter current gain of the transistor is
(A) 101
(B) 0.01
(C) 99
(D) 1.0
(3) Over a wide operating temperature range, around ambient temperature, equire-
sistance curves for a practical p-i-n diode can be considered as
(A) Straight line
(B) Parabolic
(C) Elliptic
(D) Circular
(4) Open-circuit voltage for temperature-invariant RF resistance of a silicon p-i-n
diode is approximately
(A) 0.61 V
(B) 1.15 V
(C) 2.25 V
(D) 3.8 V
648 20 Multiple Choice Questions with Answers
(5) Open-circuit voltage for temperature-invariant RF resistance of a GaAs p-i-n

diode is approximately
(A) 0.61 V
(B) 1.12 V
(C) 1.42 V
(D) 3.8 V
(6) An attenuator circuit consists of two silicon p-i-n diodes. The diodes are con-
nected in series for DC bias. The open-circuit voltage for temperature-invariant
attenuation of the attenuator is approximately
(A) 0.61 V
(B) 1.12 V
(C) 1.42 V
(D) 2.3 V
(7) An attenuator circuit consists of two silicon p-i-n diodes. The diodes are con-
nected in parallel for DC bias. The open-circuit voltage for temperature-invariant
attenuation of the attenuator is approximately
(A) 0.61 V
(B) 1.15 V
(C) 1.42 V
(D) 2.3 V
(8) 1 W power in dBm and dBW, respectively, are
(A) 30 and 0
(B) 0 and 30
(C) 1 and 0
(D) 0 and 1
(9) 10 mW power in dBm is

(A) 10
(B) 1
(C) 0.1
(D) 0
(10) 50 dBm power in Watt is
(A) 10
(B) 100
(C) 1000
(D) 10,000
20.1 Multiple Choice Questions 649
(11) 30 dBW power in Watt is
(A) 10
(B) 100
(C) 1000
(D) 10,000
(12) 30 dBm power in Watt is
(A) 100
(B) 10
(C) 1
(D) 0
(13) 0 dBW power in Watt is
(A) 10
(B) 1
(C) 100
(D) 0.1
(14) Common-emitter DC current gain of a Si transistor is 100. The current through

the 10 V Zener diode is
(A) 10.7 mA
(B) 19.3 mA
(C) 20 mA
(D) 40 mA
(15) The circuit has an ideal operational amplifier with ±10 V supply. The output
voltage is
(A) −200 mV
(B) −400 mV
(C) −600 mV
(D) −300 mV
(16) The effective resistance faced by the voltage source is
(A) 4
(B) 12
(C) 3
(D) 16
(17) The voltage across the resistor R is

√
(A) 1/√3 0° V
(B) 1/ 2 0° V
(C) 1/2√ 45° V
(D) 1/ 2 45° V
(18) A step voltage of 1 V is applied to the circuit at t 0. The currents in mA
through the resistor R just after t 0 and at steady state, respectively, are
(A) 100, 50
(B) 50, 50
(C) 100, 100
(D) 75, 75
(19) Under constant voltage bias condition, how does the bias current of a Schottky
diode change with temperature?
(A) Current decreases with the increase in temperature
(B) Current increases with the increase in temperature
(C) Current remains constant with the change in temperature
(D) At low temperature, current increases and at higher temperature, it
decreases with the increase in temperature
(20) Which of the following diodes is the most suitable to realize a voltage-variable
microwave attenuator?
(A) Schottky barrier diode
(B) Varactor diode
(C) p-i-n diode
(D) P-N junction diode
(21) A BJT is having common-emitter current gain 100. Considering 10 V supply,
V BE 0.7 V and RE 100 , what is the value of RB and RC to set the quiescent
point at I C 10 mA and V CE 8 V.
(A) RB 83 k, RC 200

(B) RB 93 k, RC 200
(C) RB 93 k, RC 100
(D) RB 83 kΩ, RC 100 Ω
(22) Amplifier-A has 6 dB gain and 3.8 dB NF. Amplifier-B has 20 dB gain and
4 dB NF. What should be the sequence of the cascaded two-stage amplifier for
achieving the lowest NF? What will be the value of the cascaded lowest NF?
(A) Input stage: amplifier-B, output stage: amplifier-A, 4.02 dB
(B) Input stage: amplifier-A, output stage: amplifier-B, 3.85 dB
(C) Input stage: amplifier-A, output stage: amplifier-B, 4.44 dB
(D) Input stage: amplifier-B, output stage: amplifier-A, 3.85 dB
(23) Input impedance of a λ/8 lossless transmission line having characteristic
impedance 13 and terminated with a load impedance of (5 + j12) is
(A) 13
(B) 65
(C) (18 + j12)
(D) (18 + j13)
(24) In an ideal Class-F microwave power amplifier, time domain voltage and current
waveforms of the device have
(A) Maximum overlap
(B) 50% overlap
(C) No overlap
(D) Less than 80% overlap
(25) The equivalent circuit of a FET is shown, where rg 100 , cgs
0.1 pF, and gm 0.1 mho. For load resistance rL 500 ,, the voltage
gain at 1 GHz is
(A) 30
(B) 20
(C) 100
(D) 50
(26) A FET power amplifier operating in Class-A with 10 V drain supply and 2 A
drain current bias. The RF load current is of amplitude 1 A. What is the DC to
RF efficiency for the load resistance of 5 ?
(A) 12.5%
(B) 25%
(C) 35%
(D) 50%
(27) A FET power amplifier operating in Class-B with 15 V drain supply. The ampli-
fier provides 12 V peak signal to a 15 load. DC to RF efficiency of the amplifier
is
(A) 8.5%
(B) 62.8%
(C) 50%
(D) 75%
(28) A high-gain MESFET packaged device of case-to-channel thermal resis-

tance has 4.5 °C/W and provides 8 W RF output, taking 20 W DC power.
What is the channel temperature of the device if the case temperature of the
device maintained at 55 °C?
(A) 145 °C
(B) 59.5 °C
(C) 109 °C
(D) 91 °C
(29) Main function of helix structure in a travelling wave tube (TWT) amplifier is
(A) To reduce the axial velocity of electron

(B) To properly focus the electron beam
(C) To reduce noise figure of TWT
(D) To reduce the axial velocity of RF field
(30) A load of the reflection coefficient ΓL is transformed to ΓIN by suitably con-

necting a circuit element as shown in the figure. The circuit element is
(A) Series-connected inductor

(B) Series-connected capacitor
(C) Series-connected the transmission line
(D) Shunt-connected inductor
(31) S-parameter matrix of a lossless transmission line of length λ/8 is

0 1 −45◦
(A)
1 −45◦ 0

0 1 −90◦
(B)
1 −90◦ 0

0 1 −22.5◦
(C)
1 −22.5◦ 0

0 1 −180◦
(D)
1 −180◦ 0
(32) A unilateral transistor has output impedance Zout (32 + j30) . Value of the
series and shunt components of the matching network (as shown in the figure)
for the complex conjugate match at the output of the device to a 50 load is
(A) X 1 j40, X 2 j10, X 3 −j40

(B) X1 j40, X2 j10, X3 +j40
(C) X1 −j40, X2 j10, X3 −j40

(D) X1 j40, X2 −j40, X3 −j40
(33) Transmission coefficient ‘S 21 ’ of the 2-port network shown in the figure, for a
50 system is
(A) 2/3 0◦
(B) 1/2 0◦
(C) 2/3 180◦
(D) 1/2 180◦

(34) Input impedance of a quarter-wavelength 50 transmission line when termi-
nated by 25 is
(A) 100
(B) 200
(C) 50
(D) 25
(35) A unilateral transistor is having input and output reflection coefficients

0.498 121◦ and 0.225 −150◦ , respectively. Its optimum source and load reflec-
tion coefficients for minimum noise figure and maximum output power are
0.455 −160◦ and 0.275 +130◦ , respectively. To design a low-noise amplifier,
source and load reflection coefficient should be, respectively,
(A) 0.455 −160◦ and 0.225 +150◦

