A novel test set design for parametric testing of analog and mixed-signal circuits

Abstract

Due to the lack of fault models and limited accessibility to internal nodes, it is difficult to test analog circuits, and many circuit specifications need to be tested in order to achieve the desired fault coverage. We introduce a mathematical technique called "factor analysis", now used primarily by social scientists to study behavioral phenomena of great complexity and diversity and mold their findings into scientific theories. Factor analysis is applied to study the correlations among the specifications for an analog circuit. An optimal set of specifications to be tested is derived based on the correlations to achieve the maximum time efficiency, and results using an implementation of the algorithm show the effectiveness of the technique.