Agilent EEsof EDA

Presentation on Adaptive Digital Predistortion of Power Amplifiers

This document is owned by Agilent Technologies, but is no longer kept current and may contain obsolete or inaccurate references. We regret any inconvenience this may cause. For the latest information on Agilents line of EEsof electronic design automation (EDA) products and services, please go to:

www.agilent.com/nd/eesof

Seminar: Gain Without Pain November 2000

Adaptive Digital Predistortion of Power Amplifiers

Shawn Stapleton Agilent Technologies 1400 Fountaingrove Parkway Santa Rosa, CA 95403

Abstract Linear modulation techniques possess good spectral efficiency. However, their fluctuating envelopes in conjunction with nonlinear power amplifiers results in spectral spreading to adjacent channels. Linearization of the power amplifier by means of predistortion is one means of compensating for these nonlinearities. One technique that is well suited to digital signal processing baseband implementations is adaptive digital predistortion. Adaptation is based on the difference between the desired modulation and the actual power amplifier output. Biography Dr. Shawn P. Stapleton has 17 years of experience in the design of RF and microwave circuits and systems. He is presently professor of electrical engineering at Simon Fraser University as well as a consultant for Agilent EEsof. He has developed GaAs MMIC components, including mixers, amplifiers, frequency dividers and oscillators. His most recent work includes digital signal processing, mobile communications and RF/microwave systems.

Agenda & Topics

Digital Predistortion of Power Amplifiers
Introduction to Adaptive Digital Predistortion Key Features: Digital Predistortion Techniques & Concepts Digital Predistortion Design Example Conclusion

This section of the workshop provides an introduction to digital predistortion. We will cover key features, technologies, and performance issues. Approaches to solving some of the design challenges will also be presented. An adaptive digital predistorter is demonstrated using the Agilent Advanced Design System.

Technology Overview
Linearization approaches:
FeedForward Linearization Based on inherently wideband technology RF Predistortion Limited accuracy of function model Implemented at RF with low complexity Cartesian Feedback Stability considerations limit bandwidth and accuracy LINC Sensitive to component drift and has a high level of complexity Dynamic Biasing Limited ACI suppression

Digital Predistortion
Limited Bandwidth (DSP implementation) Good IMD suppression
4

Of the various linearization techniques that have been developed, predistortion is the most commonly used. The concept behind predistortion calls for the insertion of a nonlinear module between the input signal and the power amplifier. The nonlinear module generates IM distortion that is anti-phase with the IM distortion produced by the power amplifier, thereby reducing out-of-band emissions. Feedforward linearization is the only strategy that simultaneously offers wide bandwidth and good IM distortion suppression. The price for this performance is higher complexity. Automatic adaptation is essential to maintain performance. RF-based predistortion offers reasonable IM distortion reduction over moderate bandwidths. Cartesian feedback is relatively less complex and offers reasonable IM distortion suppression, but stability considerations limit the bandwidth to a few hundred KHz. The LINC technique converts the input signal into two constant envelope signals that are amplified by Class C amplifiers, and then combined, before transmission. Consequently, they are very sensitive to component drift. Dynamic biasing is similar to predistortion, however the work function operates on the power amplifiers operating bias. Digital predistortion has two distinct advantages. First, the correction is applied before the power amplifier where insertion loss is less critical. Second, significant IMD reduction is achievable. Digital predistortion techniques are more complex, but provide better IM distortion suppression. However, bandwidths are low due to limited DSP computational rates.

Digital Predistortion
Digital Domain Analog Domain

Data

Modem

Complex Gain Predistorter

D A C

Quadrature Modulator

Power Amplifier

Local Oscillator

Adaptation

A D C

Quadrature Demodulator

The linearizer creates a predistorted version of the desired modulation. The predistorter consists of a complex gain adjuster that controls the amplitude and phase of the input signal. The amount of predistortion is controlled by a look-up table (LUT) that interpolates the AM/AM and AM/PM nonlinearities of the power amplifier. Note that the inputs in this adaptation process include a delayed version of the output and the input signal. The input is delayed and then subtracted from the power amplifiers output signal. The difference should contain only the distortion components.

