2019 Book MicrowaveActiveDevicesAndCircu
2019 Book MicrowaveActiveDevicesAndCircu
2019 Book MicrowaveActiveDevicesAndCircu
Microwave
Active Devices
and Circuits for
Communication
Lecture Notes in Electrical Engineering
Volume 533
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.
123
Subhash Chandra Bera
Space Applications Centre
Indian Space Research Organization (ISRO)
Ahmedabad, Gujarat, India
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.
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.
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
xix
Chapter 1
Introduction
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
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
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.
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].
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.
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.
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 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,
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.
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)
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
2.3 P-I-N Diode Characteristics 17
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
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
Fig. 2.3 Equivalent circuit of a forward biased, b zero biased, c reversed biased p-i-n diode for
f 1/2π τa
2.3 P-I-N Diode Characteristics 19
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
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
Fig. 2.4 Typical temperature dependency of V –I curve and different bias load line
22 2 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
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
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
2.6 Temperature-Invariant RF Resistance of P-I-N Diode 25
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
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.
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
∂ Ri a aT ∂ Id
− 2 (2.23a)
∂T Id Id ∂ T
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
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)
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
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
For zero temperature coefficient, suppose VS VOPT and Id Ido . Putting these
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
This leads to the value of optimum supply voltage for zero temperature coefficient
of RF resistance as
2.6 Temperature-Invariant RF Resistance of P-I-N Diode 29
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.
Here, a, b, c are constants. Derive the optimum open-circuit voltage for zero
temperature coefficient of RF resistance of the diodes.
VS Id Rb + nVd (2.25c)
Using (2.25c), the I–V relationship of the p-i-n diode can be written as
VS − I d Rb − n E b
Id bT 3 × exp (2.25d)
ncT
For zero temperature coefficient, suppose VS VOPT and Id Ido . Putting these
values to (2.25e) and using (2.25b),
Ido Ido Rb 3Ido Ido (VOPT − Ido Rb − n E b )
1+ − (2.25f)
T ncT T ncT 2
This leads to the value of optimum supply voltage for zero temperature coefficient
of RF resistance as
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
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
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
Vd j Vo + Vs cos(ωt) (3.5)
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
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
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.
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)
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
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 )
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
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
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:
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
VS Id Rb + Vd (3.21d)
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
This leads to the value of optimum supply voltage for zero temperature coefficient
of RF resistance of the Schottky diode as
VOPT φb + cT (3.21g)
References 45
References
1. Sze SM (1981) Physics of semiconductor devices, 2nd edn. Wiley Eastern Limited
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
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.
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.
(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
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
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.
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
50 4 Special Microwave Diodes
Rc
Cp
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.
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.
52 4 Special Microwave Diodes
V V V V V V
Fig. 4.4 Typical I-V characteristic of tunnel diode at various bias conditions
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
(a) I
(b) I
Fig. 4.6 Energy band diagrams of backward diodes with three different doping configurations
a no bias, b reverse bias
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
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.
56 4 Special Microwave Diodes
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)
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
Doping
p+
x
0
n+
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
58 4 Special Microwave Diodes
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
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
Vbe Vcb
B
E C
n+
n+
wb p
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
62 5 Microwave Bipolar Transistors
term is called the collector multiplication factor M. Thus, the common-base current
gain can be expressed as
αo γ αt M (5.2b)
I E IC + I B (5.3)
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
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
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.
64 5 Microwave Bipolar Transistors
Solution
Common–base current gain of the transistor
α 0.99
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
Collector resistor
(10 − 1.1 − 8)V
RC 90
10 mA
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
66 5 Microwave Bipolar Transistors
Wb
Wb WC
E C
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
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
5.2 Bipolar Junction Transistor (BJT) 67
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
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
68 5 Microwave Bipolar Transistors
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 Bipolar Junction Transistor (BJT) 69
T2 IB1
T1
VCE (Volt)
70 5 Microwave Bipolar Transistors
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.
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]
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
n+ sub-collector (Eg1)
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
72 5 Microwave Bipolar Transistors
ΔEc
(a)
EG2 Ec
EG2
ΔEv
EG1
Ev
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
5.3 Hetero-junction Bipolar Transistor (HBT) 73
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.
Emitter
n Si
n+ Si subcollector
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.
5.3 Hetero-junction Bipolar Transistor (HBT) 75
p InGaN
C n GaN C
GaN Sub-collector
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.
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)
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.
n+ n+ n+ d n+ n+ n+
a
n-type channel
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
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
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
(a) (b)
VDS VDS
S VGS G D S VGS G D
n+ n+ n+ n+
n-type channel d′
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
Vgs < 0
Vds
Breakdown voltage
9000
7000
5000
3000
1000
1013 1014 1015 1016 1017 1018 1019 1020
Doping density (cm-3)
Un-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
6.3 Hetero-Structure Field Effect Transistors (HFETs) 87
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
6.3 Hetero-Structure Field Effect Transistors (HFETs) 89
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
2DEG
EC
--
---
- EFm
--
EV
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
90 6 Microwave Field Effect Transistors
Source Drain
Gate
+
n InGaAs n+ InGaAs
n+ n+
n InAlAs donor layer
2DEG
Undoped In0.53Ga 0.47As channel
InP Substrate
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.
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).
6.3 Hetero-Structure Field Effect Transistors (HFETs) 91
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
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.
92 6 Microwave Field Effect Transistors
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
S VGS G D
n+ AlxGa1-xN n+
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.
Intrinsic FET
Rg cgd Rd
cgs cdc
gmvgs rds
ri
Rs
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
96 6 Microwave Field Effect Transistors
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
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
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
98 6 Microwave Field Effect Transistors
(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
6.5 Equivalent Circuit of Microwave FETs 99
(Ω)
= - 1.5 V
3 = - 1.0 V
10
= - 0.5V
= - 0.3V
103
102
101 (V)
0 3.0 6.0
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.
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
100 6 Microwave Field Effect Transistors
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
6.5 Equivalent Circuit of Microwave FETs 101
(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
(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.
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
102 6 Microwave Field Effect Transistors
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
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
6.5 Equivalent Circuit of Microwave FETs 103
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
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
104 6 Microwave Field Effect Transistors
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 .
6.5 Equivalent Circuit of Microwave FETs 105
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)
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
106 6 Microwave Field Effect Transistors
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
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
0.15 × 1000
fT GHz 36.73 GHz
2π × 0.65
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
L 2g
τ (6.20c)
μVd
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
108 6 Microwave Field Effect Transistors
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
1. Sze SM (1981) Physics of semiconductor devices, 2nd edn. Wiley Eastern Limited
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
110 6 Microwave Field Effect Transistors
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
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
114 7 Microwave Circuit Analysis
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)
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
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
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)
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
V − (z) V − e+ jβl
Γ (l) + − jβl Γ e+ j2βl (7.20)
V (z)
+ V e
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 |
Putting the value of Γ (l) from (7.20), the input impedance becomes
118 7 Microwave Circuit Analysis
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
7.2 Transmission Line Theory and Analysis 119
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)
120 7 Microwave Circuit Analysis
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)
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.
√
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
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.
122 7 Microwave Circuit Analysis
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
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
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
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)
Center conductor
Dielectric fill
Outer conductor
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)
b
μo μr I μo μr I l b
Φ ldx ln Ll I (7.37e)
2π x 2π a
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)
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
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
126 7 Microwave Circuit Analysis
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
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)
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.
