Abstract:

To solve the problem of low fault coverage rate in logic built-in self-test (LBIST) caused by test pattern using pseudorandom vector generator, an easily controlled and easily adjustable insertion technology of testing point has been applied. However, in the test points selection process of test point insertion (TPI), the usual criterion is ''fault coverage rate precedence'', which results in too high area overhead of some test points. To address the issue, a double filtration testing point selection strategy that can be used for LBIST has been proposed through analysis of existing mainstream selection strategies. This method first manages to obtain a single test point set of high fault coverage rate/low area overhead through pre-filtration to safeguard the whole quality of TPI. Next a single test point with high coincidence of fault coverage is filtered through filtration of global test points to complete TPI conforming to the boundary condition. Experimental results indicate that the method proposed in the paper, compared with presently novel compact unit perceived test point selection strategy, has the advantage of enhancing the fault coverage rate by 4.15%, and reducing the area overhead by 5.72%.

Key words: logic built-in self-test, test point selection, double filtration