Institutionen för systemteknik Biasing for high linearity base-station pre-driver Department of Electrical Engineering

Institutionen för systemteknik Biasing for high linearity base-station pre-driver Department of Electrical Engineering
Institutionen för systemteknik
Department of Electrical Engineering
Title
Biasing for high linearity base-station
pre-driver
Master’s thesis
Performed in Electronic Devices,
Dept. of Electrical Engineering
Linköping Institute of Technology
By Noora Solati
Reg nr: LiTH-ISY-EX--10/4423--SE
August 2013
2
Biasing for high linearity base-station
pre-driver
Master thesis in Electronic Devices,
Dept. of Electrical Engineering
at Linköping Institute of Technology
by
Noora Solati
LiTH-ISY-EX--10/4423--SE
Supervisor: Professor Ted Johansson
Huawei Company and Linköping University
Examiner: Professor Atila Alvandpour
Linköping University
4
Presentation Date
2010-06-07
Department and Division
Publishing Date (Electronic version)
Language
Department of Electrical Engineering
Electronic Devices
Type of Publication
English
Other (specify below)
Number of Pages
Licentiate thesis
Degree thesis
Thesis C-level
Thesis D-level
Report
Other (specify below)
ISBN (Licentiate thesis)
ISRN: LiTH-ISY-EX--10/4423--SE
Title of series (Licentiate thesis)
Series number/ISSN (Licentiate thesis)
URL, Electronic Version
http://www.ep.liu.se
Publication Title
Biasing for high linearity base-station pre-driver
Author(s)
Noora Solati
Abstract
This thesis is a study on different methods to cancel out the nonlinearities of class A and AB power amplifiers
(PA), using the proper biasing circuit design, or using analog pre distortion techniques.
In this thesis, the basic fundamentals, the nonlinearity sources in power amplifiers, and different biasing
circuits for PAs and their effects on nonlinearity of PAs are investigated.
In addition, ordinary static and dynamic analog distorters in GaAs based PAs are reviewed. An investigation
on static and dynamic analog distorters in silicon bipolar transistor based PAs are presented.
Some investigation on P1dB compression point, gain distortion and phase distortion, using single tone input
signal are illustrated. Adjacent Chanel Power Ratio (ACPR1) and Alternate Channel Power Ratio (ACPR2)
are also simulated using modulated input signal source.
Keywords
Amplifier, PA, Linear, Distortion
6
Abstract
This thesis is a study on different methods to cancel out the nonlinearities of class A and AB
power amplifiers (PA), using the proper biasing circuit design, or using analog pre distortion
techniques.
In this thesis, the basic fundamentals, the nonlinearity sources in power amplifiers, and different
biasing circuits for PAs and their effects on nonlinearity of PAs are investigated.
In addition, ordinary static and dynamic analog distorters in GaAs based PAs are reviewed. An
investigation on static and dynamic analog distorters in silicon bipolar transistor based PAs are
presented.
Some investigation on P1dB compression point, gain distortion and phase distortion, using single tone
input signal are illustrated. Adjacent Chanel Power Ratio (ACPR1) and Alternate Channel Power
Ratio (ACPR2) are also simulated using modulated input signal source.
8
ABSTRACT .......................................................................................................................................... 7
CHAPTER 1....................................................................................................................................... 10
Introduction ........................................................................................................................................................ 10
1.1
Overview ........................................................................................................................................... 10
1.2
Objectives ......................................................................................................................................... 10
1.3
Thesis organization ........................................................................................................................ 11
1.4
List of acronyms............................................................................................................................... 12
CHAPTER 2....................................................................................................................................... 13
Background in mobile systems and cellular communication standard .......................................................... 13
2.1 Introduction ............................................................................................................................................ 13
2.2 CDMA systems and WCDMA specification ......................................................................................... 13
CHAPTER 3....................................................................................................................................... 16
Background to the RF Power Amplifiers ......................................................................................................... 16
3.1 Power Amplifier fundamental ............................................................................................................. 16
3.1.1 Introduction......................................................................................................................................... 16
3.1.2 Pout [2].............................................................................................................................................. 16
3.1.3 Gain and Efficiency [4] .................................................................................................................... 17
3.1.4 Peak Output Power and Peak to Average Ratio [4] .................................................................... 17
3.2 Class A, AB and B Power Amplifiers ................................................................................................... 18
3.3 Nonlinearity of Power Amplifiers ....................................................................................................... 20
3.3.1 Introduction ..................................................................................................................................... 20
3.3.2 Gain Compression, 1-dB Compression Point, Harmonics, and Intermodulation [3] ............. 20
3.3.3 AM-AM, AM-PM ................................................................................................................................. 23
3.3.4 Adjacent Channel Power Ratio and Alternate Channel Power Ratio ...................................... 23
3.3.4 Error Vector Magnitude ................................................................................................................. 24
3.3.5 Volterra series model...................................................................................................................... 25
3.4 Linearization techniques...................................................................................................................... 25
3.4.1 Feed forward .................................................................................................................................... 25
3.4.2 Feedback ........................................................................................................................................... 26
3.4.3 Pre-distortion................................................................................................................................... 27
3.5 Device technologies (GaAs HBT, SiGe bip, CMOS, BiCMOS) ............................................................ 28
CHAPTER 4....................................................................................................................................... 30
Effect of biasing on PA nonlinearities ............................................................................................................... 30
4.1 Introduction ............................................................................................................................................ 30
4.2 Resistive bias circuit ............................................................................................................................. 30
4.3 Voltage source biasing .......................................................................................................................... 31
4.4 Current source biasing.......................................................................................................................... 32
4.5 Current mirror biasing circuit............................................................................................................. 33
4.6. Comparison of biasing circuit model ................................................................................................ 35
4.7 Modulated signal source measurement ............................................................................................ 37
CHAPTER 5......................................................................................................................................... 2
General Issues in Analog Pre- Distortions .......................................................................................................... 2
5.1 Introduction .............................................................................................................................................. 2
5.2 Static APD .................................................................................................................................................. 2
5.2.1 PD1 ...................................................................................................................................................... 3
5.2.2 PD2 ...................................................................................................................................................... 8
5.2.3 PD3 .................................................................................................................................................... 12
5.3 Dynamic APDs ........................................................................................................................................ 17
5.3.1. Introduction .................................................................................................................................... 17
5.3.2 IMD cancelation technique ............................................................................................................ 17
CHAPTER 6....................................................................................................................................... 21
Conclusion ........................................................................................................................................................... 21
10
Chapter 1
Introduction
1.1 Overview
PA is a key component in transmitter part and it is the most power hungry device in the whole
transceiver. There is always trade-off between the linearity and power efficiency in amplifier design.
In recent RF applications with utilizing recent communication standards that imposes stringent
requirements in linearity, the linearity of power amplifier become an interesting topic to study and
investigate. In power amplifier design bias circuit play an important role in power efficiency and
linearity. Beside the DC bias current and voltage points defined the class of PA, propose well
designed bias circuit, without utilizing linearization techniques can improve the linearity.
The main nonlinearity sources are amplitude distortion (AM-AM distortion) and phase distortion
(AM-PM distortion). There are some metrics to quantify the nonlinearity level in PA application e.g.
P1dB and ACPR.
Some well-known methods to cancel out the nonlinearity in PA are presented so far. Analog predistortion is one of the linearization methods that are well-known in power amplifier applications
because it doesn’t consume so much die area and circuit simplicity. Two types of analog pre-distorters
are introduced in recent researches and publications, named as static and dynamic pre distorters.
