SRC Forum - Message Replies
Forum: Reliability & Maintainability Questions and Answers
Topic: Reliability & Maintainability Questions and Answers
Topic Posted by: Reliability & Maintainability Forum
Organization: System Reliability Center
Date Posted: Mon Aug 31 12:47:36 US/Eastern 1998
Posted by: Ian
Date posted: Thu Aug 3 7:50:26 US/Eastern 2000
When performing burn-in on a new device I would like to know if there is a method of calculating at which point the burn-in should cease i.e. the point where "infant mortality" becomes "constant failure rate.
Can it be calculated theoretically or does it require empirical data.
Thanks in anticipation
Subject: Burn-In Testing
Reply Posted by: Bruce Dudley
Organization: Reliability Analysis Center
Date Posted: Thu Aug 3 14:04:55 US/Eastern 2000
Numerous articles and reports on burn-in have been written that indicate where time should stop. Mr. Finn Jensen wrote a complete book on this subject called "BURN-IN" published by John Wiley and Sons in 1982. This book describes analysis methods, test results, test procedures and other information pertaining to burn-in. In general, the conclusion is, one needs to perform sample testing to determine the length and effectiveness of the burn-in. Another fact found is burn-in testing is only effective if the failure distribution of the unit under test is bimodal that is; having an early infant mortality population and a larger wearout population. A paper by Mr. Gary Moen called "Trials and Tribulations of Implementing ESS" published in the 1992 Proceeding of the Institute of Environmental Sciences indicates that many equipments actually have TRIMODAL distributions which include the above two populations and a extra one called "freak distribution". This freak distribution can cause the burn-in testing to be non productive as the burn time only causes the production unit life time to be closer to the freak distribution. Mr. Moen's conclusion was that one should use an ESS(Environmental Stress Screening) test to remove early failures and freak failures. His data suggested that this approach was ten times more effective.