(B) 0.498 − 121◦ and 0.225 + 150◦
(C) 0.498 − 121◦ and 0.275 + 130◦
(D) 0.455 + 160◦ and 0.225 + 150◦
(36) The plot shows the operating load line and I-V characteristic of a Schottky diode
at two different RF power levels +10 and −10 dBm. Under this bias condition,
RF resistance of the diode
(A) Increases with the increase in RF power level

(B) Decreases with the increase in RF power level
(C) Remains constant with the RF power level

(D) None of the above
(37) Which of the following devices is the most suitable for higher order RF fre-
quency multiplier?
(A) Step recovery diode
(B) P-N junction diode
(C) Varactor diode
(D) P-I-N diode
(38) Which of the following diodes is the most suitable as linear resistance at
microwave frequency?
(A) P-I-N diode
(B) Schottky diode
(C) Varactor diode
(39) Input reflection coefficient ‘S 11 ’ of the 2-port network for a 50 system is
(A) 1/3 180◦

(B) 1/2 180◦
(C) 1/3 0o
(D) 1/2 0◦
(40) Input reflection coefficient ‘S 11 ’ of the 2-port network for a 50 system is
(A) 1/5 0◦
(B) 1/2 0◦
(C) 1/5 0◦
(D) 1/2 0◦
(41) Transmission coefficient ‘S 21 ’ of the 2-port network for a 50 system is
(A) 4/5 0◦
(B) 1/2 0◦
(C) 4/5 180◦
(D) 1/2 180◦
(42) Which of the following diodes is most suitable for detection of microwave
signal?
(A) Schottky barrier diode

(B) p-i-n diode
(C) Varactor diode
(43) Reason for superior high-frequency performance of metal–semiconductor diode

compared to p-n junction diode is
(A) No minority carrier storage effect in metal–semiconductor diode

(B) Higher conductivity of metal compared to semiconductor
(C) Metal–semiconductor junction does not rectify RF signal
(D) Hole is the majority carrier in Schottky barrier diode operation
(44) Which statement is false about microstrip line over stripline?
(A) Less radiative

(B) Easier for component integration
(C) One-sided ground plane
(D) More interaction with neighbouring circuit element
(45) A 50 transmission line is terminated by a load impedance of (100 + j100) .

Magnitude of reflection coefficient at the load is
√
(A) √5/13
(B) √1/13
(C) 5/7
(D) 5/7
(46) A unilateral transistor has output impedance Zout (10 − j10) . Value of the
series and shunt components of the matching network for the complex conjugate
match at the output of the device, for a 50 load, are
(A) X 1 −j25, X 2 +j30

(B) X1 +j25, X2 +j30
(C) X1 −j25, X2 −j30
(D) X1 +j25, X2 −j30
(47) A unilateral transistor has output impedance Zout (10 − j10). Value of the
series and shunt components of the matching network for the complex conjugate
match at the output of the device considering a 50 load are
(A) X 1 +j25, X 2 −j30

(B) X1 −j25, X2 −j10
(C) X1 +j25, X2 +j10
(D) X1 −j25, X2 +j10
(48) A unilateral transistor has output impedance Zout (8 − j8). Values of the
series and shunt components of the matching network for complex conjugate
match at the output of the device considering a 50 load are
(A) X 1 +j20, X 2 +j28, X 3 −j20
(B) X1 +j20, X2 +j28, X3 +j40

(C) X1 −j20, X2 +j28, X3 −j40
(D) X1 +j20, X2 −j28, X3 −j40
(49) DC to RF efficiency of an ideal Class-F amplifier is
(A) 100%
(B) 78.4%
(C) 50%
(D) <80%
(50) S-parameter matrix of a lossless transmission line of length λ/4 is

0 1 −90◦
(A)
1 −90◦ 0

0 1 −45◦
(B)
1 −45◦ 0

0 1 −22.5◦
(C)
1 −22.5◦ 0

0 1 − 180◦
(D)
1 − 180◦ 0
(51) Magnitude of the input impedance of a λ/8 lossless 50 transmission line

terminated with 25 is
(A) 50
(B) 25
(C) 100
(52) Gain and NF of a single-stage amplifier are 10 and 3 dB, respectively. When
two such amplifiers are cascaded, then gain and NF of the cascaded amplifier
will be
(A) Gain: 20 dB, NF: 3.2 dB

(B) Gain: 20 dB, NF: 4.2 dB
(C) Gain: 13 dB, NF: 3.6 dB
(D) Gain: 13 dB, NF: 3.7 dB
(53) A load of the reflection coefficient ΓL is transformed to ΓIN as shown in the

figure by suitably connecting a circuit element. The circuit element is
(A) Short circuit transmission line stub or shunt connected inductor

(B) Series connected capacitor
(C) Series connected transmission line
(D) Shunt connected capacitor

(A) Series connected transmission line

(B) Shunt connected inductor
(C) Series connected capacitor

(A) Series connected resistor

(B) Shunt connected resistor
(A) Series connected resistor

figure by connecting a circuit element. The circuit element is
(A) Shunt connected resistor

0.455 −160◦ and 0.275 +130◦ , respectively. To design a high-gain amplifier,
source and load reflection coefficient should be, respectively,
(A) 0.498 −121◦ and 0.225 +150◦

(B) 0.455 −160◦ and 0.225 +150◦
(C) 0.498 −121◦ and 0.275 +130◦
(D) 0.455 +121◦ and 0.225 −150◦

0.455 −160◦ and 0.275 +130◦ , respectively. To design a power amplifier with
highest possible output power and moderate gain, the source and load reflection
coefficient should be, respectively,
(A) 0.498 −121◦ and 0.275 +130◦

(B) 0.498 −121◦ and 0.225 +150◦
(C) 0.455 −160◦ and 0.225 +150◦
(D) 0.455 +121◦ and 0.225 −130◦
(60) A BJT is having common-emitter current gain 100. Considering 10 V supply

and V BE 0.7 V, the value of RC and RB to set the quiescent point at I C 10 mA
and V CE 8 V is
(A) RC 200 , RB 93 kΩ
(B) RC 2 k, RB 100 k
(C) RC 200 , RB 100 k
(D) RC 20 , RB 93 k
(61) A power MESFET operates with 10 V drain supply and 2 A drain current. What
is the maximum output power when operated in Class-A?
(A) 10 W
(B) 20 W
(C) 15 W
(D) 5W
(62) A lossless transmission line of characteristic impedance 50 is terminated in
a resistive load of impedance 100 . The VSWR is
(A) 1:2
(B) 2:1
(C) 4:1
(D) 1:4
(63) A linear amplifier of gain 30 dB is fed with 0.1 μW power, the output power of
the amplifier
(A) 1W
(B) 0 dBm
(C) 30 dBm
(D) −10 dBm
(64) 10 W RF power is transmitted with a circular polarized antenna having a gain

of 10 dB. A omni-directional receiving antenna has vertical polarization. The
path loss is 100 dB. The receiving signal is
(A) −83 dBW

(B) −80 dBW
(C) −88 dBW
(D) +80 dBW
(65) If the short-circuit and open-circuit impedance of a transmission line are 5 and
20 , respectively. Characteristic impedance of the line is
(A) 100
(B) 10
(C) 15
(D) 1000
(66) If each stage has a gain of 10 dB and NF of 10 dB, then the overall noise figure
of two-stage cascaded amplifier will be
(A) 10 dB
(B) 10.9 dB
(C) 10.37 dB
(D) 1.09 dB
(67) Transmission coefficient ‘S 21 ’ of the 2-port network for 50 system is
(A) 4/5 0°
(B) 1/3 0°
(C) 4/5 180°
(D) 1/2 180°
(68) Transmission coefficient ‘S 21 ’ of the 2-port network for 50 system is
(A) 4/5 0°
(B) 1/2 0°
(C) 4/5 180°
(D) 2/3 0°
(69) The value of shunt resistor of a T-type matched attenuator in a 50 system is

100 . The attenuation and the value of series resistors are, respectively,
(A) 2.09 dB, 11.803

