Title: Predistortion Techniques
1Predistortion Techniques
- Jonne Lindeberg
- jlindebe_at_ecdl.hut.fi
2Outline
- Introduction
- RF and IF predistortion
- Adaptive (BB) Predistortion
- Digital (BB) Predistortion
- Conclusions, References, and Homework
3Introduction
- Simplest method for RF PA linearization
- RF predistortion
- IF predistortion
- Baseband predistortion
4RF predistortion
- Traditional systems gt modest improvement (5 to
15dB) compared to feedforward systems or
cartesian feedback - Predistortion depends on the behavior of the PA
- How to fabricate circuit with inverse
characteristics - Usually only odd order distortion precompensated
- Simplest and most widely used method is to
predistort the 3rd order characteristics - If higher order predistortion needed,
predistoriting elements need to be designed
individually for each PA or at least tuned
5Cubic predistorters
- Aim to eliminate third-order distortion to same
level with higher order (usually 5th) - Good in TWT amplifiers, because these usually
generate third order distortion - Beneficial in BP systems
- Modest degrees of of linearity improvement
- Poor matching between the predistorter and PA
- Increase the level of higher-order distortion
6Cubic predistorters
-Require high degree gain and phase matching
where ?A is amplitude matching and ?? phase
matching error
7Cubic predistorters
8RF predistorters
-Series diode is the simplest predistorting
nonlinear element -Schottky diode with separate
parallel capacitor (Cp) -Adjustment with bias
resistor Rbias -ACPR improvement of 4dB for
IS-95 CDMA have been reported at 1.96GHz
9RF predistorters
-FET source drain channel used for predistorting
element -Wide variety of configurations 1
10IF predistortion
-Same predistorter can be used for different
operating frequencies -Predistorter
significantly simpler than RF -Narrower band
predistortion than RF Some methods Curve-fittin
g predistorters Digital IF Cubic Complex Adap
tive Control Predistortion (plus several others)
11IF predistortion
DIGITAL IF -Relatively little used due to
complexity and speed requirements to achieve
reasonable BW -High power consumption -Improvement
s in device technologies will increase
attractiveness -Similar to adaptive
baseband -Reported 3.rd IMD improvement of 25dB
over 45MHz (limited by the DAC and ADC speed and
resolution)
12IF predistortion
CUBIC IF PREDISTORTION
13IF predistortion
COMPLEX IF PREDISTORTION -Complex predistortion
attempt to compensate the AM/AM and AM/PM
characteristics by operating in I/Q mode (upper
Fig.) or amplitude phase mode (lower Fig.)
14IF predistortion
ADAPTIVE IF PREDISTORTION Adjacent-channel method
15IF predistortion
ADAPTIVE IF PREDISTORTION Correlation-based
control
16RF and IF predistortion
- Efficiency of Predistortion Techniques
- The class of power amplifier to be linearized.
- The power consumption of the predistorter.
- The degree of output back-off necessary to meet
required specification. - The envelope characteristics of the signal to be
amplified. - In all cases, the efficiency of the predistorter
system should be higher (usually significantly
so) than the efficiency of a backed-off amplifier
producing the same level of intermodulation
distortion (or signal vector error, if this is
more important in a given application).
17RF and IF predistortion
EXAMPLE Diode based RF predistorter
18RF and IF predistortion
EXAMPLE Diode based RF predistorter measurement
s with several modulation formats and BW.
19RF and IF predistortion
- Advantages and Disadvantages
- Simplicity of implementation
- few components gt low cost
- no stability problems (compare to FB)
- can be used in microwave region
- wide linearization BW
- - modest linearity improvement
- - difficulties with high order distortion
- gt increase higher order distortion
- - fabricate circuit with a desired transfer
characteristic -
20Baseband Predistortion
- Easier to fabricate (than the RF or IF)
- If complex versions of the signal constructed gt
broadband audio phase shift required to create I
and Q may prove difficult whit out Hilbert
transform filter in DSP - If Hilbert used gt the predistortion in DSP also
- Attractive since modern radio transmitters employ
some DSP in BB
21Adaptive (BB) Predistortion
- Long been a promising technique for narrowband
linearization - Quadrature signals for both up- and
downconversion gt gain and phase correction
22Adaptive (BB) Predistortion
Quadrature BB predistorter
-LUTs accessed by algorithms which receives its
input from downconverter -Coefficients may be
updated -Drawback gt power consumption of the
ADC, DAC and LUT memories -Power efficiency poor
until tech. is sufficiently advanced
23Adaptive (BB) Predistortion
24Adaptive (BB) Predistortion
Sampling rate considerations -Since most signals
I/Q components gt FsBimd -Bimd is the full BW
of the RF channel including all significant
orders of distortion
-Reduction of Fs gt reduce the level of the of
higher order IMD -Loop delay must be considered.
