Environmental-stress-screening using degradation measurements

Abstract

ESS (environmental stress screening) has been extensively used to reduce infant mortality by precipitating defects. The existing ESS plans precipitate defects by stressing all products for specified durations. The plans usually require long screen durations to allow nearly all defective items to fail, and thus generate excessive aging effects on good items. For some products, failures are defined in terms of performance characteristics exceeding their critical values. This paper describes the principles of using degradation measurements of performance characteristics to screen the products. In an ESS, the performance characteristics of defective items degrade considerably faster than those of good ones, yielding a bimodal distribution of the characteristics. As screen duration increases, the 2 modes of the distribution shift apart. It is possible to find a tightened critical value to weed out the defective items before they fail. This paper, based on these principles, further designs the optimal 2-level screen plans which minimize a segment of life-cycle cost and which simultaneously meet the reliability requirement. The minimum cost is achieved by choosing optimal part-level and unit-level screen durations and tightened critical values of parts. A numerical example is followed by discussion. Because the ESS regime of this paper allows defective parts to be screened out before they fail, the developed optimal ESS plans can reduce life-cycle cost, shorten part-level screen duration, and alleviate aging effects on good products.