Segmented Testing

The fraction of faults detected for a digital network is frequently high for the first few input combinations applied out of a set of test vectors. For on-line testing, there appears to be an advantage to splitting the test into segments which are applied at different times. It is shown that the expected time to error detection and the probability of an undetected double error can be reduced. The amount of reduction is dependent on the shape of the fault coverage curve. This approach may be applicable in fault-tolerant systems.