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: Sheila Prather
Organization:Northrop Grumman Corporation
Date posted: Thu May 18 10:35:54 US/Eastern 2000
Subject: PiQ Factors for Mil-Hdbk-217
This issue we're currently faced with is the appropriate use of the piQ values defined in 217F-Notice2 (for non-mil quality parts).
We currently have in house the PRISM S/W and will transition to it's use as soon as possible. But, in the interim, some prediction efforts are underway on various programs which are utilizing industrial/commercial grade product. For these efforts, one must still try and define the piQ factor to use that is appropriate and justifable yet not conservative in nature (to the degree which one is penalized).
I've reviewed the PRISM manual in hopes that some assumptions/correlations could be made to support modification of the cookbook method published in 217, but it appears that the quality factor is not factored into PRISM evaluations (or at least not to the degree as 217). And, in reading one of your previous messages (authored by Bruce Dudley), thermal cycling/upscreening is not as important as sound manufacturing processes and procedures. You also note that 217 does not accurately reflect the improved IC manufacturing techniques/processes of today (due to it's obsolecence). I understand and agree with both these notes. I also read in one of the most recent responses that a piQ = 0.25 should be used for space class V devices.
Given all of this, what is your opinion regarding the appropriate piQ factor to use for a vendor who does not screen to full 883; but, has sound processes/procedures/monitoring in place, QML house, and has provided ample HAST/test data to support use of part. Is it sound to still consider this piQ = 2? Or something better?
Thanks for your response.
Subject: PiQ Factor for Commercial Parts
Reply Posted by: Jack Farrell
Organization: Reliability Analysis Center
Date Posted: Thu May 18 15:16:04 US/Eastern 2000
It is our general recommendation that for the type of vendor you describe “a vendor who does not screen to full 883; but, has sound processes/procedures/monitoring in place, QML house, and has provided ample HAST/test data to support use of part” that a PiQ of 1 be utilized in the Mil-Hdbk-217 model. Information that we are seeing indicates that screening of microcircuits is adding very little to the overall quality of industrial high reliability grade devices. In the last ten years, the quality levels of microcircuits has risen without the need for any extra testing, in fact some extra testing actually decreases the reliability by adding unneeded stresses. In our new PRISM model for microcircuits, we are using a process grading function that evaluates the device manufacturer's process along with the assembler processes. Extra testing does not factor very high in this assessment. Consideration of use of ISO-9000 proceedures and Statistical Process Control methodology have been found to be more important in developing overall part quality.