摘要
针对短波射频功放的非线性失真及记忆效应失真问题,提出了一种直接学习结构的MP模型预失真方案,采用Filtered-X LMS(NFXLMS)算法对建立的预失真模型进行训练辨识。仿真分析,针对MP模型高功率放大器,预失真后的三阶互调分量改善了52.2 d B,五阶互调分量改善了48.85 d B,与现有的IIR Wiener预失真器相比较,进一步提高了功放输出的线性度。同时,在DSP+FPGA平台上对MP模型预失真算法进行实测,结果表明,该预失真器能有效改善实际功放的非线性失真,具有较好的线性化效果。
For the problem of nonlinear distortion and memory effect distortion of shortwave radio frequency amplifier,a new MP model pre-distortion scheme based on direct learning method is proposed,in which the scheme uses the Filtered-X LMS（NFxLMS）algorithm to train and identify the pre-distortion model. Simulating the model,the results show under the conditional of MP model high power amplifier ,the design scheme can improve the third-order inter-modulation distortion by 52.2 dB and the fifth-order inter-modulation distortion by 48.85 dB. Compared with the existing IIR Wiener pre-distorter,MP model pre-distorter can greatly improve the linearity of the amplifier output. Meanwhile,doing the actual testing for MP model pre-distortion on the DSP＋FPGA platform,the results show that the pre-distortion can effectively improve the actual power amplifier nonlinear distortion.It can bring better linear effects.
出处
《电子器件》
CAS
北大核心
2016年第6期1369-1374,共6页
Chinese Journal of Electron Devices
关键词
功率放大器
预失真
直接学习法
记忆多项式
行为模型
power amplifier
pre-distortion
direct learning
memory polynomial
behavioral models
