Water Chilling questions

[CheeseBall]

First off, do people refer to water chilling as phase change cooling, or is "phase change" soley refering to direct phase change cooling (no water).

Would it be worth it to put a TEC on my GPU, with the tec being cooled by a phase change system?

Also, how much better does direct phase change cooling work opposed to water chilling. For this comparison assume only the CPU is being cooled. Approximate temp difference in an average system would be groovey.

I'll ask more later...

[CheeseBall]

Oh yeah... What should I use to get the cold into my watercooling system (what should I run through my reservoir)? Do most people just run copper tubing in it? Would a heatercore not let the water stay in the cooling stage long enough?

[bigben2k]

A short question that requires a long answer...

First, there's the straight water cooling rig, that we're all familiar with.

Then, there's the evaporative cooling (a.k.a "bong") which works by transfering heat from the water straight to the ambient air (more details, if interested).

Then, there's phase change.

It works by transfering heat at the boiling point of a liquid. A compressor then reforms the gas into a liquid, which passes through a radiator to cool, gets heated up again (from the heat source), and evaporates. and the cycle goes on.

The trick is that phase change is most common with refrigerants (i.e. R-22, R-134, propane ...), because the boiling temps are closer to what's useful for our purposes (among other reasons).

After that, you have to consider the following: do you use a modded setup to cool the coolant and run a normal loop (i.e. the #Rotor dehumidifier mod) or do you refrigerate the electronics directly (like Vapochill products do)?

If you decide to cool the coolant (aka water), then you have a water chiller. You could use Pelts (aka TEC, Thermo Electric Junction) with it, or by themselves. The issue here is that TECs draw a lot of power (a refrigerator/compressor is much more energy efficient), and you'll need a hefty power supply.

If you phase cool the CPUs directly, then you're in for a custom made setup, and you'll need a few tools, some patience, and as much knowledge as you can get your hands on. It's not easy, by any means, and you have to be aware of the dangers (i.e. the high pressures involved, and the nature of the refrigerant).

Check out Bowman1964's work, and run some searches right here: it's out there.

I have a TEC chiller myself, but I'll be using it for the testbench (a TEC chiller gives me more control over the temperature).


Good luck!

[CheeseBall]

Thanks for the reply. Yes, I understand the basics of phase change (I think ). I was just wondering if it is accurate enough for people say they have a phase change cooled rig, when it's actually water chilled via phase change.

For now, I am wanting to just water chill, so I can still be able to take my rig on to a LAN. How? Well I needa find some fittings that when they are unplugged, they close, like those on air compressor hosing (no water loss, or very little). This way I can swap my water lines to either my heatercore, or to my chilled reservoir. Which is why I asked if I could use my heatercore to release the heat to the chilled water, or if I need something different. If so, what's going to performe the best? A fairly long strech of copper pipe like the evaporator? Or wha?

[bigben2k]

You need the self sealing fittings. We've linked to them here before. They're pricey, so you might consider a cheap valve and standard fittings.

As for transfering the heat to the chilled water, why don't you run the coolant through it directly? Otherwise, a radiator (heatercore) is fine. There are slightly more efficient ways, but expensive. Google: "heat exchanger" : either way you'll learn something.

[CheeseBall]

How will plain old dry ice in the reservoir perform compared to phase change w/ a 1/3HP compressor, R22, and a good condenser?

[MadHacker]

Originally Posted by CheeseBall

How will plain old dry ice in the reservoir perform compared to phase change w/ a 1/3HP compressor, R22, and a good condenser?

depends on how often you want to add the dry ice...
dry ice would not be a permant solution.

[CheeseBall]

Of course... what I'm looking at is just for benchmarking and playing around.

[krazy]

If you just need to cool really well for a few minutes at a time, look into liquid nitrogen

[CheeseBall]

Ok... and how does that compare w/ phase change, and dry ice? Better right? How much? Is phase change better than dry ice or not?

I really appriciate all the comments and suggestions, but it would be nice to get all my questions answered as well.

[bigben2k]

If you used dry ice the same way as liquid nitrogen (i.e. dump it right over the CPU), then it'll be better than phase change, but only temporary.

Phase change can get your CPU as low as -40 deg C, which isn't quite as good as liquid nitrogen, but still well below the CPUs operating temperature (around the freezing point).

So if you're looking for a one shot run of an extreme overclock, liquid nitrogen is your best bet. If you're looking for a permanent performance boost, (direct) phase change is your solution.

[CheeseBall]

Problem w/ that is that I would like to cool both my CPU and GPU. I could just put a pelt on the GPU but it would be great to have better cooling than a pelt. Which is why running dry ice in a res would be alright. Would dry ice be better than waterchilling? Also, if I ever do direct die... what would I have to do to cool the GPU? Would I have to do something like bowman?

[redleader]

With a complicated enough phase change system, you can get 50 or 100C colder then dry ice. However its not easy (or cheap) to build systems like that.

[CheeseBall]

Ok, ABOUT what temp will dry ice get me @ 1.95 vcore? As for the phase change system, to start off I think it would be something like this. Waterchilling, 1/5-1/3HP compressor, depending on what compressor I find R22, R404a, R12? I dunno, obviously it's kinda hard to guestimate temps when we don't even know which gas so I dunno.

[redleader]

-78.5C is the sublimation temperature for Dry ice. However since its a solid, I'm not so sure how you would apply that to the CPU. You'd need to chill some fluid that was liquid at that temp and use it to cool the die probably which is less effcient then bolting an evaporator onto your CPU.

I'd guess dry ice is probably a little better for very short term cooling (and remember to open the windows so you don't suffocate), but completely impractical for more then a few minutes to a few hours (depending how much dry ice you have).