In The Feedforward System Research And Design Of Adaptive Control

With the development of modern communication technology towards tohigh-speed, broadband, multi-carrier, it is very sensitive to the non-linear characteristicsof the communication system. In such a system, the power amplifier is an essentialsection and also the main component to produce the non-linear distortion. The poweramplifier is the key to promote the development of "energy-saving" and "green". Theadaptive-feedforward power amplifier (AFFPA) with the advantages of high linearity,broadband, unconditionally-stable and good-noise performance has attracted muchattention. It is a hot area in the power amplifier research. The AFFPA still has high-cost,complex-structure, and other disadvantages. Therefore to find a better adaptive controlscheme which could reduce the cost and simplify the structure is of critical importance.This paper described a research on the adaptive control section of AFFPA.The article describes principles and characteristics of minimized power, signalcorrelation and pilot signal control methods. Further an improved method based on thecombination of the signal correlation and pilot signal is used in the scheme. It wouldeliminate some challenges. For example, when separately using the signal correlationmethod, the convergence of α in the signal cancelation loop influences β’s in the errorcancelation loop. Another eliminated problem is, when using the pilot signal methodalone, in the signal cancelation loop the pilot signal through the main power amplifierwill produce additional distortion. The variable step adaptive algorithm——NLMScould improve the convergence performance of the adaptive control section. To analyzethe behavior of the AFFPA, some special simulations in Matlab environment wasdeveloped. Some promising results have been achieved. Compared to using signalcorrelation method alone, the improvement of the linearity in the optimized AFFPA(using the combined method) can be increased by about50%. Comparing the NLMSwith the traditional LMS with the same step value, the convergence of the optimizedAFF power amplifier can be significantly improved, and its speed is about35%faster.The frequency in the AFFPA is very high so it’s difficult to detect. The optimizedfrequency downconverting approach used in the scheme derived from the references proposed by P. B. Kenington. It can make the sampled bandwidth narrow (<5KHz) andthe frequency low to KHz level, and greatly reduce the requirements of the samplingfrequency and digital signal processing.The adaptive control section uses a FPGA as its digital signal processor because itcan be easily converted to ASIC, which could improve the AFFPA’s cost, size,complexity, efficiency and so on, after the design on the FPGA platform is verified.This paper describes in detail the way of the NLMS implemented in FPGA. A digitalhardware platform which can meet the requirements of the adaptive control has beendesigned. It is composed of FPGA, ADC, DAC and others. Finally, the logics toimplement digital signal processing into the FPGA and their functional verification havebeen described.