The static pre distorters are going to cancel out the static gain (AM-AM) and phase (AM-PM)
distortions, caused by single tone measurements. Utilizing recent wireless standard using non-constant
modulation schemes causes dynamic AM-AM and AM-PM distortions. Conventional analog predistorters don’t contain accurate model of the nonlinearity in wideband applications because they do
not consider the dependence of the nonlinearity on the modulation bandwidth [1].
Two tone test measurements with the tone spacing sweep introduced to characterize the dependency
of nonlinearity on the modulation bandwidth accurately and consequently the dynamic pre distorters
are introduced which cancel out the dynamic AM-AM and AM-PM distortions obtained by modulated
input signal measurements [1].
1.2 Objectives
The main purpose of this master thesis is to find the solution that is able to cancel out the nonlinearity
of power amplifiers using the best bias circuit design or using analog pre distortion methods. In
primary chapters the wireless standard that we have used in this work and the main power amplifier
characteristics, nonlinearity sources in RFPA applications and some general methods to cancel out the
nonlinearity in PAs have been studied. Subsequently different types of biasing circuits and their
effects on power amplifier nonlinearity investigated and simulation results with single tone and
modulated input signal measurements are presented. In later chapters different methods and circuitry
in analog pre-distortion proposed and investigated. P1dB, gain distortions, phase distortions with single
tone measurements and ACPR1 (Adjacent Chanel Power Ratio) and ACPR2 (Alternate Chanel Power
Ratio) obtained by modulated input signal measurements, for each model a designed circuit presented
and discussed. There are so many researches in nonlinearity and efficiency concept of GaAs based
PAs and not in Si based devices. As SiGe is attractive in RF Electronics application because of the
low-cost and compatibility with CMOS technology [2] all simulation results in this work have been
done in SiGe bipolar technology. Therefore the main goals of this research can be highlighted as:
_
Study the basic fundamental and nonlinearity sources in power amplifiers.
_
Study the different biasing circuits and their effect on nonlinearity of PAs.
_
Study and discuss different static analog pre distorters designed.
_
Introduce the dynamic distortion sources in power amplifiers and main idea to cancel them.
1.3 Thesis organization
Chapter 1- Introduction: presents a brief introduction and overview of general requirements for
linearity in today’s communication systems. It also shortly mentions the objectives of the Thesis and
also includes some of the terms and acronyms used in the rest of this thesis.
Chapter 2- Background in Mobile systems and cellular communication standard: Short introduction
to CDMA and WCDMA wireless standard is presented.
Chapter 3- Background to The RF Power Amplifiers: an overview in RF power amplifier
fundamentals is studied. Define the nonlinearity and metrics to measure the linearity of RF power
amplifiers. Subsequently short summery of linearization techniques in RFPA is investigated. Some
device technologies and its characteristics in each application have been studied.
Chapter 4- Effect of biasing on PA nonlinearities: Four different model of biasing the power
amplifier is introduced. The advantages and disadvantages of utilizing each bias model is introduce.
Simulation results of gain and phase distortion of each biasing method with single tone and WCDMA
signal source measurement are presented.
Chapter 5- General issues in Analog Pre Distortions: Investigates the effect of using pre-distortion
block in the power amplifier nonlinearity. Static distortion in power amplifier is defined and some
methods and circuitry of analog static pre-distorters are presented. Subsequently dynamic distortion in
power amplifier applications is defined and analog dynamic pre-distorter is introduced.
Chapter 6- The conclusion of the thesis presented.
12
1.4 List of acronyms
Terms and acronyms used in this thesis are listed as follows:
PA
Power Amplifier
RFPA
Radio Frequency Power Amplifier
CMOS
Complementary Metal-Oxide Semiconductor
ACPR
Adjacent Channel Power Ratio
APD
Analog Pre Distortion
DPD
Digital Pre Distortion
CDMA
Code Division Multiple Access
WCDMA Wide band Code Division Multiple Access
SiGe
Silicon Germanium
Si
Silicon
GaAs
Gallium Arsenide
BiCMOS
Bipolar Complementary Metal-Oxide Semiconductor
P1dB
1 dB Compression Point
IMD
Inter Modulation Distortion
UMTS
Universal Mobile Telecommunications System
Chapter 2
Background in mobile systems and cellular communication
standard
2.1 Introduction
Wireless communication standard is wide and tricky area in recent RF application and most of the
component’s specifications is affected in transmitter and receiver by the specification of modulation
formats and wireless standards. In this chapter CDMA and W-CDMA standards in which it is used in
our study in this master thesis are reviewed briefly.
2.2 CDMA systems and WCDMA specification
CDMA stands for code division multiple accesses. In CDMA systems each bit of the baseband data is
translated to a code word known by the transmitter/receiver pair. Each user uses a unique code and
codes are orthogonal to each other. Two different types of CDMA systems are direct sequence (DS)
CDMA and frequency hopping (FH) CDMA. As shown in the figure below in DS-CDMA baseband
data is multiplied (Encoded) by a pseudo noise (PN) code to encode the data.
Figure 2.1 Encoding in DS-CDMA [3].
In CDMA systems all users are sharing the same channel and same spectrum because the massages
(narrow band data) from different users are spread through the whole channel bandwidth.
14
Figure 2.2 Overlapping spectra in CDMA system [3].
In the receiver part, spread signal is multiplied by the same code and hence dispread and its
bandwidth are changing to the narrow bandwidth of the baseband data.
Figure 2.3 Dispreading operations in CDMA [3].
As depicted in the Figure 2.3 the signal from another user is remain spread after it is multiplied by the
code so the spread signal from other users is acting like a noise floor for the wanted signal. If the
number of users in the network or their transmitted power level increase, the level of the noise floor
for the wanted signal in the receiver raise as well.
Figure 2.4 Effect of strong interferer in CDMA [3].
As can be seen in the Figure 2.4 a strong interfere can make problems receiving the desired signal
even after the dispreading so there will be a need for transmission power control. CDMA systems are
utilizing a power control system to control the output power of each transmitter in the network. In this
way each transmitter only allows to transmit at a power level necessary for the base station to receive
and not more.
Third generation mobile communication systems referred to as UMTS (Universal Mobile
Telecommunications System). WCDMA (Wideband CDMA) is the one of three air interfaces for
UMTS. WCDMA system can provide Bit rates up to 2Mbps and also it has the capability of providing
Variable bit rate to offer BW on demand.
16
Chapter 3
Background to the RF Power Amplifiers
3.1 Power Amplifier fundamental
3.1.1 Introduction
In this section we present an overview of RF power amplifier fundamentals. RF power amplifiers are
used in transmitter side to drive the power to antenna. In transmitter part after passing through
modulator and other blocks the signal is too low for radio transmission so it is amplified by the PA.
As we can guess, the previous small -signal amplifier we have studied so far can’t drive enough
output power. Also there are tradeoffs among linearity, power gain, output power and efficiency. In
this chapter after introducing some performance parameters of PA, like output power ,efficiency,
PAPR, and EVM, class A,B and AB PAs are studied ,and after-wards we will discuss the PA nonlinearity and some canceling methods. Eventually, some device technologies are described.
3.1.2 Pout [2]
Consider the basic circuit in figure 3.1 [4] which shows a PA, connected to the Load that normally is
an antenna. The output power is defined as the active power delivered to the load (antenna) at the
fundamental frequency. Here we consider the antenna as a purely resistive load having a value of 50Ω
and represent the antenna as a resistor RL usually having a value of 50Ω. However, the load
impedance can be transformed to have a higher or lower value with an imaginary part by the matching
network, but here we assume that the load impedance can be represented by a resistance.
Figure 3.1 Basic PA circuit.