(B) 2.09 dB, 12
(C) 4.18 dB, 11.803
(D) 4.18 dB, 6.02
(70) Attenuation of this T-type attenuator in a 50 system is
(A) 0 dB
(B) ∝ dB
(C) 10 dB
(D) 1 dB
(71) In a 50 system, attenuation and return losses of this π attenuator in a 50

system are
(A) ∝ dB, 12.04 dB
(B) 12.04 dB, 12.04 dB
(C) 12.04 dB, ∝ dB
(D) 0 dB, 12.04 dB
(72) Attenuation and return losses of the 2-port network in a 50 system are,
respectively,
(A) ∝ dB, 12.04 dB

(B) 0 dB, 12.04 dB
(C) 12.04 dB, 12.04 dB
(D) 0 dB, ∝ dB
respectively,
(A) ∝ dB, 9.54 dB

(B) 9.54 dB, ∝ dB
(C) 9.54 dB, 9.54 dB
(D) 9.54 dB, ∝ dB
respectively,
(A) 20.98 dB, ∝ dB

(B) 9.39 dB, 20.98 dB
(C) 9.39 dB, 9.39 dB

(D) 20.98 dB, 20.98 dB
respectively,
(A) ∝ dB, 0 dB
(B) 9.03 dB, 9.03 dB
(C) 6.02 dB, 6.02 dB
(D) 0 dB, ∝ dB
respectively,
(A) ∝ dB, 0 dB
(B) 9.03 dB, 9.03 dB
(C) 6.02 dB, 6.02 dB
(D) 0 dB, ∝ dB
(77) Optimum open-circuit voltage for temperature-invariant intrinsic layer resis-

tance of a silicon (bandgap potential of Si is 1.12 V) p-i-n diode at the temper-
ature of 25 °C is
(A) 1.17 V
(B) 5.0 V
(C) 2.24 V
(D) 0.506 V
(78) Operating temperature dependency of optimum open-circuit voltage for

temperature-invariant intrinsic layer resistance of p-i-n diodes:
(A) Highly dependent

(B) Not dependent at all
(C) Weakly dependent
(D) Exponentially decreases
(79) Output power at 1-dB gain compression point of an amplifier operating in

Class-AB having third-order output intercept point 1-W is about
(A) 29.4 dBm

(B) 19.4 dBm
(C) 0.5 W
(D) 0.75 W
(80) Two-tone IM3 at 1-dB gain compression point of an amplifier operating in

Class-AB is about
(A) −21.3 dBc

(B) −10.4 dBc
(C) −15.5 dBc
(D) −30.3 dBc
(81) Two-tone third-order intermodulation power at 1-dB gain compression point

of an amplifier operating in Class-AB having third-order output intercept point
2-W is
(A) −21.3 dBc

(B) +1.1 dBm
(C) −10.63 dBm
(D) +30 dBm
(82) Two-tone IM3 of an amplifier at its 1 mW output power with Po1dB 10 mW

is
(A) −41 dBc

(B) −10 dBc
(C) −10 dBc
(D) −20 dBc
(83) Maximum power added efficiency of a practical Class-A amplifier using a

MESFET is always less than 50%. Some of the reasons are
(A) Finite loss of output network/matching circuit of the amplifier

(B) Finite gain of the amplifier

(C) Non-zero knee voltage, VK
(D) All of the above
(84) Practical limitations of the output power of a power amplifier come from
(A) Limitation of applied DC voltage across the device

(B) Limitation of current through the device
(C) Thermal dissipation inside the device
(85) Two-tone IM3 of an amplifier with PoIP3 100 mW at its 1 mW output power
is about
(A) −20 dBc

(B) −40 dBc
(C) −10 dBc
(D) −30 dBc
(86) Gain and noise figure of an amplifier are 40 and 10 dB, respectively. Total
noise power at the output of the amplifier over the bandwidth of 40 MHz will
be (To 290 K)
(A) −174 dBm
(B) −77 dBm
(C) −48 dBm
(D) −228 dBm
(87) Gain and noise figure of an amplifier are 40 and 3 dB, respectively. Part of the
total available noise power at the output of the amplifier which is added by the
amplifier over the bandwidth of 40 MHz will be (To 290 K)
(A) −174 dBm
(B) −77 dBm
(C) −58 dBm
(D) −48 dBm
(88) Gain and noise figure of an amplifier are 40 and 10 dB, respectively. Part of
the total available noise power at the output of the amplifier which is added by
the amplifier over the bandwidth of 40 MHz will be (To 290 K)
(A) −174.4 dBm

(B) −77.4 dBm
(C) −58.4 dBm
(D) −48.4 dBm
(89) Gain and noise figure of two identical amplifiers are 10 and 3.0 dB respectively.
Noise figure of the two amplifiers connected in cascade is
(A) 3.2 dB
(B) 3.0 dB
(C) 6 dB
(D) 4 dB
(90) Gain and noise figure of AMP1 and AMP2 are 6 dB, 3.0 dB and 20 dB 3.2 dB,
respectively. Noise figure of the two amplifiers connected (AMP1-AMP2) in
cascade is
(A) 3.25 dB
(B) 3.56 dB
(C) 3.05 dB
(D) 4.06 dB
(91) Gain and noise figure of AMP1 and AMP2 are 20 dB, 3.2 dB and 6 dB, 3.0 dB,
respectively. Noise figure of the two amplifiers connected (AMP1-AMP2) in
cascade is
(A) 3.00 dB
(B) 6.20 dB
(C) 3.22 dB
(D) 4.06 dB
(92) Gain and output power Po1dB of two identical amplifiers are 10 dB and 40 dBm,
respectively. The amplifiers are connected in balanced amplifier configuration
using power divider and combiner. The gain and output power Po1dB of the
balanced amplifier will be (neglect loss of the power divider and combiner)
(A) 10 dB, 40 dBm

(B) 20 dB, 80 dBm
(C) 20 dB, 40 dBm
(D) 10 dB, 43 dBm
(93) Gain and output power Po1dB of two identical amplifiers are 10 dB and 40 dBm,
respectively. The amplifiers are connected in balanced amplifier configuration
using power divider (Insertion Loss 1.0 dB) and combiner (Insertion Loss
0.5 dB). The gain and output power Po1dB of the balanced amplifier will be
(A) 10 dB, 40 dBm

(B) 20 dB, 80 dBm
(C) 8.5 dB, 42.5 dBm
(D) 10 dB, 43 dBm
(94) A broadband low-noise amplifier of gain and noise figure, 30 and 2.5 dB,
respectively, is connected in cascade with a mixer of negligible insertion loss.
In case of single-sideband signal, the effective noise figure of the cascaded
system in the absence of any filter in between the LNA and the mixer will be
(A) 2.5 dB
(B) 5 dB
(C) 5.5 dB
(D) 6 dB
(95) A broadband low-noise amplifier of gain and noise figure, 30 and 2.5 dB,
respectively, is connected in cascade with a mixer of negligible insertion loss
with an image band reject filter in between. The effective noise figure of the
cascaded system will be
(A) 2.5 dB
(B) 5.0 dB
(C) 5.5 dB
(D) 6.0 dB
(96) Noise figure of a system is 5 dB. The equivalent noise temperature of the
system is (To 290 K)
(A) 290 K
(B) 627 K
(C) 570 K
(D) 300 K
(97) A system consists of a low-noise amplifier and a passive filter of insertion loss
0.5 dB at the input of the LNA. The gain and noise figure of the LNA are 30
and 2.2 dB, respectively. The noise temperature of the system is (To 290 K)
(A) 290 K
(B) 450 K
(C) 250 K
(D) 300 K
(98) A system consists of a low-noise amplifier and a passive filter of insertion loss
0.5 dB at the output of the LNA. The gain and noise figure of the LNA are 30
and 2.2 dB, respectively. The noise temperature of the system is (To 290 K)
(A) 290.3 K
(B) 450.8 K
(C) 250.5 K
(D) 191.3 K
(99) A system consists of two amplifiers and a passive filter in between them. The
gain and noise figures of input and output amplifiers are 30, 2.7 and 40, 3 dB,
respectively. The insertion loss of the filter is 10 dB. The noise temperature of
the system is (To 290 K)
(A) 470.8 K
(B) 450.8 K
(C) 290.5 K
(D) 255.5 K
(100) Attenuation of an attenuator is 30 dB. For an input power of 10 W, power at
the output of the attenuator will be
(A) 10 mW
(B) 10 dBm
(C) −20 dBW
(101) A linear amplifier has gain of 30 dB. For an input power of 1 mW, power at
the output of the amplifier will be
(A) 1W
(B) 30 dBm
(C) 0 dBW
(102) Linear gain of a travelling wave tube amplifier (TWTA) is 50 dB. It provides
100 W output power when the input is 6 dBm. Operating gain compression of
the amplifier is
(A) 0 dB
(B) 4 dB
(C) 6 dB
(D) 2 dB
(103) Predistortion linearizer cascaded with TWTA improves
(A) Output power of the system
(B) Linearity of the system
(C) DC to RF efficiency when it operates at saturation
(104) A low-noise amplifier has gain and noise figure 40 and 3 dB, respectively. The
total noise power at the output of the LNA, over the bandwidth of 40 MHz, is
(To 290 K):
(A) −58 dBm
(B) −42 dBm
(C) −55 dBm
(D) −55 dBm
(105) A system consists of a low-noise amplifier (LNA) and an attenuator of attenu-
ation 5 dB at the input of the LNA. The gain and noise figure of the LNA are
40 and 3 dB, respectively. The total noise power at the output of the system
over the bandwidth of 40 MHz is (To 290 K)
(A) −58 dBm

