electronics

Article
A Simplified Accuracy Enhancement to the Saleh
AM/AM Modeling and Linearization of Solid-State
RF Power Amplifiers
Haider Al-Kanan * and Fu Li *
Department of Electrical and Computer Engineering, Portland State University, Portland, OR 97207-0752, USA
* Correspondence: alkanan.haider@gmail.com (H.A.); lif@pdx.edu (F.L.)




Received: 8 October 2020; Accepted: 29 October 2020; Published: 31 October 2020


Abstract: The Saleh behavioral model exhibits high prediction accuracy for nonlinearity of
traveling-wave tube power amplifiers (TWT-PAs). However, the accuracy of the Saleh model
degrades when modeling solid-state power amplifiers (SSPAs) technology. In addition, the polynomial
expansion of the Saleh model consists of only odd-order terms as analyzed in this work. This paper
proposes a novel model accuracy enhancement for the Saleh amplitude-to-amplitude (AM/AM)
model when applied to radio frequency (RF) SSPAs. The proposed model enhancement accounts
for the second-order intermodulation distortion, which is an important nonlinearity challenge in
wideband wireless communications. The proposed static AM/AM model is a three-parameter rational
function, which exhibits low complexity compared to the state-of-the-art behavioral models. A
transpose architecture of finite-impulse digital filter is used to quantify the memory effect in SSPAs.
A least-squares method is used for extracting all the model parameters. A linearization technique
using a three-parameter digital predistortion model is also calculated to compensate for the AM/AM
nonlinear distortion in SSPAs. Finally, the identification and evaluation of the enhanced Saleh model
is presented based on measurements of RF SSPAs.

Keywords: solid-state power amplifiers; behavioral models; digital predistortion; nonlinear distortion;
memory estimation

1. Introduction
The rapid evolution of wireless communications requires high dynamic range RF power amplifiers
(PAs) to efficiently operate on high-amplitude fluctuations in modern digital modulation [1]. Solid-state
power amplifiers (SSPAs) are intensively deployed devices in the front ends of mobile systems [2–4].
Therefore, empirical models for SSPAs have been utilized for decades to study the distortion effects on
wireless communications systems.
Accurate and low-complexity behavioral models for SSPAs have become important techniques
in simulating and linearizing communication systems [5–7]. In particular, an accurate estimation
of the model smoothness in the amplitude-to-amplitude (AM/AM) transition from the linear to the
saturation regions has been a topic of interest in the state-of-the-art PA modeling [8–12]. This is because
operating SSPAs near the compression region is often required for achieving high power efficiency and
transmitting signals of high peak-to-average power ratio (PAPR) [13].
Rapp, Ghorbani, and Cann behavioral models are popular mathematical functions that match
well the AM/AM characteristics of SSPAs. However, these models often consist of complicated rational
functions; therefore, starting points and iterative estimation methods are typically required for such
model estimation. In addition, these models exhibit a lack of relevance in modeling digital predistortion
(DPD) for linearizing SSPAs.

Electronics 2020, 9, 1806; doi:10.3390/electronics9111806 www.mdpi.com/journal/electronics

Electronics 2020, 9, 1806 2 of 13

The polynomial-based models are commonly used for SSPAs, because polynomials provide
sufficient modeling flexibility in estimating the parameters and adjusting the model accuracy. However,
polynomial models are subject to estimation errors such as curve under-fitting and over-fitting model
prediction when truncated nonlinear coefficients are employed.
The AM/AM characteristics of traveling-wave tube power amplifiers (TWT-PAs) exhibit high
nonlinearity and roll-over effects near the saturation region. Thus, the accuracy of the Saleh model
significantly degrades when modeling the compression smoothness of the SSPAs. The proposed
modified Saleh model by O’Droma achieves substantial accuracy improvement for modeling the
AM/AM nonlinearity for laterally-diffused metal-oxide semiconductor (LDMOS) SSPAs [9]. However,
O’Droma proposed a six-parameter rational formula, which is computationally expensive. In addition,
the direct inversion of the O’Droma model is mathematically difficult for calculating the predistortion
model. Hence, this paper presents a simplified behavioral model with an adequate AM/AM accuracy
improvement for SSPA technologies.
The amplitude-to-phase distortion is often minor in SSPAs and is assumed insignificant in this
simplified modeling approach [10]. The presented modeling approach consists of three parameters,
which allows better nonlinearity modeling for a large signal amplitude beyond the 1-dB compression
point. In addition, the presented expression is convenient to characterize the even-order intermodulation
distortion (IMD) for wideband and multiband communications.
A simplified and low complexity DPD model is also calculated to compensate for the AM/AM
nonlinearity in SSPAs. This paper is organized as follows: Section 2 describes the proposed enhancement
approach to the Saleh AM/AM model, followed by an approach to characterize the dynamic memory
effect in Section 3. Section 4 presents a linearization model for SSPA using DPD. Section 5 includes the
experiment and modeling results for both the SSPA and DPD model. Finally, conclusions are presented
in Section 6.

2. Enhanced Approach for SSPAs

The Saleh AM/AM model is a two-parameter function. The first parameter α represents the
small signal gain, and the second parameter β is used to adjust the transition sharpness (i.e., gain
compression) from the linear region to saturation region. The AM/AM Saleh model F[u(t)] can be
expressed as [8]:
αu(t)
F[u(t)] = p 2 (1)
1+ βu(t)
where u(t) is the envelope of the PA input baseband signal. The Saleh model is an odd function {F[−u(t)]
= −F[u(t)]}, and the polynomial expansion of the rational function in Equation (1) is given by:
∞
X
F[u(t)] = (−1)k αβk u(2k+1) (t) (2)
k =0

where k is an integer number {k = 0, 1, 2, 3, . . . . . . }

F[u(t)] = αu(t) − αβu3 (t) + αβ2 u5 (t) + . . . . . . (3)

The polynomial series in Equation (3) consists of the nonlinear coefficients {α, αβ, . . . , αβk , . . . }.
Equation (3) shows a mathematical relationship of a magnitude-dependency among all the polynomial
coefficients. In comparison with the typical odd-order Taylor model in Equation (4), the Taylor
coefficients are often considered statistically and linearly independent parameters. Therefore, the
Taylor model exhibits a higher degree of freedom for fitting the polynomial coefficients to the observed
Electronics 2020, 9, 1806 3 of 13

data. However, the polynomial curve fitting is subject to uncertainty and often exhibits high prediction
errors beyond the range of the observed data.

