Reliability Demonstration Testing

[Rocketman90251]

My company is planning to conduct a reliability demonstration test to show that our product, an electronic circuit board, meets it’s required meantime between failure. We have been debating whether we should count any Out-Of-Box or Dead-On-Arrival failures that occur in our calculation of mean time between failure. In other words, if a card fails, upon initial turn on and check out and before the formal start of the reliability, demonstration test, should the failure be charged and counted as a valid failure for calculating MTBF?

Does anyone have any thoughts on DOA chargeability? Are you aware of any industry, standard or mil spec? That address is judge ability of.DOA failures during reliability, demonstration, testing?

Thanks in advance for any advice.

 

[Steve Prevette]

On the question on a "burn-in" failure (also referred to as infant mortality)- you would want to gather that information and if there are enough of those to affect the MTBF then institute a burn-in prevention action of turning it on before sending it to the customer and run it through the burn-in period. Plotting the "bathtub" curve from your data would help rather than simply the single number of the MTBF. Bathtub Curve - an overview | ScienceDirect Topics

If you have taken actions to prevent shipping out the burn-in failures (Burn-in - Wikipedia.) , then I"d say ignore them from the MTBF calculation, similar to the examples in the link.

 

[Dominic Cortella]

My company is planning to conduct a reliability demonstration test to show that our product, an electronic circuit board, meets it’s required meantime between failure. We have been debating whether we should count any Out-Of-Box or Dead-On-Arrival failures that occur in our calculation of mean time between failure. In other words, if a card fails, upon initial turn on and check out and before the formal start of the reliability, demonstration test, should the failure be charged and counted as a valid failure for calculating MTBF?

Does anyone have any thoughts on DOA chargeability? Are you aware of any industry, standard or mil spec? That address is judge ability of.DOA failures during reliability, demonstration, testing?

Thanks in advance for any advice.

I agree with Steve, but, I can weigh in a little. The MTBF is a measure of the "intrinsic" failure rate that I assume your reliability demonstration test is looking to demonstrate. This is associated with the "steady-state" or "useful life" part of the bathtub curve (bottom), between the "early-life" or "infant mortality" part of the curve that burn-in & manufacturing test should weed out - including your DOA, and before the "wear out" or "end of life" part of the bathtub curve. Therefore, Steve is right - ignore the DOA in calculation of intrinsic failure/MTBF. Different part of the curve. You might consider replacing the DOA PCBA with one that starts up to maintain the sample size in your test plan, assure same hrs are accumulates, test ends same as planned.

However, if you don't successfully weed out early failure with effective manufacturing test/burn-in, these failures will show up in any field based reliability metric such as AFR, warranty return rate, etc. And your customers will hate you.