Effective generation and evaluation of diagnostic SBST programs

Abstract

Functional test and software-based self-test (SBST) approaches for processors are becoming popular as they enable low-cost production tests and are often the only solution for in-field tests. With the increasing use of volume diagnosis, efficient and cost-effective diagnosis methods are required. A high quality functional or SBST test program can be used to perform logic fault diagnosis with low-cost test equipment and therefore significantly reduce the cost of diagnosis. We present a framework for the automatic generation of functional diagnostic sequences for stuck-at faults. The framework allows a user to specify constraints imposed by the employed test environment and generates diagnostic sequences satisfying these constraints. Furthermore, the framework is able to prove the equivalence of faults under the specified constraints. This enables to compute the best possible diagnostic quality that can be reached under the given environmental constraints. Also, it gives the necessary information for implementing selective DFT techniques in order to differentiate faults which cannot be distinguished otherwise. In our experiments we evaluated a MIPS-like processor. The results show that our approach can effectively distinguish fault pairs or prove their equivalence, under different environmental constraints. To the best, of our knowledge, this is the first approach which, enables the automatic generation of diagnostic SBST, programs and allows to eectively prove the equivalence of faults in functional and SBST test environments.