Quad TECs...good idea?

[baron_iv]

I have another mad scientist idea that I've been kicking around for a few days. It seems relatively logical, but then I could be over-analyzing again (which seems to happen quite often).
I have these 4 TECs sitting here, each is 80W. I also have a massive block of copper here too. (I was a boy scout so I'm always prepared).
I got to thinking about piggy backing the TECs, then I started to think about the heat that would need to be removed from just ONE of those monsters and realized that if I piggy backed (stacked) them, each one would have to remove the heat from the first...which means they would be working FAR too hard, and it most likely wouldn't work at all, and if it DID, I think my CPU temps would go UP instead of down.
Then the idea hit me to put them side by side. Essentially, I would still need GREAT cooling, but each peltier could work for itself, it wouldn't have to try to remove the heat from other peltiers below it in a stack. I don't think I would get a full 320W of cooling power from it, but it would be greatly enhanced, assuming I could get the heat away from the hot side, plus, it wouldn't require all the power of a full 320W peltier. Less power + Less heat = good stuff.
Now, when I think about how I can cool these beasts, the only method that comes to mind is water. Sure, I could use phase change, LN, DI, etc etc etc. That's kinda defeating the purpose of cheap and easy cooling though. Chilled Water is the ticket for me, my FX-57 runs at 16C at full load @ 1.50v. You're probably wondering why I want to run TECs too since my temps are quite reasonable. It comes down to squeezing every ounce of performance out of my systems that I possibly can. I suspect that many people at the procooling.com forums do the same thing.
Anyway, the reason I need additional cooling is because I am going to add my dual opteron system to the water loop. The opterons SERIOUSLY pump out the heat. I am still not sure if I want to use the quad TEC to cool the water or to actually cool the opterons. I could easily build the copper waterblock to cool both opterons simultaneously and put 2 peltiers on top of each CPU. That's going to take some research to see what works the best. I have several ideas for a heatsink as well, but that's really another post.
So, can anyone see any problems developing if I had quad TECs side-by-side under a single heatsink (which would be VERY LARGE, and have water flowing through it) and above a single coldplate? I realize the power issues are going to be there, I have a couple of meanwells to take care of that. Obviously, the whole thing will need to be well insulated. I know that you can buy a single 360W TEC these days. I just want to use what I've got. I might be out of my mind...it wouldn't be the first time I've been told that. haha
One other thing I've been thinking of...
In a single-processor setup, to have 4 TECs, two with the cooling side down facing the CPU coldplate (directly above the CPU) and two with the cooling side down facing the CPU coldplate, but 3-4" away. We all know that heat flows from hot to cold areas. So it would be equivalent to having 2 distinct heatsinks, since the first two TECs will draw heat directly away from the CPU and up into the water block, and the second two TECs will draw heat horizontally away from the CPU through the coldplate and then up into the second set of TECs and into a separate waterblock. It's rather difficult to describe my idea, I may need to draw a diagram. I think this might be more efficient than the first idea I had of putting 4 side-by-side. This is more of a single, relatively thick coldplate, a peltier sandwich, then two individual waterblocks on top, one on top of the CPU, one 3-4" away to cool the other two peltiers.
This is what I get for taking a week off work. My brain has all this down-time to come up with these hair brained ideas.
If anybody cares to comment on either one of these...why they might work, why they might not, how crazy I am, or anything else, please feel free...I could use the constructive criticism. Maybe somebody could get me straightened out before I go crazy and turn my entire house into a super-massive heatsink.


--baron_iv

[Long Haired Git]

1. Check out Cathar's "experiments with Peltier water chillers" over at OCAU forums.

2. Remember that copper is not a perfect conductor. There will need to be temperature gradiant through the "heat spreader" (cold spreader?) to get the energy to move through the resistance.

3. It will need to be more insulated than normal, as the bigger size will result in a bigger opportunity for the heat spreader to cool the air in the case and not the CPU.

[BillA]

laterial conduction losses through the bp will kill your project
solve that and you're good to go

[baron_iv]

I think that there would be a temperature gradient between the sides of the heat/coldspreader. The second set of TECs will be cooling absolutely nothing besides the copper below them. Surely it would be cooler than the CPU side.
As far as insulation goes, I have the world's largest supply of neoprene too! I ran across a HUGE amount of it at an auction for $20 so I bought it. The little lady doesn't like it that I'm a pack rat, until she needs something that she can't find anywhere else and I say "oh, I've got one of those!". I'm going to have to buy a larger house for all the crap I have, I know I'm going end up using every bit of it at some time or other. :shrug:
It's a shame that I can't make homemade heat pipes with the (phase change) material inside, I have some interesting ideas for that too.