DYCOCO: Dynamic Configuration-Based Test Compaction

Abstract

Systems with high reliability demands require large test sets that target several or advanced fault models. To reduce the storage requirements of a large test set, it was suggested earlier that a stored test may be used for applying several different tests. Industrial applications of this observation consider compressed tests that are applied from an external tester, or stored on-chip for logic built-in self-test (LBIST). It was also suggested earlier that several different tests can be obtained by allowing the scan chains to receive test data from different outputs of the decompression logic for different tests. The connection between the decompression logic and the scan chains is referred to as a configuration. The procedure described earlier for selecting stored tests and configurations for test application is referred to as a static configuration-based compaction procedure (STCOCO) since it requires a given test set as input. This article describes a configuration-based compaction procedure that is suitable for dynamic compaction (DYCOCO). DYCOCO can be used during test generation and does not require a precomputed test set. Compared with STCOCO, DYCOCO requires fewer tests to be generated, and it is more effective in reducing the storage requirements. A comparison of the experimental results produced by STCOCO and DYCOCO for benchmark circuits in an academic simulation environment demonstrates the advantages and costs associated with DYCOCO. Both STCOCO and DYCOCO are considered in the scenario where a compressed test set is applied from an external tester.

Keywords

• Bridging faults
• single-cycle gate-exhaustive faults
• test compaction
• test data compression
• test generation