Spectrum at the Nodes

Digital Domain Analog Domain

Data

Modem

Complex Gain Predistorter

D A C

Quadrature Modulator

Power Amplifier

Local Oscillator

Adaptation

A D C

Quadrature Demodulator

Given a two-tone input signal, we can observe the spectral response at various nodes in the digital predistorter. The complex gain adjuster, once optimized, provides the inverse of the nonlinear characteristics from the power amplifier. Thus, the spectral growth from the predistorter can be observed at the input node to the power amplifier. Ideally, the IM products will be equal in amplitude, but anti-phase to the IM products created as the two tones pass through the power amplifier. The function of the adaptation process is to quickly adjust the LUT entries, so that distortion is minimized.

Design Techniques
Digital Predistortion
Generic digital predistortion techniques Complex vector mapping LUT Complex gain LUT Cartesian feedback Generic adaptation techniques Secant method Linear convergence

There are three distinct digital predistortion techniques. The complex vector mapping LUT technique translates the input vector by adding an error vector to compensate for the AM/AM and AM/PM distortion. The complex gain LUT approach multiplies the input signal by a complex gain vector optimized and stored in the LUT. The LUTs index is the envelope of the input signal. Cartesian feedback is another approach that does not require a LUT, but tends to be less stable. Adaptation based on the use of gradient signals requires a continuous computation to estimate the gradient of a three-dimensional power surface. The surface for the digital predistorter circuit is the difference between the input signal and the scaled output signal. The adjacent channel interference power is minimized when this error signal is completely suppressed. The gradient is continually updated, so deliberate misadjustment is not required. Two common gradient techniques are linear convergencea first order feedback loopand the secant method, which is based on estimating the gradient using Newtons classical method.

Complex Gain LUT

Digital Domain Analog Domain

Vm(t)
Data

Vd(t)
D A C
Quadrature Modulator Power Amplifier

Modem

|Vm(t)|
Delay

F(x)
RAM
phase shift

Local Oscillator

up-date

Ve(t)

A D C
Delay Adjust

Quadrature Demodulator

Va(t)

The complex gain LUT technique is depicted in this slide. The input signal is multiplied by the gain signal derived from the RAM. This gain term is dependent on the input-signal envelope, which is quantized to a finite number of entries (64 in this example). These entries are optimized by finding the difference between the input signal and the power amplifiers output. The results should contain only the distortion, provided that weve established the feedback delay. There are a number of techniques that operate in the time or frequency domain that are available to adaptively compensate for this delay. Updating the RAM entries can be accomplished using approaches such as linear convergence or the secant method.

Complex Gain LUT

Vd(t) = Vm(t) F{ |Vm(t)| } |V Ve(t) = Va(t) - Vm(t) |Ve(t)| = scaling error Ve(t) = rotation error
Va(ti) Vm(ti)
rotation error
9

Quadrant of I-Q Plane

scaling error

Ve(ti)

The input signal vector generated from the modem is multiplied by a gain function. The gain function is a complex quantity that is dependent on the envelope of the input signal. This dependence is because we only need to compensate for the AM/AM and AM/PM distortion in a power amplifier. The gain function can be stored in the LUT in either polar coordinate form or rectangular form. The LUT entries are derived from the resulting error vectors, which come from subtracting the input signal from the power amplifiers output signal. The result is that the power amplifiers distortion produces a scaling and rotation of the input vector.

ADS RF Predistortion Simulation

Simulation Parameters: 1) Multi-tone modulation (Fc=800 MHz, BW=25 MHz) (Fc=800 MHz) 2) 64-entry LUT 3) RAM write enable period is 0.4 ns 4) Linear convergence parameter is -0.1 5) Behavioral model for power amplifier 6) Ideal passive components assumed

10

Now well look at an example of a digital predistorter simulation based on the complex gain LUT technique, carried out using the Advanced Design System. In this case, we utilize the linear convergence technique to adjust the LUT entries to minimize the ACP. The adaptation coefficient is set to -0.1 for fast optimization. We use a 64-entry LUT to quantize the input envelope. Passive components, such as the power splitters and combiners are assumed to be ideal. For demonstration purposes, we use a ten-tone input centered on 800 MHz, spanning a bandwidth of 25 MHz.

