A new design-for-testability method based on thru-testability

Abstract

Partial scan and non-scan techniques allow test generation of high fault coverage for sequential circuits with less area overhead and less performance degradation than full scan technique. In most of these techniques, extra logic (e.g. a multiplexer introduced by partial scan) is added to permit a data transfer from a flip-flop (or input) to another flip-flop (or output). Such additional logic function is called thru function in this paper, which plays an essential role in enhancing the testability of a circuit. In this paper, we introduce a design-for-testability (DFT) technique which modifies a given sequential circuit to a thru-testable sequential circuit with acyclic test generation complexity by adding new thru functions based on the information of thru functions that may exist in the original design and the dependency among these thru functions. Thus, thru functions of a given sequential circuit should first be extracted from its high-level description. If there is a thru function that transfers data from a flip-flop to another flipflop, the latter is exempted from being considered in DFT insertion. This reduces the additional logic to be added. Using ITC'99 benchmark circuits, we show that the proposed DFT method takes up less area overhead compared to the previous scan methods in testing the difficult-to-test circuits like b07, b08 and b15. Besides, the test application time is shorter than that of previous scan methods while the test data volume is less too in difficult-to-test circuits.