Based on our circuit in figure 3.1, we can define the instantaneous output power at any particular
moment as
(3.1) with an average power of
(3.2). The PA also generates power at
frequencies other than the intended one, but these are neglected at the moment. We define
as the sinusoidal amplitude A of the signal at the fundamental frequency and
corresponding rms value. So, the power generated at the fundamental frequency will be called
(t)
as the
.
(3.1)
(3.2)
(3.3)
3.1.3 Gain and Efficiency [4]
Considering figure 3.1 again, we define the input RF power that derive the amplifier, we can define
the power gain of PA (G) as the ratio between the output power (
) and the input power ( ),
which is expressed in logarithmic scale by (3.4)
(3.4)
Another important characteristic of PA is the efficiency. As we know it has a direct impact on battery
life in portable devices and it is the matter of energy cost at base stations. The efficiency of power
amplifiers is defined by two metrics. The drain (or collector) efficiency, ŋ, that is the ratio of the
power delivered to the load (
) (at the fundamental frequency) and the supply power (
)
(3.5). Also the power-added efficiency (PAE) is equal to the difference between the input and output
powers divided by the power drawn from the supply (3.6). As easily can be seen if we have relative
large power gain, drain (or collector) efficiency is equal to PAE.
(3.5)
(3.6)
3.1.4 Peak Output Power and Peak to Average Ratio [4]
Considering the development of modulation schemes utilizing both amplitude and phase modulation,
i.e. WCDMA we need to introduce a new metric in PA efficiency named Peak to Average Power
) and Peak
Ratio (PAPR).For a signal with envelope profile A(t), the average output power (
Envelope Power (PEP) can be defined according to (3.7) and (3.8), respectively.
18
(3.7)
(3.8)
The PAPR is defined as the ratio of the average output power to the peak output power (3.9), and is
usually expressed in dB.
(3.9)
Signals with high PAPR are especially troublesome to transmit from an efficiency point of view, as
the signal requires significant signal headroom such that the peak envelope amplitudes are transmitted
without being distorted too much. Therefore, in conventional transconductance amplifiers there is
significant power dissipation as the transistor output stage is biased to handle the large power peaks,
even though the output power for most of the time is relatively low compared to the peak power level.
3.2 Class A, AB and B Power Amplifiers
There are four types of power amplifiers, distinguished primarily by bias conditions, which termed as
“classic” because of their historical precedence [5]. They are introduced as class A, B, AB and C.
Here we are going to study class A, B and AB.
Considering the basic circuit in figure 3.2 [3] RL represents the load of the amplifier and M1 can also
be replaced by a bipolar version.
Figure 3.2 Basic amplifier circuit including matching network [3].
Class A PA is the most linear model of power amplifier. In class A amplifier the transistor is biased so
that it operates linearly and it is on for the entire cycle, which means a conduction angle of 360
degrees. The conduction angle is defined as the portion of the input signal during which the transistor
conducts [4].
Figure 3.3 Drain (collector) current and voltage in class A amplifier.
Figure 3.3 shows the drain (collector) voltage and current waveforms in a class A amplifier with the
assumption of a linear relationship.
To calculate the maximum efficiency of class A amplifiers, we note that (1) if the drain (collector)
voltage is a sinusoid having a peak-to-peak voltage approximately 2Vdd , then the power delivered to
the matching network is equal to Vdd2 /2Rin , and (2) for Vx to reach 2Vdd , the RFC must provide a
current of Vdd/Rin . Since the RFC current is relatively constant, the power drawn from the supply
equals Vdd2 /Rin.
Thus, the maximum efficiency is equal to 50 % [3].
It is notable that the 50 % efficiency for the class A power amplifier is by assuming no losses in
matching network, no amplitude modulation, and that the maximum voltage swing and the efficiency
in real design is less than 40 %.
Class B PAs are more efficient and less linear than class A PAs. In class B PAs the transistor is on
just for half cycle and it is off for the other half so the conduction angle for class B PAs is 180
degrees.
As it is calculated in so many text books [1], [3], the maximum efficiency in class B PAs is equal to
π/4 ≈79 %.
Figure 3.4 shows the drain (collector) voltage and current waveform in a class B amplifier.
20
Figure 3.4 Drain (collector) current and voltage in class B amplifier.
Additionally, in class B PAs, (in theory), the gain is reduced by 6dB compared to Class-A [5].
Therefore many practical PAs are a mix of Class-A and Class-B amplifiers, named as class AB PAs,
with a conduction angle between 180 and 360 degrees, and an acceptable trade-off between linearity,
gain, and efficiency [4].
3.3 Nonlinearity of Power Amplifiers
3.3.1 Introduction
Utilizing spectrally efficient modulation formats in recent wireless communication systems and tradeoffs between the nonlinearity, makes nonlinearity an important concept in PA researches. In this
section after introducing fundamental nonlinearity concepts, some metrics to quantify the nonlinearity
level like P1dB (1-dB compression point), ACPR (Adjacent Chanel Power Ratio) and EVM (Error
Vector Magnitude) will be presented. Finally Volterra series model will be introduced to model the
nonlinearity in PAs.
3.3.2 Gain Compression, 1-dB Compression Point, Harmonics, and
Intermodulation [3]
Harmonic distortions are usually modeled by a polynomial model. If a sinusoid signal (X (t)) with
frequency ω and phase Φ (t) is applied to a nonlinear system, the output (Y (t)) consists of frequency
components that are integer multiples of the input frequency. Here we will show just up to 5th order
harmonics and neglect higher orders for simplicity (3.10)
(3.10)
We can also neglect the even order harmonic (and DC component) as they can’t cause in band
distortion if we consider the amplifier fully differential.
(3.11)
(3.12)
(3.13)
So as can be seen by (3.13), we have some components at 3ω and 5ω.
Now if we consider the amplitude of fundamental frequency, the gain is
(3.14)
With negligible harmonic distortion the gain would be equal to a1, but with increasing the input
amplitude (power), the roll of harmonic distortions will become more obvious and observe that when
a3<0, it can be easily seen that by increasing the input amplitude (power) the gain will be decreased.
This effect will be quantified by 1-dB compression point metric. The point that the gain decreases by
1dB from the ideal gain (small-signal gain) is called 1-dB compression point (P1dB ).
Another nonlinearity metric that is commonly used in Two Tone Test is the Intermediation concept.
Consider we apply two sinusoid signals at w1 and w2 frequencies, at the same time to the PA. The
output will be like
(3.15)
Then we will have some components in the Y (t) which are not harmonics of input frequencies (we
will consider up to the third order),
(3.16)
After expanding (1.16) we obtain the following intermodulation products [3]
(3.17)
22
(3.18)
(3.19)
As can be guessed intuitively some produced components are not in our band of interest and can
easily be removed by filters, but when the difference between tow tone is small, The two terms (which
we call them intermodulation products) at 2w1-w2 and 2w2-w1 Called “Third order nonlinearity” or
“Third order Intermodulation” as illustrated in figure 3.5 [3] will appear in vicinity of w1 and
w2.Assuming A1=A2=A then the amplitude (power) of intermodulation products can be increase by
degree of 3(A3) when we increase the amplitude (power) of input signal.
Figure 3.5 third order Intermodulation products [3].
Intermodulation is the most troublesome nonlinearity effect in RF amplifiers, as some of the IM
products fall in band of interest and corrupt the desired signal especially in recent wireless schemes
with multi carrier source formats i.e. WCDMA. There is a performance metric to quantify this effect
in RF systems, called “Third intercept point” abbreviated as IP3. For sufficiently small amplitude(A)
of fundamentals ,that the higher order nonlinear products are negligible and the gain is relatively
constant and equal to a1 , the fundamental increases in proportion to A, whereas the third order IM
terms increase in proportion to A3[3]. As illustrated in figure 3.6 [4], when this plotted in logarithmic
scale and extrapolated two lines, there is an intercept point measured as IIP3 (Input Third Intercept
Point) for the input and OIP3 (Output Third Intercept Point) for the output.
