Controlling Peltier Condensation – One Approach

UPDATE 8/18/01:

We had a few days of really cold and foggy days, and ‘THINK” went off line again. Once again, the unloaded peltier started to hover in the mid 30’s, which was too close for comfort. So I ran scandisk and endless loops until THINK came back on line. THINK now has a caching program so I can load many work units which will auto-switch and the system stays under constant load (and temperature).


I’ve used Peltiers before, with many condensation problems in all weather (San Francisco) and have spent too much time screwing around with it.

So……I tried something DUMB and SIMPLE.

I reversed the fan to blow into the HS and guess what, the warm air dissipated all the water! It does not get as cold as I would like and it does add to the case temp by an average of 1 degree, but no condensation.

I did this in September with a PIII Copper flipchip 850, cCO at 1030, with a 72 watt peltier and PEP 66. I’ve run it 24/7 and never found a drop of water. That’s through the fall, winter and some hot weather; it’s very consistent.

There is NO INSULATION at all.

Last week was a hot week and I had a few crashes (I’ve had all my rigs pumping Oxford/Intel’s “Think” cancer research, which heats up the CPU a lot) and so I moved up to an 85 watt peltier (and a 430 watt PS from the old 350 watt PS).

It dropped the temp back to ambient under heavy load and absolutely no condensation. Off the Think program, it drops from its loaded norm of 78-85 degrees down to 34-39 degrees…..just short of condensation.

It not a solution for the those who aspire to be king of the hill, but my system is absurdly stable (It will post way higher and run at 1088, but with occasional crashes). At 1030, it’s wonderfully reliable so that I just don’t worry anymore. It crunches about an awful lot of Think and I can still burn CD’s and other things, though I usually do most work on another system.

[ED NOTE: I emailed Glen to see if things are still OK – His response:]

Yes it’s still working.

I participate in United Devises doing Cancer research. I tweaked this particular CPU from 850 mhz to 1185 with an 85 watt peltier. When it ran a full load, it was fine. Unfortunately, UD unexpectedly shut down for maintenance and, without a load, the peltier condensed a lot of water. When UD came back on line, the ice melted and destroyed a LOT of stuff.

Since then, I backed down to a 72 watt unit. I checked to see how cold it got without a load and it also went below freezing. So… I reversed the fans and let it blow the warm air from the HS back at the board. Not an elegant solution, but simple. I left it alone for half a day without a load and at freezing, and there was no condensation.


Be the first to comment

Leave a Reply