Optimal LFSR-Coding Test Data Compression Based on Test Cube Dividing

Abstract

In this paper, a kind of test cube dividing strategy is investigated on which the optimal LFSR-coding scheme is based. The strategy utilizes the fact that the number of specified bits in different test cubes varies widely and the specified bits in a test cube clusters mostly, in which each of the test cubes with high fill rates is divided into two cubes having lower fill rates, and the guiding rule is that one cube consists of odd bits of the original test cube, whereas the second cube is comprised of even bits of the original pattern. The number of specified bits in a new cube is reduced effectively, which makes degree of LFSR needed for encoding it successfully smaller. A LFSR decoder of low complexity is used to outspread the seeds and merge two successively decoded patterns. The experimental results show that compared to the previous techniques, the proposed scheme can achieve higher test data compression ratio. The seeds for LFSR decoding being of the same length, it is convenient to communicate between ATE and circuit under test.