KR100445326B1 - Linear Power Amplifier using the Digital Signal Processor - Google Patents

Abstract

The present invention relates to an apparatus for linearizing a power amplifier using a digital signal processing device (DSP). In the predistortion linearization device, the present invention relates to a baseband QAM signal. A first AD converter for converting A / D (Analog to Digital) before modulating the component to IF; Orthogonal demodulation of the power amplifier output in an orthogonal demodulator A second AD converter for A / D converting the signal; A digital signal processing apparatus (DSP) for inputting the signals converted by the first AD converter and the second AD converter and optimizing using an RLS algorithm and extracting coefficients of the predistortion polynomial therefrom; Calculated by the digital signal processing device (DSP) It is composed of DA converter to convert the signal into an analog signal.

According to the present invention, there is provided a linearization apparatus having optimized coefficients, excellent performance, and capable of converging an amplitude error and a phase error within 1%.

Description

Linear power amplifier using the digital signal processor (DSP)

The present invention relates to a linearization apparatus for canceling distortion components of a transmission signal by nonlinear components of a power amplifier. In particular, the present invention relates to a linearization apparatus of a power amplifier using a digital signal processing device (DSP), which allows the characteristic of the power amplifier to be linearized to a saturated region.

In general, a linear modulation method is widely used to secure high bandwidth efficiency and transmit a large amount of data in a wireless communication system, and simultaneously modulates and transmits an amplitude and a phase of an RF signal having information. . In addition, since the wireless communication system must suppress interference to other communication system users as much as possible, transmission and reception should be maintained within the allocated bandwidth and energy radiation out of the bandwidth should be minimized. However, power amplifiers, which are most important in the performance of wireless communication systems, generally have the characteristic of nonlinearly distorting the signal. These non-linear characteristics include Amplitude Modulation to Amplitude Modulation (AM / AM), which indicates a decrease in output signal power as the input signal power increases, and AM / PM (Amplitude Modulation) as a change in phase of the output signal as the input signal power increases. to Phase Modulation). Due to the nonlinear characteristics of the power amplifier, distortion of the transmission signal and interference on adjacent channels are strictly regulated as a standard specification of the communication system. Accordingly, various linearization techniques have been proposed, but predistortion linearizers using polynomial predistorters are widely used to satisfy the high efficiency and linearity of power amplifiers.

1 is a view showing the principle of a conventional predistortion linearization device. According to Fig. 1, the predistorter has a nonlinear function having the inverse of the power amplifier transfer function in front of the nonlinear power amplifier. This nonlinear function is implemented by inserting a nonlinear circuit between the input signal and the power amplifier. This nonlinear circuit combines the intermodulation distortion (IMD) component of the power amplifier. An IMD component having a phase difference is generated to cancel out unnecessary IMD components. Such predistortion may be implemented at baseband, intermediate frequency (IF) or RF.

Figure 2 shows a block diagram of a power amplification system using the IF band predistortion linearizer according to the prior art. Referring to FIG. 2, the circuit of the IF band predistortion linearizer according to the related art is characterized by an in-phase component of a baseband quadrature amplitude modulator (QAM) signal. ) And quadrature A quadrature modulator 4a for converting the signal into a complex input signal, an IF local oscillator 8 for converting these signals into an IF band, an envelope detector 6 for detecting the amplitude of the predistorter input signal, and The outputs of the predistorter (1), power amplifier (3) and power amplifier are the baseband in-phase components ( ) And quadrature Orthogonal demodulator (4b) for converting the signal, RF local oscillator (7), bandpass filter (2) for filtering signals in IF and RF bands, and error estimator for extracting coefficients of predistortion polynomial 5) consists of.

The circuit flow of the IF band predistortion linearizer according to the prior art is as follows. Of the baseband Quadratue Amplitude Modulator (QAM) signal (9) Wow And converting the complex input signal of the IF band by the quadrature modulator 4a using the IF local oscillator 8, and the complex input signal is filtered by the bandpass filter 2, and then the predistorter 1 It becomes an input signal. The predistorter 1 should have an inverse characteristic of the power amplifier nonlinear characteristic, which is composed of a polynomial function, and the coefficient of the predistortion polynomial is updated by receiving the coefficient from the error estimator 5.

