Yield Optimization of Large Scale Microwave Circuits

In this paper we demonstrate for the first time the feasibility of yield optimization of large scale microwave circuits. A combined strategy to attack this difficult problem is proposed, which includes the use of supercomputers, efficient approximation to the circuit responses, and the use of fast, dedicated simulation techniques. We have successfully accomplished the yield optimization of a 5 channel multiplexer with 75 design and toleranced variables, 124 constraints and up to 200 statistically perturbed circuits. No microwave circuit design of this type and on this scale has ever been reported.