T[u(t)] = c1 u(t) + c3 u3 (t) + c5 u5 (t) + . . . . . . (4)

The magnitude of the Taylor coefficients {c1 , c3 , . . . . . . } are considered device-dependent, but
the independent coefficients are insufficient for describing the effect of the higher order IMDs on the
lower order IMDs. For example, the fifth-order IMD affects the third-order IMD and so on [14]. The
higher order Taylor polynomial can achieve better model accuracy to fit the nonlinear curvature of
the AM/AM conversion, because using additional independent coefficients increases modeling the
dimensions of the nonlinear system, but this often leads to a noisy estimate and poor coverage of
confidence intervals.
The typical output of the AM/AM Saleh model increases almost linearly in the small signal
magnitude at the rate α. However, the gain of the Saleh model decreases monotonically after the
maximum curvature (this occurs at |u| = β−1/2 ) [15] and approaches zero for very large theoretical
values of |u|. This is one of the main weaknesses in the Saleh model when applied to SSPAs, because the
saturation level in SSPAs is almost constant and specified by the supply voltage on the drain/collector
of the output transistor. Power amplifiers are typically designed to operate over a dynamic range of
the load-line with the DC quiescent point “Q-point” which is set in the active region of the transistor
on a small amplitude signal (e.g., Q-point is set to the middle of the load-line in a class-A design).
However, on a large input amplitude signal, the output signal becomes extensively large and forces
the output signal to surpass the power supply voltage and becomes flattened. Therefore, the output
signal is limited by the supply voltage in the saturation region [16].
A simple model enhancement is introduced in this work to address the model limitations by
including an additional quadratic term within the numerator of the Saleh model [17] as illustrated

αu(t) + λu2 (t)

Fes [u(t)] = (5)
1 + βu2 (t)

where Fes [u(t)] is the proposed enhanced Saleh AM/AM model. The parameter λ is a real positive
number introduced here for adjusting both gain and roll-over near the saturation region. The small
signal gain α of the model in Equation (5) is approximately equal to the linear gain of the Saleh model
as depicted:
∂(Fes [u])
lim =α (6)
u→0 ∂(u)

The parameter λ causes the function in Equation (5) to be mathematically non-monotonic and the
output increases asymptotically toward the saturation for a large signal amplitude (saturation level ≈
λ/β). Finally, the polynomial expansion of Equation (5) becomes a mixed nonlinear combination of odd
and even terms as follows:
∞
X
(−1)k βk u(2k) (t) αu(t) + λu2 (t)
 
Fes [u(t)] = (7)
k =0

Fes [u(t)] = αu(t) + λu2 (t) − αβu3 (t) − λβu4 (t) + αβ2 u5 (t) + λβ2 u6 (t) − αβ3 u7 (t) + . . . . . . (8)

Equation (8) illustrates that the magnitude of the second-order IMD is completely characterized
by the parameter λ. The other higher order even-terms in this model are characterized by the nonlinear
coefficients {λ, λβ, . . . , λβk, . . . }. In addition, the proposed model enhancement can improve the
flatness in the saturation region, because the mathematical expression in Equation (8) converges
asymptotically to the constant value (λ/β) [17].
The second-order IMD of SSPAs is an ongoing concern in wideband and multiband
communications [14,18,19]. Figure 1 shows the typical two-tone IMDs, which illustrates the locations
Electronics 2020, 9, 1806 4 of 13

of the second-order distortion (f − f1 , f2 + f1 ) with respect to the fundamental tones for wideband
Electronics 2020, 9, x FOR PEER REVIEW 2 4 of 13
communications [14].

wideband nonlinear distortion

Magnitude

Frequency

f2–f1 2f1–f2 f1 f2 2f2–f1 f2 + f1

Figure 1. Wideband output spectrum of RF power amplifier illustrating the two-tone harmonic
Figure 1. Wideband output spectrum of RF power amplifier illustrating the two-tone harmonic
distortion effect.
distortion effect.

Model Estimation
A least-squares method is used for calculating
calculating the parameters of the enhanced Saleh model as
described
described in this section. Substituting
Substituting z(t) =
= Feses[u(t)]
[u(t)] in
in Equation
Equation (5)
(5) and
and re-arranging
re-arranging the equation
variables results in
variables results in
z(t) = αu(t) + λu22 (t) − z(t)βu 2
2 (t) (9)
z(t ) = αu(t ) + λu (t ) − z(t ) βu (t ) (9)
where z(t) is the envelope of the PA baseband output signal. A matrix equation can be formulated
where z(t) is the envelope of the PA baseband output signal. A matrix equation can be formulated
from Equation (9) by substituting discrete-time samples of both u(t) and z(t) as illustrated
from Equation (9) by substituting discrete-time samples of both u(t) and z(t) as illustrated
 α 
 z(0)    z (0) u(0)  u (0) − z (0) u 2 (0) 2   α 
  
u (0)
     −z(0)u2 (0) 2
  u (0) 
 

 2 2
     − −z
z (1)   
(1u)u(1) (1) u (1) βu (1)  β 
2 2 
 z(1)  =   z (1) u (1)
  
=

  u(1)     (10)
  
 ..     ..

