PDA

View Full Version : Chilled Rad in front of a regular Rad



NaeKuh
05-19-2009, 10:02 AM
Okey..

So i know this is totally inefficient, and i am fully aware of it.

But it really feels like im still cheating thermodynamics.

So lets get a discussion going about it, and see where im missing something.


Proposed:
Get a chiller to cycle though a radiator in a sense chilling the radiator (vapor calls it a bastardization of an AC).

Have the radiator blow cold air into another radiator (yes i know this is where its inefficient).

But water has 25X the capacity of air, so were using left over capacity here.


Reason:
I was thinking, if you had a GOOD recirculating chiller (not a coolit MTEC), you could paralell Valve the radiator to control flow. If your radiator was near your gpu rads, it would exhaust cold air into the rads and do more cooling then just the cpu.


The thing is, i still feel like im cheating thermodyamics here.

In a case like this i know the rad would be a heat source instead of a heat dump. So your thoughts?

iandh
05-19-2009, 10:11 AM
One thing I've always wanted to try is buy one of those $99 window AC units at Home Depot and slap an RX480 (because of low FPI) on the front of that puppy with it turned on high. I bet that would kick at least some ass.

I know that's unrelated to what you're saying, sorry.

NaeKuh
05-19-2009, 10:13 AM
yeah that simular idea kinda got to me.

But custom built chillers have more capacity, and also are more efficient.

The PA itself is also a better condenser then an AC condensor so its really gotten me thinking.

If your not really the type to go ultra low temps with an associated chiller, why not split the load so you use that chiller more?

DeanB
05-19-2009, 10:34 AM
An LLE would be much more efficient and work much better...

email/pm/MSN me if you're interested :)

Waterlogged
05-19-2009, 10:40 AM
If this were done with your Fester stack, I think it might be a bit more efficient than we all think it could be. :shrug:

NaeKuh
05-19-2009, 10:46 AM
Well im gonna see if i can beg skinnee to test this theory with my TEC blocks.

I think he could run a 120x1 chilled and stacked a 120x1 to feed the cpu rad cold air.

Should be interesting to see if this area has potential.

Im trying to find safe ways to go on sub ambient.


Dean;3797369']An LLE would be much more efficient and work much better...

email/pm/MSN me if you're interested :)

You guys have a liquid to liquid exchanger? A HX Plate im guessing... Well using a rad would allow emergency. If your chiller was to die, a 120x1 radiator could support a cpu on idle without much trouble.


If this were done with your Fester stack, I think it might be a bit more efficient than we all think it could be. :shrug:

Well, i was thinking we start small like a 120x1. A fesser 480 can disapate 600W in a heat beat, meaning it would be a 600W heat source for my chiller and that i think would make teyber (my builder) very sad.

Realistically chillers i believe have a load capacity of around 250W-> 400W depending on your builder.

Im trying to aim for the 300-350W area right now.

Hondacity
05-19-2009, 10:49 AM
Dean;3797369']An LLE would be much more efficient and work much better...

email/pm/MSN me if you're interested :)

(chilled)liquid-liquid exchanger?

i'm interested :D

gillbot
05-19-2009, 11:01 AM
(vapor calls it a bastardization of an AC).

Basically, that's what it is. You are using the chiller to lower the ambient of the air entering the rads. I'm not sure how well the whole valved scenario would work though because no matter what, sometime will have to give because of the change on flow from one side of the loop or the other.

I THINK what you are trying to do is run the chiller at full bore and use the valving to control the temp on the CPU side to keep it above ambient while shoving the "excess cold" to the parallel rad to be dumped into the case for use on the GPU's. Sound right?

NaeKuh
05-19-2009, 11:08 AM
in short yep

repsol_23
05-19-2009, 11:22 AM
Well my current setup has some similarities to what you are proposing. I have an AC unit that I run as a chiller box with two 3.120 (swiftech) rads inside of the cooler. I have the fans paired together so that I can turn the fans on/off or set on 12/5 volts depending on dew point. I have all this run through cpu/gpu blocks (adding mosfet and nb block this weekend) and then I ran it thru a 1.120 rad located at the bottom of my case. This rad has 2x120 fans on it with fan contollers so that I can use it to not only blow cold air into my case (directly below my gpu to blow air on video card) but it also regulates the temperature of my liquid so that I can stay above the dew point. In my area of the country the dew point is usually in the negative range due to being in the desert so I rarely worry about condensation, but during monsoon season I really have to regulate my water temp.

By using the two rads in the chiller box, with 4 fans on 12V and the other 2 off, I can regularly hit liquid temps of ~6C with temps ambient temps near 36C (outside temp, where AC is located). This is with my case radiator fans ~1000 rpms, this keeps my case temps ~30C. When I have the case radiator fans off and all else the same I can reach liquid temps of -3C (lowest ever seen was -11C on liquid, all six fans in chiller box at 12V and no load on cpu/gpu).

Vapor
05-19-2009, 11:25 AM
NaeKuh, I thought about it more....didn't realize the load capacity was was 'only' around 400W of a chiller.

If that's the case....you're going to need a much weaker rad than the PA120.3. The amount of load it will apply by simply passing 25C ambient air over the 0C liquid is enormous, even with 600RPM fans it's 500+W load. :eek:

http://www.skinneelabs.com/Radiators/ThermoChill/PA120.3/PA120.3_10cDelta_Chart.jpg

So in order to get that down, you'd probably need to go to 120x1 and probably at like 0-800RPM (variable, depending on CPU load). And then in order to make sure you always have more (in terms of volumetric heat capacity--to minimize the chance of freezing) air than water, you'd need insanely low flowrates of the water (like .1GPM, maybe lower, maybe higher).

This would be a lot easier if the capacity were like 1kW :lol:

gillbot
05-19-2009, 06:34 PM
I think you are making things entirely too complex and unnecessary.

NaeKuh
05-19-2009, 06:37 PM
im just having fun pondering ideas.. ;)

skinnee
05-19-2009, 07:15 PM
:hammer:

Waterlogged
05-19-2009, 07:42 PM
Well, i was thinking we start small like a 120x1. A fesser 480 can disapate 600W in a heat beat, meaning it would be a 600W heat source for my chiller and that i think would make teyber (my builder) very sad.

Realistically chillers i believe have a load capacity of around 250W-> 400W depending on your builder.

Im trying to aim for the 300-350W area right now.

:doh: I meant to say

"If this were done like your Fester stack, I think it might be a bit more efficient than we all think it could be."

Meaning all sealed up between the rads by fans/shrouds. What about using the Black Ice GTX M92 or GTX M184? One of them should be "weak enough" ;)

Serpentarius
05-19-2009, 08:11 PM
i recall if you place cold air (not dry ice) to the rad ... the rad itself will have heavy condensation ...
unless of coz, the heat output can neutralize the cold?

wouldnt a sealed container or some sort of closed cooling will be more practical in this chilling idea? blasting cold air to the open and hoping the air will pass through the rad fins ... seems like mixing the cold air and the ambient air .. not to mention the condensation droplets later

have you tried getting an extra waterblock in the loop (or removing your gpu/nb block) and dump the waterblock in a closed EPS container filled with dry ice or AC .. wont it be more practical this way?