(B) −42 dBm
(C) −55 dBm
(D) −55 dBm
(106) A system consists of a low-noise amplifier (LNA) and an attenuator of attenu-

ation 10 dB at the input of the LNA. The gain and noise figure of the LNA are
40 and 3 dB, respectively. The total noise power at the output of the system
over the bandwidth of 40 MHz is (To 290 K)
(A) −58 dBm
(B) −42 dBm
(C) −55 dBm
(D) −55 dBm
(107) A lossless transmission line of characteristic impedance 50 is terminated in

a load of 100 resistance. The VSWR is
(A) 1.2
(B) 2:1
(C) 4:1
(D) 1:4
(108) Short-circuit and open-circuit impedance of a transmission line are 25 and

100 , respectively. The characteristic impedance of the line is
(A) 100
(B) 10
(C) 50
(D) 200
(109) The input impedance of a short-circuited line of length l, where λ/4 > l > λ/2
(A) Capacitive
(B) Resistive
(C) Inductive
(D) None of these
(110) Noise figure of an amplifier depends on
(A) Bandwidth
(B) Input power
(C) Output power
(D) None of these
(111) VSWR of a transmission line is always
(A) Greater than unity

(B) Less than unity
(C) Zero
(D) Infinity
(112) Waveguide acts like a
(A) Bandpass filter

(B) High-pass filter
(C) All pass filter
(D) Low-pass filter
(113) The wavelength of a 1 GHz electromagnetic wave propagating through a per-

fect non-magnetic dielectric of relative permittivity εr 9 is
(A) 3 cm
(B) 3 cm
(C) 100 cm
(D) 10 cm
(114) TEM wave exists in
(A) A circular Waveguide

(B) A rectangular waveguide
(C) A coaxial cable
(D) None of the transmission lines
(115) When VSWR is 3, then the magnitude of the reflection coefficient is
(A) 1/2
(B) 1
(C) 0
(D) 1/4
(116) Which of the following transmission lines is ideal for handling high RF power?
(A) Coaxial
(B) Microstrip
(C) Stripline
(D) Rectangular waveguide
(117) When a transmission line is terminated with a load equal to its characteristic
impedance, the reflection coefficient is
(A) Zero
(B) +1
(C) −1
(D) Infinity
(118) What is the importance of the noise resistance of a transistor when it is used
for LNA design?
(A) It gives the effective impedance offered by the input with respect to noise
current
(B) It offers the criterion for conjugate matching of input impedance
(C) It indicates the resistance that would generate the same amount of noise
at room temperature
(D) It indicated how the noise figure increases as we deviate from the
optimum source impedance for the lowest noise figure
(119) If a line is terminated to an open-circuit element, the VSWR of the line is
(A) 0
(B) 1
(C) −1
(D) Infinity
(120) A transmission line has characteristic impedance 50 . The line is terminated

at its one end by +j50 . VSWR produced in the transmission line:
(A) +1
(B) 0
(C) −1
(D) Infinity
(121) Major disadvantage of microstrip line compared to stripline is
(A) More likely to radiate

(B) Bulkier
(C) More expensive
(D) More complicated to manufacture
(122) The disadvantage of single-stub matching as compared to double-stub match-

ing is that
(A) The stub position has to be adjusted

(B) Only shunt stub can be used
(C) Only resistive load can be matched
(D) Useful only in two-wire transmission line
(123) A lossy open-circuit stub having a length of 1/32 of wavelength, is approxi-
mately equivalent to
(A) A small capacitor in shunt with a small resistor

(B) A small capacitor in shunt with a large resistor
(C) A small inductor in shunt with a small resistor
(D) A small inductor in shunt with a large resistor
(124) A p-n junction diode in series with a 50 resistor is forward biased so that
a current of 100 mA flows. If the voltage across the combination is instanta-
neously reversed to 10 V at t 0, the reverse current that flows through the
junction at t 0+ is approximately given by
(A) 0 mA
(B) 200 mA
(C) 50 mA
(D) 200 mA
(125) For a transistor, the manufacturer specifies maximum power dissipation

PDmax 1 W at 25 ◦ C at and the maximum junction temperature, TJ max
125 ◦ C. Its thermal resistance is
(A) 10 °C/W
(B) 100 °C/W
(C) 1000 °C/W
(D) 10,000 °C/W
(126) In a Class-AB amplifier, the current flows through the active device for
(A) Less than half of the duration of input cycle

(B) Half duration of input cycle
(C) More than half but less than full duration of the input cycle
(D) Full duration of the input cycle
(127) An earth station transmitter having power amplifier of 40 dBW saturated output
power operates at 3-dB output back-off. The total branching and feeder loss
is 3 dB, and transmit gain of antenna is 40 dB. The operating EIRP of the
transmitter is
(A) 40 dBW
(B) 74 dBW
(C) 34 dBW
(D) 80 dBW
(128) A transistor amplifier with 85% efficiency is likely to be

(A) Class-A
(B) Class-B
(C) Class-F
(D) Class-AB
(129) A satellite transponder has a receive antenna facing towards earth with a gain
of 30 dB facing towards earth. The receiver has gain of 50 dB, and equivalent
noise temperature is 26 dBK. G/T of the satellite is
(A) 1.6 dB/K

(B) −4 dB/K
(C) 26 dB/K
(D) −26 dB/K
(130) When VSWR is 3, the magnitude of the reflection coefficient will be
(A) 1/4
(B) 1/2
(C) 1/3
(D) 3
(131) A transmission line of characteristic impedance 50 is terminated by a load

impedance of 50 − j50. The magnitude of reflection coefficient at the load is
√
(A) √5/3
(B) 3/5
√
(C) 1/ 5
(132) If the RF transmission line is terminated in its characteristic impedance Zo ,

which of the following statements is wrong
(A) The input impedance of transmission line becomes Zo

(B) The transmission line acts as an infinite long transmission line
(C) The VSWR becomes one
(D) The VSWR becomes infinite
20.2 Answers of MCQs with Explanations
S. No. Answer Explanation
1 A μGaAs 8500 cm2 /V s, μGe 3900 cm2 /V s, μSi 1500cm2 /V s

