KR100498344B1 - Lenearization method for power amplifier - Google Patents

Abstract

The present invention relates to a linearization method of a power amplifier, and more particularly, to minimize the distortion caused by nonlinearity near the saturation point of the output level by making the predistortion gain of the linearizer approach infinity at the saturation point. According to an aspect of the present invention, a method for linearizing a power amplifier includes adjusting a power amplifier output level by applying a predistortion gain function close to a saturation point to a baseband input signal, and sampling the output level of the power amplifier. Calculating an error coefficient; calculating an optimal coefficient value of the post-distorter gain function with reference to the error function; and applying an optimum coefficient value of the post-distorter gain function calculated above to obtain the predistortion gain function. The updating of the coefficients may be performed by applying a recursive least square (RLS) algorithm.

Description

Linearization method of power amplifier {LENEARIZATION METHOD FOR POWER AMPLIFIER}

The present invention relates to power amplifiers and, more particularly, to a linearization method.

In general, linear modulation such as quadrature amplitude modulation (QAM) or quadrature phase shifting (QPSK) modulates amplitude and phase at the same time, and thus is affected by nonlinearity of the power amplifier.

In particular, the greater the symbol level, the greater the peak to average ratio (PAR) and the greater the nonlinear effect of the amplifier.

In addition, when the power of the input signal is changed by modulation, an intermodulation distortion component is generated by nonlinearity of the amplifier, which not only distorts the input signal channel component but also causes interference in adjacent channels.

Therefore, to reduce the effects of nonlinearity, the output level of the amplifier is back-off into a linear region or a linearizer is used.

However, if the output level of the amplifier is used by backing off the linear region, linearity is guaranteed, but the output power and efficiency are inferior. Therefore, a linearizer is employed.

Envelope feedback, Cartesian loop, polar loop, predistortion, feedforward techniques, etc., have been proposed as a method of employing a linearizer.

Among them, the predistortion method has been widely used because of its simple structure and easy implementation, and its linearization ability is relatively high. The main method is to use a look-up table based on DSP (Digital Signal Processing).

However, the predistortion method using the lookup table requires a lot of storage capacity and does not adapt well to changes in the system.

In recent years, new predistorters that adapt well to changes in linearization elements or changes in the surrounding environment introduce a postdistorter to improve performance by utilizing the superior optimization characteristics of the RLS algorithm.

1 is a block diagram of a conventional power amplifier using a predistorter. As shown in FIG. 1, an analog-to-digital converter 110 converting a baseband QAM signal into a digital signal, and the analog-to-digital converter 110 of Digital signal processor (DSP) 120 for processing the output signal, digital / analog converter 130 for converting the output signal of the DSP 120 into an analog signal, and output signal of the digital / analog converter 130 A modulator 140 for modulating the signal, a power amplifier 150 for amplifying the output signal of the modulator 140, a demodulator 150 for detecting and demodulating the output signal of the power amplifier 150, The analog / digital converter 170 converts the output signal of the demodulator 150 into a digital signal and inputs it to the DSP 120, and provides an oscillation signal to the modulator 140 and the demodulator 160. Oscillators 181 and 182.

The modulator 140 includes a quadrature modulator (QM) 141, a band pass filter (BPF) 142, a mixer 143, and a band pass filter 144.

The demodulator 160 includes a mixer 161, a band pass filter 162, a delay unit 163, and a quadrature demodulator (QDM) 164.

1, the output signal of the analog-to-digital converter 110 ( , ) Is a signal that is digitally converted (A / D: Analog to Digital) before I and Q components of the baseband QAM signal are modulated with IF.

1, the output signal of the analog-to-digital converter 170 ( , ) Is a signal obtained by digitally converting (A / D) the I and Q signals orthogonally demodulated from the output signal of the power amplifier 150.

In FIG. 1, the DSP 120 outputs an output signal of the analog-to-digital converter 110. , ) And the output signal of the analog / digital converter 170 , Optimize the amplitude of the predistorter and the coefficients of the phase polynomial function.

The optimization process in the DSP 120 uses a recursive least square (RLS) algorithm.

After the coefficients of the polynomial converge to the optimum values, the coefficients of the amplitude and phase polynomials of the predistorter are extracted, and the predistorter can accurately represent the inverse characteristics of the power amplifier nonlinearity.

If the amplitude of the signal modulated by the modulator 140 is r (t) and the phase is Assume that the non-linear output of the power amplifier 150 is shown in Equation 1 below.

here, Is the carrier frequency Wow Are the amplitude and phase transfer functions of the amplifier, respectively.

Meanwhile, a calculation process for implementing the RLS algorithm in the DSP 120 will be described with reference to the equivalent circuit diagram of FIG. 2.

The sample value of the l th operation section inputted by the DSP 120 ( , ) Is a nonlinear function ( , ) And IF are modulated by quadrature modulator 141. here and Is the l th operation period. Wow Is absolute.

The nonlinear function ( , ) Is a polynomial representing the gain of the predistorter in the l- th operation period and can be expressed as power series of the amplitude of the input signal as shown in Equations 2 and 3 below.

