High Frequency Design From May 2011 High Frequency Electronics

Copyright © 2011 Summit Technical Media, LLC

CLASS-F AMPLIFIERS

Load Network Design

Technique for Class F and
Inverse Class F PAs
By Andrei Grebennikov
Bell Labs, Alcatel-Lucent

H
ighly efficient resistance), Vcc is the supply voltage, and I0 is
This article explains the operation of the the dc current component.
design procedure for power amplifier
implementing a practical can generally be obtained Idealized Waveforms with Sinewave Drive
output network for Class-F by applying bi-harmonic In an ideal Class B operation mode with
and inverse Class-F or polyharmonic modes, output parallel resonant circuit tuned to the
power amplifiers when an additional sin- fundamental frequency, the maximum theo-
gle-resonant or multi-res- retical anode (collector or drain) efficiency
onant circuit tuned to the odd or even har- achieves only 78.5% characterized by the sinu-
monics of the fundamental frequency is added soidal anode voltage waveform and half-sinu-
into the load network. An infinite number of soidal current waveform. To increase efficien-
odd-harmonic resonators results in an ideal- cy up to maximum theoretical value of 100%,
ized Class F mode with a square voltage wave- additional resonators must be included into
form and a half-sinusoidal current waveform the load network, each of which is tuned to the
at the device output terminal, while an infi- corresponding harmonic component. In the
nite number of even-harmonic resonators case of an infinite number of harmonic res-
results in an idealized inverse Class F mode onators and an active device considered an
with a half-sinusoidal voltage waveform and a ideal lossless switch, this can result in an out-
square current waveform at the device output put half-sinusoidal current waveform and a
terminal [1, 2]. In conventional Class F or square voltage waveform (Class F mode), and
inverse Class F power amplifiers analyzed in an output half-sinusoidal voltage waveform
frequency domain, the fundamental and har- and a square current waveform (inverse Class
monic load-network impedances are optimized F, or Class F–1 mode). In this case, a Fourier
by short-circuit termination and open-circuit series for the corresponding output current i(t)
peaking to control the voltage and current and voltage v(t) in normalized forms can be
waveforms at the device output to obtain max- written as
imum efficiency. They are equal respectively to
i ( ωt ) v ( ωt )
8 Vcc π2 Vcc = =
Znet (ω0 ) = R = or (1) I0 V0 Class F −1
π 2 I0 8 I0 Class F
(4)
π N
cos nωt
Znet (2nω0 ) = 0 (2)
1 −
2
sinωt − 2 ∑
n = 2, 4 ,6,... n
2
− 1
(or ∞ for even harmonics)

Znet[(2n + 1)ω0 ] = ∞ (3) v ( ωt ) i ( ωt )

= =
(or 0 for odd harmonics) V0 I0
Class F Class F −1
(5)
4 4 N
sin nωt
where R is the fundamental-frequency load- 1 +
π
sinωt +
π
∑
n = 3,5,7,... n
network impedance (or optimum load-line

58 High Frequency Electronics

 High Frequency Design
CLASS-F AMPLIFIERS

Figure 1 · Circuit schematic of Class F GaN HEMT power amplifier with

shunt quarterwave line.

