JP2005073032A - Amplifier and method for distortion compensation - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method for compensating a distortion property including an inflection point using curvilinear interpolation. <P>SOLUTION: A distortion compensation amplifier includes a detector 11 for detecting input signal power; a distortion compensation table 12 for storing a power value and a predistortion amount correspondingly, and outputting the predistortion amount based on the detector output; distortion means (131, 132) for distorting the input signal based on the output predistortion amount; an amplifier 2 for amplifying the input signal distorted by the distortion means and outputting the amplified signal; and a controller 14 for updating the distortion compensation table 12 based on the output of the amplifier 2 and the input signal. The controller 14 performs curvilinear interpolation on a plurality of points stored by the distortion compensation table means 12 in a plurality of sections being partly overlapped, and updates the points stored by the distortion compensation table means by combining each curve obtained through the curvilinear interpolation. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

The present invention relates to a distortion compensation amplifying apparatus, and more particularly to a distortion compensation amplifying apparatus that interpolates a distortion compensation coefficient.

For example, in a base station device provided in a mobile communication system that employs a W-CDMA (Wide-band Code Division Multiple Access) method as a mobile communication method, Therefore, it is necessary to greatly amplify the signal with an amplifier. However, since the amplifier is an analog device, its input / output characteristics are nonlinear functions. In particular, after the amplification limit called the saturation point, the output power is almost constant even if the power input to the amplifier is increased. Then, nonlinear distortion is generated by this nonlinear output. In the transmission signal before amplification, the signal component outside the desired signal band is suppressed to a low level by the band limiting filter. However, the signal after passing through the amplifier causes nonlinear distortion and the signal component goes out of the desired signal band (adjacent channel). Leaks. For example, since the transmission power is high in the base station device as described above, the magnitude of the leakage power to such an adjacent channel is strictly defined. Therefore, how to reduce the leakage power of such an adjacent channel. How to do is a big problem. One of the distortion compensation methods is a predistortion method. In recent years, in order to place importance on amplification efficiency, the method is becoming mainstream instead of the feedforward method. The predistortion method is a method for compensating for distortion of an output signal of an amplifier by giving in advance an amplifier input signal with inverse characteristics of AM-AM conversion and AM-PM conversion which are nonlinear characteristics of the amplifier.

FIG. 1 shows a functional block diagram of an amplifier using a predistortion method, and its operation will be described. The input signal is input to the power detection unit 11 to detect power or amplitude, and is associated as a reference argument of the distortion compensation table 12 ′ configured by a memory or the like. The distortion compensation table 12 'stores a table for performing distortion compensation by a predistortion method. The table is the inverse characteristic of the nonlinear characteristic of the amplifier to be compensated, and is generally AM-AM conversion (amplitude) and AM-PM conversion (phase) using the power or amplitude of the input signal as an index. The distorter 13 controls the amplitude and phase according to the reference result of the distortion compensation table 12. The signal previously distorted by the predistortion method is amplified by the amplifying unit 2 ', and the output signal becomes a signal without distortion. The control unit 14 ′ updates the distortion compensation table 12 in order to adapt to the environment such as temperature change and aging change.

In the predistortion system, the frequency band is classified according to the baseband, IF, and RF, and the processing is performed digitally and analogly. However, any type can be applied in the present invention. Has a simple configuration to explain the principle of the predistortion method. Therefore, D / A converters, quadrature modulators (or quadrature modulation units), up-converters, filters, etc. that are required according to their types are not shown, and feedback signals used for the purpose of adapting to environmental changes are not shown. Existence is not shown. In order to use a feedback signal, a down converter, an oscillator, a filter, a quadrature demodulator (or quadrature demodulator), an A / D converter, or the like may be used, and an input signal may be used for distortion detection.

