Generally speaking, the dependency data compression is very useful for Intellectual Property (IP) cores and SoC. We consider the shift-in power and compression ratio in low-cost ATE environment. We propose new compression architecture with fixed length for running ones. We suppose that the ATE has not repeated function and synchronization signal. In the results, when the complexity of VLSI circuit is growing up, the number of input pins for testing is very low. The average compression ratio of our method is 63% for MinTest on ISCAS'89 benchmarks. The average of peak/WTC shift-in turns to 3×/6.6×, after comparing Selective Scan Slice (SSS) and our method. The average of hardware overhead is 6%.
Circuits and Systems (ISCAS), Proceedings of 2010 IEEE International Symposium on, pp.1587-1590