RF Power Amplifier Design

Markus Mayer & Holger Arthaber

Department of Electrical Measurements and Circuit Design
Vienna University of Technology

June 11, 2001

Contents

~ Basic Amplifier Concepts

z Class A, B, C, F, hHCA
z Linearity Aspects
z Amplifier Example

~ Enhanced Amplifier Concepts

z Feedback, Feedforward, ...
z Predistortion
z LINC, Doherty, EER, ...

2
Efficiency Definitions

POUT
~ Drain Efficiency: ηD =
PDC

~ Power Added Efficiency: η PA = POUT − PIN = η D ⋅ 1 − 1 

PDC  G

3
Ideal FET Input and Output Characteristics

IDS
VGS=0
Im

gm

VGS=VP
VGS VDS
2VP VP 0 0 VK VDD VDSmax

Ohmic Saturation Breakdown

VDD − VK
κ=
VDD

4
Maximum Output Power Match
IDS
VGS=0
Im

gm

VGS=VP
VGS VDS
2VP VP 0 0 VK VDD VDSmax

Ohmic Saturation Breakdown

VDS max − VK
ROPT =
Im
5
Class A
IDS IDS

Im Im

VGS VDS Q
2VP VP 0 0 VK VDD VDSmax 0 p 2p
VGS VDS

2p
Q

6
Class A – Circuit
VDD

G D
RL
S

η D = κ ⋅ 50%

G = G A (e.g.14 dB)

η PA = κ ⋅ 48%
7
Class B
IDS IDS

Im Im

VGS VDS Q
2VP VP 0 0 VK VDD VDSmax 0 p 2p
VGS VDS

2p
Q

8
Class C
IDS IDS

Im Im

VGS VDS Q
2VP VP 0 0 VK VDD VDSmax 0 p 2p
VGS VDS

2p
Q

9
Class B and C – Circuit
VDD

f0

G D
RL
S

Class B Class C
η D = κ ⋅ 78% η D → 100%

G = G A - 6dB (8 dB) G →1

ηPA = κ ⋅ 65% ηPA → 0%

10
Influence of Conduction Angle

11
Class F (HCA ... harmonic controlled amplifier)
IDS IDS

Im Im

VGS VDS Q
2VP VP 0 0 VK VDD VDSmax 0 p 2p
VGS VDS

2p
Q

12
hHCA (half sinusoidally driven HCA)

IDS IDS

Im Im

VGS VDS Q
2VP VP 0 0 VK VDD VDSmax 0 p 2p
VGS VDS

2p
Q

13
Class F and hHCA – Circuit
VDD

Zo(n)
0, n=1
ID Ze(n) inf, n=odd

VDS 0, n=even
RL
inf, n=even

Class F hHCA
η D = κ ⋅100% η D = κ ⋅100%

G = G A - 5dB (9 dB) G = G A + 1dB (15 dB)

ηPA = κ ⋅ 87% ηPA = κ ⋅ 96%

14
hHCA – Third Harmonic Peaking
IDS IDS

Im Im

VGS VDS Q
2VP VP 0 0 VK VDD VDSmax 0 p 2p
VGS VDS

2p
Q

15
Third Harmonic Peaking – Circuit
VDD

G D 3f0
f0 RL
S

η D = κ ⋅ 91%

G = G A + 0.6dB (14.6 dB)

ηPA = κ ⋅ 87%
16
Linearity Aspects

17
 Linearity Aspects

~ Class A ~ Class AB

~ Class B ~ Class C

18
Linearity Aspects

~ Ideal strongly nonlinear model ~ Strong-weak nonlinear model

19
Amplifier Design – An Example
~ Balanced Amplifier Configuration

Port 1
Z=50 Ohm Port 2
Z=50 Ohm

20
Amplifier Design – Simulation
~ Gate & Drain Waveforms

Gate waveforms Drain waveforms

1 1000 25 Inner Drain Voltage (L, V) Inner Drain Current (R, mA)
5000
Amp Amp

20 4000
0 500
15 3000

-1 0 10 2000

5 1000
-2 -500

Inner Gate Voltage (L, V) Inner Gate Current (R, mA)

0 0
Amp Amp

-3 -1000 -5 -1000
0 500 1000 1300 0 500 1000 1300
Time (ps) Time (ps)

21
Amplifier Design – Simulation
~ Dynamic Load Line & Power Sweep

Dynamic load line Power Sweep 1 Tone

8000 40 80
IVCurve (mA)
Output Power (L, dBm)
IV_Curve 70
Amp
6000 Dynamic Load Line (mA) PAE (R)
Amp 30 Amp 60