7.3 Microwave Transmission Lines 127
High
conducting
metal plating
Circular Rectangular
Fig. 7.7 Waveguide structures with circular and rectangular cross section
7.3.3 Waveguides
∂D
∇×H J+ (7.41a)
∂t
∂B
∇×E − (7.41b)
∂t
∇·Dρ (7.41c)
∇·B0 (7.41d)
128 7 Microwave Circuit Analysis
∂2 E
∇∇ · E − ∇ 2 E −με (7.43)
∂t 2
∂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
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
7.3 Microwave Transmission Lines 129
X
a
Z
⎡ ⎤
x̂ ŷ ẑ
⎢ ⎥
⎢ ∂ ∂ ∂ ⎥ − jωμ x̂ Hx + ŷ Hy + ẑ Hz (7.46)
⎣ ∂x ∂y ∂z ⎦
Ex E y Ez
∂ Hz
+ γ Hy jωεE x (7.49a)
∂y
∂ Hz
+ γ Hx − jωεE y (7.49b)
∂x
∂ Hy ∂ Hx
− jωεE z (7.49c)
∂x ∂y
kc2 γ 2 + ω2 με (7.50)
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
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.
∂ 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)
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 .
∂ 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
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.
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
134 7 Microwave Circuit Analysis
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)
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
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
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
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)
7.3 Microwave Transmission Lines 137
E-lines
b
H-lines
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
138 7 Microwave Circuit Analysis
Hz 0 (7.70f)
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 the values of m, n cut-off frequencies of different modes are shown in the
following Table:
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 the values of m, n cut-off frequencies of different modes are shown in the
following Table:
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.
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.
140 7 Microwave Circuit Analysis
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:
7.3 Microwave Transmission Lines 141
E field
H field
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.
142 7 Microwave Circuit Analysis
E field
H field
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
dielectric ( )
t
s s
h h
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.
7.3 Microwave Transmission Lines 145
E field
H field
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
146 7 Microwave Circuit Analysis
Bond
Wires
broadband performance as it does not need via holes for shunt and series-connected
elements.
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
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
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
is capacitive or inductive
148 7 Microwave Circuit Analysis
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)
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
is capacitive or inductive
7.4 Transmission Line Elements 149
Zin = ZL
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)
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
1 − Γr2 − Γim2
r (7.91a)
(1 − Γr )2 + Γim2
2Γim
x (7.91b)
(1 − Γr )2 + Γim2
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
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
152 7 Microwave Circuit Analysis
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)
7.5 Smith Chart Analysis 153
From (7.20), the reflection coefficient Γ (l) of a lossless transmission line can be
written as
Therefore,
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)
154 7 Microwave Circuit Analysis
1.0 j -1.0 j
0.5 j 2.0 j -2.0 j -0.5 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
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.
7.5 Smith Chart Analysis 155
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
156 7 Microwave Circuit Analysis
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.
(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.
158 7 Microwave Circuit Analysis
0.5 2
H
G
E
F
A
B
0.5 1 2
0 ∝
C D
Z Chart
0.5 2 Y Chart
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
160 7 Microwave Circuit Analysis
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 , when V2+ 0 (7.106)
V1+
7.6 Network Theory of Circuits and Transmission Lines 161
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
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
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
162 7 Microwave Circuit Analysis
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
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
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
1 + 1
In Vn − Vn− √ (an − bn ) (7.120)
Z on Z on
or
⎡ ⎤
⎡ ⎤ ⎢ S11 S12 · · · S1n ⎥⎡ ⎤
b1 ⎢ ⎥ a1
⎢ ⎥ ⎢ ⎥⎢ ⎥
⎢ b2 ⎥ ⎢ S21 S22 · · · S2n ⎥⎢ a2 ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥
⎢ ⎥⎢ ⎥⎢ ⎥ (7.126)
⎢ .. ⎥ ⎢ ⎥⎢ . ⎥
⎣ . ⎦ ⎢
⎢
..
.
⎥⎣ ..
⎥ ⎦
⎢ ⎥ a
bn ⎣ ⎦ n
Sn1 Sn2 · · · Snn
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
In matrix form,
b1 S11 S12 a1
(7.132)
b2 S21 S22 a2
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+
7.6 Network Theory of Circuits and Transmission Lines 165
√
b2 V2− Z o2 V−
S22 √ . + 2+ , when V1+ 0 (7.137)
a2 Z o2 V2 V2
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
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
166 7 Microwave Circuit Analysis
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
7.6 Network Theory of Circuits and Transmission Lines 167
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
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,
∗ ∗
Similarly, S11 S12 + S21 S22 0
Thus, the matrix elements satisfy all the conditions for a unitary matrix. Thus, the
network is lossless.
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
From the S-parameter matrix, it is clear that the network is matched and reciprocal.
Solution
The two-way power divider does not contain any active device and non-reciprocal
components. Thus, the network is reciprocal, we can write
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)
Since the network is lossless, the matrix is unitary. Thus, applying (7.145a) we
can write
and
and
∗ ∗
1/2 + S22 S23 + S23 S33 0 (7.156b)
Here, S22 and S33 are non-zero. Thus, for the 3-port lossless network, the other
ports are not matched.
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
7.6 Network Theory of Circuits and Transmission Lines 171
and
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.
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 )
172 7 Microwave Circuit Analysis
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 )
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
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
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.
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
176 7 Microwave Circuit Analysis
V1 1 0 Va
(7.169a)
I1 Ya 1 Ia
or
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
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
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
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.
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].
VL− Z L − Zo
ΓL + (7.171a)
VL Z L + Zo
Similarly,
VS ΓL VS 1 − Γ S
VL− ΓL (7.171e)
1 + ΓL ΓS + ΓL 2 1 − ΓS ΓL
1 |VS |2 |1 − Γ S |2
PincL (7.172b)
8 Z o |1 − Γ S Γ L |2
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
or
7.7 Power Transfer in Microwave Networks 181
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.
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
and
Input reflection coefficient ΓIN and output reflection coefficient ΓOUT can be writ-
ten as
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
1 |VS |2 |1 − Γ S |2
PIN 1 − |ΓIN |2 (7.177e)
8 Z o |1 − Γ S ΓIN | 2
S21 V1+
V2− (7.177g)
1 − S22 Γ L
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
184 7 Microwave Circuit Analysis
Fig. 7.51 The 2-port network with source and load impedance Z o
PL
GT |S21 |2 (7.181b)
PAVS
PL 1 1 − |Γ L |2
GP |S21 |2 (7.182)
PIN 1 − |ΓIN | 2
|1 − S22 Γ L |2
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 .
