Design of Adaptive Feedforward RF Power Amplifier

fighting

Design of Adaptive Feedforward RF Power Amplifier [Copy link]

Design of Adaptive Feedforward RF Power Amplifier

Luo Yuxia, Jia Jianhua (School of Electronic and Information Engineering, Tongji University, Shanghai)

1. Introduction The rapid development of modern wireless communication is increasingly moving towards increasing information capacity, improving channel spectrum utilization and improving linearity. On the one hand, people widely use high-power microwave transistors working in Class A and B states to improve transmission power and utilization efficiency; on the other hand, the introduction of passive and active devices, the adoption of multi-carrier configuration technology, etc., will lead to intermodulation distortion of the output signal. Therefore, when designing an RF power amplifier, it must be linearized so that the output signal can obtain better linearity. Commonly used linearization techniques include: power back-off, pre-distortion, feedforward, etc. Among them, power back-off technology can effectively improve the linearity of narrowband signals, while pre-distortion technology and feedforward technology, especially feedforward technology, have become the main technologies used to improve the linearity of broadband signals due to their advantages of high calibration accuracy, high stability and no bandwidth limitation. This paper first briefly describes the ordinary feedforward linearization technology, and then improves it on this basis, adds an adaptive algorithm, and extracts out-of-band signals for adjustment through signal envelope detection technology, so as to achieve the purpose of improving the linearity of the output signal. 2. Basic Principle of Feedforward The most basic principle of feedforward amplifier is shown in Figure 1. It consists of two loops: loop 1 consists of power divider, main amplifier, coupler 1, attenuator 1, phase shifter 1, delay line 1, and synthesizer 1. The input RF signal, that is, two pure carrier signals, is divided into two branch signals after the power divider: the upper branch is the main power amplifier branch, and the pure RF carrier signal generates an amplified carrier signal and an intermodulation distortion signal after passing through this branch; the lower branch is the auxiliary branch, and the pure RF carrier signal is delayed after passing through this branch. The nonlinear distortion signal output by the main power amplifier branch passes through attenuator 1 and phase shifter 1, and is synthesized with the signal output by the auxiliary branch in synthesizer 1. Attenuator 1 and phase shifter 1 are adjusted to make the two branch signals obtain equal amplitude, 180' phase difference and equal delay. At this time, the RF carrier signal of the main power amplifier branch can be effectively offset, and the intermodulation distortion signal generated by the nonlinear amplification of the main amplifier can be extracted. Therefore, this loop is also called the RF carrier signal elimination loop. Loop 2, also called distortion signal elimination loop, consists of delay line 2, auxiliary amplifier, attenuator 2, phase shifter 2, and coupler 2. There are also two branches: the upper branch delays the nonlinear distortion signal output by the main amplifier and sends it to coupler 2; the lower branch amplifies, attenuates, and phase-shifts the intermodulation distortion signal extracted by loop 1 and sends it to coupler 2. Attenuator 2 and phase shifter 2 are adjusted until the intermodulation distortion signal in the signal output by coupler 2 is the smallest, that is, the IMD is the smallest. At this time, the output signal is the amplified RF signal. 3. Adaptive feedforward RF power amplifier 3.1 Principle and algorithm of adaptive feedforward circuit Since the carrier signal cancellation requirement in the feedforward system is very high, changes in the internal and external environment, such as: input signal power, DC bias voltage and ambient temperature, are easy to cause the carrier signal cancellation failure. Therefore, it is very necessary to introduce adaptive technology to obtain the matching of the carrier signal in amplitude, phase and delay in time. The structure of the adaptive feedforward system is shown in Figure 2. It consists of three loops: loop 1 is mainly used to extract intermodulation distortion signals, loop 2 is mainly used to eliminate distortion signals, and loop 3 is mainly used to detect intermodulation distortion signal power. Assume that the input signal is υin(t), the output signal after the main amplifier is υρα(t), a part of υρα(t) is coupled to the vector modulator 1, and the modulation coefficient of the vector modulator 1 is represented by the complex coefficient α. At the same time, the main amplifier is simplified into a memoryless nonlinear model, and its AM/AM and AM/PM transfer functions can be simply represented by the complex voltage gain G(χ), where χ represents the instantaneous power. Then the signal υα(t) output from the vector synthesizer 1 can be expressed as: In the specific implementation structure, a power divider 2 is added after the synthesizer 1, and its purpose is to divide the signal υd(t,g, ψ) is used for power detection. Obviously, if α is adjusted so that the amplitude, phase and delay of the two input signals of synthesizer 1 are matched, then the power detected here will only be the average power of the intermodulation distortion signal υe(t), which is very small. In other words, if the power output of power divider 2 is detected to be small enough, then the adjustment of α is optimal, that is, the RF carrier signal has been eliminated to the greatest extent, and only the intermodulation distortion signal υe(t) remains. The intermodulation distortion signal entering loop 2 is amplified by the auxiliary amplifier, adjusted by vector modulator 2 (whose modulation coefficient is a complex coefficient β), and then synthesized with the output signal of the main amplifier through delay line 2 in synthesizer 2. This loop also uses adaptive technology to adjust the amplitude and phase of the intermodulation distortion signal. Its mathematical principle is as described above, but in terms of the implementation structure, it is different from loop 1. Loop 1 determines whether the RF carrier signal is offset to the minimum value by directly detecting the output signal of synthesizer 1, while loop 2 needs to introduce a third loop when determining whether the intermodulation distortion signal is offset to the minimum value. We know that for the same power output signal, the envelope of the linear signal is larger than the envelope of the nonlinear signal, and the envelope difference signal between the two is the intermodulation distortion signal. Minimizing the envelope difference signal can improve the linearity of the output signal to the greatest extent, thereby reducing IMD. The working principle of loop 3 is here. The two signals it processes are one linear signal, that is, the RF carrier signal passing through the delay line 3 and the power divider 4, and the other nonlinear signal, that is, the signal output by the synthesizer 2 after passing through the feedforward system loop 1 and loop 2. First, the loop performs power detection on the synthesized signal of the two signals and adjusts the vector modulator 3 until the detected power is minimum. At this time, it can be considered that the linear signal and the nonlinear signal have the same carrier output power. Then, the envelope detection is performed on the two signals respectively, and the envelope difference signal is applied to the vector modulator 2, that is, the out-of-band intermodulation distortion signal is minimized continuously, and a high linearity output signal is obtained. 3.2 Computer simulation application A 25 W power amplifier was designed on a computer simulation system using an LDMOS field effect transistor with a peak power of 180 W, and two carrier signals with a frequency interval of 1 MHz were input to it to generate third-order and fifth-order intermodulation distortion signals. Figure 3 shows the output of the signal when the adaptive feedforward technology is not used. At this time, IMD3 can only reach about -55 dBc, and IMD5 can only reach about -56 dBc. Figure 4 shows the output of the signal after the technology is used. At this time, IMD3 can reach about -72 dBc, and IMD5 can reach about -76 dBc. The degree of improvement is obvious. 4. Conclusion This paper uses adaptive feedforward technology and envelope detection technology to design RF power amplifiers. Since this technology takes into account the problems that may be encountered in practice, it simplifies complex problems and proves its feasibility not only in theory but also in practice. Computer simulation experiments show that this adaptive feedforward technology can indeed effectively improve the nonlinear distortion of power amplifiers. Of course, the application of this technology needs to be further strengthened.