Research On Optimal Design Of Mixed Equation-Simulation

Digital integrated circuit design has reached a high level of automation,while the development of analog integrated circuit design automation is slow,manual design is still the main method,that is the obstruction of integrated circuir design.Optimization design is an important way to achieve the design of analog circuit automation.This paper probe the mixed equation-simulation circuit optimizers.There are two kinds of circuit optimizer named equation-based optimizers and simulation-base optimizers.In equation-based optimizers,Circuit performance are calculated by the analytical formula.So it has fast optimization speed.But at the same time,the analytical formula is approximate which lead to a low precision.In simulation-based optimizers,it is on high-precision.Due to a large number calls to the simulator during optimization,it is expensive.In this paper,a mixed equation-simulation optimizers based on difference increment model is proposed,which can reduce the number of simulations and the computing costs of the circuit optimization.While at the same time providing almost the same accuracy as the simulation-based optimizers.The main idea is to regard the equation-based optimization result as a initial point,then construct the difference increment model between the accurate value of circuit performance and analytic approximation.An iterative solution to an approximate optimization problem with decreasing differences is obtained.The article finally came up with three circuit optimization example such as VCO,two-stage opamp and cascode amplifier to proved the effectivensss of the mixed equcation-simulation optimizers.