4000 50

20 40
2000
30

10 20
0
10

-2000 0 0
0 3 6 9 12 15 0 5 10 15 20 24
Voltage (V) Power (dBm)

22
 Amplifier Design – Measurements
~ Single Tone & Two Tone

60 60 PAE [%]
40 80 PAE[%]

1dBCP
35 70 50 50

30 60

P out [dBm], IMDD [dBc], Gain [dB]

40 40
P out [dBm], Gain [dB]

25 50

Pout P out
Gain IMDD
20 40 30 30
Gain
GammaIn
PAE
PAE
15 30
20 20

10 20

10 10
5 10

0 0 0 0
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35
Pin[dBm] P in [dBm]

23
Amplifier Nonlinearity
~ Gain and Phase depends on Input Signal

~ 3rd Order Gain-Nonlinearities:

24
Amplifier Nonlinearity
~ Higher Output Level (close to Saturation) results
in more Distortion/Nonlinearity

25
Nonlinearity leads to?
~ Generation of Harmonics

~ Intermodulation Distortion / Spectral Regrowth

~ SNR (NPR) Degradation

~ Constellation Deformation

26
Intermodulation and Harmonics

27
Spectral Regrowth
10
ACPR1 >60dB
ACPR2 >60dB
0
ACPR1 =16dB
ACPR2 =43dB
-10
relative power / dB

-20

-30

-40

-50

-60
-15 -10 -5 0 5 10 15
relative frequency / MHz

~ Energy in adjacent Channels

~ ACPR (Adjacent Channel Leakage Power Ratio) increases

28
Reduced NPR (Noise Power Ratio)

~ Input Signal ~ Output Signal of

Nonlinear Amplifier

~ Degradation of Inband SNR

~ „Noisy“ Constellation

29
Constellation Deformation
~ Input Signal ~ Output Signal of
Nonlinear Amplifier
(with Gain- and Phase-Distortion)

30
Modeling of Nonlinearities
~ with Memory-Effects
z Volterra Series (=„Taylor Series with Memory“)

~ without Memory-Effects
αar αΘr 2

performance
z Saleh Model f (r ) = g (r ) =
1+ βar 2
1 + βΘr 2

better
z Taylor Series
z Blum and Jeruchim Model
z AM/AM- and AM/PM-conversion

31
AM/AM- and AM/PM-Conversion
~ GaAs-PA

32
AM/AM- and AM/PM-Conversion
~ LDMOS-PA

33
How to preserve Linearity?
~ Backed-Off Operation of PA
z Simplest Way to achieve Linearity

~ Linearity improving Concepts

z Predistortion
z Feedforward
z ...

34
How to preserve Efficiency?
~ Efficiency improving Concepts
z Doherty
z Envelope Elimination and Restoration
z ...

~ Linearity improving Concepts

z Higher Linearity at constant Efficiency
Æ Higher Efficiency at constant Linearity

35
Direct (RF) Feedback

~ Classical Method
~ Decrease of Gain Æ Low Efficiency
~ Feedback needs more Bandwidth than Signal
~ Stability Problems at high Bandwidths
36
Distortion Feedback

~ Feedback of outband Products only

~ Higher Gain than RF feedback
~ Stability Problems due to Reverse Loop

37
Feedforward

~ Overcomes Stability Problem by forward-only Loops

~ Critical to Gain/Phase-Imbalances
0.5dB Gain Error Æ -31dB Cancellation
2.5° Phase Error Æ -27dB Cancellation
~ Well suited for narrowband application

38
baseband input
Cartesian Feedback
I
modulator main amp.
I RF-output
OPAs
Q
Q

local
oscillator
10
UMTS example:
original signal
I predistorted signal
Q 0

demodulator -10

relative power / dB
-20

~ AM/AM- and -30

AM/PM-correction -40

~ High Feedback-Bandwidth -50

~ Stability Problems -60

-30 -20 -10 0 10 20 30
relative frequency / MHz
39
Digital Predistortion
~ Digital Implementation of „Cartesian Feedback“
~ Additional ADCs, DSP Power, Oversampling needed
~ Loop can be opened Æ no Stability Problems

40
Analog Predistortion

~ Predistorter has inverse Function of Amplifier

~ Leads to infinite Bandwidth (!)
~ Hard to realize (accuracy)

