Predictive Methodology for Automotive Electronics Reliability in the 1980's

The automotive industry is presently experiencing a rapid escalation in the use of electronic products for the control of vital vehicle functions. Reliability of semiconductors and electronic components or, more specifically, reliability of the system to the full extent required by the automotive industry demands extraordinary levels of discipline from initial design to the manufacturing process and testing. Additionally, since product technology is far from being mature and in most instances is at the verge of state-of-the-art, a technique must be developed to accurately predict failure rates prior to the systems’ accumulating production experience. To this end an Electronics Reliability Subcommittee was formed within the Society of Automotive Engineers’ Electronics Systems Committee in October 1978 with the expressed charter of establishing procedures for the prediction and evaluation of the reliability of electronic devices, subassemblies and systems for automotive applications. Currently, the only recognized model that can be used for the prediction of reliability is MIL HDBK 217C. There are, however, many shortfalls inherent to this model. The most significant pitfalls are that the data base used in the model is not current and is not representative of the automotive environment. This paper presents some of the issues that are faced in the prediction of reliability and describes the efforts currently undertaken by the Electronics Reliability Subcommittee for the development of a predictive methodology for automotive electronics reliability in the 1980's.