2 C Common-emitter current gain β
α/(1 − α) 0.99/(1 − 0.99) 99
3 A Equiresistance curves are straight line over wide range of operating temperatures
4 B Optimum open-circuit voltage for temperature-invariant resistance of a p-i-n
diode is slightly higher than the bandgap potential of the semiconductor
from which it is made. Eg of Si is 1.12 V
5 C Optimum open-circuit voltage for temperature-invariant resistance of a p-i-n
diode is slightly higher than the bandgap potential of the semiconductor
from which it is made. Eg of GaAs is 1.42 V
6 D For series-connected p-i-n diodes, the optimum open-circuit voltage is slightly
higher than Eg multiplied by the number of diodes connected in series
7 B For parallel connected p-i-n diodes, the optimum open-circuit voltage is same as
single diode circuit, i.e. slightly higher than Eg
PdBm 10 log(PmW ) 10 log(1000) 30
8 A
PdBW 10 log(PW ) 10 log(1) 0
9 A PdBm 10 log(PmW ) 10 log(10) 10
10 B PmW 10(PdBm /10) 10(50/10) 100,000 mW
11 C PW 10(PdBW /10) 10(30/10) 1000 W
12 C PmW 10(PdBm /10) 10(30/10) 1000 mW
13 B PW 10(PdBW /10) 10(0/10) 1 W
I0.5k (20 − 10)/0.5 mA 20 mA
I10 (10 − 0.7)/10 A 930 mA
14 A
IB 930/100 mA 9.3 mA
IZ I0.5 k − IB (20 − 9.3)mA 10.7 mA
15 B Current through the 10 k: I10 k (20/1 − 40/2) μA 40 μA
Thus, output voltage: Vout −40 μA × 10 k −400 mV
16 C Current through the 4 resistor (3i/4)
The resistance of the upper branch 4Ω × ((3i/4)/(i/4)) 12
Thus resistance faced by the voltage source 4 ||12 3
17 D Current through the resistor R
IR (1 + j)/((1 + j) + (1 − j)) √
(1 + j)/2 Amp
VR IR × R (1 + j)/2 V 1/ 2 45◦ V
(continued)
20.2 Answers of MCQs with Explanations 679
(continued)
18 A At t 0+ , capacitor can be considered as short.
Thus, IR 1/10 A 100 mA
At steady state, capacitor can be considered as open.
Thus, IR 1/(10 + 5 + 5)A 50 mA
19 B
These are typical I-V characteristic of Schottky diode at two different

temperatures. At constant voltage bias condition, load line is vertical. Thus,
current increases with the increase in temperature
20 C p-i-n diode has an intrinsic layer in between p and n layer. The stored charge in
the I-layer is controllable by controlling the forward current, i.e. resistance is
controllable
Schottky diode rectifies microwave. Varactor and p-n junction diodes have high
junction capacitance and cannot work at microwave frequency
21 D IC 10 mA and VCE 8 V
Voltage across RE : 0.1 kΩ × 10 mA 1 V
Voltage across RB : (10 − 0.7 − 1) V 8.3 V
IB IC /βdc 10/100 0.1 mA
Therefore,
RB (8.3/0.1) k 83 k
RC ((10 − 8 − 1)/10) k 100
22 A AMP-A:
Gain GA 6 dB 4
Noise figure NFA 3.8 dB 2.4
AMP-B:
Gain C 20 dB 100
Noise figure NFB 4 dB 2.5
In case of input stage: AMP-B
NFBA NFB + (NFA − 1)/GB
2.5 + (2.4 − 1)/100 2.524 4.024 dB
In case of input stage: AMP-A
NFAB NFA + (NFB − 1)/GA 2.4 + (2.5 − 1)/4 2.775 4.43 dB
Z +jZ
23 B Zin(λ/8) Zo Z L−jZo
o L
Here, Zo 13, ZL 5 + j12
5+j25
Zin(λ/8) 13 1+j5 65
24 C In an ideal Class-F amplifier, there is no overlap between device voltage and its
current in time domain. Thus, efficiency is 100%
(continued)
(continued)
25 D At 1 GHz, capacitive reactance 1/2π fCgs 1592
Thus, rg 1/2π fCgs
Therefore, vgs vs
Voltage gain gm × vgs × rL /vs gm × rL 0.1 × 500 50
26 A PDC 10 V × 2 A 20 W
√ 2
RF power 1/ 2 × 5 W 2.5 W
Thus, Efficiency 2.5/20 12.5%

27 B Efficiency in Class-B operation: (π/4) × Vdrf /Vdd 62.8%
Rth 4.5 ◦ C/W

28 C
Pdissipation (20 − 8) W 12 W
Temperature increase 12 × 4.5 ◦ C 54 ◦ C
Thus, channel temperature (55 + 54) ◦ C 109 ◦ C
29 D Helix structure in TWT is used to slow down the RF field along the axial
direction to maximize interaction of RF field with the electron beam
30 A Transformed through constant resistance circle: circuit element is series
connected. Moving upwards: the component is inductor
31 A Transmission line: lossless, thus |S21 | 1 length λ/8, thus phase − 45◦
32 A ∗ 32 − j30
For the complex conjugate match Znetwork Zout
In case of (A)
Znetwork X2 + X2 ||(X2 + 50)
j10 − j40||(j40 + 50)
j10 + (32 − j40)
32 − j30
∗
Zout
S21 Transmission coefficient
33 A 2Za

Zo + 2Za
Here, Z a 50 and Z o 50
S21 2 × 50/(50 + 2 × 50) 100/150 2/3 0◦
34 A Zo 50, ZL 25
For quarter-wavelength transmission line Zin Zo2 /ZL 502 /25 100
35 A For low-noise amplifier:
Input to be matched: with source reflection coefficient for minimum noise figure.
Thus ΓS 0.455 − 160◦
Output to be matched: complex conjugate for achieving maximum gain. Thus
ΓL Γout∗ 0.225 + 150◦
36 A RF resistance is determined by (dV /dI ) of the I-V characteristic at the intercept

with load line. Here, dV /dI is more in case of +10 dBm compared to the case of
−10 dBm RF power
37 A SRD generates comblike spectrum, i.e. higher order harmonics contain more RF
power level compared to other mentioned devices. Thus, it is most suitable for
higher order frequency multiplier
(continued)
(continued)
38 A p-i-n diode has intrinsic layer in between p and n layer. The stored charge in the
I-layer is controllable by controlling the forward current, i.e. resistance is
controllable
Schottky diode rectifies the microwave. Varactor and p-n junction
diodes have high junction capacitance, and thus cannot work at microwave
frequency
39 A S 11 input reflection coefficient
(ZL − Zo )/(ZL + Zo )
Here, ZL 50||50 25
S11 (25 − 50)/(25 + 50) −1/3 1/3 180◦
40 A S 11 input reflection coefficient (ZL − Zo )/(ZL + Zo )
Here, ZL 25 + 50 75
S11 (75 − 50)/(75 + 50) 25/125 1/5 0◦
41 A S 21 Transmission coefficient 2Z2Z+Z
o
o a
Here, Za 25 and Zo 50
S21 2 × 50/(2 × 50 + 25) 100/125 4/5 0◦
42 A Schottky diode has no minority carrier storage effect thus it can rectify
microwave signals efficiently. Other junction diodes have minority carrier
storage effect and thus cannot rectify microwave signals efficiently
43 A In case of metal–semiconductor diode, only majority carrier takes part for
current conduction. Thus, there is no minority carrier storage effect
44 A Top face of a microstrip line is open, i.e. only one side has ground plane. Thus, it
radiates more RF/microwave power. Whereas, stripline has ground plane on both
the sides, thus it radiates less
45 A |S11 | (|ZL − Zo |)/(|ZL + Zo |)
Here, ZL (100 + j100),
Zo 50
|S11 | 502 + 1002 / 1502 + 1002

