An informational approach to quantizing the effectiveness of test data arrays for static memory devices

Abstract

While testing of the Static Random Access Memory (SRAM) the effective detection of potential faults is mostly determined by the structure and the algorithm of test arrays of data generation. This article proposes a multiplicative indicator W, which allows us to quantify the quality of test arrays of data. It is necessary to represent test arrays of data as a binary matrix, where each column corresponds to a certain bit of SRAM, and the rows correspond to test patterns. In this case, the process of sequential formation of test patterns is identified with the percolation process on a rectangular grid. It is assumed that percolation occurs at the moment when the maximization of the binary antagonisms both on the rows and on the columns of the binary matrix is provided. In this case, the binary sequences in all columns of the matrix must be different. Preliminary experimental data allows to suggest that the evaluation of the quality of test arrays of data based on the proposed indicator W can simultaneously be considered as the maximum efficiency of detecting a certain class of potential SRAM faults in a certain order of reading test data.