True-worst-case evaluation in circuit tolerance and sensitivity analysis using genetic algorithms and affine mathematics

New methods for circuit tolerance and sensitivity analysis (TSA) are presented in this paper. Genetic algorithms (GA) and affine arithmetic (AA) techniques have been adopted to find respectively the inner and the outer solution in true worst-case (TWC) problems for circuits where the strong uncertainty of parameters yields large changes of performances. It is shown that GA and AA allow a sharp determination of the TWC in TSA problems, even of great complexity. Some circuit examples are presented to highlight the accuracy and the efficiency of the new methods, whose application is best suited for the CAD of electronic circuits submitted to performances and regulations constraints to be fulfilled in presence of large parameter uncertainties.