Multiobjective Optimization of Class-F Oscillators

Zhan Qu,Zhenjiao Chen,Xingqiang Shi,Ya Zhao,Guohe Zhang,Feng Liang

DOI: https://doi.org/10.1109/tvlsi.2024.3449567

2024-01-01

IEEE Transactions on Very Large Scale Integration (VLSI) Systems

Abstract:To address the complex nonlinear problem of determining class-F voltage-controlled oscillator (VCO) dimensions, this article introduces an electronic design automation (EDA) framework that rapidly optimizes multiple design objectives yielding superior outcomes. The framework incorporates fast frequency determination, harmonic alignment, and extremal optimization of multiobjective particle swarm optimization with crowding distance (FHE-MOPSO-CD), an efficient algorithm we developed specifically for class-F VCOs, which includes transformer-based tank circuit strategies and extremum optimization techniques. Using a 55-nm CMOS process, this algorithm optimized various class-F VCO topologies, achieving excellent metrics and confirming its versatility. Optimization results indicate that at a 10-MHz offset, the figure of merit (FoM) is at least 8.81 dBc/Hz higher than values reported in the literature. Compared with other analog/RF dimension optimization methods, our approach yielded a higher hypervolume, indicating better convergence and greater diversity of solutions.