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
Z L − Zo
ΓL 1
Z L + Zo
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
Thus,
vo 2
S21
vg /2 3
The same result can also be obtained using (7.162c), putting Z a 50 and
Z o 50 as:
7.7 Power Transfer in Microwave Networks 187
⎡ ⎤
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
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
vg × 50||Z o vg
vo
Z o + 50||Z o 3
Thus,
vo 2
S21
vg /2 3
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
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)
7.7 Power Transfer in Microwave Networks 189
b3 0 × a1 + 1 × a2 + 0 × a3 (7.185c)
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)
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)
a3 −b3 (7.191a)
a3 −a2 (7.191b)
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)
Solution
Magnitude of reflection coefficient of load impedance Z L is given by
7.7 Power Transfer in Microwave Networks 191
Z L − Zo
|S11 |
Z L + Zo
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
Therefore,
j X 2 + j X 1 ||50 10 + j10
or
On equating real and imaginary parts equal for both the sides separately
X 1 +25
192 7 Microwave Circuit Analysis
j X 1 + j25, j X 2 − j10
and
j X 1 − j25, j X 2 + j30
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)
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)
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◦
Transducer gain G T is defined as the ratio of power delivered to the load to power
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
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
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
7.7 Power Transfer in Microwave Networks 195
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.
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
Solution
The input reflection coefficient (7.177a)
Γ L S12 S21
ΓIN S11 +
1 − Γ L S22
Transducer gain G T is defined as the ratio of power delivered to the load to power
available from the source and is given by (7.181a)
196 7 Microwave Circuit Analysis
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
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 ! !
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
7.7 Power Transfer in Microwave Networks 197
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
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.
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
(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
(a) VC (b) VC
Cbypass
RF
Cb choke Cb
Cb Cb
ZS RF ZS
choke ZL ZL
switch
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
202 8 Microwave Switches
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 .
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
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,
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))
204 8 Microwave Switches
⎡ ⎤
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
From (8.8a),
1
2Ro − jXC
2
j1326.29
2
Isolation
|S21 |2 2Ro
1 − 2 × 50
176.905
or
Thus, for a 50 system, the ON-state insertion loss is 0.343 dB and OFF-state
isolation is 22.48 dB.
206 8 Microwave Switches
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
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
208 8 Microwave Switches
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
Here, RT (Ri + Rs ) 4 , CT Cj + Cp 0.03 pF and Lp 0.1 nH.
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
From (8.8a),
2
2
1
Ro
50
Isolation
1 +
1 +
37.97
|S21 |2 2(RT + jXL )
2(4 + j2.513)
or
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.
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.
VC
Cb Cb
ZS
ZL
ZT1
ZO
ZT2 Vout ZO
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
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
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
1 Zo RT − XC XL XC Zo XL + RT XC
Isolation 1+ 2 + + 2 (8.20b)
|S21 | 2
2 RT + XL2 2Zo 2 RT + XL2
212 8 Microwave Switches
⎡ 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
or
For a 50 system, the ON-state insertion loss is 0.333 dB and OFF-state isolation
is 43.617 dB.
8.5 Compound Switch Configuration 213
VC
Cb Cb
ZS
ZL
I1 I2
ZT1 ZT1
ZO
ZT2
Vout ZO
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.
With reference to Fig. 8.15, the input and output reflection coefficient of the switch
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
214 8 Microwave Switches
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
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
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
8.5 Compound Switch Configuration 215
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
XC 2Zo RT XC − XC2 XL − Zo2 XL
+ + (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
Here, RT (Ri + Rs ) 4 , CT Cj + Cp 0.03 pF and Lp 0.1 nH.
216 8 Microwave Switches
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
Thus,
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
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.
With reference to Fig. 8.16, the input and output reflection coefficient of the switch
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
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
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
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
220 8 Microwave Switches
CDS
LS RS distributed channel RC RD LD
S D
CGS CGD
RG
LG
S D
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, 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
ZT2
ZT2 RG RG
RG
VG1
VG2
RF ZT1 RF
IN OUT
ZT1
ZT2 ZT2
ZT2 ZT2
RG
RG RG
VG2
VG1
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.
8.8 Applications of RF/Microwave Switches 223
Antenna
VC1 VC2
From To
Transmitter Receiver
RF1
RF2 RF3
RG RG RG RG
VG1
VG2
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.
224 8 Microwave Switches
RF1
RF2 RF3
RG RG
RG RG
Rs Rs
VG1
VG2
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
With reference to Fig. 8.25c, the impedance seen by the RF3 port is given by
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
With reference to Fig. 8.26c, the impedance seen by the RF3 port is given by
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,
8.8 Applications of RF/Microwave Switches 227
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
ZT1 ZT1
ZO
ZT2 ZT2 Vout ZO
or
1 2ZT 1 Zo ZT 1 ZT 1 Zo
1+ + + + 2 (8.29)
S21 ZT 2 ZT 2 Zo ZT 2
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
or
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
ZT 1 ZTH −jXC p-i-n diode OFF
j2π fCT
ZT 2 ZTL RT + j2π fLP RT + jXL p-i-n diode ON
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
2RT XC + Zo XL XL Zo XC R2T − XL2
+ − − − 2 (8.31)
R2T + XL2 Zo R2 − X 2 + (2RT XL )2
T L
or
230 8 Microwave Switches
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
RT 2XL Zo RT 2 XL Zo 2RT Zo XL 2
IL 1 + − − + + + −
ZO XC XC2 ZO XC XC XC2
2
Zo RT − 2XC XL 2Zo RT XC XL
Isolation 1 + − 2
R2T + XL2 R2T − XL2 + (2RT XL )2
2
2RT XC + Zo XL XL Zo XC R2T − XL2
+ − − − 2
R2T + XL2 ZL R2 − X 2 + (2RT XL )2
T L
or
or
The ON-state insertion loss is 0.710 dB and OFF-state isolation is 39.33 dB.
8.8 Applications of RF/Microwave Switches 231
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
RT 2XL Zo RT 2 XL Zo 2RT Zo XL 2
IL 1 + − − + + + −
Zo XC XC2 Zo XC XC XC2
2
Zo RT − 2XC XL 2Zo RT XC XL
Isolation 1 + − 2
R2T + XL2 R2T − XL2 + (2RT XL )2
2
2RT XC + Zo XL XL Zo XC R2T − XL2
+ − − − 2
R2T + XL2 Zo R2 − X 2 + (2RT XL )2
T L
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
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
Thus,
2vout2 ◦
S21 0.79863 0
vs1
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
234 8 Microwave Switches
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
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.
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
(a) (b)
Lp
Rs
Rs
Cp
Ri Ci
Ri Ci
For
The input and output return losses of the series diode attenuator are given by
238 9 Microwave Attenuators
(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
10 40
AƩenuaƟon
30
Return loss Return
Attenuation 5 20 loss
(dB) (dB)
10
0 0
0 50 100 150
(Ω)
25 AƩenuaƟon 20
20 Return loss
15
15 Return
Attenuation 10 loss
(dB) 10
5
(dB)
5
0 0
0 50 100 150
(Ω)
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.
240 9 Microwave Attenuators
VC
RF
choke
ZO
ZO
Π network Π network
RF RF
λO/4 λO/4 λO/4
I/P O/P
ZT ZT ZT ZT
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
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):
and
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
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
244 9 Microwave Attenuators
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.
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
ZT1 ZT1
ZO
ZT2 vout ZO
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 )
Z T 1 R1 , Z T 2 R2 , and Z o Ro
where
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
10 × log(N ) 3 dB
9.5 Matched Attenuator Circuits 247
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
142Ω
Thus,
N 103/10
Ro 50
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
248 9 Microwave Attenuators
VC1
Cb Cb
ZO
ZO
VC2
9.5.2 π Attenuator
or
9.5 Matched Attenuator Circuits 249
ZT1 ZT1
ZO
ZT2 ZT2 vout ZO
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
Attenuation(in dB) 20 log10
|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
Z T 1 R1 , Z T 2 R2 , and Z o Ro
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
250 9 Microwave Attenuators
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.