5/13
46 A For the complex conjugate match
∗ 10 + j10
Znetwork Zout
In case of (A)
Znetwork X2 + X1 ||50 j30 + (−j25||50)
∗
j30 + (10 − j20) 10 + j10 Zout
∗ 10 − j10
Znetwork Zout
In case of (A)
Znetwork X2 + X1 ||50 −j30 + (j25||50)
∗
−j30 + (10 + j20) 10 − j10 Zout
∗ 8 + j8
Znetwork Zout
In case of (A)
Znetwork X2 + X3 ||(X1 + 50) j28 − j20||(j20 + 50)
∗
j28 + (8 − j20) 8 + j8 Zout
49 A In case of ideal Class-F amplifier, there is no overlap in between device voltage
and with its current in time domain. Thus, efficiency is 100%
(continued)
(continued)
50 A Transmission line:
lossless, thus |S21 | 1
length λ/4, thus phase − 90◦
Z +jZ tan βl
51 A Zin Zo ZL+jZ o tan βl
o L
Thus,
ZL +jZo
Zin(λ/8) Zo Z +jZ
o L
Thus,
|Zin(λ/8) | Zo 50
NF 3 dB 2 (in factor)
52 A
Gain 10 dB 10 (in factor)
Cascaded NF 2 + (2 − 1)/10 2.1 3.2 dB
Cascaded Gain (10 + 10) dB 20 dB
53 A Transformed through constant conductance circle: circuit element is
shunt-connected
Moving upward: the component is inductive, thus it may be inductor or
short-circuited stub
54 A Transformed through constant VSWR circle: circuit element is series-connected
transmission line
55 A Transformed through constant reactance and resistance increases: circuit
element is series-connected resistor
56 A Transformed through constant reactance and resistance increases: circuit
element is series-connected resistor
57 A Transformed through constant susceptance and conductance increases: circuit
element is shunt-connected resistor
58 A For high-gain amplifier: Input and output matching: complex conjugate
Thus, ΓS Γin∗ 0.498 −121◦
And ΓL Γout ∗ 0.225 +150◦
59 A To design power amplifier: Input matching: complex conjugate

Thus, ΓS Γin∗ 0.498 −121◦
Output matching: corresponds to maximum output power
And ΓL 0.275 +130◦
RC (10 − 8)/10 k 200
60 A IB IC /βdc 10/100 0.1 mA
RB (10 − 0.7)/0.1 k 93 k
61 A PDC 10 V × 2 A 20 W
Class-A amplifier has maximum efficiency 50%
Thus, maximum output power
= 20 × 50% W 10 W
62 B Reflection coefficient:
Γ (ZL − Zo )/(ZL + Zo ) (100 − 50)/(100 + 50) 1/3
VSWR (1 + |Γ |)/(1 − |Γ |) 2
Output Power Pin (dBm) + G(dB)
63 D −40 + 30
−10 dBm
(continued)
(continued)
Received Signal P(Transmit) + G(Tx antenna) − Path loss − Polarization loss
64 A
40 dBm + 10 dB − 100 dB − 3 dB −53 dBm −83 dBW

ZL Zopen × Zshort
65 B √
20 × 5 10
NF12 NF1 + (NF2 − 1)/G1
66 C
10 + (10 − 1)/10 10.9 10.37 dB
67 B S 11 input reflection coefficient
(ZL − Zo )/(ZL + Zo )
Here, ZL (50 + 50)Ω 100Ω
S11 (100 − 50)/(100 + 50) 50/150
1/3 0◦
68 D S 21 Transmission coefficient
2Z2Z+Z
o
o a
Here, Za 50 and Zo 50Ω
S21 2 × 50/(2 × 50 + 50)
100/150 2/3 0◦
69 C For matched attenuator in a 50 system,
50 R1 + 100||(R1 + 50)
Thus, R1 11.803
Attenuation is given by
Attn 20 × log((50 + 11.803 + 100)/100) 4.18 dB
70 B The attenuator terminates the full power. Thus, attenuation is infinite
Vout 0.125 × VS
Vin 0.5 × VS
71 B S21 20 × log(Vout /Vin ) −12.04 dB
Zin 30Ω
S11 20 × log(0.25) −12.04 dB
Vout 0.125 × VS
Vin 0.5 × VS
72 C S21 20 × log(Vout /Vin ) −12.04 dB
Zin 83.33Ω
S11 20 × log(0.25) −12.04 dB
73 D Reflection Coefficient (Γ ) −0.3333
IL 20 × log(|1/Γ |) 9.54 dB
74 B Reflection Coefficients:
Γ1 −0.4286
Γ2 −0.25
IL 20 × log(2 Γ /1 + Γ2
) 9.39 dB
RL 20 × log(2 Γ /1 − Γ2
) 20.98 dB
(continued)
(continued)
75 C Reflection Coefficients:
Γ1 0
Γ2 1
IL 20 × log(2 Γ1 + Γ2
) 6.02 dB
RL 20 × log(2 Γ1 − Γ2
) 6.02 dB
76 A Reflection Coefficients:
Γ1 −1
Γ2 1
IL 20 × log(2 Γ1 + Γ2
) ∝ dB
RL 20 × log(2 Γ1 − Γ2
) 0 dB
77 A Optimum open-circuit voltage is given by
2kTo
VOPT Eb +
q
2 × 1.38 × 10−23
1.12 + (273 + 25) V
1.6 × 10−19
1.17 V
78 C Optimum open-circuit voltage is given by
VOPT Eb + 2kT
q
o
79 B Po1dB (30 − 10.63) 19.37 dBm

80 A Po1dB PoIP3 − 10.63
Thus, IMD3 (at Po1dB ) 2 × (−10.63) −21.26 dBc
PoIP3 2 W 33 dBm
Po1dB PoIP3 − 10.63 33 − 10.63 22.37 dBm
81 B
IMD3 (at Po1dB ) 2 × (−10.63) −21.26 dBc
IMD3 (at Po1dB )in dBm 22.37 − 21.26 1.11 dBm
Po1dB 10 dBm
82 A
PoIP3 10 + 10.63 20.63 dBm
Po(operating) 0 dBm, it is 20.63 dBc below PoIP3
Thus, expected IM3 (operating) −2 × (20.63) −41.26 dBc
83 D Power added efficiency is always less than 50% is due to the finite loss of output
network, finite gain of the amplifier and non-zero value of knee voltage
84 D Limitations of the output power of a power amplifier come from the limitation of
applied DC voltage across the device, limitation of current through the device,
thermal dissipation inside the device, etc
85 B PoIP3 20 dBm
Po(operating) 0 dBm, it is 20 dBc below
Thus, expected IM3 (operating) −2 × (20) −40 dBc
kT −174 dBm/Hz
86 C GA 40 dB, NF 10 dB, B 40 MHz 76 dBHz
Pno (−174 + 40 + 10 + 76) dBm −48 dBm
(continued)
(continued)
kT −174 dBm/Hz
87 C
GA 40 dB, NF 3dB, B 40 MHz 76 dBHz
Output noise due to only input thermal noise:
kTBGA (−174 + 76 + 40) dBm −58 dBm
Total output noise:
kTBGA F (−174 + 76 + 40 + 3) dBm −55 dBm
Thus, part of the total noise which is added by the amplifier:
−55 dBm − 58 dBm −58 dBm
kT −174 dBm/Hz
88 D
GA 40 dB, NF 10 dB, B 40 MHz 76 dBHz
Output noise due to only input thermal noise:
kTBGA (−174 + 76 + 40) dBm −58 dBm
Total output noise:
kTBGA F (−174 + 76 + 40 + 10) dBm −48 dBm
Thus, part of the total noise which is added by the amplifier:
−48 dBm − 58 dBm −48.4 dBm
AMP1: G1 10 dB 10, NF1 3 dB 2
AMP2: G2 10 dB 10, NF2 3 dB 2
89 A
AMP12: NF12 NF1 + (NF2 − 1)/G1 2 + (2 − 1)/10
2.1 3.2 dB
AMP1: G1 6 dB 3.98, NF1 3 dB 2
AMP2: G2 20 dB 100, NF2 3.2 dB 2.1
90 B
AMP12: NF12 NF1 + (NF2 − 1)/G1 2 + (2.1 − 1)/3.98
2.27 3.56 dB
AMP1: G1 20 dB 100, NF1 3.2 dB 2.1
AMP2: G2 6 dB 3.98, NF2 3 dB 2
91 C
AMP12: NF12 NF1 + (NF2 − 1)/G1 2.1 + (2 − 1)/100
2.10 3.22 dB
92 D G1 10 dB, PO1B 40 dBm
For balanced amplifier:
Gain 10 dB, PO1B (40 + 3) dBm 43 dBm
G 10 dB, PO1B 40 dBm
93 C
ILdivider 1.0 dB, ILcombiner 0.5 dB
For balanced amplifier:
Gain (G − ILdivider − ILcombiner ) (10 − 1.0 − 0.5) 8.5 dB
Po1B (40 + 3 − ILcombiner ) dBm 42.5 dBm
94 C In the absence of image band reject filter before mixer, noise corresponds to the
image frequency band folded and added with the in band noise at the output of
the mixer. Thus, total noise power will be doubled, i.e. 3-dB more at the output
95 A In the presence of image band reject filter before mixer, noise corresponds to the
image frequency band will be sufficiently rejected by the filter thus NF will be
same as noise figure of the LNA
(continued)
(continued)
NFr 5 dB 3.16
96 B
Te (NF − 1)To 2.16 × 290 K 627 K
97 C Noise figure of the system:
NFs ILfilter + NFLNA (0.5 + 2.2) dB 1.862
Te (NF − 1)To 0.862 × 290 K 250 K
98 D Noise figure of the system:
NFs NFLNA + (ILfilter − 1)GLNA 2.2 dB 1.66
Te (NF − 1)To 0.66 × 290 K 191.3 K
Input Amplifier: G1 30 dB 1000, NF1 2.7 dB 1.86