The output of the power amplifier 3 is down-converted to the IF band using the RF local oscillator 7 and then filtered by a bandpass filter 2 ". Using this signal and the predistorter input signal, the quadrature demodulator ( In 4b) And Generate a signal.

The envelope detector 6 detects the amplitude of the complex input signal which is the predistorter input signal and And The error estimator (5) calculates the amplitude and phase errors of the input and output signals and extracts the coefficients of the predistortion polynomial from them.

However, since the predistortion linearizer according to the prior art predistorts in-phase and quadrature components in the IF stage, a multiplier, an adder and a quadrature modulator, and an envelope detector 6 are required to detect the amplitude of the IF signal. And all of these must be configured as analog circuits, and even though the algorithm can calculate the correct convergence coefficients, The disadvantage is that the predistortion of the signal can be inaccurate. In practice, the performance of a predistortion linearization system using analog devices is known to be incompatible with the simulation results.

The present invention has been invented to solve the problems of the prior art, and all processes are processed by the DSP (Digital Signal Processor) directly without using an analog predistorter and an envelope detector. Wow In addition to predistortion at baseband, we also improve the linearizer performance and power efficiency of the power amplifier system by improving the linearizer performance in the saturation region by using the gain polynomial of the predistorter with optimized coefficients. .

1 is a view showing the principle of a conventional predistortion linearization device.

Figure 2 is a block diagram of a power amplification system using the IF band predistortion linearization device according to the prior art.

3 is a block diagram of a power amplification system using a baseband predistorter in accordance with the present invention.

4 is a detailed diagram of a DSP operation process with a predistorter and a postdistorter according to the present invention.

5 is an AM / AM and AM / PM characteristic diagram of a general TWT large power amplifier.

Figure 6 illustrates the normalized absolute error of amplitude and phase at the amplifier output according to the present invention.

7 is a view showing normalized power spectrum density of an amplifier output with and without a predistorter according to the present invention.

<Description of Symbols for Main Parts of Drawings>

1: predistorter

2, 2 ', 2 ", 12, 12', 12": bandpass filter

3, 13: power amplifier 4a, 14a: quadrature modulator

4b, 14b: orthogonal demodulator 5: error estimator

6: envelope detector 7, 17: RF local oscillator

8, 18: IF local oscillator 9: baseband QAM signal

10: Digital Signal Processing Device (DSP)

16a, 16b: AD converter 19: DA converter

20: algorithm

In order to achieve the object of the present invention, the linearization of the power amplifier using a digital signal processing device (DSP) to linearize the characteristics of the power amplifier to the saturation region in order to cancel the distortion component of the transmission signal by the nonlinear components of the power amplifier The present invention will be described in detail with reference to the accompanying drawings. Figure 3 is a block diagram of a power amplification system using a baseband predistorter according to the present invention. The present invention provides a method for canceling a distortion component of a signal caused by a nonlinear component of a power amplifier between an input signal and a power amplifier in front of the nonlinear power amplifier. Insert a non-linear circuit into the baseband of the baseband QAM signal A first AD converter which performs A / D conversion before modulating the component to IF; orthogonal demodulating the output of the power amplifier in an orthogonal demodulator; A power amplifier linearization apparatus comprising a second AD converter for converting signals into A / D, wherein signals converted by the first AD converter and the second AD converter are input and optimized using a recursive least squares (RLS) algorithm. A digital signal processing device (DSP) 10 for extracting coefficients of the predistortion polynomial; Calculated by the digital signal processing device (DSP) DA converter 19 for converting the signal into an analog signal.

According to FIG. 3, , Of the baseband QAM signal Analog to Digital (A / D) conversion of the first AD converter 16a before modulating the component to IF. , Is orthogonal demodulation of the output of the power amplifier in the orthogonal demodulator 14b. The signal is A / D converted by the second AD converter 16b. These signals enter the digital signal processing device (DSP) 10 and are optimized using the Recursive Least Squares (RLS) algorithm 20 and extract the coefficients of the predistortion polynomial therefrom. Computed at DSP 10 Is converted into an analog signal by the DA converter 19, and is converted into an IF signal by the quadrature modulator 14a using the signals and the signals of the IF local oscillator 17. The IF signal is filtered in the first bandpass filter 12 and then mixed with the RF local oscillator 17 and upconverted to an RF signal. The RF signal passes through the second bandpass filter 12 'of the RF band and becomes an input signal of the power amplifier 13. The output of the power amplifier 13 is mixed with the RF local oscillator and down-converted to the IF band, and then becomes the input signal of the quadrature demodulator 14b via the third band pass filter 12 ".