Next, the characteristics of distortion compensation will be described. The distortion compensation table 12 is sometimes called a LUT (Look Up Table), and a memory or the like is used as hardware. When the detection result of the power detection unit 11 is, for example, 10 bits, the number of addresses in the memory of the distortion compensation table 12 is 1024, and learning for adaptation becomes more difficult as the number of addresses increases. That is, as the number of addresses increases, the resolution improves, but on the other hand, two problems arise. First, the degree of influence on distortion compensation when an address is updated is small, and when using out-of-band distortion power as a feedback signal, the difference in distortion power is difficult to distinguish, which is very useful for averaging. There is a problem that it takes time. Secondly, the amplifier to be compensated is originally an analog element, and the input / output characteristics are expected to be a smooth curve, and the distortion compensation table to be obtained is preferably a smooth one, but in particular, the demodulated IQ signal is used as a feedback signal. When used, there is a problem that when the entire table is viewed due to an error included in learning of each address, minute irregularities are formed, which creates new distortion.

Therefore, generally, a method of interpolating several points is used. FIG. 2 is a diagram for explaining interpolation of a conventional distortion compensation table. In the example of FIG. 2, nine representative points are arranged at intervals of 128 addresses. When one representative point is updated, mainly the surrounding interpolation part is also updated, so that the degree of influence on distortion compensation increases. Table values are generated by linearly interpolating these points (see, for example, Patent Documents 1 and 2).

Further, as another conventional technique related to the present invention, a technique for dividing a distortion compensation table is known (for example, see Patent Document 3).

JP 2002-223171 A JP 2000-278190 A JP 2002-232328 A

However, since the conventional technique linearly interpolates between representative points, the AM-AM characteristic after distortion compensation is not smooth, and the distortion compensation performance deteriorates. Therefore, a method of interpolating between representative points with a smooth curve is conceivable, but generally in curve interpolation there is a case where the change is emphasized unnecessarily in a portion where the change in curvature is large, and the interpolated value may vibrate. If the distortion compensation table has a monotonous characteristic that does not include an inflection point, the interpolation can be appropriately performed. However, a high efficiency amplifier configured to synthesize outputs from a plurality of amplifiers having different operating points exhibits complicated distortion characteristics due to the configuration.
FIG. 4 is an operation diagram of the synthesis amplifier. For example, the combining amplifier combines an amplifier (a) that operates in class AB and an amplifier (b) that operates in class C that is adjusted so that the output suddenly increases in a region where the input is large, for example, and extends output power. Thus, a technique for improving power efficiency is used. The combined input / output characteristic (c) has an inflection point in the vicinity where the output of the amplifier (b) becomes approximately the same as the output of the amplifier (a). For this reason, when a distortion compensation table is generated by the conventional method, even if smooth interpolation is performed in order to improve the distortion compensation performance, the reverse characteristic of the inflection point cannot be accurately generated, and therefore the distortion compensation performance. There has been a problem that a contradiction occurs in that it deteriorates. FIG. 5 shows an example of the distortion compensation table of the synthesis amplifier (c) generated by the method of smoothly interpolating with the above-described curvature, but the inversion point is inaccurate.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and an object of the present invention is to provide a distortion compensation amplification apparatus and method with improved distortion compensation performance by distortion compensation table interpolation that accurately generates necessary inflection points while maintaining a smooth curve. And

The distortion compensation amplification apparatus according to the present invention includes a detection unit that detects power or amplitude of an input signal, stores the power or amplitude value and a predistortion amount in association with each other, and calculates the predistortion amount based on the output of the detection unit Distortion compensation table means for outputting, distortion means for distorting the input signal based on the outputted predistortion amount, an amplification section for amplifying and outputting the input signal distorted by the distortion means, and the amplification A distortion compensation amplifying apparatus comprising: a control unit that updates the distortion compensation table based on an output of the unit and the input signal, wherein the control unit stores a plurality of points stored by the distortion compensation table unit, The points stored in the distortion compensation table means are updated by performing curve interpolation in a plurality of partially overlapping sections and combining the curves obtained by the curve interpolation.

