When I was doing these calculations we had to agree with the customer what
standard to use. They were predictive by nature so acceleration factors had
to be agreed and lifetimes extrapolated. MIL217 was the most widely
accepted but it produced very pessimistic data for hi-tech stuff - but at
least it was comparable between suppliers all using the same rules.
The extrapolation requires a good understanding of all the possible failure
modes - like drive spindle lubricant... We had one broadcast transmitter
project where the power source was a critical point of failure so there were
redundant generators and then double redundancy on top. (ie we had 4
generators) What was omitted from the reliability prediction model was the
guy who tops up the antifreeze - they all failed more or less at the same
time one very cold night.
Jez
On 10/29/07, AG Schnozz <agschnozz@xxxxxxxxx> wrote:
>
>
> > Have you done some math that I didn't? Strictly that MTBF means
> > that by 200,000hrs (22 years) 50% will have failed.
>
> That's the beauty of "standards"--you have so many to choose from.
>
> I distinctly remember one model of Micropolis 9GB SCSI hard drives.
> 200,000 MTBF. Terrific unit, it was one of the rare ones that
> actually performed to spec.
>
> Nearly 100% failed within 12 months from spindle failure.
>
> Just tell me how they came up with 200,000 hours MTBF?
>
> AG
>
>
==============================================
List usage info: http://www.zuikoholic.com
List nannies: olympusadmin@xxxxxxxxxx
==============================================
|