The multiple observation time test strategy

The authors consider the test generation problem, for synchronous sequential circuits in the case where hardware reset is not available (or cannot be assumed to be fault free). It is shown that the conventional testing approach, in which a fault is detected at a single predetermined time unit along the test sequence and in which the response of the circuit under test is compared against a single fault-free response, valid for all initial states of the circuit, can cause detectable faults to be declared undetectable. The use of a small number of different observation times and a small number of fault-free responses can allow the fault to be detected. Based on this observation, the use of multiple fault free responses and multiple time units for observation of the response of the circuit under test is suggested and test generation algorithms under the multiple observation time test strategy are given. Experimental results demonstrate the effectiveness and practicality of the multiple-observation-time strategy in increasing the fault coverage