4 is a detailed diagram of a DSP operation process with a predistorter and a postdistorter according to the present invention. 4, input to the DSP Sample value of the first operation interval and Nonlinear function , Complex operation with and modulate IF. here and silver In the first operation interval Wow Is absolute. , silver It is a polynomial representing the gain of the predistorter in the first operation period and can be expressed as power series with respect to the amplitude of the input signal as follows.

… … … … ( One )

… … … … ( 2 )

Where M is the order of the predistorter polynomials and Is polynomial Second coefficient. The power amplifier 13 output is down-converted and then demodulated using an orthogonal demodulator 14b. Amplitude of input signal And amplitude of output signal Are the values calculated by the square root (SQRT) in the figure. Equations (1) and (2) above are the predistorter gain polynomials used when the power amplifier output level is back-off in the prior art. However, if a nonlinear function is used to linearize the power amplifier output level near the saturation point without backing off to the linear region, The predistorter gain function polynomial in the first operation section is

… … … … (3)

… … … … ( 4 )

Where M is the order of the predistorter polynomial, , and , Is polynomial Th coefficient, Is the amplitude of the input signal

Is, 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, Approaches infinity when This results in the predistorter Wow Gain function , Again, infinity, the power amplifier can be linearized near the saturation point.

and Is calculated in the Arithmetic Unit (AU) of FIG. 4 as follows.

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

The polynomial of the post-distorter in the first operation section is

here, ego And Is the coefficient of the post-distorter gain polynomial.

In the first operation interval The RLS algorithm error function composed of samples of the first operation interval is as follows.

The cost function in the first operation section is given by the following equation.

and Each and Differentiate by and use the RLS algorithm to minimize the amplitude and phase errors. Obtaining the coefficients of the predistorter gain polynomial when the amplitude error and the phase error converge within 1%, and substituting these coefficients into equations (3) and (4), the predistorter gain polynomial having the inverse function of the power amplifier nonlinearity It is to find. As a result, the power amplification system is linearized.

Hereinafter, the performance of the power amplifier linearization system according to the present invention will be described by examples. 5 is a diagram illustrating AM / AM and AM / PM characteristics of a general TWT large power amplifier. The amplifier has a higher nonlinearity than the solid state power amplifier (SSPA), which is widely used in mobile communication base stations. The input modulated signal uses a 16 QAM random modulated signal passed through a 25% raised cosine filter, and the quadrature demodulator is assumed to be perfect. Also, to evaluate the linearity in the saturation region of the power amplifier Assume 0 dB input backoff at the saturation point for peak signal power, respectively.

The predistorter gain function polynomial used in the simulation is

The coefficients of the predistorter polynomial and the coefficient vectors of the postdistorter polynomial are initialized as follows.

The phase shift constant used in this algorithm ( ) The coefficients of the polynomial after convergence of the predistorter are

Figure 6 shows the normalized absolute error of amplitude and phase at the amplifier output according to the present invention. Figure 6 shows the input and output amplitudes and phase errors normalized to the maximum of the error. The amplifier's output amplitude error converges within the 51st operation period and the phase error within 1% of the 48th operation period.

Figure 7 shows the power spectral density in decibels (dB) of the amplifier output with or without a predistorter according to the present invention. The dotted line represents the filtered 16 QAM input signal, and the solid line represents the spectral regrowth characteristics due to the nonlinearity of the power amplifier in the absence of the predistorter. At the edge of the amplification band, it exhibits about 65dB of spectral regrowth. The single-dot chain line uses the existing 7th order polynomial predistorter using equations (1) and (2). The improvement is about 27dB. Therefore, like the conventional method, the 7th-order polynomial predistorter does not have good linearization performance in the saturation region. The dashes show the linearization performance when simulated with the proposed device and technique. It can be seen that there is a spectral improvement effect of about 63 dB at the edge of the amplification band. Therefore, by using the predistortion gain polynomial having an optimized coefficient as in the present invention, the linearity of the amplifier can be greatly improved even when the amplifier is operated near the saturation point.

