Re: DS20L

["Christopher C. Chimelis" <chris debian org>]

On Fri, 31 May 2002, Maurice Hilarius wrote:

> We feel that any machines that REQUIRE a cold room should come with a 
> prominent sign on them to that effect, and that any sales literature should 
> point that out in big bold letters.

Why a prominent sign?  Why is literature that accompanies these products
not enough?  If a manual says "don't operate this outside of x temperature
range", how many more places need that information be posted?

> I also believe that if they are that sensitive to room temperature it is a 
> failing in thermal design.

I somewhat agree.  I think that, until now, only one person has noticed
that I noted the DIMMs failing at higher temperatures in the CS20.  In the
environmental chamber tests of the CS20, I definitely exceeded the
documented operational temperature range while maintaining a working
system, but again, the first failures were DIMM-related.  So, this begs
the question of why everyone has been so quick to blame thermal design of
a computer, poor environmental control, rack design (I did this as more of
a contributing factor), etc without even questioning if the DIMMs
themselves are not capable of operating within their own documented
temperature range limits.

To back this up a bit, I had DIMMs from several manufacturers in the CS20s
that I baked in that chamber and each showed a different failure
temperature (sometimes by up to 9C difference...consistent numbers).
So, the point shouldn't be whether or not the computer as a whole can
be reliable at extreme temperatures (for the record, the CS20 itself can
definitely run firmware tests reliably at temperatures in excess of those
where I saw the DIMM failures), but whether each of the component parts
can operate reliably at those temperatures.  If you had a hard drive that
failed above 20C due to heat stress, how could that be covered in the
computer's documentation or be the basis for an argument like we've seen
about the CS20?

C