Acceleration of Seed Ordering and Selection for High Quality Delay Test

Ratna Aisuwarya (1051208)

Seed ordering and selection is a key technique to provide high-test quality withlimited resources in Built-In Self Test (BIST) environment. We present a hard- to-detect delay fault selection method to accelerate the computation time in seed ordering and selection processes. This selection method can be used to restrict faults for test generation executed in an early stage in seed ordering and selection processes, and reduce a test pattern count and therefore a computation time.

We evaluate the impact of the selection method both in deterministic BIST, whereone test pattern is decoded from one seed, and mixed-mode BIST, where one seed is expanded to two or more patterns. The statistical delay quality level (SDQL) is adopted as test quality measure, to represent ability to detect small delay defects (SDDs). Experimental results show that our proposed method can signicantly reduce computation time and base set seed counts while slightly sacrificing test quality.