Correlation Between Laboratory Test & Field Part Failure Rates

In Japan, failure rates for non high-reliability electronic parts, obtained from laboratory tests, are generally about ten times those obtained in the field. The reasons for this difference have been analysed by using part failure-rate data collected from major part suppliers, equipment manufacturers, and public users in Japan. This paper numerically analyses transistor data. The two main causes of the difference between laboratory and field failure rates are 1) the difference in failure criteria between them and 2) unreasonable use of the constant failure rate without actually determining the real failure rate. By using the conversion factors, it is possible to predict electronic equipment MTBF by using laboratory failure rates with appropriate application factors pertaining to whether the units are in use on a ship board, airborne, etc. basis. Failure rates for the electronic parts with unsatisfactorily long burn-in time lapse must be treated as carefully as the rate data published in the AIST data.