Figure 3.6 Third Intercept Point product [4].
3.3.3 AM-AM, AM-PM
Two other metrics to describe nonlinearity in PA are AM-AM and AM-PM effects.
Most of the discussions so far have been about amplitude nonlinearity, that is the nonlinear
relationship between input power and output power called AM-AM effect as it is a conversion
between the amplitude modulation present on the input signal(s) and the modified amplitude
modulation present on the output signal [6]. Another translation of this effect is gain compression.
In study of any power amplifier strongly nonlinear effects we must consider phase distortion as well
as amplitude distortion, known as AM-PM conversion which is a conversion from amplitude
modulation on the input signal to phase modulation on the output signal [6].
Consider a sinusoid modulated input signal, in real amplifiers with strong nonlinear effects ,
amplitude modulation present on the input signal will cause phase modulation on output signal,
resulting in IMD products similar to those from an amplitude nonlinearity.
So in total a power amplifier suffers from both amplitude distortion (AM-AM) and phase distortion
(AM-PM) effects together.
3.3.4 Adjacent Channel Power Ratio and Alternate Channel Power Ratio
Based on the definition of the channel in communication systems, as a power of the signal in a
specific frequency bandwidth around the fundamental frequency, so the falling power of the signal out
of the frequency bandwidth can be consider as a distortion for other channels in the neighborhood. To
quantify this term of nonlinear effect, Adjacent Channel Power Ratio (ACPR) is presented. ACPR is
defined as the ratio of power in a bandwidth away from the main signal (the distortion product) to the
power in a bandwidth within the main signal [4].
The distance away from the main channel (signal) can be defined based on the application and
linearity requirements.
24
Also there is another metric in RF systems called Alternate Channel Power Ratio that is similar to the
definition of the ACPR, and is defined as the ratio of the power falling in a bandwidth farther from the
main signal, to the power in a bandwidth within the main signal.
Figure 3.7 Alternate and Adjacent channel products.
3.3.4 Error Vector Magnitude
Another signal quality measure is the Error Vector Magnitude (EVM), which is computed on I and Q
data (or amplitude and phase) measured around the carrier. In I versus Q plane (figure 3.8 [4]) each
location encodes a specific data symbol, which has a certain number of bits depending on the
complexity of the modulation schemes used. At any point in the time magnitude and phase of the
signal can be measured and mapped towards an ideal reference signal based on transmitted data
stream, clock timing, filtering parameters, etc. [4]. The difference between the measured signal and
the ideal reference signal creates the error vector, and is generally defined as the rms value of the error
vector over time [4].
The EVM measure provides the direct measure of the signal quality and the transmitter and
receiver/demodulation accuracy, and the result captures several signal impairments like AM-AM
distortion, AM-PM distortion, phase noise, and random noise [4].
Figure 3.8 Error Vector Magnitude measurements.
3.3.5 Volterra series model
Volterra series is a mathematical representation to model the nonlinear systems. The Volterra series
model is a generalization of the convolution description of linear systems.
(1.21)
The nth order kernel of (1.21), hn (τ1, τ2… τn), can be called a nonlinear impulse response of order n.
3.4 Linearization techniques
Recall from previous discussions, that nonlinearity in power amplifier becomes a critical issue due to
the need for more linear amplifier in recent wireless communication standards, with band efficient
modulation formats, especially with multi carrier systems ,i.e. WCDMA, for example in base station
transmitters. Using class A amplifiers to have more linearity is not preferable as the efficiency is low,
especially in portable devices, for which, battery life time is one of the figure s of merit. Instead, a
more nonlinear power amplifier can be used utilizing the linearization technique. Linearization allows
an amplifier to produce more output power and operate at a higher level of efficiency for a given level
of distortion [7]. There are many different ways to linearize the power amplifier and each method has
advantages and disadvantages. Feed-forwards, Feedback, and pre-distortion are the most common
forms of linearization [7]. Besides, there are more techniques that are not more common but have
been studied and introduced. In the following we will introduced the Feed-forwards, Feedback, and
pre-distortion techniques.
3.4.1 Feed forward
A nonlinear power amplifier generates an output voltage waveform that can be viewed as the sum of
linear replica of the input signal and an error signal. A feed forward topology computes this error and,
with proper scaling, subtracts it from the output waveform. Shown in figure (3.9.a) [3] is a simple
example where the output of the main PA, VM, is scaled by 1/Av, generating VN.
The input is subtracted from VN, and the result is scaled by Av and subtracted from VM . We note
which if VM=AvVin+VD , where VD represents the distortion content, VN=Vin+VD/Av , yielding
VP=VD/AV,VQ=VD , and hence Vout=AvVin .
In practice, the two amplifiers in the circuit exhibit substantial phase shift at high frequencies,
mandating the use of delay lines (figure (3.9.b) [3]) such that ∆1 compensates for the phase shift of the
PA and ∆2 for that of the error amplifier [3].
26
Figure 3.9 (a) feed forward technique [3].
Figure 3.9 (b) feed forward technique applying delay line [3].
The feed-forward technique allows the use of WCDMA base-station PAs reducing the ACPR on the
order of 20 dB [1]. However, this technique suffers from the bulky size and highly dissipated power
of the control circuits, which prevent its use in handset applications [7].
3.4.2 Feedback
The two issues related to feedback, insufficient gain and excessive phase shift can be alleviated if
most of the loop gain is obtained at low frequencies. In a transmitter, this is possible because the
waveform processed by the PA in fact originates from up-converting a baseband or IF signal. Thus, if
the PA output is down-converted, it can be compared with the original low frequency signal in a
negative-feedback loop. The concept is illustrated in figure (3.10) [3], where the frequency translation
required between the error amplifier A1 and the PA is performed by the two mixers.
Figure 3.10 feedback technique [3].
However due to the complicated implementation and also the need for extra power, this technique is
not suitable for wideband applications such as WCDMA due to problems with Instability [7].
3.4.3 Pre-distortion
Pre-distortion technique is mostly common because of its relative simplicity, small size and low cost,
mostly used in handset devices. The general idea in PD linearizers is to add a box to the amplifier
which add new nonlinearity, in opposite of the nonlinearity of amplifier to cancel it, however there are
number of drawbacks and few advantages. Most complementary distortion systems are based around
pre distorting the input signal and fall into one of the following categories [6]:
1-RF Pre-distortion—the nonlinear pre distorting element operates at the final carrier frequency.
2-IF Pre-distortion—the pre-distorting element or network operates at the convenient intermediate
frequency, thereby possibly allowing the same design to be utilized for a number of different carrier
frequencies. Alternatively, the required pre distortion components may not operate satisfactorily at the
desired carrier frequency, hence necessitating the use of an IF.
3-Baseband Pre distortion--Prior to the advantage of digital signal processors (DSP) devices, this
technique had few advantages over the IF technique. It is now, however, a useful tool. In this case, the
pre-distortion characteristic is typically stored as a table of gain and phase weighting values within a
DSP (usually in quadrature form)in order to pre-distort the baseband information before up
conversion. It is possible to use feedback to provide updating information for these coefficients,
which is called adaptive baseband pre-distortion technique.
Fundamentally APDs (Analog Pre Distortion) utilize at RF and IF technique and DPDs (Digital Pre
Distortion) use in baseband technique.
