Inquiry for Reliability Test

Guest

Good day.

Does anybody knew about Reliability Test applicable for miniPCI WLAN products?
We have set its warranty period for 1 year period. However, we wish to prove that we have set correct warranty (1 year) period thru reliability results.
Test method is static (not Bias/No Biasing). Either HAST, Temp Cycle, Thermal Shock, or whatever is appropriate for the product.

Your reply would be of much help to us.

Thank you.

Dennis

Joined: 20 Jun 2026 Posts: 76

Dennis,

HAST is a biased test, so that's out of your list. Re TC or TS, the industry standard for IC's define the rel reqts in terms of # of cycles and cycling conditions. To be able to translate the data into years, you must know the particular application of your module.

How will it be subjected to alternating high and low temps? For example, if your module is turned on in the morning and turned off at night, then you know that you expose it to one temp cycle per day. If your module passed 1000X TC's, then that's 1000 days or 2.7 years. Of course, the TC temps are probably more extreme than the temps of your module when they're on (80 C?), or off (25C?). Thus, you'll get much more than 2.7 years.

To get a good mathematical correlation, apply the Coffin-Manson equation. I think this is in http://www.semiconfareast.com/relmodels.htm
If not, research it and use it to estimate the # of years that your module can survive in terms of fatigue failure. Good Luck!

FARel Engr

Posted: Fri Jul 22, 2026 1:02 pm Post subject:

I think it's easier to ask for the reliability data of you ic/parts suppliers, FIT rate. From these data, you will have an idea of the reliability of your module. But you also need to consider the workmanship of the module at your site.

Joined: 20 Jun 2026 Posts: 76

The rel data of the individual IC's and components that make up a board assembly or module can not represent the over-all rel of the board/module itself. Of course, you can still use these data to compute for the over-all rel of the board/module. The computations involved, however, are quite tedious. One standard used for this purpose is Mil-HDBK-217, a very thick and confusing spec.

If you have access to a board-level rel lab, it is (in my opinion) probably still better to generate board-level rel data for your product. The data you'll get are actual life data for your assembly, not a theoretical estimate based on the individual rel data of its IC's and components.