  (10)
    

 .
..    .
..  . .
 
z (n()nu)2u(2n()n) u 2 (n)   λu2 (n) 
 

− −z

  z (n)
  u (n)

 
  
z(n) u(n)
 
λ

For simplicity, a matrix notation is used in representing (10)
For simplicity, a matrix notation is used in representing (10)
z = Uc (11)
where z is a column vector of ((n + 1) × 1) elements, z = Uc
c is a column vector of the model parameters (11)
(3 ×
1), and U is a matrix consisting of ((n + 1) × 3) elements of the input and output samples. Finally, the
where
vector zc is
is acalculated as: of ((n + 1) × 1) elements, c is a column vector of the model parameters (3 ×
column vector
1), and U is a matrix consisting of ((n + 1) × 3) elements of the input and output samples. Finally, the
vector c is calculated as: c = (UT U)-1 UT z (12)
T −1 T
c = (U U) U z (12)
where ()T denotes the operator of a matrix transposition [17].
where ()T denotes the operator of a matrix transposition [17].
3. Memory Modeling
3. Memory Modeling
Memory effect is a popular hysteresis phenomenon in RF PAs which causes dynamic nonlinear
Memory
dispersion effect
in the is a popular
AM/AM hysteresis
conversion phenomenon
[20,21]. Memory effectin RFisPAs which causes
considered dynamic nonlinear
device-dependent due to
dispersion
the impedances’in the variation
AM/AM conversion
of both the[20,21].
matching Memory effect is
and biasing considered
networks withdevice-dependent due
respect to operating
to the impedances’
frequency, and to the variation of both
temperature the matching
variation and biasing
in transistors’ networks
junctions with respect
as depicted to 2.
in Figure operating
frequency, and to the temperature variation in transistors’ junctions as depicted in Figure 2.
Electronics 2020, 9, 1806 5 of 13

Electronics 2020, 9, x FOR PEER REVIEW 5 of 13

Figure 2. Simplified architecture of the memory component elements of the RF power amplifier.

A digital filter is a straightforward approach for modeling the memory effect in RF PAs. In this
work a transpose finite-impulse response (FIR) digital filter is used in cascade with the enhanced
Saleh model based on Hammerstein modeling theory as depicted in Figure 3. The architecture of the
transpose FIR filter is computationally less complex than the direct FIR filter in real-time
implementation because it supports multiple constant multiplication technique.
The FIR filter
Figure
Figure 2. mimics architecture
2. Simplified
Simplified the output of
architecture signal
of the dependency
the memory
memory on theelements
component
component past samples
elements ofthe
of of power
theRF
RF the input
power signal using
amplifier.
amplifier.
a finite number of coefficients for controlling the memory depth in the model.
AAdigital
digital filter
filterisisaastraightforward
straightforward approach
approach for for modeling
modeling the the memory
memory effect
effectin inRFRFPAs.
PAs.In Inthis
this
M
workaatranspose
work transposefinite-impulse
Saleh model
model based onbased
finite-impulse
on Hammerstein
Hammerstein
ˆresponse
response
z(
modeling

n) = (FIR)
modeling
h(FIR)
theory
m=0
)Fes [digital
(mdigital
astheory
depicted
− mfilter
u(nfilter
as
j
depicted
in
u
)]ise usedis used
Figure
( n − m )
in
3.
in cascade
in cascade
Figure
The
with the
3. The
with
architecture
the enhanced
enhanced
architecture
of the
Saleh
(13)
of the
transpose
transpose
FIR FIR filter is computationally
filter is computationally less complex than less
the complex
direct FIR than filter in the direct implementation
real-time FIR filter in real-time
because
where h(m) represents the filter impulse response, M is the memory depth in the model, and 𝑧̂ 𝑛 is
itimplementation
supports multiple because it supports
constant multiple
multiplication constant multiplication technique.
technique.
the output
The FIRsignal
filterofmimics
the dynamic nonlinear
the output signalmodel.
dependency on the past samples of the input signal using
a finite number of coefficients for controlling the memory depth in the model.
M
ˆ n) =
z(  h(m)F [u(n − m)] e
m=0
es
ju ( n − m)
(13)

where h(m) represents the filter impulse response, M is the memory depth in the model, and 𝑧̂ 𝑛 is
the output signal of the dynamic nonlinear model.

Figure
Figure 3.3.Hammerstein model
Hammerstein structure
model for RFfor
structure power
RF amplifier using theusing
power amplifier enhanced
the Saleh amplitude-
enhanced Saleh
to-amplitude (AM/AM) model.
amplitude-to-amplitude (AM/AM) model.

4. Modeling
The FIR of Digital
filter Predistortion
mimics the output signal dependency on the past samples of the input signal using
a finite number of coefficients
DPDs have become popular for approaches
controlling the memory depth
for improving in the model.
the linearity of RF PAs for a reliable and
an efficient signal transmission [22,23]. The M DPD implementation in the digital domain as in Figure 4
X
is practically more efficient than ẑ(n) = predistortion
analog h(m)Fes [u(n for
− mcompensating
)]e j]u(n−m) the nonlinear distortion(13) in
Figure 3. Hammerstein model structure for RF power amplifier using the enhanced Saleh amplitude-
RF PAs. The DPD models based on Taylor m=0 polynomials are popular linearization techniques to
to-amplitude (AM/AM) model.
compensate for a weak-nonlinear distortion in RF PAs [24].
where A h(m) represents
complex the filter
structure of a impulse
DPD model is notMdesired
response, is the memory depth
in wireless in the model, and
communication, ẑ(n) isthis
because the
4. Modeling
output signal of Digital
of theand Predistortion
dynamic nonlinear model. hardware system. Hence, a predistortion model in
requires a complex high-computational-cost
DPDs
this work is have become
calculated popular
directly fromapproaches forrational
a simplified improving the linearity
function of RF PAs
of the enhanced for amodel
Saleh reliable and
using
4. Modeling of Digital Predistortion
an efficient
fewer signal transmission [22,23]. The DPD implementation in the digital domain as in Figure 4
coefficients.
DPDs have
is practically become
more popular
efficient thanapproaches for improving
analog predistortion for the linearity of RF
compensating thePAs for a reliable
nonlinear and an
distortion in
efficient
RF PAs.signal transmission
The DPD models [22,23].
based on TheTaylor
DPD implementation
polynomials areinpopularthe digital domain as techniques
linearization in Figure 4 is to
practically
compensate more
for efficient than analog
a weak-nonlinear predistortion
distortion in RFfor
PAscompensating
[24]. the nonlinear distortion in RF PAs.
The DPD models structure
A complex based on Taylor
of a DPD polynomials
model isare
notpopular
desiredlinearization techniques to compensate
in wireless communication, for
because this
arequires
weak-nonlinear
a complex distortion in RF PAs [24].
and high-computational-cost hardware system. Hence, a predistortion model in
this work is calculated directly from a simplified rational function of the enhanced Saleh model using
fewer coefficients.
Electronics 2020, 9, 1806 6 of 13

Electronics 2020, 9, x FOR PEER REVIEW 6 of 13

Figure 4. Simplified
Figure 4. Simplified diagram
diagram of
of aa digital
digital predistortion
predistortion transmitter
transmitter system.
system.