The basic circuit of a pre-distortion linearization form is shown in figure (3.10.a). The pre-distorting
function β (Vi) operates on the input signal in such a way that its output signal is distorted in a
precisely complementary manner to the distortion produced by the RF PA, F (α). The output signal is
therefore an amplified, but undistorted replica of the input signal (figure (3.10.b)) [6]
28
Figure 3.10.a Pre-distortion technique block diagram [4].
Figure 3.10.b Pre-distorter compensation [6].
3.5 Device technologies (GaAs HBT, SiGe bip, CMOS, BiCMOS)
There are various devices and technology choice in recent RF electronics application circuits, which
each one can be made with considering the component applications like power related concept,
characteristic frequencies and noise behavior.
Beside the higher current drive per unit area and higher gain characteristics that is mentioned for
bipolar transistors, generally better noise performance, better high frequency characteristics and better
analogue capability are other benefits that are introduced for bipolar transistor compare to CMOS
technology based transistors.
For RFPA applications the state of the art device should contain a good linearity and efficiency at low
voltage operation illustrated as PAE and the technology most have high enough breakdown voltage to
be able to with stand high VSWRs .
Several device technologies are introduced so far in bipolar based technologies like AlGaAs/GaAs
HBT, Si/SiGe and InGaAs/InP. Among these technologies AlGaAs/GaAs HBTs are widely used in
wireless power amplifier applications due to theirs higher breakdown voltages (relative to integration
level). SiGe processes, in the other hand, have the advantages of lower cost compare to GaAs
[8].However CMOS PA, besides some disadvantages such as low breakdown voltage, lower device
gain and linearity issue is still an interesting topic in PA research due to their lower cost and the
ability to be integrated on chip.
In our design we used SiGe bipolar transistor to find if we can get the same characteristics as GaAs
process that has better performance in PA linearity and efficiency but has higher cost.
30
Chapter 4
Effect of biasing on PA nonlinearities
4.1 Introduction
As we discussed previously operation bias point of PA is an important point in design of power
amplifiers. Beside the DC operation point which will define the operating class of PA, the bias
circuitry is also has some important roles regarding linearity and maximum output power. Recall from
the previous discussions that, the SiGe bipolar transistor has been selected due to offering high
efficiency and high power density at a low operation voltage [9] and that the PA designed as a simple
1-stage, common source power amplifier consisting of 72 elementary transistor cells in IBM process
that for each case, was matched for maximum output power and reasonable gain, operating at 2140
MHz fundamental frequency. In simulation work we have used Golden Gate in cadence simulator. As
we will discuss in later parts, the main goal of basic diode base analog pre-distortions is to keep the
base voltage of the RF power amplifier transistor constant in the large signal region. In the following
section, we will study this concept utilizing four different bias methods and circuitry, and the
advantages and disadvantages of each model will be discussed both with single tone source and
WCDMA source. Finally the optimum biasing circuit for our work will be selected.
4.2 Resistive bias circuit
Figure 4.1 is shown, basic common source, 1-stage RFPA, resistor biased (passive) as a class AB
power amplifier with Ic equal to 270 mA.
Figure 4.1 1-stage Common Source RFPA.
The circuit consists of input matching (Lin and Cin ) and output matching ( Lout and Cout). The RFC is a
big inductor that is open circuit in AC analysis, and is utilized to isolate the RF signal from Vcc .
In this circuit as the input power increases, the dc base-emitter voltage (Vbe ) decreases (figure 4.2),
causes amplitude gain compression. The explanation can be given as, when the input power increases
the dc base current of the RF transistor (Q1) increases, and the bias resistor (Rbias) causes a decrease in
the dc voltage across the Q1 base-emitter voltage (Vbe).
Of course there are different models to bias the PA using resistors and we have shown the simplest
model here, but reducing dc Vbe is the main idea in all of them.
As illustrated at the following Vbe varies with the increasing input power very much as Pin =-4 dBm
which is correspond to P1dB it reduced by more than 50 mV, and relatively falling down after this
point.
Figure 4.2 Dependence of base emitter dc voltage (
) of the
on the output power.
This circuit gives the P1dB equal to 26 dBm (Figures 4.9 and 4.10 in sections 4.6), with the gain equal
to 30 dB (Figure 4.9 in section 4.6) and the phase distortion of this model of biasing shows an
expansion behavior (Figure 4.11 in section 4.6).
As intuitively can be guesses and illustrated in figures 4.9 to 4.11, in section 4.6 this passive biasing
circuit shows the worst behavior regarding P1dB, phase distortion and ACPRs, but it is relatively easier
to match in both input and output part and has the highest gain compared to three different models.
To investigate this effect we introduce two different ideal models for biasing the transistor which offer
constant behavior for base emitter voltage (Vbe), named as Ibias and Vbias.
4.3 Voltage source biasing
In the Vbias circuit (Figure 4.3) we connect supply voltage to the base of the RF transistor,, to force
the base to have constant behavior of Vbe.
32
Figure 4.3 PA biased using supply voltage.
As it is definitely expected, this gives the constant behavior for Vbe , and as also expected much better
behavior in case of nonlinearity. However, as previously discussed, the nonlinearity and gain
compression, especially at higher input level, is affected by various factors and conditions, therefore,
still there is gain compression and non-linear behavior in the circuit besides having constant Vbe .
P1dB Improves with applying the Vbias by 6 dBm and equal to 32 dBm (Figures 4.9 and 4.10 in
section 4.6), the gain is equal to 27 dBm (Figure 4.9 in section 4.6), but the phase distortion shows a
decreasing manner but constant in wider range compared with resistor biasing model.
4.4 Current source biasing
Another ideal biasing model is forcing the current of base (instead of constant voltage) to have a
constant value, by applying a current supply to the base of Q1 as illustrated in figure 4.4.
Figure 4.4 Biasing PA, using current source.
In this model of biasing, as illustrated in figure 4.5, the dc base emitter voltage, Vbe, is constant in a
wider range but with increasing input power it starts to drop, and in turn the gain starts to fall. So
compared with previous model, biasing by applying voltage source to the base, the gain starts to fall
from lower input level. The reason of this effect can be explained by considering the small signal
model of an NPN transistor. As it is illustrated in many course books, the small signal model of a
NPN bipolar transistor consists of a small resistor Rb that represents the resistor of the base of
transistor. So as the input power increases, Rb causes an additional voltage which adds current to the
Ibias and again with increasing the current, Vbe decreases.
The value of Rb depends on various factors and parameters of transistor, like width, length and etc. but
since it is generally a small value, it is more significant at higher input level.
However Ibias bias model improves the 1-dB compression point, whereas (as we have seen) the Vbias
shows more linear behavior.
Figure 4.5 Dependence of base emitter dc voltage (Vbe) of the Q1 on the output power.
4.5 Current mirror biasing circuit
Another idea for biasing the RFPA is to utilize current mirror biasing model. Current mirror biasing is
an old and well-known model which is widely used in analog circuit topologies due to its advantages
compared with resistor biasing. Beside its insensitivity of circuit performance to variations in power
supply and temperature [10], current mirror biasing is more economical than resistor biasing because
of consuming less area on die [10].
A common used circuit of current mirror circuit is the cascade circuit. As shown in figure 4.6,
transistors Q2 and Q3 create a simple current mirror, and Q5 acts as the common-base part of the
cascade and transfers the collector current of Q3 to the output. Transistor Q4 is a diode level shifter
[10].
34
Figure 4.6 Cascade current mirror bias circuits.
Looking at the Vbe behavior of this bias model reveals the very similar manner to Vbias model. As is
behaviors of both current mirror and Vbias models are similar even up till
shown in Figure 4.7,
high input level.
Figure 4.7 comparison of
Vbe of the current mirror biasing model with Vbias model.
