Abstract
This thesis presents some test algorithms which detect neighborhood pattern sensitive faults (NPSFs), including passive neighborhood pattern sensitive faults (PNPSFs) and active neighborhood pattern sensitive faults (ANPSFs). Besides ANPSFs and PNPSFs, another fault model, the static neighborhood pattern sensitive faults (SNPSFs), can be detected if PNPSFs are detected. Traditional March tests are widely used in memory testing because of their linear time complexity and ease in built-in self-test (BIST) implementation. Although March tests do not generate all neighborhood patterns for testing the NPSFs, they can be modified by using multiple data backgrounds such that all neighborhood patterns can be generated. The proposed algorithms are based on the multiple-backgrounds approach. The proposed multiple-background March algorithms have some advantages that 1) they have shorter test length than previous proposed ones ; 2) they also can detect other popular faults with full coverage; 3) they are similar to March tests that are easy BIST implementation and reduce the test cost; 4) they can be extended to locate all SNPSFs, PNPSFs, and most ANPSFs. The diagnosis information helps memory designers and manufacturers to improve the yield.