NSGA-Based Parasitic-aware Optimization of a 5GHz Low-noise VCO


A parasitic-aware RF synthesis tool based on a nondominated sorting genetic algorithm (NSGA) is introduced. The NSGA-based optimizer casts the design problem as a multiobjective optimization problem and offers multiple solutions along the Pareto optimal front. Monte-Carlo simulations are then performed to efficiently assess sensitivity at solution points with respect to process, voltage, and temperature (PVT) variations. An example design of a 10 mW 5 GHz voltage-controlled oscillator (VCO) in 250 nm SiGe BiCMOS achieves a 12% tuning range with a phase noise of -133 dBc/Hz at 3 MHz offset. The figure-of-merit (FOM) is 188 dBc/Hz and power-frequency-tuning normalized FOM (PFTN-FOM) is -4dB.