Is this a good design? (Or what's wrong with this?)

[Metzen]

---

Thanks for all your help

[WAJ_UK]

An ideal place to start would be to measure the water temperature and the temp near to the hot side of the peltiers. just to see if the peltiers/heatsink are getting rid of the heat well enough. It is an unusual setup and I'm thinking that either the heatsink isn't big enough to dissipate the required heat as 4 peltiers will be putting out quite a bit, in addition to the cpu and video card or the small heat exchangers aren't doing a very good job of transfering the heat in the water to the heatsink.

[Incoherent]

Oh dear.

This is a terrible setup.

First of all, by sharing the one power supply, I assume this means the TEC are in series, the wattage is not halved, it is reduced by a factor of 4. This is Ohms law. You/they have halved the voltage across each TEC thus halving the current through them. P=IV, so W is a 1/2 of a 1/2 => 1/4*W (or P=I^2R.

Originally Posted by Metzen

Is the Peltiers attached to the giant heatsink+fan a ineffiecent way of transferring heat?

Yes it is highly inefficient. There have been numerous attempts to cool water using this method, it basically does not work unless done extremely thoroughly. This setup is not thorough, more of a gimmick. IMHO.

Good move doing a side by side test.

More later.

Cheers

Incoherent

[msv]

I´m trying to figure out if the peltiers are adding more heat to the radiator than they are removing from the water. In that case this setup is more of a water boiler than a CPU cooler. Not only will the radiator have to deal with the heat from the CPU, but also the unnescessary (how do You spell that one?) heat from the peltiers.
Unfortunately I´m not very experienced with peltiers.
Simple logic that even I can follow: Without the peltiers the radiator will have to remove heat from the GPU and the CPU. Add the peltiers and the same radiator (with the same capacity) will have to deal with the same CPU, the same GPU *plus* the peltiers. This is not a problem if the peltiers have an efficiency over 100%, cooling the water more than they heat the rad.
(transferring a lot of heat from the water, dump a part of the heat along the way and only leave a minor part of the heat to the rad. Sounds silly? That´s because it *is* silly)
I would say that this setup will do better without the peltiers.
regards
Mikael S.

edit:
http://forums.procooling.com/vbb/sho...astpage&t=9109
Metzen, SlightFall, do You guys know eachother?

[WAJ_UK]

MSV I don't think that is a radiator as in water doesn't flow through it. I think the large finned area is just a heatsink where the hotsides of 4 peltiers are attached and the water passes through 4 little 'water chillers' attached to the cold sides of the peltiers (2 at the top of the heatsink and 2 at the bottom). I don't think the little water chillers would be very good at removing heat from the water and even if they were I don't think that heatsink could dissipate it sufficiently. It's difficult to tell without the water temperature and hot side peltier temperatures

[Incoherent]

Is the temperature table correct?
i.e. is the GPU temp better with water cooling than air while the CPU behaves in an opposite manner? If so that is a little curious.

Anyway, as I mentioned above, the power transfered by this TEC system is in the order of 76W (1/4 of the sum of the TEC power ratings). This would be roughly equal to the power output of the system. The Heatsink ("chiller") has to take care of this, and the extra 40% or so from the innefficiencies of the peltiers. The heatsink IS capable of dissipating this , the question is, at what delta T.

In your case I think it's at about 60C water temp. Increasing the wattage of the peltiers will be a catch 22. You will be transferring more heat but the hot sides of the TECs will increase to deal with the extra heat generated by the TECs themselves, so gain may be negligible, none at all or worse.

Cather has some experience of this problem.

Cheers

Incoherent

[krazy]

I don't know much about pelts, but there seems to be a lot of air in the water system. That can't be good.

http://albums.photo.epson.com/j/View...p=66868185&f=0

Is there really a need for the antifreeze? The only reason tu use antifreeze would be if you plan on having the coolant drop below freezing temperatures. Plain water has a higher heat capacity; I'd try running pure distilled water and getting all that air out of the loop.

What kind of pump does this setup use? Judging by the size/quality of the other components, it's probably quite small and weak. I think the overall size of the system (blocks, hose diameter, radiator size, probably pump size) is too small to effectively cool that much stuff. When you have a lot of heat going into the water, you need a pretty big radiator to get that heat back out effectively. Try swapping in a bigger radiator, a faster pump, bigger tubing, better blocks, and like others are saying, ditch the pelts. In other words, this whole setup looks like a gimmick. It appears that whomever designed it didn't know much about what they were trying to do.

