Knowledge-based Semiconductor Memory Fault Diagnosis System

With the enhancement of intelligent degree of electronic device and the speedy development of computer, the type, access time and capacity of Semiconductor Memory(SM) used in a widest range and in a great quantity give rise to great change. To make SM quality conform to the system design requirement, it must be checked and tested repeatedly in every period, and diagnosed to improve its quality.Semiconductor Memory Test System(SMTS) is a national 8th-5 year Major Scientific and Technical Project, in which the author develops SMTS software and a part of hardware, and on which the design and realization of Knowlegde_Based Semiconductor Memory Fault Diagnosis System (KBSM_Diag) are based to diagnose SM fault.SMTS hardware and new generation Object-Oriented programming technique -fusion method design and realization are introduced concisely in this paper. The Fusion method given by Derek Coleman, Patrick Arnold has integrated and extended existing approach to provide a direct route from a requirement definition through to a programming-language implementation. Fusion represents the next generation of Object-Oriented software development methods. SMTS software with fusion method makes its programming technique more practical and more effective.This paper by analyzing SM fault gives a new definition of test algorithm fault coverage probability emphatically, and provides a scientific method for test algorithm evaluation. The test capabilities of conventional test algorithm are analyzed by a novel method that provides a new route for new test algorithm realization. Meanwhile, test algorithm fault coverage probability provides the diagnostic basis for KBSM_Diag. With a different design, different production process and different quality control environment, the fail models are different. Whether a specific RAM graphic sensitive fault is tested effectively or not is based on its production process and specific fail mechanism. So the fault models from all kinds of graph sensitive fault met in production should be added to the formula of fault coverage probability.