where I0 and V0 are the dc current terwave line based on a 28-V 10-W
and voltage components, respectively. Cree GaN HEMT power transistor Figure 2 · Simulated waveforms,
These idealized voltage and current CGH40010. In this case, there is no power gain and efficiency of Class
waveforms can easily be obtained by need to use an RF choke and an addi- F GaN HEMT power amplifier.
using a shunt quarterwave transmis- tional series blocking capacitor. Since
sion line with a series fundamentally- the transmission line presents the
tuned LC resonant circuit in a Class high impedance conditions to all odd tance Cin = 5 pF (measured at the
F, or a series quarterwave transmis- harmonics and zero impedance to all gate bias voltage of 8 V) at the fun-
sion line loaded by the series funda- even harmonics at its input, all odd damental frequency that resulted in
mentally tuned LC resonant circuit in harmonics provide a proper contribu- a small-signal S11 better than 25 dB.
an inverse Class F [2]. tion to the square voltage waveform The series 50 Ω resistor connected to
Ideally, control of an infinite num- assuming high reactance of the series the device gate is also included to
ber of the harmonic components resonant circuit at all harmonics. provide unconditional operation sta-
maintaining a square voltage wave- To better illustrate the drain volt- bility. The simulated drain voltage
form and a half-sinusoidal current age and current waveforms with min- (close to square blue curve) and cur-
waveform at the drain can be provid- imum effect of the device parasitic rent (close to half-sinusoidal red
ed by using a shunt quarterwave output parameters, a sufficiently low curve) waveforms are shown in Fig.
transmission line located between operating frequency of 100 MHz is 2(a) where small waveform ripples
the dc power supply and device out- chosen. In this case, a simple lossy (minimized by load-network parame-
put terminal, and a series-tuned res- RL input shunt network is used to ter optimization) can be explained
onant circuit [2, 3]. Figure 1 shows match the device input impedance to due to some effect of the device out-
the circuit schematic of a Class F a 50 Ω source and to compensate for put drain-source capacitance Cout =
power amplifier with a shunt quar- the device input gate-source capaci- 1.32 pF and package parasitics at
high-order harmonic components.
The parameters of the load-network
were slightly optimized for As a
result, a maximum drain efficiency of
84.7% with a power gain of 20 dB and
an output power of 40 dBm at a sup-
ply voltage of 28 V were obtained
with a sine-wave driving signal, as
shown in Fig. 2(b). In this case, to bet-
ter approximate the switched-mode
operation of a 10-W GaN HEMT
device, it is necessary to slightly
Figure 3 · Circuit schematic of inverse Class F GaN HEMT power amplifier increase the input signal amplitude
with series quarterwave line. to operate in a saturated mode char-

60 High Frequency Electronics

 High Frequency Design
CLASS-F AMPLIFIERS

acterized by more than 3-dB gain ation mode can be represented by CGH40010. In this case, the series-
compression point. Due to the diode- using a series quarterwave transmis- tuned output circuit presents to the
based nonlinearity of the device input sion line loaded by the series reso- transmission line a load resistance at
circuit, the amplitude harmonic ratio nant circuit tuned to the fundamen- the frequency of operation. For even
between the fundamental-frequency, tal frequency [2, 4]. Figure 3 shows harmonics, the open circuit on the
second, and third harmonic compo- the circuit schematic of an inverse load side of the transmission line is
nents at the gate terminal is Class F power amplifier with a series repeated, thus producing an open cir-
8.2:1.3:1. quarterwave line based on a 28-V 10- cuit at the drain. However, the quar-
An idealized inverse Class F oper- W Cree GaN HEMT power transistor terwave transmission line converts
the open circuit at the load to a short
circuit at the drain for odd harmonics
with resistive load at the fundamen-
tal frequency.
Similarly to the Class F GaN
HEMT power amplifier, a simple
lossy RL input shunt network is also
used to match the device input
impedance to a 50 Ω source and to
compensate for the device input gate-
source capacitance Cin = 5 pF at the
fundamental frequency, providing a
small-signal S11 better than 20 dB at
an operating frequency of 200 MHz.
The series 55 Ω resistor is necessary
to provide unconditional operation
stability. The simulated drain voltage
(close to half-sinusoidal blue curve)
and current (close to square red
curve) waveforms are shown in Fig.
4(a) where small deviations from the
ideal waveforms (with optimized
load-network parameters) can be

Figure 4 · Simulated waveforms,

gain and efficiency of inverse Class
F GaN HEMT power amplifier.
explained due to effect of the device sponding harmonic component [8, 9]. capacitor. The series transmission-
output drain-source capacitance Cout Figure 5(a) shows the idealized line section and open-circuit stub,
and package parasitics In this case, a basic configuration of the transmis- both having an electrical length of 30º
max-imum drain efficiency of 84.8% sion-line Class F load network where at the fundamental frequency, pro-
with a power gain of 12.3 dB and an a shunt quarter-wavelength trans- vide an open-circuit mode at the third
output power of 40.3 dBm at a supply mission line providing a short-circuit harmonic because an open-circuit
voltage of 24 V were obtained with a termination for even voltage harmon- stub has a quarter wavelength at the
sine-wave driving signal in a deep ic at the device output is connected to third harmonic to realize a short-cir-
saturation condition, as shown in Fig. the dc power supply with bypass cuit condition at the right-hand side
4(b).