Solution
Attenuation N is given by
Pavs
N
PL
10 × log(N ) 6 dB
Thus,
N 106/10
9.5 Matched Attenuator Circuits 251
150.5Ω 150.5Ω
Ro 50
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
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.
252 9 Microwave Attenuators
VC
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
9.5 Matched Attenuator Circuits 253
VC
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
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 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:
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
256 9 Microwave Attenuators
Id
RL
RS Id× RS
Vd
VS Ri Vdj
Ci
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
ΓcRi(%/oC)
-1.0
VOPT
-2.0
0.5 1.0 1.5 2.0 2.5 3.0
VS (volt)
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.
258 9 Microwave Attenuators
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
R1
Digital data
Analog Mux
VS
Ra V VC
Voltage OPT
regulator
Cb
Rb
Rn
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
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 )
∂ VOPTM /∂ T 0 (9.29b)
Fig. 9.29 Graphical representation of V OPTM for positive temperature coefficient of bias resistor
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
Amb Hot
Cold
Id (mA)
Optimum
Required Bias load
(Vd,Id)
Fig. 9.30 Graphical representation of V OPTM for positive temperature coefficient of bias voltage
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.
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
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
264 9 Microwave Attenuators
CDS
LS RS distributed channel RC RD LD
S D
CGS CGD
RG
LG
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
ZT2
ZT2 RG RG
RG
VG2
VG1
RF ZT1 RF
IN OUT
ZT1
ZT2 ZT2
ZT2 ZT2
RG
RG RG
VG1
VG2
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
266 9 Microwave Attenuators
VC VC VC VC
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.
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
9.9 FET-Based Attenuators 267
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
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
268 9 Microwave Attenuators
(a) (b)
50 Ω (100|| 50+50) Ω
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
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
9.9 FET-Based Attenuators 269
(a) (b)
√
(N − 1) N + 1
R1 Ro √ and R2 Ro √
4 N N −1
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).
270 9 Microwave Attenuators
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
Attenuation(in dB) 20 log10
|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|
9.9 FET-Based Attenuators 271
(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
Putting R S 5 , Ro 50 , f 10 GHz
1/2
(Ri − 45)2 + (0.28274Ri )2
|Γ |
(Ri + 55)2 + (0.34557Ri )2
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
(Ri ||Ci ) + Rs − Ro
Γ2
(Ri ||Ci ) + Rs + Ro
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
Here, a, b, c are constants. Derive the optimum open-circuit voltage for zero
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)
274 9 Microwave Attenuators
VS
Rb
Using (9.32c), the I–V relationship of the p-i-n diode can be written as
VS − I d Rb − E b
Id bT × exp
2
(9.32d)
cT
Suppose for zero temperature coefficient VS VOPT and Id Ido . Putting these
values to (9.32e) and using (9.32b),
Ido Ido Rb 2Ido Ido (VOPT − Ido Rb − E b )
1+ − (9.32f)
T cT T cT 2
This leads to the value of optimum supply voltage for zero temperature coefficient
of RF resistance as
VOPT E b + cT (9.32g)
Here, a, b, c are constants. Derive the optimum open-circuit voltage for zero
temperature coefficient of attenuation of the attenuator.
9.9 FET-Based Attenuators 275
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 (9.33a)
∂T Id Id ∂ T
or
∂ Id Id
(9.33b)
∂T T
As shown in Fig. 9.42, 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 (9.33c)
Using (9.33c), the I–V relationship of the p-i-n diode can be written as
VS − n I d Rb − E b
Id bT × exp
2
(9.33d)
cT
Suppose, for zero temperature coefficient, VS VOPT and Id Ido . Putting these
values to (9.33e) and using (9.33b),
Ido n Ido Rb 2Id Ido (VOPT − n Ido Rb − E b )
1+ − (9.33f)
T cT T cT 2
This leads to the value of optimum supply voltage for zero temperature coefficient
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
276 9 Microwave Attenuators
VS
Rb
ZO
ZO
D1 D2
aT Vd − E b
Rd and Id bT 2 × exp
Id cT
Here, a, b, c are constants. Derive the optimum open-circuit voltage for zero
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 (9.34a)
∂T Id Id ∂ T
or
∂ Id Id
(9.34b)
∂T T
As shown in Fig. 9.43, 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 Id Rb + 2Vd (9.34c)
Using (9.34c), the I–V relationship of the p-i-n diode can be written as
9.9 FET-Based Attenuators 277
VS − Id Rb − 2E b
Id bT 2 × exp (9.34d)
2cT
Suppose, for zero temperature coefficient, VS VOPT and Id Ido . Putting these
values to (9.34e) and using (9.34b),
Ido Ido Rb 2Ido Ido (VOPT − Ido Rb − 2E b )
1+ − (9.34f)
T 2cT T 2cT 2
This leads to the value of optimum supply voltage for zero temperature coefficient
of RF resistance as
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
278 9 Microwave Attenuators
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
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.
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
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.
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,
282 10 Microwave Phase Shifters
(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
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.
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.
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-
284 10 Microwave Phase Shifters
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)
and can be written as
(L 1 − L R ) (L 1 − L R )
Φ( f ) 2π 2π f (10.6)
λg vp
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
Fig. 10.8 Vector modulator diagram for the four-quadrant phase shifter
288 10 Microwave Phase Shifters
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.
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
Fig. 10.9 Quadrature hybrid matched attenuator configuration with series DC path
10.4 Realization of Phase Shifters 289
Combining (10.11a) and (10.11c), amplitude and phase of the transmission coef-
ficient can be written as
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.
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
10.4 Realization of Phase Shifters 291
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
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 ,
292 10 Microwave Phase Shifters
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.
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
10.4 Realization of Phase Shifters 293
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
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
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)
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.
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)
−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
(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
300 11 Microwave Modulators
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.
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
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.
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
Fig. 11.6 I–Q modulator circuit using two bi-phase (BPSK) balanced modulator
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.
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.
306 11 Microwave Modulators
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.
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.
(a) R1 (b)
C1 R2 R1 R2 R3
L1 L1 L2 L3
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
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
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. 12.6 A short-circuited network consists of a transmission line of length nλo /8 in between two
resistors
312 12 Amplitude Tilt Microwave Equalizers
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.
314 12 Amplitude Tilt Microwave Equalizers
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
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.
316 12 Amplitude Tilt Microwave Equalizers
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
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.
318 12 Amplitude Tilt Microwave Equalizers
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
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.
Example 12.2 Compute S-parameter of a shunt network consists of a resistance Ra
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
12.7 Versatile Equalizer with Variable Gain Slope and Insertion Loss 321
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)
b1
S11 , when a2 0
a1
V−
1+ , when V2+ 0
V1
−Z o /(Z o + 2Z sn )
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
324 12 Amplitude Tilt Microwave Equalizers
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 )
−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)
Plot of the transmission and reflection coefficients of the network for different
values of Rb are shown in Figs. 12.26 and 12.27.
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
326 12 Amplitude Tilt Microwave Equalizers
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
Z o +2Z sn Z o +2Z sn
S21 S22 2Z sn −Z o
Z o +2Z sn Z o +2Z sn
12.7 Versatile Equalizer with Variable Gain Slope and Insertion Loss 327
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.