ADS Digital Predistortion Circuit

LUT Clock Timing LUT Clock Timing

Output Output

Input Input

11

Here is the circuit schematic for the digital predistorter as displayed in the Advanced Design System. The adaptation technique is based on the linear convergence method, and the rectangular implementation is used for the complex gain adjuster. The input consists of a ten-tone modulation. Timing clocks are used to read and writ to the RAM.

LUT for Digital Predistorter

In-Phase LUT In-Phase LUT

I IError Signal Error Signal LUT I Iupdate LUT update Quadrature LUT Quadrature LUT

Q Error Signal Q Error Signal LUTQ update LUTQ update

12

The error signal derived from the difference between the input and output signals is scaled by the adaptation constant, and the result is latched in the data registers. The index for the RAM is established by passing the input envelope through an A/D converter. The in-phase (I) and quadrature (Q) signals are stored in their respective LUTs. The fixed-point summation provides the update for the new table entry based on the previous value at the corresponding index.

Optimized LUT Phase for Predistorter

Phase Phase Radians Radians (0.005/div) (0.005/div)

Signal Envelope Signal Envelope

Index Index

Time ( (20 ns/div) Time 20ns/div) ns/div)

13

These plots demonstrate the envelope for the ten-tone input signal and the corresponding LUT entries for the phase.

Optimized LUT Gain for Predistorter

Gain Gain Magnitude Magnitude

Signal Envelope Signal Envelope

Index Index

Time (20 ns/div) Time (20ns/div) ns/div)

14

These plots demonstrate the envelope for the ten-tone input signal, and the corresponding LUT entries for the magnitude. We observe that a nominal amount of gain is required to compensate for the AM/AM compression that occurs because of the power amplifier.

Multi-Tone Simulation of Predistorter

Before Before After After

-34 dBc -34 dBc -54 dBc -54 dBc dBm dBm dBm dBm

Frequency (MHz) Frequency (MHz)

Frequency (MHz) Frequency (MHz)

15

The plot on the left shows the power amplifier driven at 5dB back-off, which generates high levels of intermodulation power and high levels of harmonics. The plot at right shows the output from the digital predistorter once the LUT entries have been optimized. We can observe the spectral growth that occurs using a digital predistorter. The adjacent channel power is spread over a wider bandwidth, but the mask requirements can now be meet.

Summary
Digital Predistortion
G

Adaptive digital predistorters have moved from the research to the development phase.

Design Solutions
G

The ADS digital predistorter design example demonstrates the performance that can be achieved with linearization. System level simulation provides a solid starting point for building an implementation quickly. Designed components can be integrated into a system to witness the impact on overall performance.
16

For more information about Agilent EEsof EDA, visit: www.agilent.com/nd/eesof

Agilent Email Updates

www.agilent.com/nd/emailupdates

www.agilent.com
For more information on Agilent Technologies products, applications or services, please contact your local Agilent office. The complete list is available at:

Get the latest information on the products and applications you select.

www.agilent.com/nd/contactus
Agilent Direct
www.agilent.com/nd/agilentdirect

Quickly choose and use your test equipment solutions with condence.

Americas Canada Latin America United States Asia Pacic Australia China Hong Kong India Japan Korea Malaysia Singapore Taiwan Thailand

(877) 894-4414 305 269 7500 (800) 829-4444

1 800 629 485 800 810 0189 800 938 693 1 800 112 929 0120 (421) 345 080 769 0800 1 800 888 848 1 800 375 8100 0800 047 866 1 800 226 008

Europe & Middle East Austria 0820 87 44 11 Belgium 32 (0) 2 404 93 40 Denmark 45 70 13 15 15 Finland 358 (0) 10 855 2100 France 0825 010 700*
*0.125 /minute

Germany

01805 24 6333**
**0.14 /minute

Ireland 1890 924 204 Israel 972-3-9288-504/544 Italy 39 02 92 60 8484 Netherlands 31 (0) 20 547 2111 Spain 34 (91) 631 3300 Sweden 0200-88 22 55 Switzerland 0800 80 53 53 United Kingdom 44 (0) 118 9276201 Other European Countries: www.agilent.com/nd/contactus
Revised: March 27, 2008

Product specications and descriptions in this document subject to change without notice. Agilent Technologies, Inc. 2008 Printed in USA, November 01, 2000 5989-9086EN