99 D Output Amplifier: G2 40 dB 10,000, NF2 3 dB 2.00
Filter: Gf −10 dB 10.1, NFf 10 dB 10
Noise figure ofthe system:

NFs NF1 + NFf − 1 /G1 + (NF2 − 1)/G1 G2 1.88
Te (NF − 1)To 0.88 × 290 K 255.50 K
100 A Attenuation 30 dB
Input power 10 W 40 dBm
Output power (40 − 30) dBm 10 dBm
101 D Linear Gain 30 dB
Input power 1 mW 0 dBm
Output power (0 + 30) dBm 30 dBm 0 dBW 1 W
102 C Linear Gain 50 dB
Input power 6 dBm 4 mW
Output power 100 W 50 dBm
Operating Gain (50 − 6) dB 44 dB
Gain Compression (linear gain − operating gain) 6 dB
103 B Predistortion linearizer is used to improve linearity of the system
104 C Gain of the LNA, G 40 dB
Noise figure of the LNA, F 3 dB
Bandwidth, B 40 MHz 76 dBHz
Output noise, kTo BGF (−174 + 76 + 40 + 3) dBm −55 dBm
105 C Gain of the system, G (40 − 5) dB 35 dB
Noise figure of the system, F (3 + 5) dB 8 dB
106 C Gain of the system, G (40 − 10) dB 30 dB
Noise figure of the system, F (3 + 10) dB 13 dB
Z −Z
107 B Γ ZL +Z o 100−50 1
100+50 3
L o
108 C Zin (short) jZo tan(βl), Zin (open) jZ√
o cot(βl)
Thus, Zo Zin (short) × Zin (open) 25 × 100 50
(continued)
(continued)
109 A Zin (short) jZo tan(βl)
For, λ/4 > l > λ/2, π/2 > βl > π , tan(βl) is negative
Thus, Zin (short) is capacitive
110 D Noise figure of an amplifier does not depend on its bandwidth, input power or
output power
111 A VSWR is defined as the ratio of maximum voltage to minimum voltage of the
standing wave. Thus, it is always greater than unity
112 B Waveguide does not allow low frequencies to pass. Thus, it acts as a high-pass
filter
√ √
113 D λ c/ f εr 3 × 1010 / 1010 × 9 cm 10 cm
114 D TEM wave can exist in a coaxial cable and not in circular or rectangular
waveguides
115 A |Γ | (ρ − 1)/(ρ + 1) 1/2
116 D Rectangular waveguide can handle more RF power than coaxial, microstrip or
stripline
117 A Γ (ZL − Zo )/(ZL + Zo ) 0
118 D Noise resistance of a transistor indicates how the noise figure increases as the
source impedance move away from the optimum source impedance corresponds
to lowest noise figure
119 D Here, (ZL − Zo )/(ZL + Zo ) 1
120 D Γ (ZL − Zo )/(ZL + Zo ) (j50 − 50)/(j50 + 50) (j − 1)/(j + 1)
Thus, |Γ | 1
Therefore, VSWR (1 + |Γ |)/(1 − |Γ |) ∝
121 A The signal bearing conductor is open for microstrip line. This it is more likely to
radiate. Otherwise, it is less expensive, easier to fabricate and less bulky
compared to stripline
122 A In case of single-stub matching, the stub position from a load end cannot be kept
at any distance. It has to be adjusted as per the requirement to match a load to
particular source impedance
123 B An open-circuit stub having less than λ/4 line length can be approximated to a
small capacitor and lossy characteristic can be represented by a large resistor in
shunt with the capacitor
124 D Due to the charge storage effect, just after altering the forward-biased diode to a
reverse bias, it will remain in forward biased and the voltage across the diode will
be negligible. Thus, the current through the diode will be 10 V/50 200 mA
125 B PDmax 1 W at 25 ◦ C and TJ max 125 ◦ C
Increase of temperature (125 − 25) ◦ C 100 ◦ C
126 C In a Class-AB amplifier, device current flows for more than half but less than full
duration of input cycle
(continued)
(continued)
127 B EIRP (40 dBW − 3 dB − 3 dB + 40 dB) 74 dBW
128 C Maximum efficiency of Class-A: 50%
Maximum efficiency of Class-B: 78.5%
Maximum efficiency of Class-AB: <78.5%
Maximum efficiency of Class-F: 100%
Thus, among Class-A, B, AB and F, only Class-F amplifier can provide 85%
efficiency
129 A Antenna gain: G 30 dB
Equivalent noise temperature of receiver: Te 26 dBK 398.1 K
Total noise temperature: T 398.1 K + 290 K 688.1 K 28.4 dBK
Thus, G/T (30 − 28.6) dB/K 1.6 dB/K
130 B |Γ | (ρ − 1)/(ρ + 1) (3 − 1)/(3 + 1) 2/4 1/2
Γ (ZL − Zo )/(ZL + Zo ) (50 − j50 − 50)/(50 − j50 + 50)
131 C
(−j50)/(100 − j50) (−j)/(2 − j)
√
Thus, |Γ | 1/ 5
132 D A RF transmission line terminated by its characteristic impedance does not
generate standing wave. Thus, VSWR is unity
Index
A Class-AB amplifier, 435

ABCD parameter, 174 Class-B amplifier, 435
Absorptive-type switch, 200 Class-C amplifiers, 435
Active equalizers, 311 Class-F power amplifier, 452, 454
ALC dynamic range, 615 Closed-loop limiter, 495
AM/AM conversion, 503 Coaxial transmission lines, 123
AM/PM conversion, 503 Commandable gain setting, 613
Ambipolar lifetime, 14 Communication transponder, 597
Amplifier classification, 435 Compensation mechanisms, 23
Amplifier dynamic range, 353 Complex conjugate matching, 417
Amplifier stability, 363 Compound switch, 209
Amplifiers, 349 Conductivity modulation, 15
Amplitude modulator, 296 Constant gain circles, 400
Amplitude tilt equalizer, 307 Constant reactance arcs, 151
Attenuators, 7, 235 Constant resistance circles, 151
Automatic level control, 612 Constant VSWR circles, 153
Available power gain, 184, 194, 196, 387 Cut-off frequency, 65
Average power, 334
D
B Detectable power level, 361
Backward diode, 54 Diffusion capacitance, 65
Backward diode detector, 345 Digital driver circuits, 258
Balanced frequency multipliers, 551, 552 Digital phase shifters, 282
Balanced power combining, 123, 469 Digital predistortion linearizer, 524
Balun, 571 Diode driver circuits, 256
Bi-phase modulators, 298, 300 Diode linearizer, 508
Bipolar Junction Transistor (BJT), 59 Double-balanced mixers, 574
Branch line coupler, 471 Driver amplifier, 612
Breakdown voltage, 424 Dynamic load line, 422
Broadband linearizers, 520
Butler matrix, 624 E
Einstein relationship, 69
C Electron diffusion coefficient, 65
Channel amplifier, 612 Energy band diagrams, 54
Class-A power amplifier, 424 Equalizer, 6