328 12 Amplitude Tilt Microwave Equalizers
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 )
Transmission and reflection coefficients of the shunt network of Fig. 12.30 are
shown in Fig. 12.31 over the frequency range of 5–15 GHz.
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 )
330 12 Amplitude Tilt Microwave Equalizers
Transmission and reflection coefficients of the shunt network of Fig. 12.32 are
shown in Fig. 12.33 over the frequency range of 5–15 GHz.
References
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
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.
and
Here, θ is the phase difference in between the voltage and current. The instanta-
neous power across the load is given by
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)
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
336 13 Microwave Detectors
p(t)
p(t) - pav
pav
t
T
0, 0 T/2
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.
Solution
Using (13.4a),
100 mW
100 mW Power(in dBm) 10 × log
1 mW
2
10 × log 10 dBm
20 dBm
13.2 Microwave Power 337
Using (13.4b),
100 mW
100 mW Power(in dBW) 10 × log dBW
1W
2
10
10 × log dBW
103
−10 dBW
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
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.
338 13 Microwave Detectors
ATTN
15 dB
ATTN
13 dB
Solution
100 W
100 W Power(in dBW) 10 × log dBW 20 dBW
1W
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
400 W Power(in dBW) 10 × log dBW 26 dBW
1W
Power
TP
P
PP
Pavg
time
20 W
20 W Power(in dBW) 10 × log dBW 13 dBW
1W
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
340 13 Microwave Detectors
τP
Duty Cycle (13.6)
TP
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)
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
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
Vd j Vo + Vs cos(ωt) (13.9)
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
Id ( A)
20
10
Vd (mV)
-40 -30 -20 -10 10 20 30 40
-10
-20
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
13.3 Diode Detectors 343
Id ( A)
Detected
voltage Diode current
Vd (mV)
Input voltage
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
(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
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
13.3 Diode Detectors 345
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.
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
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-
348 13 Microwave Detectors
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
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
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
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.
RECEIVER TRANSMITTER
Mixer
6 GHz Low High High 4 GHz
Noise Gain Power
Amp Amp Amp
Rx Tx
2 GHz
Antenna Antenna
Local
Oscillator
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
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
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)
14.2 Types of Microwave Amplifiers 355
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
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
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,
356 14 Microwave Solid-State Amplifiers
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
Pni k To B (14.9b)
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
Therefore, the total output noise power over the 4.0 kHz bandwidth is as follows:
Therefore, the minimum detectable input signal level PiMDS can be expressed as
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
Using (14.10), (14.11) and considering X (in dB) 3 dB, the dynamic range can
be written as
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.
358 14 Microwave Solid-State Amplifiers
Solution
The higher limit on input power level of the dynamic range:
Therefore,
DR(in dB) Pi1dB (in dBm) − PiMDS (in dBm) −13 − (−90) 77
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:
14.2 Types of Microwave Amplifiers 359
Signal
Frequency
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
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
14.2 Types of Microwave Amplifiers 361
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)
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)
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
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
and
From (14.19a) and (14.19b) the condition for unconditional stability for a two-
port network can be derived as [1–3]
and,
364 14 Microwave Solid-State Amplifiers
where
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
Solution
From Table 14.1, the S-parameters of the device at 6.0 GHz are
14.3 Stability of Microwave Amplifier 365
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
366 14 Microwave Solid-State Amplifiers
To calculate ||:
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
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
14.3 Stability of Microwave Amplifier 367
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
From Table 14.1, the S-parameters of the device at 14.0 GHz are
To calculate ||:
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
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
14.3 Stability of Microwave Amplifier 369
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
or
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
370 14 Microwave Solid-State Amplifiers
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
or,
|S11 − ΓL | |1 − 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
14.3 Stability of Microwave Amplifier 371
∗ ∗
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
Example 14.6 Compute the stability circles at 6.0 GHz of the device for S-parameters
which are given in Table 14.1.
372 14 Microwave Solid-State Amplifiers
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
14.3 Stability of Microwave Amplifier 373
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.
374 14 Microwave Solid-State Amplifiers
Solution
For input stability circle at 14.0 GHz:
∗ ∗
S11 − S22
cS
|S11 |2 − ||2
| |< 1 | |< 1
(a) (b)
= 2.145
= 3.227 − 114.42o
= 9.126 − 4.6o
= 10.263
Input stability circle
Output stability circle
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.
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
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,
378 14 Microwave Solid-State Amplifiers
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
14.4 Single-Stage Amplifier Design 379
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 |
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 |
From Table 14.1, the S-parameters of the device at 14.0 GHz are
380 14 Microwave Solid-State Amplifiers
Therefore,
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
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 |
Example 14.9 Compute maximum stable gain at 6.0 GHz of the device (Ref.
Table 14.1).
Solution
From Table 14.1, the S-parameters of the device at 6.0 GHz are
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)
382 14 Microwave Solid-State Amplifiers
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.
1 − |ΓS |2 1 − |ΓL |2
GT × |S21 | 2
× GS × Go × GL
|1 − ΓS ΓIN |2 |1 − S22 ΓL |2
1 − |ΓS |2
GS (14.43a)
|1 − ΓS ΓIN |2
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
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
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
384 14 Microwave Solid-State Amplifiers
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
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:
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,
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
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.
386 14 Microwave Solid-State Amplifiers
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
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
14.5 Amplifier with Specific Gain 387
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
388 14 Microwave Solid-State Amplifiers
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
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 )
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
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
390 14 Microwave Solid-State Amplifiers
Solution
∗
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,
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.
14.5 Amplifier with Specific Gain 391
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
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
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
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
And,
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∗ .
14.5 Amplifier with Specific Gain 393
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.
394 14 Microwave Solid-State Amplifiers
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.
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 )
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
or
√
k − k2 − 1
gP max (14.62)
|S12 S21 |
|S12 S21 |
RP (14.64b)
|S22 |2 − ||2
Solution
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
Using the values of these parameters, centre and radius of constant operating
power gain circles given by (14.60a) and (14.60b) corresponding to 12.2, 10.0, 8.0,
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.
Solution
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
Operating 1.0 j
gain (dB)
0.5 j 2.0 j
6.0
8.0
10.0
12.2
Locus of centre
of the constant
GP circles -0.5 j
-2.0 j
-1.0 j
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
Therefore,
And
Using the values of these parameters, centre and radius of constant operating
power gain circles given by (14.64a) and (14.64b) corresponding to 16.0, 14.0, 12.0,
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.
14.5 Amplifier with Specific Gain 399
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.
400 14 Microwave Solid-State Amplifiers
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.
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.
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
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
Expressing GSource , the real part of YSource in terms of the reflection coefficient and
substituting in (14.67):
This is a constant, for a given noise figure and set of noise parameters NFmin for
ΓOPT . Thus, (14.69) can be written as
402 14 Microwave Solid-State Amplifiers
ΓS ΓS∗ + ΓOPT ΓOPT
∗ ∗
− ΓS ΓOPT − ΓS∗ ΓOPT N 1 − ΓS ΓS∗ (14.70)
or,
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
Solution
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):
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
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:
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
406 14 Microwave Solid-State Amplifiers
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:
Solution
From the S-parameters,
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.
