Fast Algorithm For The Multi-objective Optimization Of CMOS Opamp

The design of analog integrated circuits is a bottleneck in the field of integrated circuit design,and how to improve the level of automated design of analog integrated circuits is an urgent problemnowadays. This paper researches a key issue in the automatic synthesis of analog CMOS unit circuit,namely, the optimization of the design parameters.Single-objective function optimization is used in traditional circuit optimization. In order toachieve the design search in the full performance parameter space, the multi-objective optimizationbased on genetic algorithm has been introduced in recent years. Due to a large number ofcomputation of objective function during the optimization process, the simulation-basedoptimization approach costs a lot of calculation time. As for small size devices, althoughequation-based optimization approach is very fast, its accuracy is very poor.This paper has studied mainly on the problem mentioned above and put forward a method afteranalysis of the factors affecting the accuracy of equation-based optimization approach. The methodhas established BSIM3v3level numerical models of a single transistor’s parameters by adoptingfuzzy logic. In this way, it can avoid the error of the electrical parameters of the small size devices.Therefore, it can improve the accuracy of the equation-based optimization approach. As for theCMOS two-stage operational amplifier, the paper has proposed a way to improve the analyticalcalculation of frequency domain performance, such as bandwidth. All these efforts have raised theaccuracy of equation-based optimization approach to the extent that very close to the accuracy ofsimulation-based optimization approach, and provided a great convenience for the application of thecircuit multi-objective optimization methods.An optimization example of CMOS two-stage operational amplifier is given in this paper, andthe results show that the method proposed in this paper is very effective.