Where M is the order of the predistorter polynomials and Is the kth coefficient of the polynomial.

Thereafter, the modulator 140 up-converts the quadrature-modulated signal from the mixer 143 and inputs the signal to the power amplifier 150.

The output signal of the power amplifier 150 is down-converted by the mixer 161 and then demodulated by the quadrature demodulator 160.

2, the amplitude of the input signal ( ) And the amplitude of the output signal ( ) Are computed square root (SQRT).

and Is calculated as shown in Equation 4 and Equation 5 below in the Arithmetic Unit (AU) of FIG. 2.

here, and Are the phases of the amplifier output signal and the predistorter input signal, respectively. Is the phase shift constant that affects the convergence speed.

The gain function of the post-distorter in the Nth operation period is polynomial ( , ) Is expressed as Equation 6 and Equation 7 below.

Where M is the order of the post-distorter polynomial, , Is the kth coefficient of the postdistorter polynomial , Is the M + kth coefficient.

However, the power amplifier has the best efficiency when operating in the saturation region, but the prior art shows that the performance of the polynomial predistorter is poor because the inverse nonlinear characteristic of the power amplifier diverges infinity near the saturation point and cannot be approximated by polynomial. There is a problem of falling near the saturation region.

Therefore, there is a need for a technique for linearizing the power amplifier to near the saturation point.

An object of the present invention is to provide a linearization method of a power amplifier designed to minimize the distortion caused by nonlinearity near the saturation point of the output level by bringing the predistortion gain of the linearizer to infinity at the saturation point in order to solve the conventional problems. There is this.

According to an aspect of the present invention, there is provided a method of linearizing a power amplifier, the method comprising: adjusting an output level of a power amplifier by applying a predistortion gain function close to a saturation point to a baseband input signal; Calculating an error function by sampling the output level; calculating an optimum coefficient value of the post-distorter gain function with reference to the error function; and applying the optimum coefficient value of the post-distorter gain function calculated above to the transpose The updating of the coefficient of the distortion gain function is performed by applying a recursive least square (RLS) algorithm.

That is, the present invention provides a new nonlinear function to minimize distortion caused by nonlinearity of the saturation point of the power amplifier. Is applied to the predistortion gain function of the linearizer to implement a method of linearizing the power amplifier to near the saturation point by approaching the gain infinitely near the saturation point.

This allows the predistorter to accurately implement the inverse of the amplifier nonlinearity for amplitude and phase without having to back off the output level at the saturation point.

Hereinafter, the present invention will be described in detail with reference to the drawings.

The power amplifier for the embodiment of the present invention is configured in the same way as the block diagram of FIG. 1, and the operation process of the DSP 120 is the same as that of FIG. 2, but the actual operation process is different.

In the embodiment of the present invention, the objective function of the RLS optimization algorithm is , Squared of and the gradient of the coefficients of the postprocessor polynomial is calculated. This enables global optimization and stability for the coefficients of the postprocessor polynomial.

In the embodiment of the present invention, in order to linearize the output level of the power amplifier 150 to the vicinity of the saturation point without backing off to the linear region, a nonlinear function ( ) By using the l-th group in the operation section predistortion gain function it is expressed by a polynomial, such as Equation 8, Equation 9 below.

here, Is Will look like this, Is the maximum value of the input amplitude applied near the saturation point of the amplifier.

The derivative of is the maximum value of the input amplitude, When we approach infinity, the gain function of I and Q of the predistorter , ) Can also approach infinity and linearize the power amplifier near the saturation point.

The post-distorter gain function in the nth operation section is expressed by Equation 10 and Equation 11 below.

here, to be.

n error function of the RLS algorithm consisting of a sample of the l-th operation period in the second operation period ( , ) Is as shown in [Equation 12] and [Equation 13] below.

here to be.

if, If the coefficients of the post-distorter polynomial converge to the optimal values in Two error functions in , ) Is '0', and the following [Equation 14] and [Equation 15] are established from [Equation 12] and [Equation 13].

And, In this case, it is assumed that Equation 16 and Equation 17 below hold.

[Equation 18] and [Equation 19] below are established based on the same assumptions as [Equation 16] and [Equation 17].

However, the amplitude of the amplifier distortion output function ( ) And phase ( ) Is a single valued function, so the post-distorter gain function ( , ) Is also a cost function.

Therefore, Equations 20 and 21 below hold.

here, to be.

Accordingly, after convergence ego Becomes That is, the characteristics of the power amplifier are linearized.

However, after convergence, the coefficients of the postprocessor polynomial are the same as the coefficients in the previous calculation interval. ego Becomes

Therefore, the following Equations 22 and 23 are established.

Accordingly, the following relation can be obtained by substituting [Equation 8], [Equation 9], [Equation 10], and [Equation 11] into [Equation 22] and [Equation 23].

here, to be.

The coefficient of the predistortion polynomial can be updated using the relational expression.

The objective function in the nth operation section is given by Equations 24 and 25 below.

