Compensarea şi mascarea erorilor în conceptul DALG-II

Abstract

Functionality increase led to maximizing of the integrated circuits (IC) complexity. The time for test generation reached up to 10-15 months, and some of the faults of the combinational circuits (CC) with convergent fan-outs (CFO), according to [1], could nor be generated within the frame of DALG-I concept for activation the unique path in the circuit. For overrunning this impasse, development of the new efficient test generating algorithms was proposed; these algorithms should diminish the number of the tests and the time needed for their generation. Schneider [1] brought an example of CC with CFO – for the fault 6 ≡ 0 the test could not be generated based on the DALG-I concept of unique path activation, even if this test exists. As a result, the algorithm DALG-II of test generation [2] was proposed based on the simultaneous activation of all the CFO paths. DALG-II is an efficient algorithm, well formalized mathematically, allowing generation of the tests in all the cases when it is possible. These advantages of DALG-II compared to DALG-I should not undermined the fact that DALG-I is based on the singular constant error method (SCE) and activation of the single path through CC, while DALG-II is based on simultaneous activation of all FOC paths and implicit it is based on the model of multiple constant errors (MCE). At the same time, DALG-I allows obtaining of diagnosis tests, while DALG-II allows obtaining only the fault detection tests. More than that, in DALG-II and MCE case the faults interaction could occur, as well as fault compensation and/or fault masking. The paper presents the results of an interaction study in case of MCE and simultaneous activation of all CFO paths.