A complex
The structure
proposed of a DPD
DPD model canmodel is not desired
be deployed for theinPA wireless communication,
of strong nonlinearity because this
over a wide
requires a complex and high-computational-cost hardware system. Hence, a predistortion
amplitude range. The mathematical inversion of the enhanced Saleh model is calculated as follows: model in
this work is calculated directly from a simplified rational function of the enhanced Saleh model using
fewer coefficients. u (t ) = Fes−1[ d (t )] (14)
The proposed DPD model can be deployed for the PA of strong nonlinearity over a wide amplitude
where d(t) and u(t) are the DPD model input and output magnitudes, respectively, as depicted on the
range. The mathematical inversion of the enhanced1 Saleh model is calculated as follows:
simplified transmitter architecture in Figure. 4. Fes is the proposed enhanced DPD model, which
is calculated by exchanging the input and u output
(t) = Fvariables
−1 of the enhanced Saleh model [11] Fes[u(t)]
es [d(t)] (14)
in Equation (5) and simplifying the expression as:
where d(t) and u(t) are the DPD model input and 2output magnitudes, respectively, as depicted on the
(dβ − λ)u − αu+d = 0 (15)
simplified transmitter architecture in Figure 4. F−1 es [ ] is the proposed enhanced DPD model, which is
calculated bythe
Finally, exchanging
DPD modeltheisinput and output
calculated variables
by solving of the enhanced
a quadratic Saleh
formula in model(15)
Equation [11]with
Fes [u(t)] in
respect
Equation (5) and
to the variable u.simplifying
An equationthe expression
solution as:
of a negative sign is used in this model, since the DPD input
and output magnitudes are real and normalized in this work.
(dβ − λ)u2 − αu + d = 0 (15)
α − α 2 − 4 βd 2 + 4dλ
Fes−1[d ] = (16)
Finally, the DPD model is calculated by solving2 (adβ − λ)
quadratic formula in Equation (15) with respect
to the variable u. An equation solution of a negative sign is used in this model, since the DPD input
Equation
and output (16) represents
magnitudes are real the
and DPD model,inwhich
normalized outputs a magnitude of the predistorted
this work.
baseband signal. The maximum input amplitude of the DPD model is calculated to satisfy a positive
p
real value inside the root term of Equation
−1 α − as αfollows:
(16) 2 − 4βd2 + 4dλ
Fes [d] = (16)
2 2(dβ − λ2)
α +4dmax λ ≥ 4βdmax (17)
Equation (16) represents the DPD model, which outputs a magnitude of the predistorted baseband
The maximum amplitude of the DPD input |dmax| to satisfy Equation (17) is given by:
signal. The maximum input amplitude of the DPD model is calculated to satisfy a positive real value
inside the root term of Equation (16) as follows: λ + λ 2 + βα 2
d max = (18)
2 2β 2
α + 4dmax λ ≥ 4βdmax (17)
where |dmax| represents a clipping threshold of the DPD baseband input signal. Therefore, a clipping
modelTheinmaximum
Equation amplitude of the DPD
(19) is required on the input |dmax
input of | the
to satisfy
DPD Equation
model to (17) is given
maintain theby:
condition in
Equation (17). p
λ + λ2 + βα2
|dmax |=d (t ) , if d (t ) < d (18)
2β max

dˆ (t ) =  (19)
where |dmax | represents a clipping thresholdof d the, DPD
if d (tbaseband
) ≥ d input signal. Therefore, a clipping
 max max
model in Equation (19) is required on the input of the DPD model to maintain the condition in
where |𝑑(17).
Equation 𝑡 | is a clipped amplitude version of the time domain signal d(t). Figure 5 describes the
characteristics of the clipping model in Equation (19) for the amplitude relationships between the
input and output signals. This model forces the large amplitudes of the input signal to a constant
threshold according to the value of |dmax|, which is specified by the model parameters (α,λ,β).
Equation (18) illustrates that the level magnitude of the clipping model is directly proportional to the
power amplifier gain (α,λ) and inversely proportional to the parameter β. Although the model in
Electronics 2020, 9, 1806 7 of 13

   
if d(t) < |dmax |

   d(t),
dˆ(t) = 

(19)


  
 |dmax |, if d(t) ≥ |dmax |
 
where dˆ(t) is a clipped amplitude version of the time domain signal d(t). Figure 5 describes the
characteristics of the clipping model in Equation (19) for the amplitude relationships between the
input and output signals. This model forces the large amplitudes of the input signal to a constant
Electronics 2020, 9, x FOR PEER REVIEW 7 of 13
threshold according to the value of |dmax |, which is specified by the model parameters (α,λ,β). Equation
(18) illustrates
Electronics
Figure 52020, 9,that
is linearx FORthe level
PEER
with magnitude
gain in theofamplitude
REVIEW
unit the clipping model
range (0 tois ddirectly proportional to the power
max), the clipping model is another
7 of 13
amplifier
nonlineargain (α,λ) and
function inversely
required proportional
on the input of thetoDPD model. Theβ.model
the parameter Although the model
clipping in Figureare
characteristics 5
isFigure
linear 5with
is linear
unit within
gain unit
the gain in the range
amplitude amplitude
(0 to range
d max ), (0
thetoclipping
dmax), the
depicted in Figure 6, which describe the variation in the input signal PAPR with respect to the
clipping
model is model nonlinear
another is another
nonlinear
function function on|required on
ofthe
the input of the DPD model.clipping
The model clipping characteristics are
clippingrequired
level |dmax the input
using DPDfrequency
an orthogonal model. The model
division multiplexing characteristics
(OFDM) signal.are depicted
The outputin
depicted
Figure in
6, which Figure 6,
describe which describe
thetovariation the variation
in the input in
signalfor the
PAPR input signal
withclipping PAPR
respect level with respect
to theasclipping to the
level |dmax |
amplitude becomes equal a non-clipped amplitude a large illustrated.
clipping
using level |dmax|frequency
an orthogonal using an orthogonal frequency division
division multiplexing (OFDM)multiplexing (OFDM)
signal. The output signal. The
amplitude output
becomes
amplitude
equal becomes equal
to a non-clipped to a non-clipped
amplitude for a large amplitude for aaslarge
clipping level clipping level as illustrated.
illustrated.

