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: IC
Date posted: Thu Apr 18 9:43:06 US/Eastern 2002
Subject: Is 217 Obsolete
I have came across several articles stating that 217F is longer being used to the same extent for theoretical analysis purposes. If this the case what will replace it?
How are engineers now calculating reliability figures? If there is no in-service/field data available how do we now calculate a realistic failure rate?
I would be grateful for any pointers in this matter.
Thanks in Advance.
Subject: Is 217 obsolete
Reply Posted by: c.jayaraman
Date Posted: Thu Apr 18 11:32:16 US/Eastern 2002
Because 217Fn2 has not been updated for a while it seems useless. RAC promotes Prizm which has additional algoritms. I have not fully evaluate it. But you can continue to use 217FN2. Before you sign off on the numbers I would suggest evaluating the Pi factors closely and see if they are teneable. Modify them if you have reason to believe that the parts quality is beeter than indicated. Also use vendor test data wherever possible. If you follow the derating guidelines strictly and keep internal ambients low, the end item shd track pretty close to the prediction or even better. Also allow 20% design margin for failures during design and EDM phases to ensure the rel goals can be met. Good luck.
Subject: Mil-HDBK-217 Alternatives
Reply Posted by: Bruce Dudley
Organization: Reliability Analysis Center
Date Posted: Thu Apr 18 14:11:44 US/Eastern 2002
There are a number of alternative reliability prediction models that are available to you. These models include the RAC PRISM Program, Telcordia Reliability Prediction Procedure, and Handbook of Reliability Data for Electronic Components RFD-2000. The RAC also has individual component data in the electronic and non-electronic databases as shown in NPRD-95 and EPRD-97. I do not recommend the use of modified terms to the old 217 models as the models were built on an existing database and modifying this base without restructuring the model would appear to invalidate the model. This includes extrapolating the numbers of gates, power levels, and other stress factor. Quality levels can always be applied to the component model, as you believe the device is built.