A Novel Fault Diagnosis Mechanism based on MBIST for Embedded Memories

This paper proposed a novel memory built-in self-diagnostic approach to fault localisation and fault type identification for failed memory cells. The efficiency of memory built-in self-test (MBIST) depends on the complexity of the test algorithm and its fault coverage, and the diagnosability rate of the test algorithm also determines the fault diagnosis capability. Therefore, it is necessary to analyse the test and diagnostic capability of the test algorithm to determine its diagnosability rate. In order to avoid the tedium of manual analysis, this paper describes an automated analysis procedure that enables the performance analysis of memory test algorithms. When a memory cell fails, it is necessary to diagnose the error and analyse its error address, data, current algorithm operation steps and other information in order to locate the error and make a quick and correct remedy for subsequent repair. Therefore, a fault diagnosis mechanism is proposed that realises the identification of fault types by capturing the march syndrome recognised during MBIST execution.