Abstract :
Two ideas are Coupled to help determine the Pseudorandom test length in a self-testing environment. The first is that yield, defect level after test, and test coverage are related. Tbe second is that Fault Coverage as ta function of the number of pseudorandom test patterns can be approximated on semilogarithmic paper an exponential curve, with statistical confidence intervals. Merging these two concepts allows one to relate the shipped defect level as a function of the number of radomp pattterns and yield, With this knowledge, the test length of pseudorandom patterns Can be Predicted in a self-test environment.
