Abstract:

The probable confounding syndrome occurring in the group sequence, net sequence and shifted net -sequence (GNS) algorithm is first analyzed. Then, a proper net-grouping process which can avoid the confounding yndrome on the condition of the three nets shorted together is presented and proved. Furthermore, a net-roup rule which can decrease the probability of the confounding yndrome is proposed, which requires putting the nets with a higher probability to be shorted together in a group. According to this rule, a heuristic net-roup method based on net-short-elationship graph (NSRG) is proposed. In this method, an NSRG which reflects short probabilities among all nets is first built. The net-group problem is described in terms of the graph theory and then the optimal function for the problem is introduced. Because of the complexity of the non-deterministic polynomial (NP) complete problem, a heuristic netgroup algorithm is proposed. The results show that the algorithm can find a good resolution in a comparatively short time. It reduces the probability of confounding syndrome in the GNS algorithm to improve its test performance.

Key words: interconnect test, fault diagnosis, short, confounding yndrome, graph theory