[Zogthetroll]

I'll add my $0.02 as well here. I don't dislike pelts, but i'll never use them in my setup since they're a waste of energy, and probably the most inneficient way of cooling your pc that there is. a couple of observations, if that is antifreeze or similar, you'll only need that if (as stated above) your coolant goes below freezing, which isn't going to happen unless you get a phase-change system, or if you mix metals, ie. copper and aluminum, which I would steer away from if at all possible. its just one more thing that can potentially go wrong with your setup. There have been numerous threads devoted to the pelt-chiller topic, do a search of the forum for some good debates. the general consensus seems to be that the best way to use pelts is to use watercooling, and stick them between the waterblock and the cpu. to improve that setup you have there, i'd replace the heatsink with a similarly sized radiator or heatercore, get rid of the pelts (saving yourself a LOT of energy). Also I wouldn't run the cpu block in parallel with the blocks on the vid card(that's what it looks like from what i can make of the mess of tubing you've got there), you're killing the flow rate to the cpu that way, which would explain the great temps for the vid card. your tubing and unit installation in general would be simpler that way. tubing from reservoir, to cpu, to radiator, to gpu, back to reservior.

[Groth]

That the CPU does worse on water than air, while the GPU does better on water make sense if you look at the strange plumbing.

Assuming Metzen was correct that water goes to the GPU first, then the tee at the outlet of the lower TECs sends the majority of the water back to the reservoir. Only a small portion of the water would go through the upper TECs and the CPU.

[Zogthetroll]

oh CRAP, please tell me the top connection on the res is the outlet. please, please, please. otherwise the cpu is basically plumbed out of the loop that way. if the flow does go to the gpu first, you can magically drop cpu temps with their existing setup by making the outlet pipe be the one that goes to the T connector, thus making equal flow to both the cpu and gpu blocks, sorta. wow, the more i learn about it, the more i despise this thing.

[Metzen]

---

Thanks

[Zogthetroll]

well, in your case antifreeze is needed to help prevent galvanic corrosion since you've got mixed metals in your system, aluminum and copper. adding glycol (alcohol based) will LOWER the boiling point as well as the freezing point of the coolant, but again, neither is a concern since even in a well-designed system (unlike your demo unit there), you will never see either of those extremes in temp, unless you use refrigeration.

[MMZ_TimeLord]

Three things...

1. The use of peltiers in this configuration will be VERY ineffective. Especially when on the same power supply.

2. The plumbing arrangement is questionable at best. Unless there are two outlets and one inlet on the reservior, it will only succeed in doing what you say it's doing as the CPU block is in a split loop with part of it returning to the reservior.

3. The use of aluminum as opposed to copper for those blocks is not going to give you the performance you need with peltiers in that arrangement.

I would suggest replumbing it so that the pump outlet ( I presume there is only one with the black ring around it at the reservior ) goes to the CPU first, then the GPU, then loops through all the peltiers THEN back to the reservior. Then cap off the second return.

If you don't see a big enough improvement with that, put each of the peltiers on thier own power supply.

My 2¢ worth...

[Metzen]

---


Thanks

[Groth]

If you can, measure the temperature of the hot side of the TECs (or of the heatsink as close to the TECs as possible). The temperature difference between the coolant and the hot side will give you an idea of whether the TECs are accomplishing anything at all. Frankly, I think that with a 55° C water temp most of the heat is being dissapated by the tubing and reservoir and that TECs are doing nothing useful.

For experimental purposes, go ahead and use another power supply so that the TECs can have full power. That'll let you figure whether the current heatsink can handle the heatload and help determine what size TECs you want for the final product. Even better, buy a variable voltage supply and use it for your tests.

Information on the pump would be good - max flow, max head, model?

Are you committed to the TECs (for their l33t value)? Even a small radiator would beat your current setup.

When our forums have finished designing your system, you're going to donate one for us to raffle off to help out Brian and Heather, right?

