Dry Ice and Liquid Cooling?

[CanadaRox]

So after pushing my QX9650/4870 to their limits on water (and stock voltage for the 4870). So since my brother is getting my some soldering equipment for Christmas, I plan to voltmod my 4870 to get some more juice out of it. On hwbot, the top single 4870 score 22512, done with a QX9650 (and 4870 of course) under LN2. Now I can't afford a different pot for my CPU and for my GPU (my current CPU pot has too thin of a base for LN2), nor can I afford a dewar and LN2. So my next option would be just getting a dry ice pot for my GPU and using my current pot for my CPU. But another atleast $100 is still pretty steep for a pot that I might not be able to carry forward. So here comes my idea, I get a bucket deep enough to submerge my radiator completely. Then I fill the bucket with alcohol and dry ice. I would have to replace the coolant for the cooling system with something along the lines of anti-freeze. I think alcohol would get too cold and would become thick which could potentially damage the pump. Then I could either use my CPU pot with dry ice, or just have both my CPU and GPU connected to the rad. How do you think this would work? Would it be able to provide similar level cooling as a dedicated pot? And what type of fluid would be the best to use in the tubing so that it doesn't thicken and damage the pump.


TL;DR... rad in bucket of dry ice + alcohol, connected to cpu/gpu with anti-freeze coolant. Thoughts?

[enaberif]

Dry Ice or LN2 is not for long term use.

They are designed for short burst runs and if that is what your wanting to accomplish then that is what you will need.

[CanadaRox]

Originally Posted by enaberif

Dry Ice or LN2 is not for long term use.

They are designed for short burst runs and if that is what your wanting to accomplish then that is what you will need.

I am perfectly aware of how dry ice and LN2 work as I have used dry ice on a few occasions. I just want to have a bench session so I can attempt to take a higher position on hwbot for single 4870's in 3DMark06. I was hoping someone might have a suggestion for how I could maximize the cooling, and prevent the coolant from becoming too thick.

[matsta31]

Coolant wise, you really think the water will go lower than freezing point ? Not sure, but I would not bet on it. If not, why not use some anti freeze ! or a mix with water. Altough, I am not sure your block will like it.

I would also be afraid of condensation around the CPU and GPU.

[magelan]

I can't comment on your question but it looks like you're wanting to go extreme. If so, why not simply have an extended water loop and put the radiator inside a fridge or freezer? Of course, you'd have to cover the rad with some kind of plastic or material so it doesn't condense or simply freeze. Way cheaper and easier to do.

[miggs78]

Very true.. ^^. Seen people do it, and yea the temps aren't below freezing point, but still you can get much out of it.

[biff]

The rads would be above the ambient so they shouldn't freeze up at all.

[CanadaRox]

Originally Posted by matsta31

Coolant wise, you really think the water will go lower than freezing point ? Not sure, but I would not bet on it. If not, why not use some anti freeze ! or a mix with water. Altough, I am not sure your block will like it.

I would also be afraid of condensation around the CPU and GPU.

I'm pretty sure the coolant would be below freezing as dry ice is around -78°C. I am mostly worried about the coolant either ruining my water blocks or becoming too thick and damaging the pump.

Originally Posted by magelan

I can't comment on your question but it looks like you're wanting to go extreme. If so, why not simply have an extended water loop and put the radiator inside a fridge or freezer? Of course, you'd have to cover the rad with some kind of plastic or material so it doesn't condense or simply freeze. Way cheaper and easier to do.

From what I've read at XS, trying to use a fridge or freezer will cool very poorly and damage the fridge as they are not meant to have such a high heatload. They are meant to keep things producing almost no heat cool. A computer would pump out 200-300W of heat that the fridge wouldn't be able to cool quickly enough and it would cool no better than putting you rad into an airtight box.

[Eldonko]

Your best bet is to run ice water through your loop or dunk the rad in ice bath. For dice you need a container to put it in, you cant just dump it in the fluid. Not to mention all the insulating required. The price you will pay for antifreeze, dice, and risk to the h20 components is not really worth it in my books. Ice bath or buy a dice pot is my recommendation.

[BrainEater]

heeheh.



Welcome to the world of aerospace engineering my friend.
-------

What you suggest is certainly possible , but you need to throw out your current idea's about running DI.Here's what I mean :

-'antifreeze' aka ethylene/propylene glycol can't handle Dry Ice , It'll freeze solid.

-Thing's like standard waterblocks and rads will contain several different metals soldered/welded/fitted together.The thermal contraction that happens during a 100 degree temp shift can totally destroy these seals.

-Pumps for watercooling just won't work.they will fail forsure.Same reasons....

I have a worklog here about my attempt at DI cooling.mebbie you've seen it.

-----------

I tend to agree with Eldonko.You still have a lot of experiments with watercooling to do.Especially if it's only benching runs you wanna do.

I assume you know what 'open loop' watercooling is ? ( cold water tap into blocks , output to drain.)....Start there .Then try some peltier coolers.I've achived loaded temps of -21c with this method(load side).

Start there.Then you can try pre-cooling an open loop system with normal ice and salt.

-----

If yah just wanna giver , I can certainly understand.

You need to look into building a lot of your own components , and spending cash.

Think 'alcohol loop'....custom pump.

--------