Water cooling TEC concept. Please advise!

3fingeredjack

3fingeredjack

3fingeredjack

Associate
Joined
25 Jul 2013
Posts
5
Right, I need some help regarding a water cooler system I’m plotting. I’m new to all this so bare with me, I welcome all comments regarding how utterly ridiculous this idea is, and why it will result in the eventually destruction of my PC, or my death.

The system works around a sealed heat exchanger tank. The tank is split into two chambers in the middle by an insulated bulkhead. This bulkhead seats the TEC.

Effectively a line runs from the CPU to the topmost (hot) chamber. Another line runs from this chamber to an external radiator and fan. The radiator then runs into the lower (cold) chamber. This chamber is filed with some sort of antifreeze not in the rest of the system, and a length of aluminium coiled tube. The radiator line runs into this aluminium tube. Another line leads from the exit of the aluminium tube, to the pump and back to the CPU to close the system.

So there it is, I’ll try and up load some images of my design to make that a bit easier to follow.
 
Its a good concept, what I was considering doing...

A pair of universal GPU blocks (these are usually high flow) with a TEC slapped between them, the cold side block would be linked into the active loop, the hot side block would run on its own loop to move the by-product heat.

Word of warning though, if you plan to use this on an every day system you'll have to consider insulation and the amount of power a TEC will pull.
 
Would it not be preferable to have the pump between rad and coil? Just to remove the heat dissipated by the pump?

Are tecs reasonably efficient? I'm just wondering if the amount of heat you're dumping into the hot side of the loop would overcome the cooling power of the tec in the bottom part. For example if the tec heats the water to 50c up top and after rad water is say 45c and the tec is only capable of removing 20c its not going to be any improvement. (I'm a noob so forgive me if this is complete bull.)

Can tecs be passively cooled?

Great idea I'm looking forward to seeing this develop :)
 
You've effectively got the cold side and hot side of the TEC both in your loop there - all you'll do is heat your water up even more, by dumping the TECs waste heat in your loop.

What you need is two separate loops, one on the cold side with the pump and CPU going to/from your coil, then on the hotside another pump cooling the TEC with the external rad.
 
Mynight said:
Would it not be preferable to have the pump between rad and coil? Just to remove the heat dissipated by the pump?

Are tecs reasonably efficient? I'm just wondering if the amount of heat you're dumping into the hot side of the loop would overcome the cooling power of the tec in the bottom part. For example if the tec heats the water to 50c up top and after rad water is say 45c and the tec is only capable of removing 20c its not going to be any improvement. (I'm a noob so forgive me if this is complete bull.)

Can tecs be passively cooled?

Great idea I'm looking forward to seeing this develop :)
Click to expand...

I can't be 100% sure, but I have seen someone cooling the hot side with a regular PC HSF (looked like a tuniq tower) for a makeshift chiller for a fish tank.

Efficient they are not, they consume absolutely shed loads of power, so much so it would even be an idea to run it from a separate PSU.

Edit: I'd take Jokesters advice where possible, he's pretty much a fluid wizard :D
 
I was wondering if some form of dual tec could be implemented to disapate the heat in the top half from the lower tec.

Would be a waste if they're that inefficient
 
If you add another TEC you end up in the same position as before, as you'll still have a hot side to cool.

They are inefficient, but as a TEC can be had for under £20, they're ideal cheap chillers for bench runs, designed to be ran for a few mins at a time durring runs.
 
Very true hence the question if they could be passively cooled :( was hoping the second one could just be left to vent into the room.

Could you attach them to a rad? Or does it need direct contact to liquid?
 
It's funny seeing these ideas pop up every now and then after faffing about with TECs along with the rest of the early water coolers around 7 years ago. :cool:

Unless there have been any quite frankly epic leaps in technology pelt wise and you're aiming at sub ambient with dual loops for giggles, you'd be better off with a phase change unit IMO
 
Hmmm, phase changer you say? Looked it up, seems like a mini-air conditioner to me. And expensive for what it is...does it just cool the entire PC case or something? Rather than specific components as per watercooling loops?

Lots of good advice there, thanks a lot!

Jokester:
What you are saying about the hot and cold sides of the loop; my thinking was that the hot side, (top chamber, which would be insulated against the lower chamber) would have its exit pipe leading to a radiator and fan combo dumping the bulk of its heat into the room (or, in this specific case, the chimney), bringing the temperature down to ambient in that line before it enters the lower chamber, and chills down to the minus figures we're looking for. The drawing doesn't indicate the length of the radiator lines, which are looking to be about 1500mm each way.

From what research Ive done, and information I CAN gather from the net, (being as there are no technical diagrams or how to guides for TEC cooling as far as I can tell) the hot side of the TEC needs to be allowed to get hot in order for the cold side to get cold. I was hoping that I might be able to use the pump to kind of round off the cooling rates, which I suspect will be erratic and bumpy if just controlled by the voltage. By leading the hot back into the cold, then the energy exchanged should be based on the speed the water is flowing through the aluminium tube in the cold chamber.

But, this got damn complicated fast!

Second idea: I've got a mini fridge somewhere, capable of cooling a single beer can down. What if I drilled some holes and just put a fan/radiator inside?