Figure 5. Signal amplitude limiter model required on the input of the digital predistortion (DPD)
model.
Figure5.5.Signal
Figure Signal amplitude
amplitude limiter
limiter model
model required
required oninput
on the the input of the predistortion
of the digital digital predistortion (DPD)
(DPD) model.
model.

Figure6.6.The
Figure Themodel
modelclipping
clippinglevel
leveland
andthe
thepeak-to-average
peak-to-averagepower
powerratio
ratio(PAPR)
(PAPR)ofofthe
theclipped
clippedsignal.
signal.
Figure 6. The model clipping level and the peak-to-average power ratio (PAPR) of the clipped
The impact of the clipping nonlinearity on the distortion effects is not severe, because the large
signal.
peaks The impact
typically of the
occur withclipping nonlinearity
low probability. on the distortion
Therefore, effects isapproaches
different clipping not severe,have
because
beenthe large
widely
peaks
used in typically occur with lowtoprobability.
wireless communication reduce the Therefore,
signal PAPR, different
and theyclipping approaches
are considered have been
an important
The impact of the clipping nonlinearity on the distortion effects is not severe, because the large
widelyaspect
design used when
in wireless communication
operating RF PAs on OFDMto reduce the In
signals. signal PAPR,
addition, and they
reducing theare considered
signal PAPR can an
peaks typically occur with low probability. Therefore, different clipping approaches have been
important design aspect when operating RF PAs on OFDM signals. In addition,
improve the power efficiency in RF PAs, because the maximum power efficiency is obtained whenreducing the signal
widely used in wireless communication to reduce the signal PAPR, and they are considered an
PAPR can
operating RFimprove
PAs near the power efficiency
the compression regionin[14].
RF Therefore,
PAs, because the maximum
applying power
signal clipping efficiency is
in combination
important design aspect when operating RF PAs on OFDM signals. In addition, reducing the signal
obtained
with whenleads
DPD often operating RF PAs near
to performance the compression
improvement in powerregion [14].
efficiency Therefore,
and applying signal
linearity [13,25–27].
PAPR can improve the power efficiency in RF PAs, because the maximum power efficiency is
clipping in combination with DPD often leads to performance improvement in power efficiency and
obtained when operating RF PAs near the compression region [14]. Therefore, applying signal
linearity [13,25–27].
clipping in combination with DPD often leads to performance improvement in power efficiency and
linearity [13,25–27].
5. Evaluation Results
Two different
5. Evaluation SSPA technologies, gallium arsenide (GaAs) (ZFL-1000LN from Mini-Circuits,
Results
Brooklyn, NY, USA) and gallium nitride (GaN) (ZX60-8008E from Mini-Circuits) devices under test
Two different SSPA technologies, gallium arsenide (GaAs) (ZFL-1000LN from Mini-Circuits,
(DUT) are used in this experiment for model evaluation and linearity enhancement of the DUT PA.
Electronics 2020, 9, 1806 8 of 13

5. Evaluation Results
Two different SSPA technologies, gallium arsenide (GaAs) (ZFL-1000LN from Mini-Circuits,
Brooklyn, NY, USA) and gallium nitride (GaN) (ZX60-8008E from Mini-Circuits) devices under test
(DUT) are used in this experiment for model evaluation and linearity enhancement of the DUT PA.
The experiment set-up consists of a signal generator (E4438C from Keysight Technologies, Santa
Rosa, CA, USA), a spectrum analyzer (RSA6120A from Tektronix, Beaverton, OR, USA), a high-power
attenuator, and a computer (Dell computer with Intel Xeon CPU E3-1241 v3 @ 3.50 GHz and 16 GB of
ram) with MATLAB (R2018a from MathWorks, Natick, MA, USA) as shown in Figure 7. A 64-QAM
Electronics 2020,
dual-band 9, x FOR PEER
modulated OFDMREVIEW 8 ofis
signal was applied on the input of the SSPA, and the output signal 13RF
down-converted, demodulated, and acquired by the spectrum analyzer.

(a) (b)
64-QAM dual-band modulated OFDM signal was applied on the input of the SSPA, and the output
Figure 7. Experimental validation for modeling and digital predistortion of a solid-state power amplifier
signal is RF down-converted, demodulated, and acquired by the spectrum analyzer.
(SSPA). (a) Schematic diagram. (b) Photograph of the test set-up.
Figure 7. Experimental validation for modeling and digital predistortion of a solid-state power amplifier
5.1. SSPA(a)Modeling
(SSPA). SchematicPerformance
diagram. (b) Photograph of the test set-up.
The AM/AM conversion of a GaAs RF PA is measured by sweeping the input amplitude of a
5.1. SSPA Modeling Performance
two-tone signal from the signal generator at a 1 GHz center frequency and a 10 KHz tone spacing. The
AM/AM Thenonlinear
AM/AM conversion
conversion of the
a GaAs
SSPARFmeasurements
PA is measured by sweeping
as well the input
as the model amplitude
estimation using ofboth
a
two-tone signal from the signal generator at a 1 GHz center frequency and a
the Saleh and enhanced Saleh models is illustrated in Figure 8. The enhanced Saleh model shows 10 KHz tone spacing.
The AM/AM
improved nonlinearresults
curve-fitting conversion oflinear
in the the SSPA
andmeasurements as well as
compression regions the modeltoestimation
compared using
the original Saleh
both the
model. InSaleh and enhanced
addition, Saleh
the residual models
square is illustrated
errors between thein Figure 8. Theand
measured enhanced
model Saleh (|zmeas | −
model shows
conversions
improved
|zmode curve-fitting
|)2 clearly show lowerresults in the linear
fluctuation errorsand
andcompression regions compared
accuracy improvement to the original
of the enhanced SalehSaleh
model
model. In addition, the residual
with respect to the input amplitude. square errors between the measured and model conversions
(|zmeas
The |−|z mode|)2 clearly show lower fluctuation errors and accuracy improvement of the enhanced
model evaluation for predicting the nonlinear distortion in a frequency domain is shown in
Saleh 9,
Figure model
usingwith
the respect to the input
power spectrum amplitude.
density of a dual-band long-term evolution (LTE) signal. Table 1
presents the obtained accuracy improvement in the enhanced Saleh model using the normalized mean
square-errors (NMSE) based on GaAs and GaN DUT PA technologies. These results show a model
improvement of around 4dB NMSE compared to the original Saleh model. In addition, the enhanced
model achieved better model accuracy compared to the fifth-order Taylor model.
The AM/AM nonlinear conversion of the SSPA measurements as well as the model estimation using
both the Saleh and enhanced Saleh models is illustrated in Figure 8. The enhanced Saleh model shows
improved curve-fitting results in the linear and compression regions compared to the original Saleh
model. In addition, the residual square errors between the measured and model conversions
(|zmeas|−|z
Electronics mode|)
2020,
2 clearly show lower fluctuation errors and accuracy improvement of the enhanced
9, 1806 9 of 13
Saleh model with respect to the input amplitude.

