Temperature Cycle

[Qimonda]

Hi all, this is my first post.

I have doubts on Temperature Cycle. I tried to find the reason why most of the semiconductor company are using 500 cycle as gating. i believe that this is not constant. Either using Weibul or Arhenius, this we can change the cycle and commit certain years to the customer...

Any reason why most of the company make 500 cycle as standard (jedec) for all products?

None of here could agree that each product will have same life time rite...

another question, anyone can help me to determine the correct monitoring sample size for this test and what is the frequency of the monitoring?

What is the normal goal for this test? zero fails after 500 cycle or we expect a fail?

Thanks

[nimm12]

the purpose of doing a reliability test is to determine the reliability R(t) of a product at some given time and some given operating condition. eg R(500hr)=0.9 at operating temperature 40degC is a reliability target that can be calculated from the MTTF value . to obtain this MTTF value, you need to perform some accelerated tests eg TC or 85/85 or HAST ,etcetc to gather empirical data - time to failure. if your product has a constant defect rate, use the exponential chi-sq distribution, provided your production process has already filtered out all infant motality failures. if not you have to use the weibull distribution to estimate MTTF. the chi square distribution is very simple to use, and MTTF = 1/defect rate . always remember to multiply your total device hours with the acceleration factor

[Pantelis]

And just as an added clarification -- the MTTF is the reciprocal of the failure (defect) rate ONLY when the exponential distribution is used. If you use Weibull (or anything else) to compute the MTTF this is NOT the case. Also the Chi squared mentioned, again only applies when an exponential distribution is used.