Water Cooled Peltiers

ok, refering to http://www.ocforums.com/showthread.php?t=326328 . I installed a dual 120 MM heatercore and Delta Fans on it. I kept the original radiator aswell. so I have 3 120MM fans pumping onto this water.

I get temps at idle being 2* on Proc. 1 and 7* on Proc. 2 . Stressed I am getting about 30* on 1 and 34* on 2. What am I missing here?

Any help would be greatful

Your idle temps are great...

Sounds to me like you may have too much restriction and your pump is underpowered.

Are you using coldplates and do you have enough clamping pressure?

What size tubing and what pump?

Are you choking the inlet of your pump?

*goes to read other thread

I'm using the Via aqua 1300 with 1/2" ID tubing, Im not using cold plates, since I have the understanding of the cold plate is being a heatspreader, and I have the Tualatin which already has the heat spreader, unfortunatly not copper. And I am clamping it down as much as possible.

I also just cleaned up the tubing, not one kink.

Thanks for the help

You still need coldplates 4-8 mm thick, CPU lids are too thin to be effective. TECs need 200-300 pounds of clamping, you can't get that from just the water block mount.

any suggesttions on how I can apply more force.

Ok, this is obviously your problem and the last thread should have been halted once this was mentioned. You have 2 problems

1) Coldplate

You must use a cold plate... A coldplate is NOT just a heatspreader, it allows the entire power of the TEC to be applied to the smaller surface of the CPU. I'm not sure what the exact dimensions of your CPU and TEC are, but I am almost certain they do not match up near perfectly - Your CPU does not completely cover the TEC, correct?

Here is an article telling you how thick a coldplate should be and some useful facts:

http://www.overclockers.com/articles305/

2) Clamping pressure

In order to be most effective, the TEC needs a lot of clamping pressure. In order to achieve this, you will need to fashion some sort of assembly which squeezes the TEC between the coldplate and the waterblock.

How you create this assembly is purely up to you. I envision a coldplate with parts slighlty larger than the TEC and waterblock so that they can be drilled, and a bar can be put over the waterblock, and then bolts can be run from the coldplate to this bar and squish everything together.

Correct these, and you should see some results.

I love water cooled Peltiers. Good luck I hope you work it out.

ok, what you are describing is a heat spreader. And actually the TEC only is over by about 2-3MM and Remmebering right, and now that I see how large 8MM is, this is obviously two large, I'll try going with a smaller plate, more aroun 2-3MM, which seemed to be great for similar wattage, which I had experienced 4 years ago with my brand new p3 600 coppermine's. I was able to test this on a couple of them, and worked great.

As far as the pressure I was at the max, to see the assembly, refer to previous post, I had attached a picture. I never measured how much pressure I had applied, but I think 300Lbs of it on small processor and computer like this is not advisable or realisticly possible. I could be wrong since I have not measured any of these attempts.

I was already at an more then ideal pressure. Actually I snapped one of the sockets and if I have the assembly attached, I am unable to post since not all pins are making contact and it is causing the post error. So now I am down to one processor now.

The motherboard I can not seem to find anywhere now, and Since I am between jobs I am out of the game and out near half the processing power of my machine.

I found part of the problem, some of the foam was laying under the peltier and was not perfectly aligned (I know, a novice mistake. In the haste, something that I had overlooked and is why I started asking for help.) I do and was posting temps of -10 idle for proc 1 and -4 for proc 2. Load was some where in 20's I think, High 20's. The cold plate may have helped.

Just in case,if anyone has a p3 Ghz processor (dual or not) , can you post up for comparsion. And would anyone know of somewhere that stocks the MSI (9105) Pro266TD Master-LR or know someone who has it for sale. Would be very helpful.

Well If I can not recieve this MSI board, I am looking at doing a Dual Xeon 1.7Ghz system in place. If anyone is interested in hearing further let me know, email or ICQ: 108-577-050

No, read the link, you can call it a heat spreader, but that is wrong here - a heat spreader on a CPU doesn't spread heat, it interferes with the interface between CPU and cooling block, but protects from crushing. (Why would the heatspreader be better at spreading heat than a block or heatsink in direct contact with the die? It wouldn't.) A cold plate has a purpose, it allows the entire TEC to be utilized instead of a processor sized portion - heat does not conduct horizontally across the surface of the TEC.

The difference between heat spreader and cold plate would be unimportant, if it weren't for the fact that heat spreader is a poor euphamism for chip insurance.

Clamping pressure has to be on the cooling assembly (waterblock, tec, coldplate) - The TEC has a much higher requirement than the CPU. You need a hold down to hold the assembly on the motherboard, and a clamping assembly to hold the coldplates and TEC together - these are two seperate things, the pressure on the assembly, applies no pressure on the CPU. Does anyone have a picture for this guy?

I would not recommend a coldplate as thin as you are considering, especially because you may find that it has a certain amount of flex at that thinness, not to mention the thickness can improve cooling. Reading that article should help you understand why thinner isn't better - the person who wrote it knows his stuff, and its a technically objective outcome.

This may require some work, so if you want an easy solution, TEC's were a mistake in the first place. Please read the link I gave you, it is excellent.