Electronics 2020, 9, x FOR PEER REVIEW 9 of 13

The model evaluation for predicting the nonlinear distortion in a frequency domain is shown in
Figure 9, using the power spectrum density of a dual-band long-term evolution (LTE) signal. Table 1
presents the obtained accuracy improvement in the enhanced Saleh model using the normalized
meanFigure
square-errors
Figure 8.
8. Measured
(NMSE)
Measuredand and
basedstatic
modeled
modeled
onstatic
GaAs
AM/AM andresults
AM/AM
GaN ofDUT
results the PA technologies.
of SSPA usingusing
the SSPA
These
a two-tone results
test. The
a two-tone
show a
square
test. The
model improvement
residual errors of
square residual of around
the Saleh
errors 4dB
of themodel NMSE compared
and enhanced
Saleh model to
Saleh model
and enhanced the original Saleh
are overlaid
Saleh model model.
with respect
are overlaid In addition,
to the to
with respect the the
input
enhanced model
amplitude. achieved
input amplitude. better model accuracy compared to the fifth-order Taylor model.

9. Measured
Figure 9. Measuredand
andmodeled
modeledoutput
output spectrum
spectrum of of
thethe solid-state
solid-state power
power amplifier
amplifier usingusing a dual-
a dual-band
LTE
bandsignal.
LTE signal.

Table 1. Accuracy comparison in normalized mean square-errors (NMSE) of the model evaluation
Table 1. Accuracy comparison in normalized mean square-errors (NMSE) of the model evaluation
using gallium arsenide (GaAs) and gallium nitride (GaN) technologies.
using gallium arsenide (GaAs) and gallium nitride (GaN) technologies.
Intermodulation Distortion NMSE NMSE (dB)
Model Number of Coeff. (dB)
Number of Intermodulation
(IMD)Distortion
Model GaAs GaNGaN PA
GaAs PA
Coeff. (IMD)
Saleh 2 odd terms PA
−27.12 PA −28.54
Saleh
Enhanced Saleh 23 odd
odd andterms
even terms −27.12
−31.37−28.54−32.71
Enhanced Saleh
5th order Polynomial 35 odd and even terms
odd and even terms −31.37
−22.16−32.71−21.47
5th order
5 odd and even terms −22.16 −21.47
Polynomial
The overall accuracy in NMSE of the enhanced Saleh model in cascade with FIR filter is depicted
in Figure 10, with respect to the FIR filter number of coefficients (i.e., memory depth). This shows
The overall accuracy in NMSE of the enhanced Saleh model in cascade with FIR filter is depicted
the obtained optimal model accuracy and the required filter number of coefficients to quantify the
in Figure 10, with respect to the FIR filter number of coefficients (i.e., memory depth). This shows the
obtained optimal model accuracy and the required filter number of coefficients to quantify the
memory effect of the SSPA. The FIR filter in this modeling approach reaches the optimal accuracy at
a memory depth around 15 for practical implementation, and the accuracy in NMSE changes slightly
and becomes almost flat after in the higher order memory depth as illustrated in Figure 10. The
dynamic enhanced model achieves a NMSE below −47dB.
Electronics 2020, 9, 1806 10 of 13

memory effect of the SSPA. The FIR filter in this modeling approach reaches the optimal accuracy at a
memory depth around 15 for practical implementation, and the accuracy in NMSE changes slightly and
becomes almost flat after in the higher order memory depth as illustrated in Figure 10. The dynamic
Electronics 2020,
Electronics
enhanced 2020, 9, xx FOR
9,
model FOR PEER REVIEW
PEER
achievesREVIEW
a NMSE below −47dB. 10 of
10 of 13
13

Figure 10.
Figure 10. The
The model
model accuracy
accuracy in
in normalized
normalized mean
mean square-error
square-error (NMSE)
(NMSE) ofof the
the dynamic
dynamic enhanced
enhanced
Saleh model in terms of the memory depth of the finite-impulse
Saleh model in terms of the memory depth of the finite-impulse response
finite-impulse response (FIR)
response (FIR) filter.
(FIR) filter.
filter.

5.2. Results
5.2. Results of
of Digital
Digital Predistortion
Predistortion
The static
The static AM/AM
AM/AM conversion
conversion causes most the
causes most the nonlinear
nonlinear distortion
distortion in in RF
RF PAs. Therefore, the
PAs. Therefore,
PAs. the
efficiency of the DPD to compensate for the AM/AM distortion is a significant factor
efficiency of the DPD to compensate for the AM/AM distortion is a significant factor in linearizing in linearizing
the nonlinearity
the in communication
nonlinearity in systems. The
communication systems. The DPD
The DPD model
model based
based onon the enhanced Saleh
the enhanced model is
Saleh model is
evaluated in the time and frequency domains using a dual-band LTE signal. The
evaluated in the time and frequency domains using a dual-band LTE signal. The AM/AM conversionAM/AM conversion
of the
of the DPD
the DPDmodel
DPD modelexhibits
model exhibitssignificant
exhibits gain
significant
significant gain
gainexpansion
expansion
expansioncharacteristics overover
characteristics
characteristics a wide
over rangerange
aa wide
wide of theof
range ofPA input
the
the PA
PA
amplitude
input as
amplitudeshownas in Figure
shown in11. In addition,
Figure 11. In the DPD
addition, model
the compensated
DPD model for the
compensated
input amplitude as shown in Figure 11. In addition, the DPD model compensated for the gain gain compression
for the gain
behavior
compression
compressionof the SSPA and
behavior
behavior of improved
of the SSPA the
the SSPA andbalance
and improved
improvedbetween in-phasebetween
the balance
the balance and quadrature-phase
between and in
in-phase and
in-phase the signal
quadrature-
quadrature-
constellation
phase in
phase in the diagram,
the signal
signal as shown diagram,
constellation
constellation in Figure as
diagram, 12.shown
as shown inin Figure
Figure 12.
12.

Figure 11.
Figure The dynamic
11. The dynamic AM/AM
AM/AM conversion
conversion of
of the
the RF
RF power
power amplifier
amplifier and
and DPD
DPD model
model conversions
conversions
based inverting the enhanced Saleh model.
based inverting the enhanced Saleh model.
Electronics 2020, 9,
2020,
Electronics 2020, 9, 1806 11 of 13
Electronics 9, xx FOR
FOR PEER
PEER REVIEW
REVIEW 11 of 13

(a)
(a) (b)
(b)
Figure 12.
Figure 12. Constellation
Constellation diagram
Constellation diagram of
of aaa 64-QAM
diagram of 64-QAM baseband
64-QAM baseband output
baseband output signal
output signal of
signalof the
ofthe RF
theRF power
RFpower amplifier,
poweramplifier, (a)
amplifier, (a)
(a)
without DPD, and (b) with DPD.
without DPD, and (b) with DPD.

The linearization
The
The linearizationcapability
linearization capability
capability of the
of the
of the DPD
DPD DPD model
modelmodel in aa frequency
in a frequency
in frequency domain
domain shows
domain shows significant
significant
shows significant
mitigation
mitigation
of
mitigation of the
the spectrum
of theregrowth
spectrum
spectrum regrowth
inregrowth
the adjacentin the
in the adjacent
channels
adjacent channels
of channels
the of the
dual-band
of thespectrum
dual-band
dual-band spectrum
as spectrum
depicted in asFigure
as depicted
depicted13.
in Figure 13. The numerical evaluation results of this linearization approach is depicted in Table 22
in
TheFigure 13.
numerical The numerical
evaluation evaluation
results of this results of
linearization this linearization
approach is approach
depicted in is
Table depicted
2 using thein Table
adjacent
using the
channel
using thepower
adjacent
adjacent channel
ratio (ACPR)
channel power ratio vector
and error
power ratio (ACPR)
(ACPR) and error
error vector
magnitude
and vector magnitude
(EVM) magnitude
for GaAs and (EVM)
GaNfor
(EVM) for GaAs
SSPA.
GaAs and GaN
GaN
Finally,
and the
SSPA.
obtainedFinally, the
linearizationobtained linearization
performance of the performance
enhanced DPD of the
model enhanced
achieved an
SSPA. Finally, the obtained linearization performance of the enhanced DPD model achieved an DPD model
adequate achieved
improvement an
adequate
in time andimprovement
frequency in time
domains. and frequency
adequate improvement in time and frequency domains. domains.

Figure 13.
Figure 13. Spectrum
Spectrum of
of aaa dual-band
Spectrum of dual-band LTE
dual-band LTE output signal
LTE output signal without
without DPD
DPD in
in the
the red
red trace
trace and
and with
with DPD
DPD in
in
the black
the black trace.
trace.

2. DPD
TableTable modelmodel
2. DPD
DPD performance in adjacent
performance channel
in adjacent
adjacent powerpower
channel ratio (ACPR) and error
ratio (ACPR)
(ACPR) andvector magnitude
error vector
vector
Table 2. model performance in channel power ratio and error
(EVM) for linearizing both
magnitude (EVM) GaAs
(EVM) for and GaN
for linearizingpower
linearizing both amplifiers
both GaAs
GaAs and (PAs).
and GaN
GaN power
power amplifiers
amplifiers (PAs).
(PAs).
magnitude
GaAs
GaAs GaNGaN
Evaluation
Evaluation
Evaluation Measurewithout DPD GaAswith DPD
Measure
Measure
GaN
without DPD with DPD
without DPD
without DPD with DPD
with DPD without DPD
without DPD with DPD
with DPD
ACPR in Upper
ACPR
ACPR
band/Lower band
in
in
(dBc)
−38.73/−39.35 −52.10/−52.81 −41.86/−42.55 −56.13/−57.41
−38.73/
−38.73/ −52.10/
−52.10/ −41.86/
−41.86/ −56.13/
−56.13/
Upper band/
band/
EVMUpper
(%) −39.35
9.12−39.35 −52.81
1.86
−52.81 −42.55
8.76
−42.55 −57.411.54
−57.41
Lower band
Lower band (dBc)
(dBc)
EVM (%)
EVM (%) 9.12
9.12 1.86
1.86 8.76
8.76 1.54
1.54
Electronics 2020, 9, 1806 12 of 13

6. Conclusions
This paper analyzed the Saleh model mathematically using polynomial expansion for modeling
the odd-order IMD and described the Saleh weakness for modeling SSPAs. Therefore, a new model
enhancement was proposed to improve the model accuracy for the Saleh AM/AM model when
deployed for SSPA technologies in communication systems. In addition, the polynomial expansion of
the enhanced Saleh model consists of both odd-order and even-order nonlinear terms for characterizing
the PA nonlinearities in the AM/AM conversion. The Hammerstein-based model structure was adopted
to quantify the dynamic effect in the AM/AM conversion. A least-squares approach was employed for
calculating all the parameters of the enhanced Saleh model. The evaluation results showed an accuracy
improvement of the enhanced Saleh model to reflect the measured AM/AM characteristics of SSPAs
over a wide range of the PA input magnitude. A linearization approach using DPD was developed
and evaluated for mitigating the nonlinear distortion in SSPAs.

