[ShrimpBrime]

Originally Posted by kikicoco1334

don't remove that thing dude! it's gonna yank the whole socket out!

OK I wont!!! thanks for the heads up too! Really not risk it, that's why I asked!!

So I drew up a little basic paint picture. Basically what I have in mind for setting up a 3 TEC cooling system with 2 loops and about 390w of chilling power, 130w of which will be right on the Cpu, or on the cold plate over the cpu rather.

I've ordered a 2 pack 12715 TECs (already have 2 of them)

Also ordered a 1 Kilo Bar of Pure .9995 Copper Bullion for a cold plate on the Cpu.
The dimensions are 5 1/4" x 2 3/4" x 1/2 ". I needed to find between 1/4" and 1/2" think bar of the purest copper for best heat dissipation on that Cpu.

The cold plate also need to be slightly bigger, but not bigger than the water block surface. I think the GTZ block looks to be around 45mm-50mm at the bottom, but having the bar makes it easy so I can make the block w/e size I need.

I would have went with Silve Bullion, but I don't have your guys's moneys
______________________________Just kidding there. ^^

Here's an example of what I'd like to try. notice how there's 2 TECs on the Chilling block. Well...... that's how much it takes to get water cold fast. In reality I should use 2 or 3 chiller blocks, but that get's into using ton's of power, may as well go SS at that point....

Air cooler on each hot side water block FTW. I only drew the one.

Dig it?

[rickss69]

Bring us some pic's when you get it mocked up...good luck and happy chillin'!

[ShrimpBrime]

I really need the luck. Ambient is always a problem. 50f air temps are sweet for this kind of heat dissipation. Each TEC can produce over 250w of HEAT not chilling.

In order to run 40f with a Peltier, the hot side will always be double or more, so 80f on the hot side.... generally with air cooling, it's much hotter than that. That's why liquid is the only way.... 2 loops......

The Chilling plate on the Cpu is the "buffer" point. It may be freezing at the top, and at the bottom get the Cpu socket temp into the 30-40c range and hold it there, much like your LN2 pots you guys use. It holds heat because of mass. So the plate needs to be there.

I will ahve pictures. you can bet your bottoms on it. This won't start until mid March to early spring. China and Japan are far.

I also need to pick up 3 dimmer switches as well for individual TEC control. The Cpu TEC will always have to bee on cause the Cpu will soak the plate and it wont dissipate through the TEC to the water block either. I've tried it, didn't end well in the past.

[Neuromancer]

I assume you TEC guys have already read this page, but for those that have not. Here are a TON of TEC questions and answers

http://www.tellurex.com/technology/peltier-faq.php

[kikicoco1334]

wow nice! are those 180w ones?

[Bones]

Shrimpy, if I understand it correctly, you'll be using water to cool the hot side of the TEC right?

Instead of the regular rad type setup, you could think about what's called a "Bong Cooler" that is a type of rad setup that can cool the water to below ambient temps, sometimes called an evaporative cooler.

It works just like a Port-A-Cool fan in your shop would, air passes through a sheet of water that runs over a structure that's setup to thin out or diffuse/spreadout the water stream as it is returned down to the pickup tube in the res and is recycled. Since the water is diffused into a wide thin sheet instead of just a solid stream, more heat is removed with the airflow going through it.
More efficient than a typical rad & res setup would be - I've always wanted to build one just to try it and might do it one day.... If I can find the time. One thing to be aware of is it's also an evaporative-type setup so you'll need to keep an eye on the water level of the res to maintain it or just hook up a small supply line with a toilet type float valve that will regulate the water level in the res as needed.

If you're going with a slushbox type setup, that would be better but just throwing an idea out there on my part.

[rickss69]

^ I remember those too Bones...never did get around to trying it out. May be a fun project to try...all it takes is a little pvc, a fan and a showerhead. http://www.overclockers.com/nuclear-...water-cooling/

[ShrimpBrime]

That's stuff is cool. haven't tried it nope.

@Kiki - They are 136w TECs.

@Neuromancer - great read. Lots of good q&a in there!

[Neuromancer]

Originally Posted by Bones

Shrimpy, if I understand it correctly, you'll be using water to cool the hot side of the TEC right?

Instead of the regular rad type setup, you could think about what's called a "Bong Cooler" that is a type of rad setup that can cool the water to below ambient temps, sometimes called an evaporative cooler.

Originally Posted by rickss69

^ I remember those too Bones...never did get around to trying it out. May be a fun project to try...all it takes is a little pvc, a fan and a showerhead. http://www.overclockers.com/nuclear-...water-cooling/

Bong coolers are great, and depending on location can really be effective. (Dry climates will work better than humid ones).

they are a bit more work though because you will be adding dust and stuff directly to teh water, so your loop and blocks will have to be cleaned more often, plus you will need to add water to it frequently.
they are also noisy-ish as well, although it would preclude the need for a whitenoise generator

Still for those used to LN2DIceing for benching, the maintenance will be pretty minor

I would really like to do one of them some day

[Bones]

You could set it up so that it uses a AC paper type air filter on the air intake side of the box, that will help keep crap from getting into the cooler. As long as the fan is running, no water will get to it.
I also have to say if you can get the return water coming back as a mist rather than a stream, it will become more efficient.

If I can get the materals together, I can do that as a project later and would love to set one up sometime just to see how well it works.