Basic Practical Load Network

Configurations
In practice, it is usually enough to
apply the second- or third-harmonic
peaking to achieve maximum drain
efficiency of more than 80% in Class F
or inverse Class F mode that can
result in a sufficiently simple circuit
schematic convenient for practical
realization [5, 6]. For example, by pro-
viding a proper high-impedance sec-
ond-harmonic peaking and load
matching at the fundamental, the col-
lector efficiency of 83% was achieved
for a 435-MHz microstrip power
amplifier [7]. Generally, the optimum
impedance matching at the funda-
mental frequency with the second-
and third-harmonic tuning can be
provided by using the series trans-
mission-line sections and shunt open-
circuit stubs to obtain open-circuit
peaking and short-circuit termination
seen by the device output at corre-

Figure 5 · Idealized transmission-

line Class F load networks.
 High Frequency Design
CLASS-F AMPLIFIERS

of the series transmission line having a quarter wave-

length at the third harmonic as well. Such a Class F load
network implemented using coplanar lines into a 24-GHz
MMIC GaAs pHEMT power amplifier had contributed to
a drain efficiency of 59% at an output power of 20 dBm
[10].
The load-network impedance Znet seen by the device
output at the fundamental frequency can be written as

RL ( Z2 − Z1 tan 2 θ ) + jZ1 Z2 tan θ

Znet = Z1 (6)
Z1 Z2 + j ( Z1 + Z2 ) RL tan θ

where θ = θ1 = θ2 = 30º, Z1 and θ1 are the characteristic

impedance and electrical length of the series transmis-
sion line, and Z2 and θ2 are the characteristic impedance
and electrical length of the open-circuit stub. Hence, the
impedance matching with the load at the fundamental
can be provided by proper choice of the characteristic
impedances Z1 and Z2.
Separating Eq. (6) into real and imaginary parts and
taking into account that ReZnet = R and ImZnet = 0, the
system of two equations with two unknown parameters is Figure 6 · Idealized transmission-line inverse Class F
obtained by load network.

Z12 Z22 − RL2 ( Z1 + Z2 ) ( Z2 − Z1 tan 2 θ ) = 0 (7)

vides zero impedance at the second harmonic and infinite
( Z1 + Z2 ) RL2 R tan 2 θ − Z12 Z22 ⎡⎣ RL (1 + tan 2 θ ) − R ⎤⎦ = 0
2
(8) impedance at the third harmonic components [11]. The
short-circuit impedances at right-hand side of the series
which enables the characteristic impedances Z1 and Z2 to quarter-wavelength transmission line is provided by a
be properly calculated. This system of two equations can short-circuited shunt quarter-wavelength transmission
be explicitly solved as a function of the parameter r = line at the second harmonic and an open-circuit λ/12 stub
RL/R resulting in at the third harmonic. In this case, to provide an
impedance matching at the fundamental frequency, it is
Z1 4r − 3 necessary to use a separate matching circuit since the
= (9)
RL r series quarter-wavelength transmission line with charac-
teristic impedance Z1 can match only real part of the
Z1 ⎛ r −1⎞ equivalent fundamental-frequency device output
=3 ⎜ ⎟ (10)
Z2 ⎝ r ⎠ impedance equal to Class F optimum load resistance R.
However, there is no additional parameter to compensate
Consequently, for the specified value of the parameter for its imaginary part.
r with the required Class F optimum fundamental-fre- Figure 6(a) shows the idealized transmission-line
quency load resistance R and standard load resistance inverse Class F load-network schematic where a shunt
RL = 50 Ω, the characteristic impedance Z1 is calculated λ/6 transmission line grounded through the bypass capac-
from Eq. (9) and then the characteristic impedance Z1 is itor provides low impedance for the third harmonic and
calculated from Eq. (10). For example, if the required certain impedances for the fundamental and second har-
Class F optimum load resistance R is equal to 12.5 Ω monics at the de-vice drain terminal [2, 12]. The second-
resulting in r = 4, the characteristic impedance of the harmonic peaking is achieved by using an open-circuit λ/8
series transmission line Z1 is equal to 45 Ω and the char- stub which creates a low impedance at the right-hand
acteristic impedance of the open-circuit stub Z1 is equal to side of the series λ/12 transmission line at the second har-
20 Ω. monic. As a result, this series λ/12 transmission line with
Figure 5(b) shows the simplified Class F load network electrical length of 60º at the second harmonic (combined
with the second- and third-harmonic tuning using a in parallel with the shunt λ/6 transmission line with elec-
series quarter-wavelength transmission line which pro- trical length of 120º at the second harmonic) represents a