in Electrical Engineering 533, https://doi.org/10.1007/978-981-13-3004-9
690 Index
Equiresistance curves, 24, 25, 41, 257 Low-noise amplifiers (LNAs), 5, 349, 606
Even- and odd-mode analysis, 465
M
F Matched attenuators, 244
Feedback linearization, 504 Metal–semiconductor field effect transistors
Feed-forward linearization, 504, 505 (MESFETs), 80
FET multiplier, 549 Metamorphic HEMT (mHEMT), 92
Field effect transistor (FETs), 79 Microwave communication, 1
Fixed gain mode, 612 Microwave frequency, 1
Flicker noise, 355 Microwave power combing, 463
Four-quadrant phase shifter, 287, 289 Microwave power module, 622
Frequency doubler, 534 Microwave switches, 199
Frequency mixer, 555 MIMO, 590
Frequency multipliers, 6 Mixer conversion gain, 560
Mixer noise figure, 562
G Mixers, 6
Gain slope equalizer, 314 Mmwave frequency, 1
GaN HEMT, 92 Mobile communication, 589
5G mobile communication, 590 Modified optimum voltage, 260
Ground receiver, 5 Modulation doped field effect transistor
Ground transmitter, 3 (MODFET), 80, 85
Modulator, 295
H Multiport amplifier, 623
HHBT, 70
Hetero-junction bipolar transistor, 59, 70 N
Hetero-structure field effect transistor (HFET), Negative differential resistance, 52
80 Network analysis, 158
High-electron-mobility transistor (HEMT), 80, Noise factor, 355
87 Nonlinear distortion, 502
High-gain amplifier, 411 Nonlinear power amplifiers, 445
High power amplifiers (HPAs), 5, 349 Non-linearities, 3
I O
Idler frequencies, 529 Ohmic contact, 34
Image frequency, 558 Operating power gain, 184, 394
Image rejection mixers, 575 Optimum supply voltage, 257
Impedance transformer, 148 Optimum voltage, 25
InP HEMT, 91 Output conductance, 97
I–Q modulator, 302 Overdrive controller, 495
K P
Kirk effect, 68 Parabolic gain slope equalizer, 313
Passive equalizers, 308
L Peak envelope power, 339
Left-hand circular polarized, 644 Phase modulators, 305
Limiter, 5, 479 Phase shifter, 7, 279
flat leakage, 481 PI configuration, 248
spike leakage, 480 P-i-n diode, 9, 13
Limiting amplifier, 492 attenuators, 235
Linear distortion, 502 equivalent circuit, 17
Linear power amplifier, 424 limiter, 482
Linearized TWTA, 621 nonlinearities, 19
Linearizer, 5, 620 switches, 219
Load pull technique, 434 temperature effect, 20
Index 691
Pinch-off point, 424 Spurious-free dynamic range, 358

2-port networks, 158 Square-law detector, 342
Power amplifiers, 420 SRD multiplier, 537
Power back-off, 441 Step recovery diode (SRD), 47, 48
Power detector, 333 Sub-harmonic mixer, 576
Power robbing, 634 Switch, 7
Power transfer theorems, 179 Switched-line phase shifters, 283
Pre-distortion linearizers, 506 System noise power, 593
Pre-select filter, 4
Pseudo-morphic HEMT (pHEMT), 92 T
Pulse power, 339 TEE attenuators, 244
TEE type switch, 213
Q Time delay, 279
Quadrature hybrid attenuator, 252 Time shifter, 281
Transceiver, 352
R Transconductance gain, 62, 95
Receiver, 606 Transducer gain, 183, 194, 195
Reflective switch, 200 Transducer power gain, 376
Right-hand circular polarized, 643 Transistor multiplier, 540
Rms detector, 346 Travelling wave tube amplifier, 616
Tunnel diode, 51
S Two-dimensional-gate FET, 85
Satellite transponder, 616
Saturation flux density, 593 U
Scattering parameter, 159 Unconditionally stable transistor, 377
Schottky barrier diode, 33
Schottky diode detector, 334, 341 V
Series-configured switch, 202 Varactor diode, 56
Shot noise, 355 Varactor multiplier, 536
Shunt-configured switch, 206 Variable attenuators, 255
Signal distortion, 502 Vector modulator, 302
Signal-to-Noise Ratio, 356 Voltage standing wave ratio, 117
Single-balanced mixer, 570
Single-ended mixer, 566 W
Small signal amplifiers, 400 Waveguides, 127, 139
Smith chart, 149 Webster effect, 67
Snap-off-diode, 48 Wilkinson splitter, 464
Solid-state power amplifiers, 618
Space segment, 4

2019 Book MicrowaveActiveDevicesAndCircu

Uploaded by

Copyright:

Available Formats

2019 Book MicrowaveActiveDevicesAndCircu

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2019 Book MicrowaveActiveDevicesAndCircu

Uploaded by

Copyright:

Available Formats

What are the different types of microwave devices discussed?

What are the different types of microwave devices discussed?

What are some applications of microwave communication?

What are some applications of microwave communication?

Lecture Notes in Electrical Engineering 533

Subhash Chandra Bera

Board of Series editors

• Communication, Networks, and Information Theory

More information about this series at http://www.springer.com/series/7818

Microwave Active Devices

ISSN 1876-1100 ISSN 1876-1119 (electronic)

Library of Congress Control Number: 2018958497

© Springer Nature Singapore Pte Ltd. 2019

Ahmedabad, India Tapan Misra

Ahmedabad, India Subhash Chandra Bera

4.3 Tunnel Diodes . . . . . .............. . . . . . . . . . . . . . . . . . 51

7.5 Smith Chart Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

10.3 Types of Phase Shifter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281

14 Microwave Solid-State Ampliﬁers . . . . . . . . . . . . . . . . . . . . . . . . . . 349

16 Microwave Linearizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501

19 Microwave Communication Systems . . . . . . . . . . . . . . . . . . . . . . . . 583

Subhash Chandra Bera received his B.Sc. degree

Microwave frequency region is generally considered from 1–30 GHz in electro-

1.1 Microwave Communications

Fig. 1.1 Block diagram of a mobile communication system

Fig. 1.2 Block diagram of satellite microwave communication system

The space segment is a repeater in space with a sizable amount of

1.2 Microwave Active Circuits

In microwave communication systems, various types of amplifiers are used in receive

intentionally or unintentionally exposure of high power levels. Basically, two types of

1.3 Microwave Active Devices

1.4 Microwave Circuit Analysis and Measurements

1.5 Book Outline

The book consists of 20 chapters including this introduction.

Chapter 17 describes microwave frequency multipliers using various types of

A microwave p-i-n diode behaves as current-controlled resistance at microwave fre-

2.2 Basics of P-I-N Diode

© Springer Nature Singapore Pte Ltd. 2019 13

Fig. 2.1 Impurity profile of

2.3 P-I-N Diode Characteristics

Therefore, Ri at 10 μA (27.8 × 100) ≈ 2.8 k

At low-frequency operation, the p-i-n diode behaves as simple junction diode.

In general, under sufficient forward-bias condition, Cd  C j . Thus, putting the

Fig. 2.2 Equivalent circuit

At frequencies well below the cut-off frequency f c , the voltage-variable junction

Therefore, this diode can behave as voltage-variable RF resistance at its forward-

a p-i-n diode as RF resistance not only determines by only f c . It also depends on

2.4 Nonlinearity of P-I-N Diode

For controlling RF and microwave signals, p-i-n diode-based voltage/current-

capacitance–voltage relationship than thinner diodes. This results in a lower rate

2.5 Temperature Behaviour of P-I-N Diode

RF resistance, Ri , of p-i-n diodes depends on forward-bias current (Ido ) and electronic

Therefore, from (2.4), (2.11) and (2.12), temperature dependency of high-

Here, A is a temperature-independent constant depending on area and electronic

Fig. 2.5 Typical

and η 2 [8]. Operation of diodes under constant-voltage-bias condition never

2.6 Temperature-Invariant RF Resistance of P-I-N Diode

For (n − m) ~1 and neglecting reverse saturation current compared to the forward

Fig. 2.6 Simple bias circuit

In general, under sufficient forward-bias condition, Cd C j . Thus, putting the

Here, A is a temperature-independent constant. The temperature and current

Under small signal operating condition (Vs Vt ), the RF resistance of Schottky