Sign in with
Sign up | Sign in
Your question

Watercooling: Radiator or Condenser?

Last response: in Overclocking
Share
December 16, 2012 1:43:10 AM

Hi,
Frozencpu has this interesting tool, the condenser:
http://www.frozencpu.com/products/6233/ex-vap-15/Cooler...

Now, my question is this. Suppose that instead of using a traditional radiator which uses *Air* to cool off water, I tried something different.

Take a reservoir and using a water-proof rubber seal, via a hole you insert the cooling tip of the condenser into the reservoir to cool off the water. So instead of pumping through the radiator, the water is cooled inside the reservoir. Weather this is practical is irrelevant. Would this work? Would this work better than a radiator solution? Is it even possible? Can the water possibly begin freezing around the tip (if that's the case, it would be important for a pump to circulate water inside the reservoir itself). Has anybody tried it before?

Reason is, I believe the tip cooled to -48C would be cold enough to easily cool water in a reservoir to well below ambient temperatures (which is a limit of a radiator) and be able to cool components far more effectively. I could be wrong though, as I've never seen this tried.
a c 78 K Overclocking
December 16, 2012 1:55:01 AM

http://www.tomshardware.com/reviews/5-ghz-core-i7-980x-...

read on this, similar unit just a year older - though the only difference is a socket compatibility.

* correct me if I'm wrong but won't phase change make the water sub zero in temps - i.e turning the water in the res into ice? or maybe water with highest density @ 4C?
m
0
l
December 16, 2012 2:02:09 AM

Usually TECs are used as direct die contact, with a watercooler cooling one side of the TEC plate...

As previously posted, if you used this (at the -48 deg C) temp you would likely turn your radiator into a ice block or ice slush, which would not end very well for many pumps.. not to mention that if you did get this to work, there would likely be quite a bit of condensation forming on all your fittings, which electronics typically dont like.

I guess if you really wanted to go this way, you could get a non-water coolant that won't freeze at those temps, then insulate all your fittings with foam to prevent condensation... But it probably won't be worth it for the result.

You don't want your parts cooled too much below ambient temps, especially in humid environments, without proper insulation.

blackhawk1928 said:
Hi,
Frozencpu has this interesting tool, the condenser:
http://www.frozencpu.com/products/6233/ex-vap-15/Cooler...

Now, my question is this. Suppose that instead of using a traditional radiator which uses *Air* to cool off water, I tried something different.

Take a reservoir and using a water-proof rubber seal, via a hole you insert the cooling tip of the condenser into the reservoir to cool off the water. So instead of pumping through the radiator, the water is cooled inside the reservoir. Weather this is practical is irrelevant. Would this work? Would this work better than a radiator solution? Is it even possible? Can the water possibly begin freezing around the tip (if that's the case, it would be important for a pump to circulate water inside the reservoir itself). Has anybody tried it before?

Reason is, I believe the tip cooled to -48C would be cold enough to easily cool water in a reservoir to well below ambient temperatures (which is a limit of a radiator) and be able to cool components far more effectively. I could be wrong though, as I've never seen this tried.

m
0
l
Related resources
a c 78 K Overclocking
December 16, 2012 2:24:47 AM

adding to what you posted, OP would then need to use antifreeze coolant, this means more restriction(anti-freeze coolant is viscous, compared to regular water or distilled) inside the loop even if OP managed to get the temps close to freezing but the pump would give up the ghost.

* it depends on how much you're willing to spend on the whole thing, pop by 4Ryan6's sub ambient thread and you'll see he's already put alot of time and effort into doing this. These kinda stuff as well as watercooling isn't mean't for halfhearted people/approaches.
m
0
l
December 16, 2012 4:33:57 AM

Hmm, well I don't think the water would become frozen. If your resorvoir is large enough and your water circulation is fast enough within reason, it won't be exposed long enough to become frozen. That cooled tip is quite small.

On top of that, I believe you can adjust its temperature so if -48 is too much, perhaps higher.

And suppose I use a more beefy pump that can handle viscosity of anti-freeze.

Either way, wouldn't the method be far more effective than a radiator? Even the water in the loop was effectively cooled to just above freezing rather then ambient temperatures it would improve temperatures tremendously. So instead of using that unit for just your CPU, it can be shared by all components: CPU/GPU(s)/Ram...etc.
m
0
l
a c 78 K Overclocking
December 16, 2012 4:44:19 AM

effective is not the word you're looking for, more like expense bearing.
m
0
l
a b K Overclocking
December 16, 2012 5:09:55 AM

One can use the same method of a type of rubber-based putty that LN2 overclockers use to prevent condensation around the sensitive electronics.