64 High Frequency Electronics

 High Frequency Design
CLASS-F AMPLIFIERS

second-harmonic tank.
It is also useful and very practical for inverse Class F
power amplifiers to consider an alternative transmission-
line inverse Class F load network with a series λ/8 trans-
mission line as a first element, as shown in Fig. 6(b) [13].
In this case, the series λ/8 transmission line short-circuit-
ed at its right-hand side by the shunt quarter-wavelength
transmission line at the second harmonic provides an
open circuit at the second harmonic seen by the device
output, while the combined (λ/8 + λ/24 = λ/6) series
transmission line short-circuited at its right-hand side by
the open-circuit λ/12 stub provides a short circuit at the Figure 7 · Idealized transmission-line inverse Class F
third harmonic seen by the device output. load network.
The load-network impedance Znet at the fundamental
can be written similarly to Eq. (6) as
vide a short circuit at the second harmonic and open cir-
RL ( Z2 − Z1 tan 30D tan 60D ) + jZ1 Z2 tan 60D cuit at the third harmonic in a Class F mode, then it is
Znet = Z1 (11)
Z1 Z2 + j ( Z1 tan 30D + Z2 tan 60D ) RL necessary to use a series λ/8 line to provide an open cir-
cuit at the second harmonic and a series λ/6 line to pro-
where Z1 is the characteristic impedance of the combined vide a short circuit at the third harmonic in an inverse
series transmission line and Z2 is the characteristic Class F mode. To combine effects of these two uniform
impedance of an open-circuit λ/12 stub. Separating Eq. transmission lines, the non-uniform stepped-impedance
(11) into real and imaginary parts, the system of two transmission line can be used. It is known that the band
equations with two unknown parameters is obtained by properties of a resonant circuit containing a short-circuit-
ed transmission line can be improved if, instead of a uni-
( Z1 + 3Z2 ) RL2 R − 3Z12 Z22 ( 4 RL − 3R ) = 0
2
(12) form transmission line, a non-uniform transmission line
is used, the spectrum of natural frequencies of which is
much non-equidistant [14, 15]. Among the resonant sys-
3 Z12 Z22 − RL2 ( Z2 − Z1 ) ( Z1 + 3Z2 ) = 0 (13) tems with non-uniform transmission lines, the resonant
systems with multi-section line having alternating high
which allows direct calculation of the characteristic and low characteristic impedances of its uniform sections
impedances Z1 and Z2. This system of two equations can are ideal from the standpoint of maximum frequency tun-
be explicitly solved as a function of the parameter r = ing bandwidth. A sensitivity variation of such a multi-sec-
RL/R resulting in tion line is determined by a spectrum rarefaction near
resonant frequency for a tuning band widening and spec-
Z1 4r − 1 trum narrowing in the case of frequency stabilization. In
= (14)
RL 3r this case, a simple combination of two loaded transmis-
sion lines with high and low characteristic impedances
can provide the fundamental impedance matching and
Z1 r − 1
= (15) second-harmonic tuning to a high impedance nearby open
Z2 r condition simultaneously [16].
Figure 7 shows the simplified inverse Class F load
Consequently, if the required optimum load resistance network with the second and third harmonic tuning using
and standard load resistance are equal to R = 20 Ω and a series stepped-impedance transmission line which can
RL = 50 Ω, respectively, resulting in r = 2.5, then the char- provide infinite impedance at the second harmonic and
acteristic impedance of the series transmission line cal- zero impedance at the third harmonic components, where
culated from Eq. (12) is equal to Z1 = 35 Ω and the char- θ1 and θ2 are the electrical lengths of the stepped-
acteristic im-pedance of the open circuit stub calculated impedance transmission-line sections and M is the char-
from Eq. (13) is equal to Z2 = 58 Ω. acteristic impedance ratio. Using the shunt quarter-wave-
length transmission line is convenient to directly supply
Load Network with Stepped-Impedance dc current to the active device from the power supply
Transmission Line since it is RF grounded at its end. For the simplicity of
As it follows from Fig. 5(b), if it is enough to use a sin- analytical calculations, the transmission-line lengths can
gle series quarter-wavelength transmission line to pro- be set equal as θ = θ1 = θ2.

