Algorithm in Walsh function basis for analysis of integrated microcircuits on MOS-structures

Use of a Walsh function basis facilitates computer-aided design and performance analysis of integrated microcircuits, especially of those built on MOS structures for discrete processing of analog signals. As the integration scale increases here, so does the variance of correspondingly smaller geometrical dimensions and so does the deviation of electrophysical parameters. An analysis with direct application of the Walsh function basis then becomes difficult and unwieldy so that a faster convergence is needed for coverage of a wide range of simultaneously varying circuit parameters and external conditions, while taking into account all possible nonlinearities in the output characteristics. An algorithm is proposed to meet this need for any number of variable parameters, an algorithm which optimizes the convergence by minimizing the norm of differences between output (solution) vectors obtained in successive iterations. This is achieved by conversion of the deviations matrix into the Gray code, that can be regarded as a diadic vector. The procedure is demonstrated for the case where deviations of circuit components, signal sources, and power supplies are describable by Rademacher functions.