One hundred percent abrupt failure between two radiation levels in step-stress testing of electronic parts

A rigorous analysis is developed for treating a common problem in hardness assurance where all the parts fail between two consecutive test levels in a step-stress test. The method, which is the mathematically correct way to calculate confidence and probability of part survival, gives significantly more precise and conservative estimates of survivability than do previous treatments of the same problem. The method is satisfying because it is a general approach that includes both overtesting and LTPD (lot tolerance percent defective) testing. It further suggests a still more general type of overtest which should give better results than current methods and that would be useful even when there are no reliable estimates of maximum standard deviations