Reducing the number of specified values per test vector by increasing the test set size

Abstract

Test sets consisting of incompletely specified test vectors for full-scan circuits have applications in input test data compression and power reduction. Earlier procedures for reducing the percentage of specified values in a given test set maintained the test set size. A procedure is described that starts from a given (compact) test set and reduces the percentage of specified values by replacing a selected test vector with a subset of test vectors that have fewer specified values per test vector and together detect the same subset of faults. By applying this replacement process iteratively, the procedure provides a series of solutions with increasing test set sizes and decreasing numbers of specified values per test vector. The importance of considering a series of solutions is demonstrated in the application of input test data compression.