In addition, in the conventional polynomial predistorter, when the amplifier operates near the saturation region, the performance of the linearizer is significantly reduced because the inverse nonlinear characteristics of the amplifier near the saturation region cannot be approximated by the polynomial. In order to solve this problem, the coefficient of the predistorter gain polynomial is optimized to linearize the characteristics of the power amplifier from the polynomial predistortion linearizer to near the saturation region. Using this optimized coefficient, the system's predistortion gain function can sufficiently compensate for the nonlinear characteristics of the power amplifier even near the saturation point. Therefore, it is possible to linearize around the saturation region without backing off the input level of the power amplifier. Since the power amplifier has the best power efficiency when operating in the saturation region, the power amplifier system as in the present invention has high linearity, optimum efficiency and High output power can be obtained.

In addition, a stable characteristic can be obtained by predistorting at baseband, and in a narrow band system such as GSM (Global System for Mobile Communications) or in many other cases, out-of-band ACPR is sufficient to be below -71dB, so that the computation time Some of the gain functions of the predistorter can be omitted to reduce the

As described above, according to the present invention, the linearizer performance in the saturation region can be improved by adopting a predistortion structure at baseband and using a gain polynomial of a predistorter having an optimized coefficient. Conventionally, the QAM signal modulated by IF is predistorted using an analog device. However, in the present invention, a method of pre-distorting the in-phase component and quadrature component of the baseband QAM signal through DSP operation is adopted. Since the circuit with the predistorter and the envelope detector has difficulty in fabrication and tuning and its characteristics are unstable, in the present invention, the implementation of the system is simpler and the linearity is improved by allowing most operations including the predistortion to be performed by the DSP. According to the present invention, the linearization performance and power efficiency can be greatly improved by introducing not only a nonlinear amplification system such as a TWT amplifier but also an amplifier such as SSPA which is widely used in a mobile communication system. .

Claims (4)

A nonlinear circuit is inserted between the input signal and the power amplifier in front of the nonlinear power amplifier in order to cancel the distortion component of the signal caused by the nonlinear component of the power amplifier. A first AD converter for A / D converting the component before modulating the IF; Orthogonal demodulation of the power amplifier output in an orthogonal demodulator A power amplifier linearization device comprising a second AD converter for A / D converting a signal, the conventional power amplifier linearization device for canceling a distortion component of a signal by a nonlinear component of the power amplifier, A digital signal processing device (DSP) for inputting the signals converted by the first AD converter and the second AD converter, optimizing using an RLS algorithm, and extracting coefficients of the predistortion polynomial therefrom; Calculated by the digital signal processing device (DSP) Linear amplifier of the power amplifier using a digital signal processing device (DSP), characterized in that it comprises a DA converter for converting the analog signal. The method of claim 1, Anything in the digital signal processing device (DSP) The gain of the baseband predistorter in the second operation section is a linearizer of the power amplifier using a digital signal processing device (DSP), characterized in that calculated by the following polynomial. (here Is the order of the predistorter polynomial, , and , Is polynomial Th coefficient, Is the amplitude of the input signal, in the polynomial Where Is the maximum input amplitude applied near the amplifier's saturation point. The method of claim 1, In the digital signal processing apparatus (DSP), the gain of the baseband post-distorter in any n- th operation period is calculated by the following polynomial, and the linearization device of the power amplifier using the digital signal processing apparatus (DSP) . ( here, ego And Is the coefficient of the post-distorter gain polynomial, Where Is the amplitude of the output signal, Is the phase of the amplifier output signal, Is the phase of the predistorter input signal, Is the phase shift constant that affects the speed of convergence) The method of claim 2 or 3, In the first operation interval RLS Algorithm Error Function Calculated by the formula of, The cost function in the first operation section is Calculated by the formula of, and Each and A linearization device of a power amplifier using a digital signal processing device (DSP), characterized in that the partial differential, and to minimize the amplitude error and phase error using an algorithm.