An analysis of test effectiveness via surrogate simulation of a commercial IC

Abstract

Choosing which test sets to apply to a commercial design is important in order to efficiently achieve low defective part levels. This work compares a test set created by the DO-RE-ME process utilizing the MPG-D defective part level model that maximizes observation at problem sites with a commercial automatic test pattern generation test set that utilizes stuck-at faults to generate tests and to estimate defective part level. Incorporated in this comparison are stuck-at fault predictions, MPG-D predictions, and bridging surrogate simulation results. In contrast to standard commercial practice, the MPG-D model also allows analysis on a fault-by-fault basis and on a pattern-by-pattern basis.