14.6 Small Signal Amplifiers 407
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.
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
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
Solution
Synthesis of input matching network:
The desired normalized source impedance is
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
410 14 Microwave Solid-State Amplifiers
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
14.6 Small Signal Amplifiers 411
= 1 2 3
AMP1 AMP2 AMP3
1 2 3
1 2 3
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.
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
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
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
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:
Solution
Noise and gain in factors of the individual amplifiers are
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
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
414 14 Microwave Solid-State Amplifiers
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◦ .
14.6 Small Signal Amplifiers 415
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:
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
416 14 Microwave Solid-State Amplifiers
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
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◦
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.
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
418 14 Microwave Solid-State Amplifiers
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 j
-2.0 j
-1.0 j
Fig. 14.25 Source and load reflection coefficients for conditional stable device using constant
operating gain approach
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
Solution
Synthesis of input matching network:
The desired normalized source impedance corresponding to the ZSdesired is
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
along the constant VSWR circle as shown in the diagram. The electrical length of
the series transmission line is 25.65° (0.0713λ).
420 14 Microwave Solid-State Amplifiers
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
along the constant VSWR circle as shown in the diagram. The electrical length of
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.
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
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
422 14 Microwave Solid-State Amplifiers
Non-linear
Low
=
<0
Increasing
Linear
approximation
of vs High
characteristic
>0
Pinch off
voltage
dependency
Low on
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
14.7 Large Signal Amplifiers 423
Large signal
non elliptic
load curve
Small signal
elliptic load
curve
( , )
0
−
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
( , )
− 0
−
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
VdgB − Vgso − VK
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
( , )
− 0
−
= < >
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
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
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
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
14.7 Large Signal Amplifiers 429
= >
VdgB − Vgso − VK
RoOPT
IF
20 − 4
16
1
The optimum load impedance is
rds × RoOPT
RLOPT
rds − RoOPT
100 × 16
19.05
100 − 16
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
430 14 Microwave Solid-State Amplifiers
idrf gm vs
(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
( )
( )
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
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
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
14.7 Large Signal Amplifiers 433
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
Now, the output power at 1-dB gain compression point Po1dB can be expressed as
Here, GLdB is the linear gain of the amplifier. Combining (14.85), (14.87) and
(14.88), the Po1dB can be expressed as
or
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).
434 14 Microwave Solid-State Amplifiers
Solution
Output power at 1-dB gain compression point is given by
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.
PM PM PM
DUT
INPUT OUTPUT
TUNER TUNER
VECTOR NETWORK
ANALYZER
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.
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
436 14 Microwave Solid-State Amplifiers
1.0 j
0.5 j 2.0 j
Maximum
efficiency
Maximum
output power
-0.5 j
-2.0 j
-1.0 j
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
can be determined by
14.7 Large Signal Amplifiers 437
Class-A Class-AB
/2
< /2
0 0
Class-B Class-C
0 0
0,0 2 5 /2 3 4
2 1
3 /2 7 /2
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
Therefore,
IF −IF cos θ1
Idp and, IdQ (14.95)
1 − cos θ1 1 − cos θ1
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
14.7 Large Signal Amplifiers 439
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 )
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
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
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
440 14 Microwave Solid-State Amplifiers
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
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)
2
IF 2VDC IF VDC
PBrf max √ (14.100c)
2 2 IF 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.
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
442 14 Microwave Solid-State Amplifiers
/2
0 0
Class-A Class-B
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 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
14.7 Large Signal Amplifiers 443
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
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:
444 14 Microwave Solid-State Amplifiers
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
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.
Here, ω is the angular frequency of the RF signal a0 , an and bn are constants and
can be determined by
446 14 Microwave Solid-State Amplifiers
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
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
14.7 Large Signal Amplifiers 447
⎡ 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.
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
448 14 Microwave Solid-State Amplifiers
( , )
0
−
voltage capacity which ensures symmetric clipping of the output current waveform,
i.e.
θ1 θ2 α (say)
IQ IF /2
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
−
0
( −2 + − )/2
extent by clipping the signal waveform in the preceding amplifier stage before feeding
it to the high power amplifier.
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)
3.0 90
Efficiency (%)
2.0 85
PRF (dB)
1.0 80
0.0 75
0 45 90 135 180
Clipped Angle (2 1)
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
452 14 Microwave Solid-State Amplifiers
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
can be determined by
14.7 Large Signal Amplifiers 453
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
vdrf (θ ) 0, over 0 ≤ θ ≤ π
vdp , over π ≤ θ ≤ 2π
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
454 14 Microwave Solid-State Amplifiers
or
2vdp
bn − nπ
for all odd values of n
0 for all even values of n
∞
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
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-
14.7 Large Signal Amplifiers 455
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
456 14 Microwave Solid-State Amplifiers
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.
14.7 Large Signal Amplifiers 457
1.0 100
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
Fig. 14.48 RF output power and efficiency dependence on amplitude of third harmonic component
(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
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
14.7 Large Signal Amplifiers 459
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
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
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.
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
460 14 Microwave Solid-State Amplifiers
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
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
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
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
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,
14.8 FETs Output Power Capability 463
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.
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
464 14 Microwave Solid-State Amplifiers
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.
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
14.8 FETs Output Power Capability 465
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
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:
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
466 14 Microwave Solid-State Amplifiers
Fig. 14.55 Schematic block diagram of corporate-type power combining technique for combining
8 transistors
Thus,
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
14.8 FETs Output Power Capability 467
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
RF RF
100 Ω 100 Ω
IN OUT
50Ω 50Ω
o /4 o /4
o /4 o /4
50Ω 50 Ω
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
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.
14.8 FETs Output Power Capability 469
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.
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
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
14.8 FETs Output Power Capability 471
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
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
472 14 Microwave Solid-State Amplifiers
(a) o/4, Z1
1 2
A
o/4, Zo o/4, Zo
4 o/4, Z1 3
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
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
14.8 FETs Output Power Capability 473
Z12
Zo +jZo
+jZo Zo
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.
474 14 Microwave Solid-State Amplifiers
Fig. 14.62 Schematic diagram of balanced power amplifier using branch line couplers
Solution
Gain of the amplifier-A and amplifier-B are
GA 6 dB 4 (in factor)
GB 20 dB 100 (in factor)
Thus, cascaded lowest noise figure is 4.024 dB where amplifier-B is used as first
amplifier in cascaded two-stage amplifier.
14.8 FETs Output Power Capability 475
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
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:
Pdissi (20 − 8) W 12 W
Increase of temperature:
476 14 Microwave Solid-State Amplifiers
T 12 W × 4.5 ◦ C/W 54 ◦ C
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.
IEEE Trans Microw Theory Tech 54(3):993–1000
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
478 14 Microwave Solid-State Amplifiers
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
15.1 Introduction
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.
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)
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.
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
484 15 Microwave Limiters
Fractional power absorbed, i.e. dissipated within the diode impedance is given by
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
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
15.3 P-I-N Diode Limiters 485
Transmitted
Reflected
1.0
0.5
0.0
1 10 100 1000
Diode Resistance (Ω)
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.