41
Analog Predistortion
~ Possible Realizations:

42
LINC (Linear Amplification by Nonlinear Components)

s1(t) Ks1(t)
K
K(s1(t)+s2(t))
s(t) signal =Ks(t)
separation

s2(t) Ks2(t)
K

UMTS example:
10
~ AM/AM- and s(t)
ACPR1 >60dB
ACPR2 >60dB

AM/PM-correction
0 s 1 (t)
ACPR1 =18dB
ACPR2 =29dB
-10
~ Digital separation required

relative power / dB
(accuracy!) -20

~ High Bandwidth, -30

oversampling necessary -40

~ Stability guaranteed -50

-60
-30 -20 -10 0 10 20 30
relative frequency / MHz
43
Doherty Amplifier
~ Auxiliary amplifier supports main amplifier during saturation
~ PAE can be kept high over a 6dB range

44
 Doherty Amplifier
~ Gain vs. Input Power ~ Efficiency vs. Input Power
POUT

A2)
A1+
(
n
tio main amp. (A1)
a
gur
i
onf
c
e rty aux. amp. (A2)
h
do

PIN

~ No improvement of AM/AM- and AM/PM-distortion

~ Behavior of auxiliary amplifier very hard (impossible) to realize
~ Stability guaranteed

45
EER (Envelope Elimination and Restoration)

~ Separating phase and magnitude information

~ Elimination of AM/AM-distortion
~ Application of high-efficient amplifiers
(independent of amplitude distortion)
~ Stability guaranteed
amplitude information

RF input signal
separation
phase information RF output

high efficiency
power amplifier

46
 EER (Envelope Elimination and Restoration)
supply voltage
~ Analog realization peak detector amplifier

z Limiter hard to build

z Accuracy problems limiter
z Feedback necessary RF input RF output

high efficiency
peak detector power amplifier

~ Digital realization
z Oversampling + high D/A- amplitude information
conversion rates required D
supply voltage amplifier

digital baseband input

z High power consumption I
A

of DSP and D/A-converters digital D modulator

signal
z Possible feedback Q processor A I RF output
Q
elimination D
high efficiency
A
z Compensation of AM/PM- power amplifier
phase information
distortion possible local oscillator

47
 EER (Envelope Elimination and Restoration)

~ Bandwidth of Magnitude- and ~ Five times (!) oversampling

phase-signal have higher than necessary to achieve standard
transmit signal requirements

UMTS example: UMTS example:

10 10
full bandwidth ACPR1 >60dB
M agnitude
P hase 3⋅B 0 bandwidth ACPR2 >60dB
0 0
5⋅B 0 bandwidth ACPR1 =33dB
7⋅B 0 bandwidth ACPR2 =40dB
-10 -10
relative power / dB

ACPR1 =51dB

relative power / dB
ACPR2 =36dB
-20 -20
ACPR1 =53dB
ACPR2 =49dB
-30 -30

-40 -40

-50 -50

-60 -60
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30
relative frequency / M Hz relative frequency / MHz

48
Adaptive Bias
~ Varying/Switching of Bias-Voltage depending on
Input Power Level
~ Selection of Operating Point with high PAE
~ Applicably for nearly each type of Amplifier
peak detector
bias
control

RF input RF output

high efficiency
power amplifier

49
 Adaptive Bias
~ Single tone PAE for switched ~ Simply to implement Concept
VDD with VG kept constant ~ Stability guaranteed
90 ~ Possible problems:
80 z DC-DC converter with high
efficiency necessary
power added efficiency / %

70

60 z Possible Linearity Change

50 (can increase and decrease)
40
especially for HCAs
V D =3.5V
30 V D =4.5V
V D =6.5V
20
32 33 34 35 36 37 38 39 40
output power / dBm

50
Summary
~ Digital Realization required to achieve Accuracy

~ Problem of Stability for high Bandwidth Application

~ Higher Bandwidths (Oversampling) necessary,

depending on Order of IMD cancellation

~ Predistortion gives best Results while keeping

Efficiency high (valid for high Output Levels > 40dBm)

51
Figure References
~ F. Zavosh et al,
“Digital Predistortion Techniques for RF Power
Amplifiers with CDMA Applications”,
Microwave Journal, Oct. 1999

~ Peter B. Kenington,
“High-Linearity RF Amplifier Design”,
Artech House, 2000

~ Steve C. Cripps,
“RF Power Amplifiers for Wireless Communications”,
Artech House, 1999

52
Contact Information

DI Markus Mayer DI Holger Arthaber

 +43-1-58801-35425  +43-1-58801-35420

 markus.mayer@tuwien.ac.at  holger.arthaber@tuwien.ac.at

53