In the distortion compensation method of the present invention, some of the table points of the distortion compensation table for storing the input power value or amplitude value in correspondence with the compensation amount are used as representative points for learning, and the representative points are used as representative points. A distortion compensation method of a predistortion method that generates a distortion compensation table by interpolating points other than representative points based on each curve obtained by interpolating the distortion compensation table in a plurality of different sections. Generate by combining.

According to the distortion compensation amplifying apparatus and the distortion compensation method of the present invention, since the vicinity of the inflection point is accurately interpolated with a smooth curve, the distortion compensation performance can be improved.

Hereinafter, the present invention will be described by way of examples. However, the following examples do not limit the claimed invention, and all combinations of features described in the examples serve as means for solving the invention. It is not always essential. A part of the configuration of the embodiments or any combination of the configurations of the embodiments may also be included in the present invention.
FIG. 6 is a diagram for explaining the combination of the distortion compensation tables in the best mode. The point indicated by the dotted line is the inflection point of the amplifier. The control unit 14 adjusts so that the coupling point of the interpolation curve coincides with the arrangement point of the amplifier. By using the present invention, a distortion compensation table near the inflection point can be accurately generated.

FIG. 7 is a configuration diagram of a distortion compensation amplifying apparatus in the present embodiment. In this embodiment, analog predistortion is performed on RF. The same components as those in the prior art are denoted by the same reference numerals and the description thereof is simplified.
The power detection unit 11 inputs an input signal (RF signal) of the predistorter 1, measures the power of the input signal, and outputs it. The hardware is realized by an A / D converter that operates at a frequency that is at least equal to or twice the bandwidth of the input signal.
The distortion compensation table 12 is a table having distortion compensation (predistortion) characteristics. When the measurement result of the power detection unit 11 is input, a compensation amount corresponding to the measurement result is output. The distortion compensation characteristic is an inverse characteristic of the nonlinear characteristic of the amplitude and phase generated in the amplifying unit 2, and is generally AM-AM conversion and AM-PM conversion using the power of the input signal as an index. That is, the distortion compensation table 12 stores an amplitude compensation amount and a phase compensation amount in association with one power value, and the amplitude compensation amount and the phase compensation amount always correspond to the measurement result of the power detector 11. Works to output. The predistorter 1 operates at a frequency at least equal to or twice the bandwidth of the input signal. The distortion compensation table 12 is realized by a RAM having, for example, a 10-bit address space as hardware.
The attenuator 131 receives an amplitude compensation amount corresponding to AM-AM conversion from the distortion compensation table, and controls the amplitude (attenuation amount) of the RF signal according to the input amplitude compensation amount. Note that this RF signal is an input signal (RF signal) of the predistorter 1 so that the RF signal whose power is detected by the power detector 11 and the RF signal whose amplitude is controlled by the attenuator 131 correspond temporally. Is appropriately delayed.
The phase shifter 132 receives the phase compensation amount corresponding to the AM-PM conversion from the distortion compensation table, and controls the phase of the RF signal output from the attenuator 131 according to the input amplitude compensation amount.
By giving distortion compensation characteristics by the predistorter 1, the amplifier 2 outputs a signal without distortion. A part of the output of the amplifying unit 2 is fed back to the control unit 14, and the distortion compensation table 12 is updated by using an algorithm adapted to the secular change and the environmental change, such as a method of extracting distortion and a method of demodulating.

FIG. 9 is a block diagram of the control unit 14 and describes a method of using out-of-band distortion power as a feedback signal. The feedback control unit 141 controls the oscillation frequency of the VCO 142 so that the BPF 144 can extract distortion power in a desired band. The output of the amplifier 2 is frequency-converted by the mixer 143, and the desired band is extracted by the BPF 144. The frequency converter 145 converts the signal to a frequency in the vicinity of DC so that the A / D converter 146 can capture the signal. The signal converted into a digital signal by the A / D converter 146 is an amount (distortion amount) representing distortion generated in the amplifier 2 and is input to the table update unit 147. The operation of the table update unit 147 will be described below.

