>
> Date:    Tue, 29 Sep 1998 13:41:37 -0400
> From:    "Max A. Lupton" <[log in to unmask]>
> Subject: Re: cooling amd
>
> I was just wondering what the issues with "too much cooling" might be.  I
> seem to recall that most integrated circuits have lower temp ratings but I
> always assumed that this was just a packaging constraint since you don't
> want to freeze the plastic encapsulation material or it might become
> brittle and crack due to the thermal stress between the "hot" chip and the
> cold outside.  Mil-Spec chips with ceramic encapsulation have much wider
> temp ranges.  I've always assumed the lower low end was due to the better
> thermal stress characteristics of ceramic (lower coefficient of thermal
> expansion) and the higher high end was due to higher thermal conductivity
> of the particular encapsulation ceramic when compared to the epoxy plastics
> normally used.  I thought all the microprocessors these days were in
> ceramic packages so that the low temperature rating would be virtually
> non-existant.
>
> My question is, are thermal stress limitations the only reason for low end
> temperature requirements with IC's (obviously with display devices like
> LCD's freezing can wreak havoc, but I'm interested in microprocessors).  I
> suppose issues like condensation could be a problem (I know for a fact that
> computers operate very poorly when cooled with coffee, even if it is cold)
> but I'm wondering about inherent temperature limitations  not ancillary
> problems like condensation or thermal stress.  Anybody know of this type of
> problem with current microprocessor technology?
>
> Max
>
> >On 23 Sep 98 at 17:08, Roberto Safora wrote:
> >> Talking about cooling. Is there any special thing when we are dealing
> with
> >> amd k6 266 or 300 mhz?
> >> do the cooling fans I was using with P166 mmx fit them(amd)?
> >  They will fit, but they may not be sufficient.  The fans that I use
> >on most of my machines ("AOC CPU Cooler") are 25mm "thick", but
> >there's a 35mm version that I'm using on my Cyrix MX-PR233+, which I
> >intend to replace soon with a K6-2/300.
>  > The older K6s may be okay with the slightly smaller fans.  [You
> >*can* put too much cooling on a CPU, but only with a Peltier
> >device....]
> >David G

Liquid nitrogen, etc. will also stop electronics cold - it has been
used by bomb squads (it also kills the battery).

Referring to SAMS "Reference Data for Engineers" - 1993, p.18-10,
 "Equation 21 is called the diode equation, which describes the total
  current through the diode for either forward or reverse bias."


                                             qV/kT
     I = q A [ (D  /L )p   +  (D  /L )n  ] (e       - 1)
                 p   p  n       n   n  p

Assuming V>>kT/q, qV/kT>>1.0 and the equation becomes essentially
exponential.
As the Absolute temperature T increases from T1 to T2, qV/kT decreases
and I decreases by a factor of e raised to the power of T1/T2

                Range          T1/T2      factor
 commercial    0 to 70 C       1.256       1.292
 military   -55 to 125 C       1.825       2.281

If you push the temperature range too far either way, the "resistance"
of the components change too much, and the device goes "out of spec."
and may even stop working.  You can think of a typical bipolar
transistor as two "back-to-back" diodes, although this is an over-
simplification.

Do you have a rough idea how many million transistors there are in
your computer?  If something gets too hot, your first sign of
trouble may be "garbage" on the screen.  If you are lucky, you may
be able to switch off in time to save the system, and replace the
burned-out fan or take the lint out of the heatsink.  Other times,
the first symptom is "sudden death" of some chip.

Boyd Ramsay

[log in to unmask]

                                  -----
       **Need help with PCBUILD mailing list? Send an Email to:**
        Bob Wright<[log in to unmask]> or Drew Dunn<[log in to unmask]>