COMPRESSED SKEWED-LOAD DELAY TEST GENERATION BASED ON EVOLUTION AND DETERMINISTIC INITIALIZATION OF POPULATIONS

摘要

The current design and manufacturing semiconductor technologies require to test the products against delay related defects. However, complex system-on-chips (SOCs) require low-overhead testability methods to keep the test cost at an acceptable level. Skewed-load tests seem to be the appropriate way to test delay faults in these SOCs because the test application requires only one storage element per scan cell. Compressed skewed-load test generator based on genetic algorithm is proposed for wrapper-based logic cores of SOCs. Deterministic population initialization is used to ensure the highest achievable transition delay fault coverage for the given wrapper and scan cell order. The developed method performs test data compression by generating test vectors containing already overlapped test vector pairs. The experimental results show high fault coverages, decreased test lengths and better scalability in comparison to recent methods.