First, update (learning) of representative points by the table update unit 147 will be described. In this embodiment, the representative points are updated by the perturbation method. In a simple form of the perturbation method, in step 1, a minute value is added to the value of a certain representative point, and the value is intentionally shifted. Next, in step 2, interpolation is performed for each section divided by the interpolation method described later based on the updated value, the interpolation of each section is connected, and the result is written in the distortion compensation table 12. Next, in step 3, distortion compensation is performed by the updated distortion compensation table, and the resulting distortion amount is input from the A / D converter 146. Next, in step 4, the input distortion amount (current value) is compared with the distortion amount before adding a minute value to the representative point value in step 1 (previous value). If the current value is smaller, step The value obtained by adding a minute value at 1 is validated, and if the current value is large, the value obtained by adding the minute value is invalidated at Step 1 and at least the sign of the minute value is inverted for the next update. All the representative points are learned by performing the above steps 1 to 4 for each representative point. The comparison between the current value and the previous value may be performed after waiting for the amount of distortion (current value) to be obtained when the power value is equal (with a resolution similar to the interval between representative points). The average distortion amount obtained at an interval sufficiently slower than the operation speed of the predistorter 1 may be compared. Further, the absolute value of the minute value may be fixed, or may be adjusted according to a change in the amount of distortion.

Next, the interpolation of representative points by the table updating unit 147 will be described. In this embodiment, representative points are interpolated by root Nyquist interpolation. Nyquist interpolation is realized by a function having a Nyquist characteristic (for example, a sinc function) and an FIR (product-sum operation) of representative points. In the present embodiment, for example, 1024 table points of the distortion compensation table 12 are grouped into 8 and 8 to 9 representative points are set at equal intervals (128-point intervals), and the intervals are Nyquist characteristics for product-sum operation. Equal to the zero crossing interval. By the root Nyquist interpolation, a table value is obtained that forms a curve that passes through the representative points and smoothly connects the representative points. Interpolation is performed for each of the amplitude compensation amount, the phase compensation amount, and the divided sections.

Next, division of a table section by the table updating unit 147 will be described. The section is divided by detecting the inflection point of the table. The inflection point appears as a point where the curvature abruptly changes in the distortion compensation table 147, for example, by combining amplifiers with different operating points. In contrast, the position of the inflection point also changes slowly but needs to be updated. FIG. 13 is a flowchart of inflection point update, which is an example of a perturbation method in which distortion power or error before and after inflection point update is compared and converged so that they become smaller. When there are a plurality of inflection points, the control in FIG. 13 is performed independently for each inflection point.

FIG. 11 is a diagram for explaining the interpolation, section division, and combination of the present embodiment according to the above description. FIG. 11 shows a case where the inflection point is a point between the representative points, the inflection point is included in each of the divided sections (a) and (b), and the end of each section is the representative point. The section is overlapped by one section so as to end. Further, when joining, each section is cut at the inflection point and joined. Since the inflection points are updated so as to minimize the distortion power or the error according to the flowchart as shown in FIG. 13, the table values are almost continuous in the combined portion without special consideration such as boundary conditions. FIG. 12 is a diagram for explaining a distortion compensation table having a plurality of inflection points. When there are n inflection points, interpolation is performed by dividing into n + 1 sections.

FIG. 3 shows an amplifying unit (combining amplifier) of this embodiment. Reference numeral 21 denotes a distributor that distributes signals to be input to the subsequent carrier amplifier and peak amplifier with an appropriate phase, and is, for example, a 3 db coupler. A carrier amplifier 22 operates in class A or class AB and amplifies the input level up to a medium level. Reference numeral 23 denotes a peak amplifier that operates in a class B or C class and amplifies an intermediate or higher input level. A synthesizer 24 synthesizes the outputs of the carrier amplifier 22 and the peak amplifier 23, and is composed of, for example, a 4 / λ line.

