The Application And Research Of Transform Compression Based On Multiple Scan Chains

With the development of electronic manufacturing technology,the scale and complexity of integrated circuits are increasing,including more and more functions.This brings a lot of convenience to people's lives,but also makes the testing of integrated circuits become more and more complex.The increasing complexity of integrated circuits has led to an increasing number of possible failures,In order to ensure a high fault coverage,the amount of test data will increase significantly.Large amount of test data stored in the automatic test equipment(ATE),not only brings a very expensive hardware storage costs,but also lead to longer test time.The aim of test data compression is to reduce the amount of test data,reduce the test power consumption and shorten the test application time.In this paper,the method of testing excitation compression is studied.There are two main work based on the transform of decomposing.(1)A test vector compression method based on Hadmard transform using shared prominent component is proposed.In this method,transform compression is applied to the test environment of multiple scan chains.Modify the transform object as the adjusted test vector,namely small test set.Then decompose small test set into small prominent components and small residual components based on the principle of least number of 1 in small residual components.There is a certain degree of circuit correlation between bit-streams in test data set.Experimental results show that the optional prominent components of scan slices in same group exists intersection.We can effectively compress prominent component sets by means of Huffman coding the shared prominent components selected by greedy algorithm based on sharing degree.On the other hand,smaller Hadmard matrix can get higher matching degree and reduce the number of 1 in residual components,so improved the compression ratio of residual component sets.Experimental results show the participation rate of sharing prominent component up to 75.91%and the average of compression ratio is 70.47%.(2)Test data compression method for improving compression ratio by decomposing transform and bit-reversion is proposed.The basic idea of this method is to improve the compatibility of test vectors.Using the transformation and bit-reversion to get the original test vector closer to the broadcast vector.The prominent component selection principal is the least incompatible bits in residual components.Then bit-reversion is used to further reduce the number of incompatible bits without changing the fault coverage of test set.Finally,the position of the incompatible bits is marked by the position map which is compressed by FDR coding.Experimental results show that the average compression ratio of incompatible bits up to 66.19%and the average compression ratio of test data is 76.82%.