The reason of the similar Vbe behavior can be explained as: The Q3 base-collector diode is forward
biasing and as the input power increases the rectified dc current of the Q3 increases and the dc voltage
across the Q1 base emitter tends to increase (to compensate for the drop of dc voltage) and makes it
show almost the same behavior as an straight line like Vbias. In fact the diode connected Q3 transistor
act as local feedback, as when the dc voltage drop at the base of Q1 causes the drop dc voltage of the
base of that it causes itself to increases the dc voltage of the collector of Q3 and keep the Vbe of Q1
almost constant.
Because of the similar Vbe manner to that of the Vbias model, most of the characteristics of this two
biasing models are similar.
4.6. Comparison of biasing circuit model
Figure bellow shows the comparison of the four biasing models.
Figure 4.8 Dependency of DC base emitter voltage on Pout with four different biasing models.
Figure 4.8 shows the Vbe variations versus Pin, the early drop at resistor bias model and relatively
constant behavior with current mirror bias design can be easily recognized.
Figures 4.9 and 4.10 show the gain and Pout versus Pin characteristics. As illustrated the resistor bias
model has the higher gain but it starts to decrease from lower input level.
36
Figure 4.9 Gain characteristics of a PA under different biasing model.
Figure 4.10 Pout versus Pin with for different model biasing.
Figure 4.11 shows the comparison of phase distortions between four different biasing models.
Figure 4.11 phase distortion characteristics versus Pout in four different biasing model.
As it is shown in figures 4.8 to 4.11 the behavior of current mirror biasing is very similar to Vbias that
shows the advantages of using this bias model compared to the resistor bias model in terms of
linearity issues.
4.7 Modulated signal source measurement
In the next step, we simulated each biasing circuit model applying WCDMA source signal and studied
the Adjacent Channel Power Ratio, which we call it ACPR1 within 5MHz bandwidth from the
fundamental frequency and Alternate Channel Power Ratio named ACPR2 within 10MHz bandwidth
from fundamental frequency based on the 3GPP standard.
Figures 4.12 and 4.13 reveal that Ibias biasing circuit has the best behavior in case of ACPR1 and the
current mirror biasing has the best manner in terms of ACPR2. However the Ibias and Vbias biasing
circuits are too ideal to be used in real designs, they need to be implemented as circuit so the
performance will be degrade, and as a matter of fact the current mirror biasing model shows relatively
acceptable linearity in terms of both ACPR1 and ACPR2.
38
Figure 4.12 ACPR1 characteristics versus Pout in four different biasing circuits.
Figure 4.13 ACPR2 characteristics versus Pout in four different biasing circuits.
2
Chapter 5
General Issues in Analog Pre- Distortions
5.1 Introduction
As it is discussed previously, the design of a power efficient and linear power amplifier is a
major problem in transmitter researches. Especially, in recent communications systems with
an increase in utilization of non-constant envelope modulation techniques, linearity of power
amplifier becomes an interesting field to research. Some linearization techniques are
introduced so far with some advantages and dis-advantages, like feed-forward, feed-back and
pre-distortion. However, among these techniques along pre- distortion technique is more
renowned regarding its simple circuitry and almost no increase in die area. In the following,
in first section, some schematic design of static analog pre-distortion (linearized static AMAM and AM-PM) using a diode connected transistor is introduced. The pre-distorters
operation principles are studied from a basic model to an advance one. In most recent
communications systems, with using the modulation schemes which put highly demands on
the PA linearity to have low Adjacent Channel Power Ratio and to keep modulation accuracy,
the conventional static APDs cannot illustrate the nonlinearity of PA accurately. This is
because they do not consider the dependency of nonlinearity on the modulation bandwidth
[11]. In the second section in this chapter, Dynamic APDs are investigated using two-tone test
analysis to model the dynamic AM-AM and AM-PM, and the dynamic nonlinearity is
investigated by study on 3rd Inter -Modulation product.
5.2 Static APD
As it is discussed previously, in RFPA design, in order to have a power efficient amplifier, the
dc current of power amplifier is one of the important figures of merits to consider, since it
determines the efficiency of PA [4]. As previously mentioned we choose a bipolar transistor
due to higher gain and lower leakage current, but like any other nonlinear devices it shows
some nonlinear characteristic like gain and phase distortion. Due to the utilization of nonconstant envelope modulation techniques in recent communications standards, the need of
high linear and high efficient RFPAs becomes more evident. So to accomplish high efficiency
with high linearity it is essential to improve the gain compression and phase distortion of the
transistor. Some linearization techniques are introduced so far that have some dis-advantages,
such as not compensating for high dc current consumption, or having an insertion loss of
about 3dB or suffering from bulky size [9].
So to avoid these features, simple analog pre distortion, proposed a linearization technique
using the bipolar base collector junction diode is introduced [9].
Compared with other linearization techniques, the novel linearization technique has important
advantages as:
_
Improvement in both gain compression and phase distortion.
_
Simple in circuitry and design.
_
No additional dc current consumption.
_
And no addition in die area.
Since in non-constant envelope modulation schemes signal is varying continuously in
amplitude and phase, any device nonlinearity can distort the ideal spectrum [4]. As the input
power increases the transconductanc (gm) decreases, causing gain compression (AM-AM
distortion) and variations of Cbc during the device operating cycle over its load line, which in
turn can cause phase distortion (AM-PM distortion) [12]. The nonlinearities of device
parameters are assumed as a weak nonlinearities and not a strong one [12], so it is called as
static nonlinearity (static AM-AM and AM-PM) So the APDs that cancel the static AM-AM
and AM-PM are named as static APDs.
5.2.1 PD1
An effective method to improve the gain compression caused by device nonlinearity is to
increase the base-emitter dc bias voltage, or in fact to keep it constant, (Vbe) slightly as the
input power increase. Consider the basic circuit in figure 5.1 that shows a conventional 1stage common-source PA. In this circuit the PA biased through the resistor connected to the
base of the transistor. In this circuit, with increasing the input power, the rectified dc current
of the Q1 base-emitter diode increases. Respectively, the bias resistor (Rbias) is caused to
decrease of the dc voltage across the Q1 base-emitter voltage (Vbe ).
4
Figure 5.1 circuit topology of conventional common source resistor bias PA.
Figure 5.2 circuit topology of novel diode linearizer.
In order to compensate for the Vbe drop, the circuit topology of the linearizer illustrated in
figure 5.2 [9] is introduced. In this circuit the RF transistor (Q1) is biased with the diode
connected transistor Q2 and bias resistor Rb. In this circuit, with increasing the input power the
rectified dc current of diode connected Q2 increases causing the dc voltage across the diode to
decrease. So the dc voltage across the base-emitter diode of Q1 stays constant or increases
slightly. The capacitor Cb does not have any role in case of the dc analysis but at the presence
of high frequency RF signal, since it is a very small value, it forms an RF short node at the
base node of the active bias diode connected transistor Q2 [13].
To investigate this concept, some circuit topologies are studied in the following in this
chapter.
We applied the linearizer illustrated at figure 5.2, called PD1, on the resistor biased common
source PA that we had been studied in previous chapter (section 4.1). After returning the PA
in case of input and output matching, the following results are achieved, illustrated in figures
5.3 to 5.6.As shown in figure 5.3, the linearizer improves the Vbe versus input power
characteristic as a small value, so not a considerable improvement in case of P1dB , the value
of gain is 28 dB and P1dB is at 27 dBm (figure 5.4) improvement in phase distortion and
ACPR1 (5 MHz from fundamental frequency bandwidth) and ACPR2 (10 MHz from
fundamental frequency bandwidth) with applying WCDMA source signal.