According to the above-described embodiment, it is possible to automatically follow the arrangement point of the table even when the position of the arrangement point of the amplifier fluctuates due to aging, aging, temperature, or the like. Further, by dividing the table into a plurality of sections, the curve becomes monotonous, and a highly accurate table value can be obtained without using an optimization method with a large amount of calculation. In this embodiment, a synthesis amplifier is used. However, the amplifier is not limited to the type of amplifier, and a single amplifier may be used. The learning method is not limited to the perturbation method, and the interpolation method is not limited to Nyquist interpolation, and the present invention can be realized by other methods.

This embodiment is different from the first embodiment in that a predistorter that operates in a baseband is used, and the description of the same parts as the first embodiment is omitted. FIG. 8 is a configuration diagram of the distortion compensation amplifying apparatus in the present embodiment. The inputs to the predistorter 1 are I and Q complex signals. The functions of the attenuator 131 and the phase shifter 132 of the first embodiment are performed by a vector calculator 133 (such as a complex multiplier). The predistorter output is converted into an RF signal by the quadrature modulator 31 and the up-converter 32 and input to the amplifier 2.

FIG. 10 is a block diagram of the control unit 14 of the present embodiment. In this method, a demodulated IQ signal is used as a feedback signal. The output of the amplifier 2 is converted into IQ digital data by the quadrature demodulation 148 unit. The error detection unit 149 obtains a difference from the predistorter input signal delayed so as to correspond to the output of the quadrature demodulation unit 148, and outputs the difference as an error signal.

The distortion compensation table interpolation is substantially the same as in the first embodiment, but also differs from the first embodiment in that two tables are overlapped in connection. That is, in FIG. 11, a weighted addition is performed so that a place where each table data does not exist is set to “0” to generate a composite table having an inflection point. In FIG. 11, two interpolation curves are combined for simplicity, but two or more may be used.

Functional block diagram of amplifier using predistortion method The figure explaining the interpolation of the conventional distortion compensation table Synthetic amplifier configuration diagram Synthetic amplifier operation diagram Example of distortion compensation table with incorrect inflection point reverse characteristics The figure explaining the combination of the distortion compensation table in the best mode Configuration diagram of distortion compensation amplifying apparatus in embodiment 1 Configuration diagram of distortion compensation amplifying apparatus in embodiment 2 The block diagram of the control part in Example 1 The block diagram of the control part in Example 2 The figure explaining the interpolation in Example 1, the division | segmentation of a section, and a coupling | bonding The figure explaining the distortion compensation table which has several inflection points Flow chart of inflection point update in embodiment 1

Explanation of symbols

1: Predistorter, 2: Amplifying unit,
11: Power detector, 12: Distortion compensation table, 13: Distorter, 14: Control unit,
147: Table update unit

Claims (2)

A detection unit for detecting the power or amplitude of the input signal;
A distortion compensation table means for storing power or an amplitude value and a predistortion amount in association with each other, and outputting a predistortion amount based on an output of the detection unit;
Distortion means for distorting the input signal based on the output predistortion amount;
An amplifying unit for amplifying and outputting the input signal distorted by the distortion means;
A distortion compensation amplifying apparatus comprising: a control unit that updates the distortion compensation table based on an output of the amplification unit and the input signal;
The control unit performs curve interpolation on a plurality of points stored in the distortion compensation table means in a plurality of overlapping sections, and combines the curves obtained by the curve interpolation to thereby combine the distortion compensation table. A distortion compensation amplifying apparatus, wherein the points stored by the means are updated. Of each table point of the distortion compensation table that stores the input power value or amplitude value in correspondence with the compensation amount, some points are used as representative points for learning, and points other than the representative points are interpolated based on the representative points. And a distortion compensation method of a predistortion method for generating the distortion compensation table,
A distortion compensation method characterized in that the distortion compensation table is generated by combining respective curves obtained by curve interpolation in a plurality of different sections.

Images