Reliability prediction: the turn-over point

The main objective of this paper is to present an approach for a practical, easy to implement reliability prediction procedure. The proposed procedure is based upon the combined use of a number of reliability data sources, calculation methods and deterministic models. The procedure includes "tailoring" of the models according to their defined criteria, and with emphasis on the use of deterministic models whenever appropriate and possible. The criteria for reliability data sources and models are defined. The paper emphasizes, as most important, the use of inside manufacturing data from different production stages. It details mathematical principles for the use of information from incoming inspection, from environmental stress screening (ESS) and from the field. A reliability test of PCB leads for "parametric" deterministic reliability modeling, is illustrated. Some advantages of the recommended methodology are: improved accuracy of reliability prediction; simplicity of use; and incorporation of product line information. This approach is useful not only for reliability prediction, but also for providing engineering feedback for reliability improvements.