Simulated Annealing Particle Swarm Optimization for High-Efficiency Power Amplifier Design

Abstract

In this article, a method for design automation high-efficiency power amplifiers (PAs) based on an optimization-oriented is proposed. The advantages of using particle swarm optimization to design PAs in electronic design automatic tools are analyzed, and then, simulated annealing is combined to form simulated annealing particle swarm optimization (SA-PSO). In this article, the perfectly inelastic collision used in conventional PSO is replaced by the perfectly elastic collision to improve the searching ability through reducing kinetic energy consumption. A stepped fitness function is defined to avoid the performance pit at a single-frequency point, and the concept “missing and returning” is introduced to solve the nonconvergence problem in simulation. Based on the improved SA-PSO algorithm, a broadband 10 W and a triband 6-W PA are designed. The broadband 10-W PA achieves an average drain efficiency (DE) of 62.5% and an output power of 41.2 dBm in the 0.8-3-GHz frequency band, respectively; the triband 6-W PA achieves an average of 57.4%/47.9%/47.3% DE and 37.7-/38.3-/37.4 dBm output power in 3.35-3.55-/5.75-5.95-/7.65-7.95-GHz frequency bands, respectively, while the gains of most stopbands reach less than 0 dB. With digital predistortion technology, the adjacent channel power ratio (ACPR) can reach less than -45 dBc.