Research On Test Data Compression Coding Method For Integrated Circuit

With the development of SOC (System-on-a-Chip), the integration of IP cores on a single chipand performace and complexity of chip has increased dramatically, which has brought great challenges to SOC test.Improving the performance of the automatic tester (ATE)can alleviate these problems to some extent, but the price of ATE equipment is quite expensive and the cost of testing is often unacceptable.How to effectively reduce the cost of testing, in particular, the test data storage overhead, has become a hot issue in academic and industrial research.In this paper we focus on the SOC test data compression, introduce the theoretical knowledge of SOC testing and several common test data compression technology in detail, improve the existing test data compression schemes, and propose two new solutions:Firstly,this thesis presents a new hybrid coding compression method of test vector based on one identification.The mixed coded test data is divided into a plurality of data blocks.Each data block contains two sub-blocks of data, the first sub-block of data with dual-prefix single run-length coding, the second sub-block of data with an identification method.When the decompression, according to the length of first sub-data block,we can calculate the length of the second sub-data block.Thus, it can reduce the part data needing to be encoded.Experimental results show that:the average compression rate can reach66.70%.In order to obtain better compression effect, according to their vector containing unrelated X ratio of the size, we has made the different treatment.The entire set of tests were divided into two groups, unrelated large ratio of X a set of vectors with hybrid coding,unrelated X ratio smaller set of vectors with two-run double prefix coding.Experimental results show that:the average compression ratio increased by2.04%.Secondly,this thesis presents a new difference encoding compression method based on test vectors reordering. We sort of test vectors based on the compatibility between the vectors from the two directions, then the differential.Finally, using the FDR code compression, the compression ratio is compared with a single direction of the sorting method has inreasing2.30%.