• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Homemade TEC mem cooler???

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

MIAHALLEN

Senior Member
Joined
Aug 29, 2007
Hey guys....I've got a 245W TEC sitting around that needs some work to do. I've also got some Hyper IC based memory that I'd like to freeze for some SuperPI 32M scores.

So I was thinking about taking some think pieces of aliminum (2mm thick) and cutting a piece to mount over the memory heatsinks. I'll cut off the top of the heatsinks with a dremel, and mount the aluminum sheet with AS Adhesive. Then I'll use it as the coldplate, and put a waterblock on top. The water loop is powerful, and when I put it outside I should get a water temp around 5C at night next week.

Like a home-made one of these, but without the controller:
http://hothardware.com/Articles/Corsair-Cooling-Ice-T30-/

Is this a stupid idea? If I undervolt the TEC to 5V will it perform as well? Is this feasible? Is the coldplate sufficient? Anything I'm overlooking?

I'm very aware of condensation, and I'll insulate as needed. I'm wanting to do this to compete for GOOC 2010, and I need the best SPI32M time I can get (shooting for sub 6 min with my 980X on LN2).
 
TECs undervolt well, I know that isn't an issue. The main thing you will want to do is keep the distances short when cutting the metal. That will keep the RAM cooler.

Once everything gets down to the proper temperature, there won't be a large load on the TEC and, consequently, won't be a large load on the water loop.
 
Tecs get more efficient the further they're undervolted, so that isn't an issue.

The water loop will have to deal with the TECs heat output regardless of how much additional heat it's moving, but a decent water loop shouldn't have an issue with that.
I would use copper where ever possible, it's almost double the conductivity (thermal and electrical) of aluminum.

You want to make sure the top of the ram heatspreaders are all totally flat and paralel, i would clamp them together (gently...) in a small box or somesuch to keep the connectors all lined up flat, and then attack them with a belt sander.

One high spot on one stick and you have bad contact on all three sticks.
Though with 2mm aluminum you can get away with a bit as it'll flex.

I personally would use maybe 3mm copper, you'll get a much better spread of cold that way, as compared to freezing the hell out of whatever is right under the TEC and not so much the rest with thin aluminum.
 
At 12V the TEC should be operating at about 245W according to the specs...at 5V it should be around 100W right?

That means that it won't be capable of cooling as big of a load...but since my load is small anyway, it should be fine.

Am I correct on all that?

If so, my last question is, with the temp difference between the hot and cold plate be about the same whether at 245W or 100W?
 
Ninja'd me Bob :D So, if I undervolt to 5V, a Megahalems would be plenty to cool the hot plate right?

I'd love to use copper, and I think I'll have a look around, but I have a feeling I'll be stuck with this aluminum :(
 
Temp difference in vacuum is the same regardless of voltage/wattage.
Here in the real world, lower voltage means less heat is being taken across, which in turn means the cold side is being warmed by the world at large and is a bit warmer.

Far as i can tell, a TEC can move about half the wattage it's eating, so at 100w it can shuffle 50w across the gap.

My 130w (@12v, running at 11.5 cause it was on a lousy PSU) kept my e1200 overvolted to the moon and running at 3.2ghz at 20*c p95 load. The mobo doodad says the e1200 was drawing 50-60w, for whatever that is worth.
 
Back