Investigating the performance of cross-coupled CMOS LC-VCOs using genetic algorithm

Abstract

A fast simulation-based and parasitic-aware LC-VCO optimizer CAD tool, using Elitist Non-dominated Sorting Genetic Algorithm (NSGA-II) is presented. The proposed CAD tool optimizes the power consumption and phase noise of each conventional LC-VCO (e.g. NMOS-only, PMOS-only and complementary), simultaneously. In order to evaluate the fitness of the LC-VCO, the harmonic balance analysis is performed by HSPICE RF and then the optimizer core, implemented in MATLAB, manipulates the fitness and suggests some optimized designs. This CAD tool has the ability to optimize Figure of Merit of each configuration too. Design methodologies and comparison of performance of each structure, based on trade-off curves, are also presented and in a case study the cross-coupled LC-VCOs are optimized for 2.4-GHz central frequency (802.11 IEEE) in a commercial 0.18μm RF-CMOS technology.