Input-Aware Statistical Timing Analysis-Based Delay Test Pattern Generation

Bao Liu and Lu Wang
University of Texas at San Antonio


Abstract

Delay test vector generation has emerged as an increasingly critical problem in high performance VLSI designs. Existing techniques achieve estimated timing critical paths by STA or SSTA, for which a subsequent traditional ATPG method finds corresponding test vectors. In this paper, we propose a new delay test vector generation method, which achieves estimated timing critical paths by more accurate vector-based statistical timing analysis, achieves input vectors by back-tracing, and verifies the estimated timing critical paths under the input vectors by logic simulation. We observe that as a MAX operation leads to the maximum signal arrival time at the output of a gate, multiple simultaneous signal propagation paths lead to the worst case performance of a combinational logic block in the presence of performance variation. We propose delay test generation for a signal propagation network instead of a signal propagation path. Our experimental results based on 7 ISCAS'89 benchmark circuits show that while the state-of-the-art SSTA-TQM-BnB technique achieves an average of 20.5%, 18.5%, and 18.5% delay fault coverage, our SPSTA-Network technique achieves an average of 94.2%, 92.7%, and 96.0% delay fault coverage for a test size of 20, 50, and 100, respectively.