Re: Typical failure rate assignment for COTS microprocessors
- From: "Paul E. Bennett" <peb@xxxxxxxxxxxxxxxxxx>
- Date: Fri, 25 Nov 2005 19:38:39 +0000
Dave Rahardja wrote:
>>Are we speaking of the microprocessor chip alone or as part of an
>>assembled board?. I am guessing the latter. Does the board manufacturer
>>not publish an MTBF figure at all?. If you are buying from a source that
>>does not have this information worked out then you may have more work to
>>do than the savings of any cost reductions.
>
> We usually fabricate custom boards using COTS components, so the failure
> rate data has to be derived from component failures.
So you are the manufacturer. In which case all the chip manufacturers
publish reliability data. This may not b given on an individual chip by
chip basis but the data is there.
> (I assume you mean 1e-2 failures per year as opposed to 10e-2). 1e-2
> failures per year is generous, and translates to roughly 2.3e-6 failures
> per hour. Numbers in those ranges are easy enough to work into a safety
> assessment.
You are right it is 1e-2 per year (speed typing on the numeric pad led to
that typo).
> How do you define a "channel" in this case?
A channel is the sensors, wiring, processor, power supply, relays and
actuators that form the control and feedback system. It could also include
the enclosure and cooling fans if these are separate. I only mentioned
channels as with many high integrity systems we will double or triple up on
sensors and processors to provide additional means to check system
integrity. The additional channels combine to improve overall system
integrity when it takes more than one channel failing to leave the system
in an unsafe condition.
> Are these numbers an industry-standard quantity? Is there a document that
> I can refer to?
The only document I recall it being mentioned in was the original
specification documents from the client. The figures were standard maximum
permitted claim of integrity for a processor channel within the nuclear
power organisation at that time.
>>What is your application?
>
> Rail transportation signaling and control.
Having also worked in the rail industry in the UK I know that the Railway
Industry Association used to produce a set of standards that laid out the
various practices to be used. You may find http://www.riagb.org.uk/ useful
for some elements of your work.
--
********************************************************************
Paul E. Bennett ....................<email://peb@xxxxxxxxxxxxxxxxxx>
Forth based HIDECS Consultancy .....<http://www.amleth.demon.co.uk/>
Mob: +44 (0)7811-639972
Tel: +44 (0)1235-811095
Going Forth Safely ..... EBA. www.electric-boat-association.org.uk..
********************************************************************
.
- References:
- Typical failure rate assignment for COTS microprocessors
- From: Dave Rahardja
- Re: Typical failure rate assignment for COTS microprocessors
- From: Paul E. Bennett
- Re: Typical failure rate assignment for COTS microprocessors
- From: Dave Rahardja
- Typical failure rate assignment for COTS microprocessors
- Prev by Date: Re: Typical failure rate assignment for COTS microprocessors
- Next by Date: safety boots
- Previous by thread: Re: Typical failure rate assignment for COTS microprocessors
- Next by thread: safety boots
- Index(es):
Relevant Pages
|