It can get quite cold at times, even producing ice. External power source, nice loud fans inside and out, jobs a goodun.
 
also, someone mentioned putting the pump after the CPU, yeah you're probably right, the temperature after the CPU block will be better for the pump than the very cold or very hot temps directly either side of the TEC chambers.
 
3fingeredjack said:
What you are saying about the hot and cold sides of the loop; my thinking was that the hot side, (top chamber, which would be insulated against the lower chamber) would have its exit pipe leading to a radiator and fan combo dumping the bulk of its heat into the room (or, in this specific case, the chimney), bringing the temperature down to ambient in that line before it enters the lower chamber, and chills down to the minus figures we're looking for. The drawing doesn't indicate the length of the radiator lines, which are looking to be about 1500mm each way.
Click to expand...
The issue you have is that your radiator just won't be able to cool your water down that much - it might drop it 0.5degC or so. All you will have done is dumped another 200W from the TEC into your loop on top of the 100W from the CPU. If your CPU was originally giving you a deltaT over ambient of say 5degC, the addition of the TEC into the loop will conceivably make it 15degC over ambient and your CPU 10degC hotter than it was previously. You could even actually end up killing the TEC because they prefer to operate with a big delta T across them, but the way your loop works it will never achieve that.
 
Jokester said:
The issue you have is that your radiator just won't be able to cool your water down that much - it might drop it 0.5degC or so. All you will have done is dumped another 200W from the TEC into your loop on top of the 100W from the CPU. If your CPU was originally giving you a deltaT over ambient of say 5degC, the addition of the TEC into the loop will conceivably make it 15degC over ambient and your CPU 10degC hotter than it was previously. You could even actually end up killing the TEC because they prefer to operate with a big delta T across them, but the way your loop works it will never achieve that.
Click to expand...

Wikipedia has educated me to the point I now understand most of what you wrote there... Are you a thermodynamics tech then? And you are basically saying this won't work?

The part about the 5deg over ambient can't be right though, you know TECs give massive heat exchange in the region of say 100 degrees either way if left unchecked?

Right, lets add some figures and you can see where I'm going wrong.

Lets say the CPU puts out 40degrees, the top chamber reaches 100degrees, and the ambient is 30, lower chamber is -10 degrees. My thinking was that the heat from the CPU will meet the top chamber's 100, causing it to fall to about 90 or something given the wildly different volumes. The water entering the radiator will be 90degrees, and you're saying that it won't cool to the ambient 30degrees whilst passing through? If it did, then by passing through the copper coil in the cold chamber at -10 degrees it should lose its remaining heat and come out somewhere between 0 and 10deg?

I just can't see why you are saying the radiator won't be able to dump enough heat into the ambient. Isn't that their job?
 
I studied fluid and thermodynamics many moons ago, but my interest these days is purely related to watercooling.

Don't think of heat as temperature, it's energy/power. A CPU pumps out for sake of ease 100W and in a loop at equilibrium the radiator then takes that 100W and transfers it to the air. You can actually work out the temperature change in the water due to that 100W being added to it and it's typically 0.1degC or less for normal flowrates you would expect in a loop (I seem to remember that 250W is roughly 0.1-0.2 degC). What this means is that for all intents and purposes the water is the same temperature everywhere in the loop.

When you first switch a loop on, the CPU pumps out 100W, the radiator doesn't remove any heat at all, as you're trying to cool ambient water with ambient air. As the water heats up over time, the radiator starts to remove more and more heat until the water heats up to the point there's a big enough difference in air/water temperature that it can now remove 100W. Double the deltaT means you can dissipate double the heat and so on.

You now add your TEC to the loop, ignoring the actual internals of the TEC, all a TEC does is create a temperature differential across itself, which then results in a heat flow from the cold side to the hot side, but in doing this it also expands energy in itself (200-300W depending on model), this extra heat must now also be dumped into the air by your radiator. In simple terms, yes the TEC will cool the cold side by removing heat, but it immediately dumps that heat, and substantially more into the otherside of it straight back into your loop. The 100W cooling effect is cancelled out by the heating effect, but you also now have the heating effect of another 200+W. This is additional heat the radiator must get rid off, whilst it was able to previously cool the 100W at whatever water temperature it had reached before, dumping the extra 200W in, means the water temperature has to increase substantially to the point where the deltaT is sufficient large to cope.

Here's some figures of an actual loop:-
http://martinsliquidlab.org/2011/04/09/i7-2600k-cpu-koolance-cpu-370/5/

Using a quad rad he's getting a deltaT of slightly more than 3degC at I guess 150W of heat from the CPU. For roundness sake if your TEC dumps 300W of heat that'll push up your deltaT about 10deg. What will happen is, that rather than the water being 0.1degC colder leaving the rad, it'll be 0.3degC colder, but rather than 2.9degC over ambient, it'll now be 9.7degC over ambient.

But yes, ultimately it won't work. You need to keep the hotside of the TEC in a separate cooling loop to the cold side - you're using the TEC to cool your CPU and the radiator to cool your TEC only. The radiator might still be sat 10degC over ambient dealing with your 450W of heat ouyput quite happily - but the TEC will be sucking heat out of the water on the CPU side without that additional 300W to worry about, or be limited by the radiator only being able to cool down to ambient temperature (put a radiator into a loop with sub-ambient water will actually heat the water using heat from the air!).
 
You must log in or register to reply here.
Back
Top Bottom