On 2 Jul 00, at 23:51, John R Pierce wrote:
> chilling the CPU below the point of condensation will cause
> corrosion.
Nah! Lowering the temperature reduces the rate of chemical and
physical reactions which constitute corrosion.
Over-rapid temperature changes can cause cracking due to thermal
stress. Also, metals become very brittle at low temperatures and may
be subject to fatigue cracking if mechanical vibration is not
controlled adequately. These are _real_ problems but are relatively
easily controlled.
> Chilling it further below freezing will turn the whole thing
> into a ball of ice.
Not if you keep it dry. Immersion in a nonconductive fluid (freon or
liquid nitrogen, depending on the required temperature) should fix
that, and provide a means for waste heat to be removed from the
operating components. Disconnecting the CPU fan & relying on the heat
sink alone would seem to be indicated.
> The chip maker specifies a maximum AND minimum operating temperature
> within which they guarantee the chips timing specs. stray too far from
> these limits in either direction can/will cause problems.
They usually specify 5C / 90% RH in order to make sure that there is
no condensation buildup. Without a controlled environment it's
difficult to prevent condensation with temperatures below 0C.
Actually the physics work in favour of cooler temperatures.
Resistivity falls and the circuits can switch faster with the same,
or even a lower, power consumption. This effect continues until
unwanted superconductivity sets in at a few degrees above 0K. (Say
somewhere below -250C, even without specially selected materials.)
The downside is that the reduced values of resistors in delay
circuits results in smaller decay time constants. Therefore, if you
run a chip _very_ cold you _may_ find that you have to _reduce_ the
core voltage slightly, or actually _increase_ the clock rate, in
order to get reliable operation.
The worst effect of seriously cooling a whole motherboard is likely
to be that the CMOS battery will not be happy (to say the least!).
Providing an external 3V power source should fix that problem. (Even
the same 3V battery mounted "warm" & connected via long leads will
do).
I would venture to suggest that the size, mechanical noise and power
consumption of a suitable nitrogen liquifying plant are the reason
why such techniques are not widespread.
Radio astronomers have been running electronics (RF detectors)
immersed in liquid _helium_ for decades. The practical problems are
solvable, at a price.
Regards
Brian Beesley
_________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