Figure 5.3 Comparison of the Vbe in conventional PA and PA with applying the diode linearizer PD1.
Figure 5.4 Comparison of the gain in conventional PA and PA with applying the novel diode linearizer
PD1.
6
Figure 5.5 Comparison of the out-put power in conventional PA and PA with applying the novel diode
linearizer PD1.
Figure 5.6 Comparison of the phase distortion in conventional PA and PA with applying the novel
diode linearizer PD1.
As figure 5.6 shows, utilizing PD1 improves the phase distortion of PA by 9 degrees at 24
dBm output power (3 dBm back-off from P1dB).
By applying the WCDMA signal source we get the following result in case of ACPR1 (figure
5.7) and ACPR2 (figure 5.8). In case of ACPR1 the linearizer improves it by 1dBm at 14
dBm out-put power that the ACPR1is 45 dBc and higher (this is the defined standard for
WCDMA standard at base station). Also it improves the ACPR2 by 2 dB, however
improvement in ACPR1 is in higher priority.
Figure 5.7Comparison of the ACPR1 in conventional PA and PA with applying the novel diode
linearizer PD1.
Figure 5.8 Comparison of the ACPR2 in conventional PA and PA with applying the novel diode
linearizer PD1.
8
5.2.2 PD2
In the next step to improve the PA linearity characteristics more, we apply the circuit
topology shown in figure 5.9 [14], named as PD2. The fundamental linearization technique is
similar to PD1 and the capacitor Cb and HBT diode connected transistor Q2 form the
linearizer, however in this circuit two diode connected transistors applied to the base of the
linearizer transistor Q2.
Figure 5.9 Circuit topology of PD2.
In this circuit, in presence of RF signal the impedance seen from the resistor and the parallel
connection of two diode connected transistors D1 and D2 is much higher than the impedance
that capacitor Cb sees, so all the RF signal at base node of Q2 crosses through Cb, on the other
hand, some amount of current increasing because of the RF power that change the dc
operating point of the base emitter voltage, leaks to the node V1 that is shorted by capacitor
[14].
Figures 5.10-5.15 show the result of utilizing the PD2.It can be seen from the figures that we
reach a good linearity improvement utilizing PD2. The first figure (5.10) shows the
comparison of the dc base emitter voltage of RF transistor Q1, with and without presence of
the novel linearizer and it shows that the Vbe drop is relatively improved in this circuit. The
Vbe characteristic versus input power shows slightly increasing behavior of Vbe instead of
decreasing.
Figure 5.10 slightly increasing behavior of Vbe using PD2 compare with conventional PA.
Figure 5.11 Gain compression behavior of the linearized and conventional PA.
10
Figure 5.12 Pout Versus Pin characteristic of the linearized PA in compare with conventional PA.
The P1dB point increases with applying PD2 with amount of 4 dB in this case, as can be seen
from figure 5.11 the gain is linear in wider range in comparison with conventional PAs.
Also phase distortion is relatively decreases by applying PD2 (figure 5.13).
Figure 5.13 Comparison of the phase distortion in conventional PA and PA with applying the novel
diode linearizer PD2.
Applying WCDMA signal sources gives following results. As is shown in figures 5.14 and
5.15 this model of PA linearizer is relatively effective in case of ACPR1 and ACPR2 .Using
this linearization improves ACPR1 by 8 dB means from the output value of almost 22 dBm
and lower the ACPR1 of the PA has a value of 45 dBc based on the WCDMA standard in
base station. Also for the ACPR2, a relatively considerable improvement can be seen from the
figure.
Figure 5.14 considerable improvement of the using PD2 in case of ACPR1.
12
Figure 5.15 considerable improvement of the using PD2 in case of ACPR2.
5.2.3 PD3
Some other novel linearizer models also have been introduced so far, by small changes in the
circuit topology of PD2. One of them is using LC resonant circuit instead of using capacitor
Cb in the linearizer named as PD3. As illustrated in figure 5.16 capacitor Cb in series with the
inductor Lb resonates at the fundamental frequency (2140 MHz in our case) and provides the
RF short node for the linearizer circuit [15].
Figure 5.16 linearizer circuits utilizing the LC resonant circuit used for shorting inserted RF signal.
To choose a suitable value for the Lb and Cb following calculation is presented:
(5.1)
Equation 5.1 shows the way of calculating resonant frequency in a series LC circuit, choosing
the value of the Cb equal to 1pF gives the value of the inductor, Lb.
The following figures show the effect of applying PD3 in the PA linearity characteristics.
Figure 5.17 shows that Vbe characteristic improve slightly more with this design linearization
compared with PD2 model. Also P1dB improves slightly more compare with PD2. However, as
illustrated in figure 5.18 ACPR1, the behavior of PA using PD2 shows better linearity
compared with the circuit model of PD3 using series LC resonant circuit. The ACPR2
behavior of PD3 is similar to PD2 (figure 5.19).
Figure 5.17 Comparison of Vbe between the conventional PA and PD2 and PD3.
14
Figure 5.18 Comparison of ACPR1 between the conventional PA and PD2 and PD3.
Figure 5.19 Comparison of ACPR2 between the conventional PA and PD2 and PD3.
In all investigated diode linearizer technique designed circuit, the circuit schematic is consist
of resistor biasing and also capacitor and even inductor in last model introduced utilizing LC
series resonant circuit. This all circuit component, besides consuming relatively more die
area, by using a resistor in biasing circuit, makes it sensitive to variations of power supply and
temperature [10].
Considering the effective design of current mirror biasing model in the next step we are going
to evaluate the current mirror biasing circuit investigated in previous chapter (section 4.4)
with the novel diode linearizer designs in this section.
Figure 5.20 Comparison of the Vbe characteristics of current mirror biasing PA model and linearized
PA using PD2.
As figure 5.20 shows the Vbe behavior of the current mirror biasing PA shows the most
constant manner in comparison with the linearized PA using PD2 and PD1 and because of
that the P1dB in the current mirror biasing model has the highest value of 31 dBm.
16
Figure 5.21 Comparison of the gain characteristics of current mirror biasing PA model and linearized
PA using PD2.
Figure 5.22 Comparison of the ACPR1 characteristics of current mirror biasing PA model and
linearized PA using PD2 and PD3.
Figure 5.23 Comparison of the ACPR2 characteristics of current mirror biasing PA model and
linearized PA using PD2 and PD3.
As can be seen from the figures, PD2 just in case of ACPR1 shows more linear behavior than
the current mirror biasing model of PA. However as current mirror biasing meets the WCMA
standard for the ACPR1 with suitable back-off power from P1dB it is still acceptable
considering the benefits of this biasing circuit and also considers that using current mirror
biasing circuit, in some sense, there is no need to linearizer.
5.3 Dynamic APDs
5.3.1. Introduction
Conventional PDs that we investigated in the previous section, have been designed based on
the cancelation of the static AM-AM and static AM-PM distortions obtained by single tone
source measurements, however single tone measurements doesn’t consider the dependency of
the nonlinearity on the modulation bandwidth in non-constant modulation schemes in recent
communication standards like WCDMA [1]. In this section, some new methods and their
circuit topology based on the cancelation of dynamic AM-AM and AM-PM is introduced and
studied.
5.3.2 IMD cancelation technique
Respectively to face the dynamic AM-AM and AM-PM distortions problems in RFPA, IMD
cancelation technique is introduced. In this technique, As illustrated in figure 5.24 (a), the
idea is to combine two amplifier that one of them has the dynamic compression gain
18
characteristic and other one has dynamic gain expansion nonlinearity that cancel out each
other. In the other hand in this two stage amplifier the first stage (driver stage) produces antiphase intermodulation products that is going to cancel out the in-phase intermodulation
products in PA stage (out-put stage), (figure 5.24.b). In fact there is a relation between the
phase of the IMD and dynamic gain deviation in an amplifier that is shown in table 5.1[8].