Author Contributions: Conceptualization, H.A.; Methodology, H.A. and F.L.; Software, H.A.; Supervision, F.L.;
Writing—original draft, H.A.; Writing—review & editing, H.A. and F.L. All authors have read and agreed to the
published version of the manuscript.
Funding: Publication of this article is funded by Portland State University’s open access fund.
Acknowledgments: The authors would like to acknowledge James Morris from the Department of electrical and
computer engineering at Portland State University for contributions on proofreading this article.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Popovic, Z. Amping Up the PA for 5G: Efficient GaN Power Amplifiers with Dynamic Supplies. IEEE Microw.
Mag. 2017, 18, 137–149. [CrossRef]
2. Singh, S.K.; Suthar, L.; Singh, R.; Kumar, A. GaN Based S-Band 500W Solid State Power Amplifier (SSPA)
Module for Troposcatter Communications. In Proceedings of the 2019 IEEE MTT-S International Microwave
and RF Conference (IMARC), Mumbai, India, 13–15 December 2019; pp. 1–4.
3. Davis, C.; Hawkins, J.; Einolf, C. Solid state DTV transmitters. IEEE Trans. Broadcast. 1997, 43, 252–260.
[CrossRef]
4. Hu, S.; Xu, K.; Hu, Y.; Wu, K.; Qian, J.; Gao, Y.; Song, D.; Wang, Z.; Zhou, B.; Cai, Z.; et al. Design of a compact
chip filter with two transmission zeros using 0.35 µm GaAs HBT. Microelectron. J. 2020, 95, 104661. [CrossRef]
5. Sun, J.; Shi, W.; Yang, Z.; Yang, J.; Gui, G. Behavioral Modeling and Linearization of Wideband RF Power
Amplifiers Using BiLSTM Networks for 5G Wireless Systems. IEEE Trans. Veh. Technol. 2019, 68, 10348–10356.
[CrossRef]
6. Ghannouchi, F.; Hammi, O.; Helaoui, M. Behavioral Modeling and Predistortion of Wideband Wireless Transmitters;
Wiley: Hoboken, NJ, USA, 2015; pp. 43–63.
7. Cann, A.J. Improved nonlinearity model with variable knee sharpness. IEEE Trans. Aerosp. Electron. Syst.
2012, 48, 3637–3646. [CrossRef]
8. Saleh, A. Frequency-Independent and Frequency-Dependent Nonlinear Models of TWT Amplifiers. IEEE
Trans. Commun. 1981, 29, 1715–1720. [CrossRef]
9. O’Droma, M.; Meza, S.; Lei, Y. New modified Saleh models for memoryless nonlinear power amplifier
behavioural modelling. IEEE Commun. Lett. 2009, 13, 399–401. [CrossRef]
10. Ochiai, H. An Analysis of Band-limited Communication Systems from Amplifier Efficiency and Distortion
Perspective. IEEE Trans. Commun. 2013, 61, 1460–1472. [CrossRef]
11. Nguyen, T.; Yoh, J.; Lee, C.; Tran, H.; Johnson, D. Modeling of hpa and hpa linearization through a predistorter:
Global broadcasting service applications. IEEE Trans. Broadcast. 2003, 49, 132–141. [CrossRef]
12. Al Kanan, H.; Yang, X.; Li, F. Improved estimation for Saleh model and predistortion of power amplifiers
using 1-dB compression point. J. Eng. 2020, 2020, 13–18. [CrossRef]
13. Van Nee, R.; Prasad, R. OFDM for Wireless Multimedia Communications; Artech House: Norwood, MA,
USA, 2000.
14. Cripps, S.C. Advanced Techniques in RF Power Amplifiers Design; Artech House: Nonvood, MA, USA, 2002.
Electronics 2020, 9, 1806 13 of 13

15. Al-Kanan, H.; Yang, X.; Li, F. Saleh model and digital predistortion for power amplifiers in wireless
communications using the third-order intercept point. J. Electron. Test. 2019, 35, 359–365. [CrossRef]
16. Xu, K. Integrated Silicon Directly Modulated Light Source Using p-Well in Standard CMOS Technology.
IEEE Sens. J. 2016, 16, 6184–6191. [CrossRef]
17. Al-kanan, H. Power Efficiency Enhancement and Linearization Techniques for Power Amplifiers in Wireless
Communications. Ph.D. Dissertation, Portland State University, Portland, OR, USA, 2020.
18. Singh, A.; Scharer, J.; Booske, J.; Wohlbier, J. Second- and Third-Order Signal Predistortion for Nonlinear
Distortion Suppression in a TWT. IEEE Trans. Electron Devices 2005, 52, 709–717. [CrossRef]
19. Ding, L.; Zhou, G. Effects of Even-Order Nonlinear Terms on Power Amplifier Modeling and Predistortion
Linearization. IEEE Trans. Veh. Technol. 2004, 53, 156–162. [CrossRef]
20. Moon, J.; Kim, B. Enhanced Hammerstein Behavioral Model for Broadband Wireless Transmitters. IEEE
Trans. Microw. Theory Tech. 2011, 59, 924–933. [CrossRef]
21. Al-Kanan, H.; Li, F.; Tafuri, F.F. Extended Saleh model for behavioral modeling of envelope tracking
power amplifiers. In Proceedings of the 2017 IEEE 18th Wireless and Microwave Technology Conference
(WAMICON), Cocoa Beach, FL, USA, 24–25 April 2017; pp. 1–4.
22. Ren, J. A New Digital Predistortion Algorithms Scheme of Feedback FIR Cross-Term Memory Polynomial
Model for Short-Wave Power Amplifier. IEEE Access 2020, 8, 38327–38332. [CrossRef]
23. Jang, K.; Ryu, H.-G.; Ryu, S.B.; Lee, S.G. Spectrum characteristics and predistortion gain in the nonlinear
high power amplifier (HPA). In Proceedings of the 2017 International Conference on Information and
Communication Technology Convergence (ICTC), Jeju Island, Korea, 18–20 October 2017; pp. 931–934.
24. Zhou, G. Analysis of spectral regrowth of weakly nonlinear power amplifiers. IEEE Commun. Lett. 2000, 4,
357–359. [CrossRef]
25. Wang, J.; Xu, Y.; Zhu, X. Digital predistorted inverse class-F GaN PA with novel PAPR reduction technique.
In Proceedings of the 2011 IEEE MTT-S International Microwave Symposium, Baltimore, MD, USA, 5–10
June 2011.
26. Chen, S. An Efficient Predistorter Design for Compensating Nonlinear Memory High Power Amplifiers.
IEEE Trans. Broadcast. 2011, 57, 856–865. [CrossRef]
27. Wang, Y.; Akansu, A.N. Low-complexity peak-to-average power ratio reduction method for orthogonal
frequency-division multiplexing communications. IET Commun. 2015, 9, 2153–2159. [CrossRef]

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional
affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).