Loop delay is dominated by the D/A converter
reconstruction filters which change with
temperature
25Adaptive (BB) Predistortion
- Quantization and Resolution
- Source (input) signal
- Calculation of the table index
- The LUT coefficient
- The modulator error correction process
26Digital (BB) Predistortion 2
- Adaptive Linearization of PA in Digital Radio
Systems, Saleh and Salz, 1982. - Linearizer utilizes a real-time data-directed
recursive algorithm for predistorting the signal
constellation - Predistortion accomplished within a digital
memory, which is used to generate the desired
baseband signal
27Digital (BB) Predistortion 2
28Digital (BB) Predistortion 2
- LUT and DAC have to operate full signaling rate
- ADC and linearizing processor can operate much
reduced rate since updating is needed for
compensating drifts that occur much lower rate - System assumes that the amplifier is memoryless
gt No filtering inside the loop gt not good
29Digital (BB) Predistortion 3
- Linear Amplification Technique for Digital
Mobile Communications, Nagata, 1989. - Suitable for Narrow band system
- Linearization by adding a Predistorting signal to
BB modulating signal - Adapts to the fast and frequent changes of
nonlinearity - Power consumption can be reduced small enough for
use in mobile station
30Digital (BB) Predistortion 3
31Digital (BB) Predistortion 3
Simulation results
32Digital (BB) Predistortion 4
- A GSM-EDGE High Power Amplifier utilising
Digital Linearisation, Kenington, Cope, Bennett,
and Bishop, 2001. - Method of improving both efficiency and signal
vector error - Simple third-order linearizer will not achieve a
sufficient level of mask improvement for EDGE
33Digital (BB) Predistortion 4
- Input output comparison made in frequency domain
34Digital (BB) Predistortion 4
35Digital (BB) Predistortion 5
- Digitally Controlled RF input/output
Predistortion for A Class AB Power Amplifier,
Teikari, Vankka, and Halonen, 2003. - Digitally controlled analog predistortion system
- Linearizes an class AB PA
- Signal is 16-QAM, with 18kHz BW around 420MHz
36Digital (BB) Predistortion 5
-Phase and amplitude detection -Comparison made
in time domain
37Conclusions
- In RF and IF predistortion problem is low
improvement - In digital predistortion problem becomes the
power consumption - Predistortion methods interesting in the future
with the development of different technologies
38References
- 1 Peter B. Kenington, "High-Linearity RF
Amplifier Design, Chapter XXX. - 2 A.A.M. Saleh and J. Salzm, Adaptive
Linearization of PA in Digital Radio Systems,
BSTJ, Vol. 62, No. 4, April 1983. - 3 Yoshimori Nagata, Linear Amplification
Technique for Digital Mobile Communications, - 4 P. B. Kenington, M. Cope, R. M. Bennett, and
J Bishop, A GSM-EDGE High Power Amplifier
utilising Digital Linearisation. - 5 I. Teikari, J. Vankka, and K. Halonen,
Digitally Controlled RF input/output
Predistortion for A Class AB Power Amplifier,
URSI 2003.
39Homework
- Your PA has 3rd. IMD at level 30dBc. Using RF
cubic predistortion with amplitude error of 0.5
dB and phase error of 1 degree respect to ideal
case. What is the theoretical maximum of 3rd. IMD
result you can achieve with these parameters? - Your job is to choose a linearization method for
two different transmitter. What method would you
prefer and WHY? - You need to transmit 4 WCDMA carriers (20MHz BW),
and your transmitter chain is 3dB over
specifications (ACPR)? - You need to transmit GSM carrier and the signal
does not fit to the mask at the antenna (3rd. and
5th. order distortion causes signal to rise 20dB
over mask limit) - (Hint. There are no correct solutions. Just
explain your desicions.)