I do believe that simply shoving the tip of the cooling unit into a reservoir to be rather crude, but as long as the water keeps being pumped there should not be any freezing with some anti-freeze solution.
(I don't know if there is a proper solution to use, that won't cause problems with the watercooling tubes or metals.)

The main problem here is for the device to cool down the water enough, it will need to be run for a while. It will probably take 5-10 minutes to drop the entire water temperature by 1C, and you can NOT leave it on to chill the water beforehand, or even with antifreeze the reservoir should freeze up.

So leaving the pump and cooling unit on 24/7 would be VERY power inefficient and impractical I would say.
m
0
l
December 16, 2012 3:41:39 PM

Well, I can't find exact specifications, but it seems like the unit runs around 400watts, but that would be under full load aka minimum temperature which I suppose would be much too overkill. Now, if you subtract the power that would be used from all those fans for a high-end radiator and such, it may not be that bad.

Nevertheless, thats not what I am comparing this too. What I was comparing this solution too, is that many people who use an air conditioner with tubing to use cold air in order to cool off computers.

The benefits of water with an evaporator would be not that much more power consumption if not less than an AC, dust free, and far more effective, less space consumption, and probably far quieter.

Thanks for the input so far BTW!
m
0
l
a c 190 K Overclocking
December 16, 2012 6:04:12 PM

I'm currently waiting on a few bits to hook my waterchiller back into my loop,
48w peltier is what cools the box and the tubing from my loop takes the water from it through the loop,
Idid experience a 2-3'c drop in running temps but i wanted another pump to keep up my flow rate so keep an eye on my thread and see what the addition of another pump does
http://www.tomshardware.co.uk/forum/279524-11-chiller-p...
its not quite what you're after doing, but the principle is the same, chill the water used in the loop,
I use my rads to remove the heat created, but the chiller provides cooler water for the loop
Moto
m
0
l
December 16, 2012 9:27:30 PM

Thanks, I'll definitely read through that and follow the thread.
m
0
l
a b K Overclocking
December 17, 2012 3:11:26 PM

i tried the peltier on top of my block with a heatsink and fan and somehow performed worse than with it underneath the block heating the liquid cooler. my radiator was freezing cold and had a hair higher temps than just the liquid cooler alone. so possibly it was freezing whatever concentration they used in my sealed unit
m
0
l
a c 190 K Overclocking
December 17, 2012 3:20:25 PM

You do have a hefty Peltier though, I wouldn't have mounted that directly onto or under the block :) 
you want a reservoir to drop that into, chill the mass and take the coolant from it, this also helps prevent any freezing up issues, its quite easy to freeze a block or 11mm diameter tube, but not so easy to freeze a 2L mass of constantly moving water,
I know thats not so simple with your modified Aio setup, but from my experiments and paying attention to Ryans thread, it is the safest way to approach using a Peltier to cool W/c loops

Moto
m
0
l
a b K Overclocking
December 17, 2012 3:32:06 PM

Motopsychojdn said:
You do have a hefty Peltier though, I wouldn't have mounted that directly onto or under the block :) 
you want a reservoir to drop that into, chill the mass and take the coolant from it, this also helps prevent any freezing up issues, its quite easy to freeze a block or 11mm diameter tube, but not so easy to freeze a 2L mass of constantly moving water,
I know thats not so simple with your modified Aio setup, but from my experiments and paying attention to Ryans thread, it is the safest way to approach using a Peltier to cool W/c loops

Moto




very true man i dont even know how much that pump actually moves. it does great underneath the liquid cooler and the block doesnt get that hot and ran real good on my old system. thats what im actually trying by modifying this and adding a res to this which idk if anyone has ever tried before with a sealed unit.
m
0
l
a c 190 K Overclocking
December 17, 2012 3:37:09 PM

I've seen your thread and answered in there but I believe Amuffin and Lutfij are the only two Hxx modders I know of
Moto
m
0
l
a c 78 K Overclocking
December 17, 2012 6:54:43 PM

anyone say my name ?!?! :sol: 
@ modo+06yfz450ridr - the Hxx units have a 1400rpm pump inside the chamber. If you go downt o the watercooling sticky, closed loop cooler section, there is an in depth review on the coolit vs H50 units and the pump specs are tested. Though there is no official statement from Corsair about their pump specs. The pump is actually the weakest link in the whole chain/loop. Its able to work on its own cos of the rather slim dia of the tubes and the small channels inside the rad. Give it slightly fatter tubing and a larger rad (+ 240rad) -> we have a looser in our hands.
m
0
l
!