66 High Frequency Electronics

 High Frequency Design
CLASS-F AMPLIFIERS

The input network impedance Znet as a function of

electrical length θ of equal sections of the stepped-
impedance transmission line is written as
⎛ 1 ⎞
ZL (1 − M tan 2 θ ) + jZ0 ⎜ + 1 ⎟ tan θ
Znet = Z0 ⎝M ⎠ (16)
⎛ tan 2 θ ⎞
Z0 ⎜ 1 − ⎟ + jZL ( M + 1) tan θ
⎝ M ⎠

where ZL is the load impedance including the short-cir-

cuit and open-circuit stubs.
For ZL = 0,

1
+1
Znet = jZ0 M tan θ (17)
tan 2 θ
1−
M

whose zeroes correspond to θ0k = (π/2)k and poles corre-

spond to

θpk = kπ ± tan −1 M (18) Figure 8 · Frequency response of load network with

second- and third-harmonic tuning.
where k = 0, 1, 2, … . Thus, by choosing the proper char-
acteristic impedance ratio M, the first pole θp1 can be
moved from initial value of 45 Ω for uniform line towards
the second zero at θ02 = 90º and set at θp1 = 60º for the
stepped-impedance transmission line with M = 3. As a
result, by using such a stepped-impedance transmission
line, an open circuit at the second harmonic and short cir-
cuit at the third harmonic of the fundamental frequency
can be provided if the fundamental frequency corres-
ponds to θ = 30º.
Figure 8(a) shows the design example of a stepped-
impedance transmission-line inverse Class F load net-
work with harmonic tuning which provides an impedance
matching at the fundamental frequency between the
input impedance of 20 Ω and load impedance of 50 Ω at
an operating frequency f0 = 2 GHz. In this case, the
stepped-impedance transmission line with equal section
lengths of 30º loaded by the short-circuited quarter-wave-
length line and open-circuit λ/12 stub creates an open cir-
cuit at the second harmonic and short circuit at the third
harmonic seen by the 20-Ω source at its input. Using of a
subsequent short-length transmission line as an element
of the output funda-mental-frequency matching circuit is
required to compensate for the reactive part of the modi-
fied source impedance (any source impedance can be
transformed to a real load impedance using a λ/8 trans-
former whose characteristic impedance is equal to the
magnitude of the source impedance [17]) followed by the
quarter-wavelength transmission line which provides the
transformation of the real part of the modified source Figure 9 · Transmission-line load networks with shunt
impedance to the standard load of 50 Ω. The frequency capacitance.

68 High Frequency Electronics

 High Frequency Design
CLASS-F AMPLIFIERS

response of such a stepped-impedance transmission-line

inverse Class F load network with harmonic tuning is
shown in Fig. 8(b), which demonstrates the fundamental-
frequency impedance matching at 2 GHz, second-har-
monic open circuit at 4 GHz, and third-harmonic short
circuit at 6 GHz.

Effect of Shunt Capacitance

However, the efficiency of the power amplifier can be
limited by the transistor output capacit-ance Cout (mostly
represented by drain-source capacitance Cds or collector
capacitance Cc) if this capacitance is not absorbed into
multiharmonic load network without compromising the
ability to properly terminate the second and third har-
monic components. Figure 9(a) shows the simplified cir-
cuit schematic of the transmission-line transistor power
amplifier, whereas the equivalent load network corre-
sponding to Class F mode with a shunt short-circuited
quarter-wavelength transmission line which provides a
short circuit at even harmonics is shown in Fig. 9(b) and
equivalent load net-work corresponding to inverse Class Figure 10 · Transmission-line inverse Class F power
F mode with a shunt short-circuited λ/3 transmission line amplifier and its equivalent circuit.
which provides a short circuit at the third harmonic at
the device output is shown in Fig. 9(c).
In Class F mode with a load network shown in Fig. mission line TL2 and ω0 is the fundamental frequency.
9(b), the electrical length of an open-circuit stub TL3 is However, when effect of the device output parasitic
chosen to have a quarter wavelength at the third har- series inductance becomes significant, it leads to an addi-
monic to realize short-circuit condition at the right-hand tional circuit complexity and some problems with tuning
side of the series transmission line TL2, whose electrical accuracy. In this case, it is possible to use either an addi-
length θ should provide an inductive reactance to res- tional series compensation line [19] or an additional open-
onate with the device output capacitance Cout at the third circuit transmission-line stub [20] that generally results
har-monic. Such a load network is very practical when the in excessive losses at harmonic frequencies and lower effi-
dc drain voltage is very high to increase the device output ciency.
impedance, or operating frequency is sufficiently low, and
bare die is used instead of packaged transistor to easily Effect of Series Inductance
absorb bondwire inductances by the load network [18]. In In a power amplifier using the packaged device at
inverse Class F mode with a load network shown in Fig. high frequencies or with high output power, the presence
9(c), both transmission lines TL1 and TL2 having overall of a transistor output series bondwire and lead induc-
inductive reactance in a parallel connection are tuned to tance Lout creates some problems in providing an accept-
the parallel resonance condition with the device output able second- or third-harmonic open- or short-circuit ter-
capacitance Cout to form a parallel resonant circuit to mination. Even if Lout can be placed between the device
realize an open-circuit condition for the second harmonic output capacitance Cout and shunt short-circuited trans-
at the device output [2]. mission line TL1 shown in Fig. 9(a) and a short-circuit ter-
As a result, the electrical lengths of the series trans- mination is provided at its right-hand side, the open-cir-
mission line TL2 at the fundamental frequency can be cuit condition will fully depend on a physical length of
obtained by this inductance which is difficult to set very accurately in
practical hybrid or monolithic integrated circuits.
1
θ= tan −1 ( 2 Z0ω0Cout ) for Class F (19) In this case, it is convenient to use a series transmis-
3 sion line as a first element of the load network connected
to the device output, as shown in Fig. 10(a), where the
−1
1 ⎡⎛ 1 ⎞ ⎤ transmission line TL1 is placed between the device drain
θ= tan −1 ⎢⎜ 2 Z0ω0Cout + ⎟ ⎥ for Class F
−1
(20)
2 ⎢⎣ ⎝ 3 ⎠ ⎥⎦ terminal and shunt short-circuited quarter-wavelength
transmission line TL3. However, if in a Class F mode with
where Z0 is the characteristic impedance of the series trans- short circuit at the second harmonic and open circuit at