[Metzen]

---

Thanks

[Groth]

A company that sells luxury PCs can't afford a variable voltage supply for its design/testing people? :shrug:

And I'm not suggesting you personally donate a system, but the company that will be profitting from our help. The good will and good press would help expand your client base. And besides, you guys have had enough pre-production press attention that the l33t kiddies would be all over even this half-assed model.

[Metzen]

---

Thanks

[MMZ_TimeLord]

From left field:

To all the experienced members, a suggestion for the Peltier portion... Would it be better (to reduce power and waste heat) if you utilized a smaller Peltier with lower power requirements and a smaller plate?

My thought is that it should only be a little bigger than the die size you are cooling. You can always encase the sides of the Peltier in a 100% silicone gasket to keep condensation down to a minimum.

This way your water blocks would be cooling a surface more equal to the die size and the Peltier advantage would cool the die while creating less waste heat and creating less power consumption.

Or am I needing to increase my medication again? :shrug:

[Incoherent]

Leave the antifreeze in. If you somehow accidently get the computer going without any load (e.g. you put a HSF on the CPU to test it or something), the liquid WILL freeze, even with the TEC's at 1/4 power.
The new heatsinks will do nothing to help. That's my prediction. The problem is not that the heatsink is inefficient, it is that the heatsink is not capable of dissipating the heat at the temperatures that you want this system to run at.
Here is an analogy.
A vertical mineshaft is dug 100m up the side of a mountain. 50m down the side of the mountain there is a small lake, 25m deep. The lake is drained by small cracks and tunnels in the rock to the sea, 50m below the level of the lake.
Everyone is happy until the mineshaft starts filling up with water from an underground spring. It ends up totally flooded. (the computer is turned on without any heatsink) It fills up with water (temperature rises) until the pressure increases so much that the water forces itself through cracks in the rock to the sea. The level stops rising (temperature has stablised via secondary losses) but the mine is unusable (dead cpu) so it is filled with big rocks and closed.
The miners are pretty clever, they decide to try again, dig a new mine but this time they drill a drainage pipe (A heatsink is installed) through the rock to the sea (ambient air). It is a pretty narrow hole because they don't have any wide drills which will reach that far. (its a pretty crappy heatsink)
The underground spring starts flooding the new mine, leaving the old one to stagnate but the drainage pipe holds the water level to a usable level, but they need better, the vein of gold they are mining goes deeper.
They come up with the idea of drilling a much bigger but shorter hole (installing a waterblock) and draining the water (heat energy) from the mine (CPU) to the lake (Liquid coolant), which is closer than the sea. They also drill a tunnel from the lake to the sea (a radiator), but environmentalists living on the beach protest and force them to accept a limit on the size of the drainage pipe. However, despite the problems this seems to work well, the water level is kept low because the wide drainage pipe takes the water away as fast as it enters the mine. And so they dig. All is well.
A travelling salesman comes by, selling waterpumps. (TECs) "A great invention" he says, "with this pump you can pump the water out of your mineshaft to a depth of 76m, more gold for everyone" and proceeds to demonstrate by pumping the old, rock filled mine dry in a matter of minutes. "it runs on waterpower, really cheap to run"
"Great" everyone says, "lets do it, but rather than pumping the mine directly we'll pump out the lake, then the water level will be lower and we'll be able to drain the deeper mine into the lower lake level."
And so they did, they diverted a mountain river to power the water wheel driven pump, calculating that the extra water entering the lake would be offset by the increase in drainage capacity given by the pump...
But the environmentalists would not let them enlarge their lake drainage. The extra water added by the pump backed up and the backpressure caused by the too narrow outlet pipe made the slow moving waterwheel stall, the lake filled up, the the mine drainage failed and the mine filled up to a higher level than when they had the narrow pipe to the sea.
"Oh" said the salesman, "it shouldn't do that. Try this smaller, more efficient pump"
"Try adding more water wheels" said a passing engineer "OK" said the miners and they did.
The pump pumped furiously, the water wheels churned and the lake level reduced a small amount but not lower than where they had started.
A passing mass murderer killed the environmentalists on the beach, they drilled a bigger drainage tunnel (bigger radiator), added extra pumps, (TECs) added more water wheels (TECs got full power), and the mine was able to be dug deeper than ever before, even below the sea level.

Unfortunately the miners went broke.

Cheers

Incoherent