Improving Fault Coverage of Random Pattern Test using a Scalable Distance Function

In this paper, we proposed a new method to improve the fault coverage of random test patterns generated based on a scalable distance function. The proposed patterns are made by effectively mixing pseudorandom and antirandom patterns according to a distance function. Antirandom patterns are generated by calculating hamming distance among candidates, and pseudorandom patterns are by H/W logic such as LFSR. Experimental results show the proposed method can improve the fault coverage by at least $1 \%$ up to $7 \%$ compared to the previous methods fixing the distance function. Since the proposed method can easily obtain new patterns using only the relationships between predetermined patterns, and can generate random test patterns in real time at the chip and board levels regardless of DUT size and structure, it can be widely utilized as a technology for test logic embedded in system semiconductors or semiconductor test equipment.