70 High Frequency Electronics

 High Frequency Design
CLASS-F AMPLIFIERS

current source at the second harmonic is shown in Fig.

11(a), with shorting effect of the quarterwave short-cir-
cuited stub TL3. Here, the series transmission line TL1
provides an open-circuit condition for the second harmon-
ic at the device output by forming a second-harmonic tank
together with Cout and Lout. Similar load network at the
third harmonic is shown in Fig. 11(b), due to the open-cir-
cuit effect of the short-circuited quarter-wave line TL3and
shorting effect of the open-circuit stub TL4 at the third
harmonic. In this case, the combined transmission line
TL1 + TL2 (together with the series inductance Lout) pro-
vides a short-circuit condition for the third harmonic at
the device output being shorted at its right-hand side. The
electrical lengths of the transmission lines TL1 and TL2,
assuming the same characteristic impedance Z1 for both
series transmission-line sections, can be defined from
Figure 11 · Load networks seen by the device output
at second and third harmonics. 1
2ω0Cout − =0 (21)
2ω0 Lout + Z1 tan 2θ1

3ω0 Lout + Z1 tan 3 ( θ1 + θ2 ) = 0 (22)

with the maximum total electrical length θ1 + θ2 = π/3 or

60º at the fundamental frequency or 180º at the third har-
monic component when Lout = 0.
Figure 12 shows the equivalent representation of an
inverse Class F load network (including the device output
parameters Lout and Cout) by a lumped low-pass π-type
matching circuit where

C = tan30º/(ω0Z2)

L~
= (Z1/ω0)sin (θ1 + θ2) + Lout

Figure 12 · Equivalent representations of load network due to the sufficiently short length of the combined trans-
at fundamental frequency. mission line TL1 + TL2, typically much less than 60º at
the fundamental frequency depending on the device out-
put parameters. As a result,
the third harmonic, the length of combined series trans-
mission line TL1 + TL2 becomes very long and additional ω0 ( L − Lout )
Z1 ≅ (23)
fundamental-frequency matching circuit is required, then sin ( θ1 + θ2 )
such a load network in an inverse Class F mode is com-
pact, convenient for harmonic tuning, and very practical.
1
Figure 10(b) shows the transmission-line inverse Class-F Z2 = (24)
load network, where the combined series transmission ω0C 3
line TL1 + TL2 (together with an open-circuit capacitive
stub TL4 with electrical length of 30º) provides an For a lumped low-pass π-type matching circuit with RL
impedance matching between optimum fundamental-fre- > R and Q = ω0CoutR,
quency load resistance R and standard load resistance RL
by proper choice of the transmission-line characteristic
RL
impedances Z1 and Z2, where Cout and Lout are the ele- QL =
R
(1 + Q 2 ) − 1 (25)
ments of the matching circuit [21].
The load network seen by the device multiharmonic

72 High Frequency Electronics

 High Frequency Design
CLASS-F AMPLIFIERS

Figure 14 · Simulated results of

transmission-line inverse Class F
GaN HEMT power amplifier.