Re-arranged [Equation 24], [Equation 25] can be expressed as shown in [Equation 26], [Equation 27] below.

here, : Reference Distortion In-Phase Signal

: Reference Distortion Quadrature Signal

Data vector

: Coefficient vector of postprocessor polynomial

: Coefficient vector of the postprocessor polynomial, Means transpose.

And, represented by the above [Equation 26], [Equation 27] and Each and , By partial differential, the following [Equation 28] and [Equation 29] can be obtained.

If [Equation 28] and [Equation 29] are each '0' , In this case, Equation 28 and Equation 29 are expressed as Equation 30 and Equation 31 below.

In this case, when [Equation 30] and [Equation 31] is replaced with [Equation 32], [Equation 33], and [Equation 34] below, it can be expressed as [Equation 35] and [Equation 36]. have.

Substituted procession( ) Is not singular Therefore, the coefficient of the postprocessor polynomial is expressed as Equation 37 and Equation 38 below.

From this, the RLS algorithm obtains relations as shown in Equations 39 to 44 below.

here, Is the gain vector, , Is an estimation error.

[Equation 42], [Equation 43], [Equation 44] is , To initialize with Is a very small constant, silver Zero vector.

On the other hand, the simulation results for the above process in the embodiment of the present invention are as follows.

First, it is assumed that the input modulated signal uses a 16 QAM random modulated signal passed through a 25% raised cosine filter and that quadrature demodulation (QDM) is perfect.

Also, to evaluate the linearity near the saturation point of the power amplifier, we assumed '0' dB input backoff at the saturation point for the I, Q peak signal power, respectively.

Input Amplitude and Phase Transfer Function of Power Amplifiers Predistorted Wow, Is expressed as Equation 45 and Equation 46 below.

In addition, the predistortion polynomial used in the simulation is represented by Equations 47 and 48 below.

The coefficients of the predistortion polynomial and the coefficient vectors of the postprocessor polynomial are initialized as follows.

, , , , , , ,

, , , , , , ,

Therefore, in the embodiment of the present invention, the phase shift constant ( )end After convergence of the ramen predistorter, the polynomial is expressed as the following equation.

Figure 3 shows the gain function of the predistorter , 4 and 5 show the AM / AM characteristics and AM / PM characteristics of the linearized amplifier.

Thus, as shown in Fig. 3, it can be seen that the gain function approaches infinity at the maximum input value, and the predistorter is very effective in implementing the inverse of the amplifier nonlinearity with respect to amplitude and phase.

Accordingly, a power amplifier system having a very high linearity up to a saturation point can be realized even if the power output level is not backed off to the linear region.

Figure 6 shows the amplifier output power spectrum with or without predistorter. As shown in FIG. 6, it can be seen that there is an improvement of about 40 dB or more at the edge of the amplifier output power spectrum band.

That is, in the embodiment of the present invention, the signal having the output power spectrum as shown in Fig. 6 restores the power spectrum of the baseband I and Q input signals to almost the original state.

As described in detail above, the present invention can minimize the distortion caused by the nonlinearity of the saturation point by applying the nonlinear function to the predistortion gain function of the linearizer so that the power amplifier is linear to almost the saturation point.

Therefore, even if the output level is not backed off at the saturation point, the predistorter accurately implements the inverse function of the amplifier nonlinearity with respect to the amplitude and phase, thereby achieving a very high output power and optimum efficiency even near the saturation point.

In addition, the present invention is difficult to manufacture and tune a circuit having an analog predistorter and an envelope detector, and inaccuracies occur due to the inaccuracy of the circuit. It is effective.

1 is a block diagram of a power amplifier using a predistorter.

2 is an equivalent circuit diagram of the DSP in FIG.

3 is a waveform diagram showing a gain function of a predistorter;

4 is a waveform diagram showing AM / AM characteristics of a power amplifier.

5 is a waveform diagram showing AM / PM characteristics of a power amplifier.

6 is an output spectral waveform diagram with or without a predistorter of a power amplifier.

Explanation of symbols on the main parts of the drawings

120: DSP (Digital Signal Processor)

140: modulator 150: power amplifier

160: demodulation unit

Claims (3)

In the linearization method of the power amplifier, Adjusting the output level of the power amplifier by applying a predistortion gain function close to the saturation point to the baseband input signal; Sampling an output level of the power amplifier to calculate an error function; Calculating an optimal coefficient value of a post-distorter gain function with reference to the error function; And updating the coefficient of the predistortion gain function by applying the optimum coefficient value of the postdistorter gain function calculated above. The predistortion gain function of claim 1 wherein Nonlinear functions ( The linearization method of the power amplifier, characterized in that to perform the operation as shown in the following equation. here, Is Will look like this, Is the maximum value of the input amplitude applied near the saturation point of the amplifier. The method of claim 1, wherein the recursive least square (RLS) algorithm is established by the following relationship. here, Is the gain vector, , Is an estimation error. only, , Initialize to, Is a very small constant, silver Zero vector.