486 15 Microwave Limiters
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)
15.3 P-I-N Diode Limiters 487
−Z o 2
pr |S11 |
2
2Z T + Z o
−50 2
0.25
50 + 50
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 )
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)
Vc
Rb
RF RF
3-dB 90o Coupler
IN OUT
Rs
D1 D2 Rd Cd
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
490 15 Microwave Limiters
RF IN
RF Out
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)
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)
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 .
492 15 Microwave Limiters
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)
(a) (b)
POUT
Limiting characteristic
PDC
PIN POUT
PDISSI
PIN
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
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
Output
Power
Cold
Ambient
Hot
Input power
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.
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
-V
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
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
498 15 Microwave Limiters
(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.
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
14. 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
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
AM/AM AM/PM
Pout SSPA
Gain
TWTA
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.
504 16 Microwave 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.
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)
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.
b Gb
Delay-2 Coupler-2
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.
RF Predistortion Power RF
IN Linearizer Amplifier OUT
gain phase
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.
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
0.0
-0.50 -0.25 0.00 0.25 0.50 0.75 1.00
Vd (V)
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.
510 16 Microwave Linearizers
Ci
L
Cb1 Cb2
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
diode can be written as
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
16.4 Implementation of Predistortion Linearizers 511
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.
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.
512 16 Microwave Linearizers
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
16.4 Implementation of Predistortion Linearizers 513
|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)
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.
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
514 16 Microwave Linearizers
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.
16.4 Implementation of Predistortion Linearizers 515
Rip
Rbp Rsp
VSD, ISD
VO
Cip Pin diode
λ o/4 Ca λo/4
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
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
16.4 Implementation of Predistortion Linearizers 517
S21 (deg.)
-10 |S21| 50
-15 0
10 100 1000
Rsd ( )
RF IN RF Out
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
518 16 Microwave Linearizers
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
T n−m
Rd (Id (T ), T ) z p (16.12a)
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
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
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
520 16 Microwave Linearizers
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.
Predistortion linearizers are cascaded with a power amplifier such as solid-state power
amplifiers (SSPAs) and travelling wave tube amplifiers (TWTAs) to improve nonlin-
16.4 Implementation of Predistortion Linearizers 521
Non-linear Phase RF
RF Attenuator Shifter
OUT
IN Linear
Equa-1 Equa-2
Attn-2
Linear
Attn-1
Temp. dependent
resistive network
VS
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,
16.4 Implementation of Predistortion Linearizers 523
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.
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
524 16 Microwave Linearizers
Control signals
Envelop
Detector DAC DAC
Fig. 16.21 Schematic block diagram of a digital predistortion linearizer variable amplifier and
phase shifter
3 dB in phase
power combiner
Q-VVA
3 dB 90o RF
power divider OUT
RF I-VVA
IN
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.
IEEE Trans Microw Theory Tech 34(12):1327–1332
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
10. Yamauchi K, Mori K, Nakayama M, Mitsui Y, Takagi T (1997) A microwave miniaturized
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
526 16 Microwave Linearizers
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 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
( ) ( )
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
2π
1
an vo (θ ) cos nθ dθ (17.2a)
π
0
2π
1
bn vo (θ ) sin nθ dθ (17.2b)
π
0
530 17 Microwave Frequency Multipliers
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
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
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
532 17 Microwave Frequency Multipliers
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
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
534 17 Microwave Frequency Multipliers
I
current waveform
V
0
voltage waveform
Li1 Lo1
Ci Co
Li2 Lo2
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
17.3 Diode Multiplier 535
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.
Capacitance
Diode voltage
0
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
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
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
538 17 Microwave Frequency Multipliers
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-
17.3 Diode Multiplier 539
Input
voltage
Time
Input
Current
Time
Output
voltage
Time
(a) (b)
Amplitude Amplitude
15th
Frequency Frequency
Fig. 17.14 Typical spectrum at the output of a impulse generator and b BPF
(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
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)
542 17 Microwave Frequency Multipliers
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
The bias point of the transistor can be derived from (17.8a). For FET multiplier,
gate bias voltage can be derived from:
IQ
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
17.4 Transistor Multiplier 543
Ion an2 + bn2 (17.11)
⎣ 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⎢
−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
544 17 Microwave Frequency Multipliers
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⎢
−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,
17.4 Transistor Multiplier 545
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)
546 17 Microwave Frequency Multipliers
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;
17.4 Transistor Multiplier 547
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.
548 17 Microwave Frequency Multipliers
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
VG VDD
RG RD
Output
matching
Input network
matching
network
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
id1
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.
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
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].
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
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
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
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,
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-
558 18 Microwave Frequency Mixers
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.
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
18.2 Working Principle of Frequency Mixers 559
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
560 18 Microwave Frequency Mixers
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)
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
562 18 Microwave Frequency Mixers
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.
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.
18.2 Working Principle of Frequency Mixers 563
Image noise
Signal noise
Output
Signal
Input
Signal Ideal
noise-less
mixer
Image reject
filter
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
564 18 Microwave Frequency Mixers
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
Therefore, ratio of the output signal power to the output noise power is
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
18.2 Working Principle of Frequency Mixers 565
Therefore, ratio of the output signal power to the output noise power is
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.
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
566 18 Microwave Frequency Mixers
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
RF IF
input output
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
568 18 Microwave Frequency Mixers
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
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.
570 18 Microwave Frequency 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.
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
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
572 18 Microwave Frequency Mixers
Fig. 18.14 Schematic circuit of a single-balanced diode mixer using 180° hybrid
AL AR
vd 2 − √ cos ωL t + √ cos ωR t (18.28b)
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
LO IN RF IN
D4 D1
IF
Balun1 Balun2
D3 D2
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.
mixer-1
RF
LPF LSB
IN
o o
3-dB 90 3-dB 90
Hybrid LO IN Hybrid
LPF USB
mixer-2
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
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
576 18 Microwave Frequency Mixers
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
Filter &
matching Filter &
matching
Filter &
matching
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)
578 18 Microwave Frequency Mixers
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
The mixer is lossless. Thus, signal power at the output of the mixer is the same
and can be written as
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
Therefore, ratio of the output signal power to the output noise power is
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
580 18 Microwave Frequency Mixers
NFIM 3 dB 1.995
GIN 30 dB 1000
GIM 32 dB 1585.893
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
19.1 Introduction
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.
Suppose
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)
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
19.2 Mobile Communication Systems 587
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
588 19 Microwave Communication Systems
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.
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
590 19 Microwave Communication Systems
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.
592 19 Microwave 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
594 19 Microwave Communication Systems
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
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)
PFD × GR × λ2R
> kTR B (19.4)
4π
or
4π TR
PFD > ×k ×B×
λR
2 GR
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
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
596 19 Microwave Communication Systems
GR
> −116−
TR dB/K
> −115 dBW/m2
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
19.3 Satellite Communication Systems 597
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
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:
Calculate insertion loss and return losses when such two networks connected in
cascade.
Solution
Insertion loss (IL) of the network is
[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
600 19 Microwave Communication Systems
2
Cascaded S21
Ac + Bc Yo + Cc Zo + Dc
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◦
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:
Calculate insertion loss and return losses when such two networks connected in
cascade.
Solution
Insertion loss (IL) of the network is
[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
0.911 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◦
2
Cascaded S21
Ac + Bc Yo + Cc Zo + Dc
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◦
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:
and
Calculate insertion loss and return losses when such two networks connected in
cascade.
