[Tfug] peltier?

[Sean Warburton]

I have been contemplating the use of alternate cooling systems for quite
some time now (air is just so...primitive...considering what it is cooling-a
shining example of humanities' collective computational power, all contained
in a 1.25 inch square blob of silicone). Take, for example, liquid cooling.
While it is neat to put small objects (like glitter or even just watching
the bubbles) in the tubes and watch them make the cycle, it has some
physical limitations. For example, it will do very well under load (massive
amounts of heat generated from a severely overclocked CPU) for about an
hour, but then the temperature will rise to points where elements become
unstable and spontaneously decay, and I have to turn it off. Why did it get
this hot? The thermal compound is Arctic Silver 5, the name brand stuff. It
has the smallest molecule size of any product in the market (that I am aware
of), meaning it can get in the little grooves of the water block and the
CPU, maximizing efficiency in thermal transfers. I figured for a 1/2 inch ID
Tygon tubing system with a 320 GPH pump, with enough non conductive coolant
(not sure of chemical composition) should provide more than adequate water
volume traveling over the water block. So heat isn't being built up in the
CPU. (well, I am sure some is, but an insignificant amount) This leads me to
think the water in the tubes is heating up enough so the water rushing over
the CPU, which should be cool enough to whisk away the heat, is heating up
to past the point of effectiveness. A bigger reservoir would stop the
problem for a while, but weight almost becomes a problem (I do like to go to
LAN parties).
Now, I heard of an interesting physical property called the peltier effect,
which is (as I understand it) the direct conversion of electricity into
temperature changes.  This makes sense, because  when I overclock my CPU, I
force more power than normal into it, and a higher temperature is achieved.
This system must work the other way, where it can make things cold. Some
research yielded a product that looks like a ceramic plate and one side gets
cold while the other gets hot when a current is applied. Furthermore, the
temperature difference is dependent on the ambient temperature, right?
I can only really foresee two downfalls to this cooling method. The first
one (and a primary concern of mine) is burning out the peltier block. If I
dont cool the hot side quick enough, it will burn itself out and my CPU will
overheat. The other problem I see is power. I only have an 850 watt power
supply, and I imagine this peltier system will take an incredible amount of
power. Does anyone have any suggestions as to how I can overcome these
shortcomings, or a better cooling system? Liquid Nitrogen and dry ice
doesn't count, because it evaporates, and I can''t be refilling a dewar
attached to my motherboard every three days. My CPU under load consumes
about 225 watts of power, and a portion of that (65 watts) is transfered
into heat. I will have .9246 degrees celcius per watt to remove, and for a
65 watt processor, that comes out to 60.1 degrees per unit of time (not sure
as to what that is in this instance...). So I would have to remove that much
heat in that much time....Like I said, suggestions are welcome and
encouraged! Thanks in advance,
     Sean


-- 
FreeBSD v.1.4 (beta)
ASUS P5N32-SLI Premium
Intel Core 2 Duo 6600
dual eVGA 7900 GT OCs (full x16 SLI)
2 gigs DDR2 PC2-6400 (OCd to 866MHz)
250 gig RAID 1 (mirroring)
custom Liquid cooling :)
four 17" CRTs (uber widescreen)
7.1 surround sound (296 watts)
one happy gamer