Overclock.net banner
1 - 6 of 6 Posts

· Registered
1 Posts
Discussion Starter · #1 ·
The reviews I've seen of TECs have all used them exclusively as a CPU coolant, and found that they aren't powerful enough to dissipate the CPUs heat. I therefore am wondering whether it isn't better to use a standard radiator to dissipate the bulk of the heat from the CPU, and then run the water out the cold end of the radiator and then over a TEC cooled waterblock. Then the TEC would have a much lower delta to overcome and basically just serves to bring the ambient water to below ambient. Using a low-flow high-turbulance block on the CPU could also reduce the workload on the TEC. Have you tried this? Would it work?
Thanks. First post, hope the format is okay.

· High Clocker
3,762 Posts
Its not worth the power unless your just doing benchmark runs.
And condensation builds up if your not careful.

Iv done it with 2X 200W ebay peltier, The raiting they give you is how much heat they put out. So it needed its own water loop to cool them.
Drew 600W from the wall. And during a run only made about 5C difference.
What you needed to do tho was let your computer idle for 10mins while it cooled the loop right down. Use a controller so it switches off before dew point.
That gives you a good ammount of time to get though a run before the water temp warms up.

Flow I had was GPU, radiator, peltier, CPU, radiator, res, pump, peltier,

A more effective way is to just buy a 800-1000W AC unit and blow that though the radiators.
Makes sure the whole case is cooled so you have low risk of condensation

· Registered
57 Posts
I had the same reasoning and did this to a P4 on an old Koolance 3/8" (low flow) system.

It was years ago so details are a bit sketchy. For eg. I dont remember the specs of the TEC. It came out of a mini fridge/cooler thing.

Testing was promising with a white layer of ice forming on the cold block in short order.
But in the loop the temperature drop was minimal and my PSU was crying to the point the system was unstable iirc, so out it came.

I recall that:
Torquing the TEC down to quite a high number greatly improved the efficiency of the TEC.
Stainless steel has the lowest thermal conductivity of metals, so its what one should use for the torquing bolts, with as much distance between where said bolts touch the hot and cold metals, if ever...
Also a sealed vacumed TEC was a bit more efficient, but never got that far. Just a piece of closed cell foam.

Some pics of the plock I milled, with a P4 iirrc HS on the hot side:
Passive circuit component Circuit component Hardware programmer Electronic instrument Electrical wiring

Building Wood Flooring Floor House

Wood Musical instrument accessory Gas Metal Fastener

Circuit component Wood Audio equipment Gas Hardwood

Much better was to use an Ultrasonic Mist Maker/s to make a mist that is sucked through the rad.
Here are some old, unsorted pics and temp drop screenshots:

NB that you need to live in a dry climate and have the room/windows open so the the humid, moisture saturated air can leave the room.
If this saturated air is recycled; cooling stops as there isnt any more 'room' in the air for evaporated water.
1 - 6 of 6 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.