Solution
Insertion loss (IL) of the network P is
In similar way, ABCD parameters in matrix form of the 2-port Q network are
19.3 Satellite Communication Systems 605
[S] parameters of the cascaded 2-port network can be derived using (7.125)
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◦
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◦
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.
606 19 Microwave Communication Systems
19.4 Receiver
pHEMT
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
608 19 Microwave Communication Systems
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
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
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
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.
(a)
(b)
Fig. 19.16 a Block diagram of a multiplier-based local oscillator. b Simple block diagram of a
PLL-based local oscillator
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.
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.
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
614 19 Microwave Communication Systems
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.
19.5 Satellite Transmitter 615
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.
616 19 Microwave Communication Systems
Fig. 19.22 I-O characteristic of CAMP in ALC mode operation for different attenuation in DAT1
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
19.5 Satellite Transmitter 617
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.
618 19 Microwave Communication Systems
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
19.5 Satellite Transmitter 619
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
620 19 Microwave Communication Systems
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
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.
622 19 Microwave Communication Systems
Fig. 19.30 Typical I–O, gain and phase characteristic of TWTA and Linearized TWTA
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
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.
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.
624 19 Microwave Communication Systems
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) (b)
Fig. 19.34 4 × 4 Butler matrix using 3-dB 90° hybrid couplers a with crossover, b planner structure
without crossover
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
626 19 Microwave Communication Systems
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
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
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
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.
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
628 19 Microwave Communication Systems
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
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
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
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)
[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
[IM3RXTX ]dB 10 log 10−(10/10) + 10−(10/10)
630 19 Microwave Communication Systems
10 log(0.1 + 0.1)
−6.99 dB
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)
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
632 19 Microwave Communication Systems
Fig. 19.39 Power levels under two-tone excitation of equal input power levels of Example 19.8
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
19.8 Multiport Amplifier (MPA) 633
A 8.94 V, B 4.48 V
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
634 19 Microwave Communication Systems
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
19.8 Multiport Amplifier (MPA) 635
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)
A 9.535 V, B 3.015 V
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
636 19 Microwave Communication Systems
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 Multiport Amplifier (MPA) 637
Fig. 19.42 Power levels under two-tone excitation of unequal input levels by 10 dB of Example
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
(a) For equal power levels of both the carriers with PiTotal 0 dBW 1 W. From
(19.23),
A B 7.071 V
Fig. 19.44 Power levels under two-tone excitation of equal input levels of Example 19.9
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
A 9.535 V, B 3.015 V
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
19.8 Multiport Amplifier (MPA) 641
Fig. 19.45 Power levels under two-tone excitation of unequal input levels by 10 dB of Example
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
642 19 Microwave Communication Systems
Fig. 19.46 Gain of the carriers with unequal input levels by 10 dB of Example 19.9
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)
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
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
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
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
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
Similarly, at the port-2 (say, z 0), the signal voltage can be written as
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
646 19 Microwave Communication Systems
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
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. Yamauchi K, Mori K, Nakayama M, Mitsui Y, Takagi T (1997) A microwave miniaturized
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
(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
(A) 30 and 0
(B) 0 and 30
(C) 1 and 0
(D) 0 and 1
(A) 10
(B) 100
(C) 1000
(D) 10,000
20.1 Multiple Choice Questions 649
(A) 10
(B) 100
(C) 1000
(D) 10,000
(A) 100
(B) 10
(C) 1
(D) 0
(A) 10
(B) 1
(C) 100
(D) 0.1
(A) 10.7 mA
(B) 19.3 mA
(C) 20 mA
(D) 40 mA
650 20 Multiple Choice Questions with Answers
(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
(A) 4
(B) 12
(C) 3
(D) 16
√
(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.
652 20 Multiple Choice Questions with Answers
(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%
(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
(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) 2/3 0◦
(B) 1/2 0◦
(C) 2/3 180◦
(A) 100
(B) 200
(C) 50
(D) 25
(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
656 20 Multiple Choice Questions with Answers
(C) 1/5 0◦
(D) 1/2 0◦
(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?
(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
(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
(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
20.1 Multiple Choice Questions 659
(A) 100%
(B) 78.4%
(C) 50%
(D) <80%
(A) 50
(B) 25
660 20 Multiple Choice Questions with Answers
(C) 100
(D) None of the above
(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) 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
664 20 Multiple Choice Questions with Answers
(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
(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°
20.1 Multiple Choice Questions 665
(B) 1/3 0°
(C) 4/5 180°
(D) 1/2 180°
(A) 4/5 0°
(B) 1/2 0°
(C) 4/5 180°
(D) 2/3 0°
(A) 0 dB
(B) ∝ dB
(C) 10 dB
(D) 1 dB
666 20 Multiple Choice Questions with Answers
(72) Attenuation and return losses of the 2-port network in a 50 system are,
respectively,
(75) Attenuation and return losses of the 2-port network in a 50 system are,
respectively,
(A) ∝ dB, 0 dB
(B) 9.03 dB, 9.03 dB
(C) 6.02 dB, 6.02 dB
(D) 0 dB, ∝ dB
(76) Attenuation and return losses of the 2-port network in a 50 system are,
respectively,
(A) ∝ dB, 0 dB
(B) 9.03 dB, 9.03 dB
(C) 6.02 dB, 6.02 dB
(D) 0 dB, ∝ dB
(A) 1.17 V
668 20 Multiple Choice Questions with Answers
(B) 5.0 V
(C) 2.24 V
(D) 0.506 V
(84) Practical limitations of the output power of a power amplifier come from
(85) Two-tone IM3 of an amplifier with PoIP3 100 mW at its 1 mW output power
is about
(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)
(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
670 20 Multiple Choice Questions with Answers
(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)
(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
(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
20.1 Multiple Choice Questions 671
(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
672 20 Multiple Choice Questions with Answers
(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
(D) All of the above
(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
(D) All of the above
(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
(D) All of the above
(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)
20.1 Multiple Choice Questions 673
(A) 1.2
(B) 2:1
(C) 4:1
(D) 1:4
(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
(C) Zero
(D) Infinity
(112) Waveguide acts like a
(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?
20.1 Multiple Choice Questions 675
(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
(A) 0
(B) 1
(C) −1
(D) Infinity
(A) +1
(B) 0
(C) −1
(D) Infinity
(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
(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
(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
(B) −4 dB/K
(C) 26 dB/K
(D) −26 dB/K
(A) 1/4
(B) 1/2
(C) 1/3
(D) 3
(continued)
S. No. Answer Explanation
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
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)
680 20 Multiple Choice Questions with Answers
(continued)
S. No. Answer Explanation
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%
(continued)
S. No. Answer Explanation
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
47 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
48 A For the complex conjugate match
∗ 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)
682 20 Multiple Choice Questions with Answers
(continued)
S. No. Answer Explanation
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◦
(continued)
S. No. Answer Explanation
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)
684 20 Multiple Choice Questions with Answers
(continued)
S. No. Answer Explanation
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
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)
20.2 Answers of MCQs with Explanations 685
(continued)
S. No. Answer Explanation
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)
686 20 Multiple Choice Questions with Answers
(continued)
S. No. Answer Explanation
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
(continued)
S. No. Answer Explanation
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)
688 20 Multiple Choice Questions with Answers
(continued)
S. No. Answer Explanation
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
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