Figure 13 · Circuit schematic of transmission-line inverse Class F GaN

HEMT power amplifier.

mission line for implementation con-

QL
C= (26) venience and because of the effect of
ω0 RL
the series inductance Lout formed by
drain bonding wires and package
Q + QL
L= RL (27) lead. In this case, the device input
1 + QL2 and output package leads as exter-
nal elements were properly modeled
where the parameters Lout, Cout, R, to take into account the effect of
and RL are fixed [22]. By subsequent their inductances, and their models
calculation of the load quality factor were then added to the simulation
QL from Eq. (25) and then the capac- setup.
itance C from Eq. (26), the character- Figure 14 shows the simulated
istic impedance Z2 can be directly results of a transmission-line inverse
obtained from Eq. (24). After calcu- Class F GaN HEMT power amplifier
lating the inductance L from Eq. using a nonlinear device model sup-
(27), the remaining parameters plied by Cree and based on the tech-
including the electrical lengths θ1 nical parameters for a 30-mil Figure 15 · Measured results of
and θ2 and characteristic impedance RO4350 substrate. The maximum transmission-line inverse Class F
Z1 can be finally obtained from a sys- output power of 41.3 dBm, power power amplifier at 2.14 GHz.
tem of three equations given by Eqs. gain of 13.3 dB (linear gain of about
(21) to (23). 17 dB), drain efficiency of 80.3%, and
power-added efficiency (PAE) of ferent electrical lengths and charac-
Design Example 76.5% are achieved at an operating teristic impedances according to the
Figure 13 shows the simulated frequency of 2.14 GHz with a supply simulation setup shown in Figure 13.
circuit schematic which approxi- voltage of 28 V and a quiescent cur- Figure 15 shows the measured
mates a transmission-line inverse rent of 40 mA. results of a transmission-line inverse
Class F power amplifier based on a The test board of the transmis- Class F GaN HEMT power amplifier
28-V 10-W Cree GaN HEMT power sion-line inverse Class F GaN HEMT which demonstrate the maximum
transistor CGH40010F(P) and power amplifier using a 28-V 10-W output power of 41.0 dBm, drain effi-
transmission-line load network with Cree GaN HEMT power transistor ciency of 76.0%, and PAE of 72.2%
the second- and third-harmonic tun- CGH40010P in a metal-ceramic pill with a power gain of 13.0 dB at an
ing, as shown in Fig. 10. The input package was fabricated on a 30-mil operating frequency of 2.14 GHz
matching circuit provides a complex- RO4350 substrate. The input match- (gate bias voltage Vg = 2.8 V, quies-
conjugate matching with the stan- ing circuit, output load network, and cent current Iq = 50 mA, and drain
dard 50 Ω source. The load network gate and drain bias circuits (having supply voltage Vdd = 28 V), achieved
was slightly modified by optimizing bypass capacitors on their ends) are without any tuning of the input
the parameters of the series trans- fully based on microstrip lines of dif- matching circuit and load network.

