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: Ümit BÜLBÜL
Date posted: Mon Mar 14 5:02:00 US/Eastern 2005
I am an hardware design engineer in a firm called Aselsan Inc, mostly serving in military areas. I have a general question related to the applicability and validity of MIL-STD-217F as regards the new technological progress taken in electronic/electrical components. Could you share your thoughts about my comments(or questions) given below and suggest alternative methods if possible :
Question_1) We design many products consisting mostly MOSFETs, ceramic capacitors, opto-couplers etc. But according to MIL-STD-217F BJT transistors seem to be superior than MOSFETs. But as you know MOSFET technology has improved for the past years and I think MOSFETSs now are more rugged than BJTs (if you take necessary consideration against ESD disadvantage of MOSFETs). Also according to MIL-STD-217F chip ceramic capacitors seem to be worse than chip tantalum capacitors. Apart from the cracking problems observed in ceramic capacitor I think that ceramics are better than tantalum.
Question_2) If MIL-STD-217F standard is a good approximation to component failure rates what is the reason of the big difference between MIL-STD-217F Notice1 and Notice2 regarding failure rate equations and different coefficients suggested in these revisions?
Question_3) According to MIL-STD-217F to get a good MTBF data for a complex system (e.g. tanks, gun systems etc.) we should use at least JANXXX type components which are becoming more difficult to find and even in case we find, it is nearly impossible to satisfy the other system criteria if we use JANXXX (because JANXXX type components are somehow old and not competetive regarding packet size, efficiency and other advantages of todays' high technology components. What is your suggestion on this ?
Question_4) For the MTBF data of mechanical components which standard should we use for military projects?
In summary in our military designs we have some difficulty to satisfy the MIL-STD-217F Notice2 specifications coming from customer side. Therefore your comments are valuable for us to reevaluate these requirements and provide alternative but reasonable solutions to MTBF requirements of our customers?
Reply Posted by: bwd
Date Posted: Tue Mar 15 14:59:33 US/Eastern 2005
The RAC does not recommend the use of MIL-HDBK-217 for new system procurements. This handbook last updated in 1995 does not accurately reflect the reliability of new electronic system designs. Significant errors in the capacitor and transistor models exist that will never be corrected. Specifically, the ceramic chip capacitor has a base failure rate that is ten times the leaded ceramic capacitor and the MOSFET model is incorrect by a factor of at least four. All the models have shortcomings in the quality and environmental factors. The RAC PRISM methodology was developed to address shortcomings that existed in MIL-HDBK-217 and other prediction methods.
It should be remembered that a reliability prediction should be used as a tool to assess competing designs and is not a crystal ball that will provide a one to one correspondence between a predicted and field failure rate. The RAC believes that PRISM methodology provides an accurate prediction assessment when used properly. Any valid prediction methodology should account for the processes used in design and development otherwise companies utilizing poor processes will have the same reliability figure of merit as an organization that has good processes. While some in industry have modified or gamed the results in MIL-HDBK-217 to approximate their field experience it is not a standardized approach which will work for all users. PRISM provides a standardized approach that accounts for and quantifies the variability in processes that affect system reliability.
Several independent papers that validate the PRISM approach with comparisons to MIL-HDBK-217 analyses have been written. These papers are available at the RAC PRISM web site:
Mil-HDBK-217 is not considered a good approximation as it is outdated. The differences in notice 1 and notice 2 are the result of additional data being added to the database resulting in parameter modifications and changes to the structure.
If you use PRISM, you will notice that the effects of quality are taken care of at the vendor line level and the manufacturer assembly area. MIL-STD parts are not any more reliable then quality industrial grade components.
Mechanical data and models are not easily obtained or performed. The U.S. Navy has a handbook that provides mechanical reliability prediction models for 19 component types. Reliability data on some mechanical component is available in the RAC product NPRD-95 and the data base included in the PRISM prediction tool.