TECHNET Archives

April 2013

TechNet@IPC.ORG

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Louis Hart <[log in to unmask]>
Reply To:
TechNet E-Mail Forum <[log in to unmask]>, Louis Hart <[log in to unmask]>
Date:
Fri, 12 Apr 2013 20:40:07 +0000
Content-Type:
text/plain
Parts/Attachments:
text/plain (1 lines)
Almost 20 years ago, I became certified as a reliability engineer by the American Society for Quality. GE had an expert in reliability statistics: Wayne Nelson. He published, through Wiley, two books, the names of which I forget but can look up if someone really wants to know. You can make some informed comment on MTBF or reliability in the absence of any failures, as Dr. Wayne showed. O'Connor likewise has a good reliability book.

One point worth noting is that a confidence interval on MTBF, or any statistic, may be more useful than a single-sided confidence level.

Regarding historical data, there is an approach described as Bayesian Reliability Analysis, to which Martz and Waller devoted a book about 25 years ago. The above is as much as I can say for the moment. Bob or anyone interested, contact me if you want more information.  Louis Hart

-----Original Message-----
From: TechNet [mailto:[log in to unmask]] On Behalf Of Robert Kondner
Sent: Friday, April 12, 2013 1:51 PM
To: [log in to unmask]
Subject: Re: [TN] Reliability Predictions

Ahne,

  Can you give me a reference for the formulae? I am curious about this.

  Here is a link: http://thequalityportal.com/glossary/rc.pdf of what I was thinking about. Sampling and errors rates required to establish levels of reliability.

  Now remember, I am claiming total ignorance here. But if MTBF is the average time to failure for a device, and we require a 50% confidence in that MTBF number, then we would expect zero failures in a test of 1 devices for 1 hours for each declared MTBF hour.

 So if we test 400 devices for 100 hours can we declare the MTBF at 40K Hours?

 I probably screwed this up terribly, I should go back to the text books, but the idea here is having zero failures with a certain lot size for a certain time must, I think, tell us something about the MTBF. I just don't know exactly. I can see where getting failures helps to define a failure distribution. 

Thanks,
Bob K.


______________________________________________________________________
This email has been scanned by the Symantec Email Security.cloud service.
For more information please contact helpdesk at x2960 or [log in to unmask] 
______________________________________________________________________

ATOM RSS1 RSS2