74 High Frequency Electronics

 High Frequency Design
CLASS-F AMPLIFIERS

References istics of transistor voltage-controlled oscillators with dis-

1. F. H. Raab, “Class-E, Class-C, and Class-F power tributed parameters,” Int. J. Circuit Theory Appl., vol. 27,
amplifiers based upon a finite number of harmonics,” pp. 393-414, July 1999.
IEEE Trans. Microwave Theory Tech., vol. MTT-49, pp. 15. A. Grebennikov, RF and Microwave Transistor
1462-11468, Aug. 2001. Oscillator Design, John Wiley & Sons, 2007.
2. A. Grebennikov and N. O. Sokal, Switchmode RF 16. Y. Yamasaki, Y. Kittaka, H. Minamide, Y.
Power Amplifiers, Newnes, 2007. Yamauchi, S. Miwa, S. Goto, M. Nakayama, M. Kohno, and
3. V. A. Borisov and V. V. Voronovich, “Analysis of N. Yoshida, “A 68% efficiency, C-band 100W GaN power
Switched-Mode Transistor Amplifier with Parallel amplifier for space applications,” 2010 IEEE MTT-S Int.
Forming Transmission Line (in Russian),” Radiotekhnika Microwave Symp. Dig., pp. 1384-1387.
i Elektronika, vol. 31, pp. 1590-1597, Aug. 1986. 17. D. H. Steinbrecher, “An interesting impedance
4. M. K. Kazimierczuk, “A New Concept of Class F matching network,” IEEE Trans. Microwave Theory Tech.,
Tuned Power Amplifier,” Proc. 27th Midwest Circuits and vol. MTT-22, p. 382, June 1967.
Systems Symp., pp. 425-428, 1984. 18. D. Schmelzer and S. I. Long, “A GaN HEMT Class
5. F. Raab, “Class-F power amplifiers with maximally F amplifier at 2 GHz with >80% PAE,” IEEE J. Solid-
flat waveforms,” IEEE Trans. Microwave Theory Tech., State Circuits, vol. SC-42, pp. 2130-2136, Oct. 2007.
vol. MTT-45, pp. 2007-2012, Nov. 1997. 19. S. Dietsche, C. Duvanaud, G. Pataut, and J.
6. C. J. Wei, P. DiCarlo, Y.A. Tkachenko, R. McMorrow, Obregon, “Design of high power-added efficiency FET
and D. Bartle, “Analysis and experimental waveform amplifiers operating with very low drain bias voltages for
study on inverse Class-F mode of microwave power use in mobile telephones at 1.7 GHz,” Proc. 23rd Europ.
FETs,” 2000 IEEE MTT-S Int. Microwave Symp. Dig., vol. Microwave Conf., pp. 252-254, 1993.
2, pp. 525-528. 20. A. V. Grebennikov, “Circuit design technique for
7. W. McCalpin, “High efficiency power amplification high efficiency Class F amplifiers,” 2000 IEEE MTT-S Int.
with optimally loaded harmonic waveshaping,” Proc. RF Microwave Symp. Dig. vol. 2, pp. 771-774.
Tech. Expo'86 (Anaheim, CA), pp. 119-124, 1986. 21. A. Grebennikov, “High-efficiency transmission-line
8. F. Giannini and L. Scucchia, “A complete class of GaN HEMT inverse Class F power amplifier for active
harmonic matching networks: synthesis and application,” antenna arrays,” Proc. 2009 Asia-Pacific Microwave Conf.,
IEEE Trans. Microwave Theory Tech., vol. MTT-57, pp. pp. 317-320.
612-619, Mar. 2009. 22. A. Grebennikov, RF and Microwave Power
9. M. Helaoui and F. M. Ghannouchi, “Optimizing loss- Amplifier Design, New York: McGraw-Hill, 2004.
es in distributed multiharmonic matching networks
applied to the design of an RF GaN power amplifier with Author Information
higher than 80% power-added efficiency,” IEEE Trans. Andrei Grebennikov received his Dipl. Ing. degree in
Microwave Theory Tech., vol. MTT-57, pp. 314-322, Feb. radio electronics from Moscow Institute of Physics and
2009. Technology and Ph.D. degree in radio
10. R. Negra, F. M. Ghannouchi, and W. Baechtold, engineering from Moscow Technical
“Study and design optimization of multiharmonic trans- University of Communications and
mission-line load networks for Class-E and Class-F K- Informatics in 1980 and 1991, respec-
band MMIC power amplifiers,” IEEE Trans. Microwave tively. He has extensive experience
Theory Tech., vol. MTT-55, pp. 1390-1397, June 2007. working with Moscow Technical
11. K. Honjo, “A simple circuit synthesis method for University of Communications and
microwave Class-F ultra-high-efficiency amplifiers with Informatics (Russia), Institute of
reactance-compensation circuits,” Int. J. Solid-State Microelectronics (Singapore), M/A-COM (Ireland), Infineon
Electronics, vol. 44, pp. 1477-1482, Aug. 2000. Technologies (Germany/Austria), and Bell Laboratories
12. T. Heima, A. Inoue, A. Ohta, N. Tanino, and K. Sato, (Ireland) as an engineer, researcher, lecturer and educator.
“A new practical harmonics tune for high efficiency power He has lectured as Guest Professor at the University of
amplifier,” Proc. 29rd Europ. Microwave Conf., pp. 271- Linz (Austria) and presented short courses and tutorials as
274, 1999. an Invited Speaker at International Microwave
13. P. Aflaki, R. Negra, and F. M. Ghannouchi, “Design Symposium, European and Asia-Pacific Microwave
and implementation of an inverse Class-F power amplifi- Conferences, Institute of Microelectronics, Singapore, and
er with 79% efficiency by using a switch-based active Motorola Design Centre, Malaysia. He is an author and
device model,” Proc. 2008 IEEE Radio and Wireless coauthor of more than 80 papers, 15 European and US
Symp., pp. 423-426. patents and patent applications, and 5 books. Dr.
14. A. Grebennikov, “The frequency tuning character- Grebennikov can be reached at: grandrei@yahoo.com

76 High Frequency Electronics