Superior reliability prediction in design and development phase

Abstract

High demanded reliable components or systems projects necessitate reliability prediction in early design and development phase. Historically reliability predictions are performed using standardization of the technique such as MIL-HDBK-217 or similar handbook, which in many cases has produced inaccurate reliability results. Today's complex designs, which have intricate interfaces and boundaries, cannot rely on these methods to predict reliability. This paper presents superior reliability prediction approach for design and development of projects that demands high reliability where traditional prediction approach has failed to do the job. The effectiveness of the proposed approach is illustrated with a suitable example of solder ball failures on a silicon chip and circuit board. Reliability of solder ball has been accurately predicted. Sensitivity analysis, which determines factors that can mitigate or eliminate the failure mode(s) has been performed. Probabilistic analysis such as the burden capability method is employed for assessing the probability of failure mode occurrences, which provides a structured approach to ranking of the failure modes, based on a combination of their probability of occurrence and severity of their effects. The effective number of simulation runs for various random variables and how to compute the life is also presented.