Linearization techniques and RF predistorter tuning for maximum PA efficiency

“When targeting high-efficiency amplifiers, overcoming the nonlinear nature of RF signals is very challenging. In this application note, linearization techniques and RF predistorter tuning are presented to achieve maximum PA efficiency using minimum components. The MAX2009/MAX2010 analog RF predistorter is used in this application to neutralize nonlinearity without sacrificing power amplifier efficiency and performance.

“

When targeting high-efficiency amplifiers, overcoming the nonlinear nature of RF signals is very challenging. In this application note, linearization techniques and RF predistorter tuning are presented to achieve maximum PA efficiency using minimum components. The MAX2009/MAX2010 analog RF predistorter is used in this application to neutralize nonlinearity without sacrificing power amplifier efficiency and performance.

introduce

Linear modulation schemes such as WCDMA allow higher data rates and multiple wireless connections per carrier, but they also introduce higher carrier-to-signal ratios. Unlike constant envelope modulation schemes that can drive PAs (power amplifiers) for compression, amplifiers must now be pulled back considerably to meet adjacent channel leakage constraints. Since the more the PA efficiency is reduced, the more the PA is backed off, a linearization technique is applied to combine maximum efficiency with minimum IM (intermodulation).

Well-known linearization techniques such as feedforward (FFW) and digital predistortion (DPD) are expensive and require considerable space. Therefore, there is a need for a method that uses few components and is easy to handle.

Compared to FFW or DPD, the MAX2009/MAX2010 analog RF predistorter requires few external components, is easy to adjust, and has considerable linearity.

The MAX2009/MAX2010 rely on AM-AM and AM-PM curve correction at RF frequencies to improve IM3 and ACPR performance. Internally, the chip measures the signal power and distorts the phase and gain predistortion based on the current signal amplitude. Although AM-AM and AM-PM corrections rely on memoryless Circuits, Class AB amplifiers can still benefit from the negative distortion produced by Maxim® devices and can significantly improve performance.

As with all linearization techniques, a good signal clipping algorithm reduces the peak-to-average ratio of the signal before the PA (without exceeding the EMV limit), helping to simulate predistortion. Using the MAX2009/MAX2010 with proper signal clipping is a good combination.

The general predistorter theory is

Given a sinusoidal RF input signal, the compression distortion of an amplifier at RF frequencies is typically shown in Figure 1. A predistorter distorts the input signal to counteract the distortion added by the amplifier. The result is a net linear transfer function.

Amplitude Distortion Transfer Function.

Phase distortion works pretty much the same way. Most amplifiers tend to delay the input signal more as the amplitude increases. This means that the phase of the output signal decreases as the amplitude increases. The phase part of the predistorter does the opposite by reducing the variation of delay with amplitude. The end result is a constant delay transfer function.

Phase Distortion Transfer Function.

The previous figures show the instantaneous VIN/VOUT characteristics. This is difficult, if not impossible, for RF amplifiers. In systems without memory, the nonlinear behavior of amplifiers can be fully characterized by simply drawing AM-AM and AM-PM diagrams. The input signal is a single frequency; the x-axis is the input power; the AM-AM and AM-PM plots show the magnitude and phase of the gain, respectively. Note that phase compression begins before amplitude compression occurs. This is important for choosing the correct analog predistortion method.

How to prepare an amplifier for predistortion

The general function of the MAX2009/MAX2010 is to expand the phase and gain to compensate for the phase and gain compression of the amplifier. This process corresponds to linear mapping, where each point of the power transistor compression curve is assigned a single value for phase and gain correction. In fact, amplifiers suffer from memory effects to some extent. Like every semiconductor component, the characteristics of power transistors change with temperature. Since power amplifiers have limited efficiency, most of the power will be converted to heat. This is done with several different time constants. It may take a few minutes to heat up the entire amplifier. It may take a few seconds to heat the transistor package, but the time constant for heating the LDMOS channel is in the microsecond range. ¹ Therefore, if the envelope power of the signal changes very rapidly (as in WCDMA, for example), the temperature of the active channel will not remain constant, but will vary with modulation. This can lead to memory effects. Simply put, the amplifier behaves differently when compressing the curve up and down because its channel is warmer when driven down from the peak. For CMDA signals, this may affect more than one of the following data chips, which means there will be a lot of EVM and intermodulation products. Because its channel temperature is higher when driving down from the peak. For CMDA signals, this may affect more than one of the following data chips, which means there will be a lot of EVM and intermodulation products. Because its channel temperature is higher when driving down from the peak. For CMDA signals, this may affect more than one of the following data chips, which means there will be a lot of EVM and intermodulation products.