Consider the relation between the amplifier input and output expressed by Taylor series
expansion as equation 3.10,
(5.2)
That Y (t) is output voltage and X (t) is input voltage and a0… an are bias dependent
coefficients. For two tone test we have
(5.3)
So for the output we have equation (5.4) at the fundamentals frequency w1 and w2 and for the
3rd intermodulation products at 2w1-w2 and 2w1-w2 we have 5.5 respectively
Figure 5.24[3] Principle of IMD cancelation technique: (a) principle of combining two amplifier with
reverse AM-AM and AM-PM characteristic. (b) Pre distortion based on the IMD cancelation
technique.
With considering A1=A2=A we have
(5.4)
(5.5)
In equation 5.4 a1A is the linear amplitude and other terms represent the nonlinear gain
distortion (expansion or compression) of the amplifier. As intuitively someone can guess, if
the amplifier shows gain expansion so the sign of a3/a1 and a5/a1 must be positive it means that
the carrier frequency and IMD (3rd and 5rd order) are in phase.
Respectively when the amplifier shows gain compression characteristic, we have negative
values for a3/a1 and a5/a1 , consequently the fundamental frequency and IMD are in different
phase [16].The conclusion represented in table 5.1 [16].
Table 5.1 Gain deviation (AM-AM distortion) and IMD phase relation.
Some circuit topology are introduced recently utilizing a GaAs HBT device to produce IMD
products that they cancel out the dynamic AM-AM and AM-PM produced by modulatedinput signal measurements to characterize the dependency of nonlinearity on modulation band
width. One of the introduced techniques that are used to model the Dynamic AM-AM and
AM-PM is the IMD products in which it is produced using two-tone test measurements with
tone spacing sweep.
20
Some new ideas are introduced in some paper [1] like introducing a broadband dynamic
analog pre distortion, using tow-tone test measurements with tone spacing around 4 MHz
which is accurate for 97 % of the WCDMA bandwidth of 3.84 MHZ [1]. In the other hand
they have shown that using a two-tone signal with 4 MHz spacing actually approximates a
simulation with a full WCDMA signal very well, since most of the carriers in the WCDMA
are separated by 1 MHz or more.
We have tried to implement their concept for APD and to set up the same simulation as they
did using the same SiGe bipolar devices as in the previous simulations of biasing in the
previous sections, but in fact it was very tricky to find correct component values to reproduce
their findings, and we did not succeed in proving if it would work for our circuits within the
scope of this thesis work.
Chapter 6
Conclusion
In this master thesis we have studied different biasing circuit to investigate the effect of
biasing circuit in Power Amplifier linearity and we have shown that choosing suitable biasing
circuit have important role in linearity of power amplifier. We presented current mirror
biasing circuit that has a high 1dB compression point and good improvement with modulated
input signal source measurement in case of ACPR1 and ACPR2 without using any linearizer
compare with resistor biasing circuit. In the next step we studied static and dynamic AM-AM
and AM-PM distortion in power amplifier and introduced diode based analog pre-distorters to
cancel out the static gain (AM-AM) and phase (AM-PM) distortion in power amplifier. We
investigated the effect of diode based APDs with three different APD circuit and compare the
result with the current mirror biasing circuit.
Current mirror biasing circuit shows higher 1dB compression point compare with using APD
with resistor bias amplifier but instead using APD gives higher improvement in case of
ACPR1.
Compared with other linearization techniques, the diode based APD linearization technique
has important advantages as:
_
Improvement in both gain compression and phase distortion.
_
Simple in circuitry and design.
_
No additional dc current consumption.
_
And almost no addition in die area.
In the next step we introduced dynamic amplitude and phase distortion in power amplifier and
try to investigate the efforts have been done so far using GaAs device technology, utilizing
SiGe device technology in our case, with two tone test measurement with sweeping the tone
spacing. However it is really hard or almost impossible to reach the same linearity
improvement with SiGe in compare with GaAs.
22
References:
[1] S. Yamanouchi, Y. Aoki, K. Kunihiro, T. Hirayama, T. Miyazaki, H. Hida, (2007).
Analysis and Design of a Dynamic Predistorter for WCDMA Handset Power Amplifiers.
IEEE Trans. Microw. Theory Tech, Vol. 55, No. 3, pp. 493-503.
[2] R. Singh, D.L. Harame, and M.M. Oprysko, “Silicon germanium:Technology, modeling,
and design”, Wiley, Hoboken, NJ, 2004.
[3] B. Razavi, ” RF Microelectronics”, Prentice Hall Inc., Upper Saddle River, NJ,
2008.
[4] J. Fritzin, “Power Amplifier Circuits in CMOS Technologies”, Ph.D. Thesis, Technical
Report No. LiU-TEK-LIC-2009:22, Linköping University, 2009.
[5] T.H. Lee, “The Design of CMOS Radio-Frequency Integrated Circuits”,
Second Edition, Cambridge University Press, 2004.
[6] P.B. Kenington, “High-Linearity RF Amplifier Design“, Norwood, MA, USA,
Artech House, 2000.
[7] A. Katz, (2001, December). Linearization: Reducing Distortion in Power Amplifiers.
IEEE Microw. Mag., pp. 37-49.
[8] O. Esame, Y. Gurbuz ve I. Tekin, (2004). Performance Comparison of State-of-the-Art
Heterojunction Bipolar Devices (HBT) based on AlGaAs/GaAs, Si/SiGe and InGaAs/InP.
Microelectronics Journal, pp. 901-908.
[9] T. Yoshimasu, M. Akagi, N. Tanba, and S. Hara, “An HBT MMIC Power Amplifierwith
an Integrated Diode Linearizer for Low-Voltage Portable Phone Applications”, IEEE J SolidState Circuits, VOL. 33, NO.9, 1998.
[10] P. R. Gray and R. G. Meyer, “Analysis and Design of Analog Integrated Circuits”, 4th
Edition, Wiley, 2001.
[11] S. Cripps, “RF Power Amplifiers for Wireless Communications”, Second Edition, Artech
House, MA, USA, 2006.
[12] T. Yoshimasu, “High power AlGaAs/GaAs HBT’s and their application to mobile
communications systems”, IEEE IEDM Tech. Dig., 1995, pp. 787–790.
[13] Haiwen Liu, Toshihiko Yoshimasu, Satoshi Kurachi, Nobuyuki Ito, and Koji
Yonemura,(2006). A Novel Diode Linearizer For SiGe HBT Power Amplifier. Microwave
and Optical Technology Letters, 48(8): 1535-1537.
[14] Youn Sub Noh, Chul Soon Park, (September, 2002). PCS/W-CDMA dual band MMIC
power amplifier with a newly proposed linearizing bias circuit. IEEE Journal of Solid State
Circuits, vol. 37, issue 9, pp. 1096-1099.
[15] J. H. Kim, K. Y. Kim, C. S. Park, (2008, May). Linearity Improvement of a Power
Amplifier Using a Series LC Resonant Circuit. IEEE Microwave and Wireless Components
letter, vol. 18, no. 5.
[16] Y. Aoki, K. Kunihiro, T. Miyazaki, T. Hirayama, and H. Hida, “A 20-mA Quiescent
Current Two-Stage W-CDMA Power Amplifier Using Anti-Phase Intermodulation
Distortion” IEEE RFIC Symp. Dig. , pp.357-360, June 2004.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertising