Using Physics of Failure to Predict System Level Reliability for Avionic Electronics

Abstract

Today's analyses of electronics reliability at the system level typically use a “black box approach”, with relatively poor understanding of the behaviors and performances of such “black boxes” and how they physically and electrically interact. Box level analyses tend to use simplistic empirical predictive models, and the effort is typically driven by cost and time constraints. The incorporation of more rigorous and more informative approaches and techniques needs to better understand and to take advantage of the advances in user interfaces and intelligent data capture, which will allow for a broader range of users and for similar resource allocation. Understanding the Physics of Failure (PoF) is imperative. It is a formalized and structured approach to Failure Analysis/Forensics Engineering that focuses on total learning and not only fixing a particular current problem. It can involve material science, physics and chemistry; also variation theory and probabilistic mechanics. The approach necessitates an up-front understanding of failure mechanisms and variation effects. In this paper we will present an explanation of various physical models that could be deployed through this method, namely, wire bond failures; thermo-mechanical fatigue; and vibration. We will provide insight into how this approach is being accepted by system assemblers, as it allows for failure oriented accelerated testing, for substitution or “what if” analyses in lieu of the traditional accelerated life testing. This paper will also provide insight into a process to develop viable test plans and a tool that facilitates the entire process so that minimal testing is performed, thus reducing costs and schedule impacts. Examples of this approach will be presented.