What if ... a new use for an SS

[MaadDaawg]

Started thinking today, which is always dangerous, and hit upon an idea that may or may not work. Depending water flow I think.

What if instead of a radiator, you hooked a CPU water block up to the head of an SS and used that to cool the GPU loop? In this case, the thought occurred to me regarding chilling the water for my quad sli 580s. The Feser quad does a good job, and I have a DYI chiller unit, but since I have a spare SS and CPU water block, my mind wandered into this idea.

Would it work so long as the water flow was maintained? I think so, but I'm thinking I might be SOL if the rig needs to reboot too often while tuning the OC.

huh??

[Kal-EL]

I'm guessing the coolant would freeze in the block before leaving and stop the flow.

[Witchdoctor]

you would have to introduce a glycol solution / AKA antifreeze

take the rads out of the loop

I woould also keep the flow as slow as possiple to maximixe heat transfere at the block, but yes it would work under those circumstances ........ :thumbsup:

you will probably create condinsation so insulating the cards would not be a bad idea

but it depends on what type of heat transfer you get across the block and flow will determin

I would have to think with a 20 run time you could drop the loop temp considerably

it may have to catch up between benches as 4 of those will put off some heeat

[GFDuke]

I don't think there would be enough surface area on a water block to keep the loop cold. i guess that would depend on the lenght of the loop. You could always try it and let us know.

[MaadDaawg]

I think WD is right, and disassembling and erasering the cards isn't something I think I want to spend time on right now. The heat of the 4 gpus should keep a glycol solution mostly liquid but I had been thinking water like Supes. Surface area is another good concern.

oh well, back to the drawing board :D

[Witchdoctor]

Kmart

32 quart cooler

Add 4/120 radiator ....... insert 30 lbs. of ice on top of it

turn on system and comence benching FTW ............. :thumbsup:

[MaadDaawg]

Kmart

32 quart cooler

Add 4/120 radiator ....... insert 30 lbs. of ice on top of it

turn on system and comence benching FTW ............. :thumbsup:

I have one of those big beer chest coolers with a collapsable handle in which I've wound 50ft of 1/2 copper tubing connected to a 300gph fish tank pump, with a little resv to fill and monitor it. It's hooked to about 10ft of hose on each side. Fits 60lbs of wet ice. I've used it before to great effect, I think my VRs were down in the 7C range and the GPUs were about 14C. May break that out again but again condensation is a worry :D

[Mr.Scott]

you would have to introduce a glycol solution / AKA antifreeze
Gets too thick that cold. Windshield juice.;)

[Witchdoctor]

Gets too thick that cold. Windshield juice.;)

Great idea, didn'6t even think of that ............. :thumbsup:

[MaadDaawg]

With the wet ice chiller water works cause you never get below freezing :thumbsup:

[Mr.Scott]

With the wet ice chiller water works cause you never get below freezing :thumbsup:
Salt your ice. Your loop will go below freezing. Just like making ice cream.;)

[MaadDaawg]

Salt your ice. Your loop will go below freezing. Just like making ice cream.;)

Awesome idea Scotty :thumbsup: So cold out today and had the window open I didn't really need anything more than the quad :D

[punx223]

the problem here MD is thermal wattage.... the unit will actually overload trying to keep the loop cool pending you have a large enough surface area contacting the loop where you can avoid freezing or slushing the coolant which will inevitably kill the pump also.

the idea is cool in theory but first the Evap would have to be mounted onto a decent sized plate with enough copper tube coverage to allow coolant to flow and be chilled without icing over....

Worst case scenario: loop freezes no flow...

other scenario: loop slushes = low flow and pump gets shit kicked outta it.

[punx223]

Salt your ice. Your loop will go below freezing. Just like making ice cream.;)

sorry didnt read this but Awesome idea and it will work too.

[MaadDaawg]

jee whiz, you always ruin my fun :rofl:

hadn't thought about the pump though, not even close to worth screwing with.. oh well, I've had stupid ideas before, and, will hopefully have more in the future :laughing:


the problem here MD is thermal wattage.... the unit will actually overload trying to keep the loop cool pending you have a large enough surface area contacting the loop where you can avoid freezing or slushing the coolant which will inevitably kill the pump also.

the idea is cool in theory but first the Evap would have to be mounted onto a decent sized plate with enough copper tube coverage to allow coolant to flow and be chilled without icing over....

Worst case scenario: loop freezes no flow...

other scenario: loop slushes = low flow and pump gets shit kicked outta it.

[punx223]

jee whiz, you always ruin my fun :rofl:

hadn't thought about the pump though, not even close to worth screwing with.. oh well, I've had stupid ideas before, and, will hopefully have more in the future :laughing:

im sorry bro :-/

[Bones]

I've tried supercooling a loop before with DICE and learned a few things.

As Witchy said earlier, take the rad out of the loop, your coolant will freeze in it and stop coolant flow. I also tried a mix of antifreeze and water, got too thick for the pump to move it and since I could not find anything else to use at the time, poured a spare bottle of 190 proof (Everclear) in the res and that did the trick. :cheers:
Be sure to start off if using an open res as I do, You'll need to have the coolant being used chilled FIRST.
I was starting with a dry res and I would pour the coolant being used in, then slowly add the DICE - It creates a boiling effect as it sunlimes and could cause the same thing as a boilover that hot water could do in a pot for example. Start the system to bleedoff any air bubbles though before actually adding the DICE to the coolant. Once you see the air bubbles are gone, add the DICE.

Correct me if I'm wrong but I think it's denatured alchohol that will do the same thing and it's cheaper too.

The pieces of DICE themselves created an "Iceball" in my res that took a long time to melt away and as it did, the remaining DICE would refeeeze the water around it creating a small block of ice that lasted for sometime = Good for longer runs.
I actually had several of these blocks in it and they kept coolant temps low for sometime. Just having one or two didn't make alot of difference but having several of these going in the res did the job. Had to do some testing to see what the best "Balance" of these iceballs would be, obviously the more you have the colder your temps will be but even with the coolant treated as such, too much DICE will cause freezing problems unless the coolant simply won't freeze at DICE temps.
That wasn't the case when I was testing (The coolant would freeze but not badly) and I also noted once these were present, temps did come up but stayed low enough to do some good benching.
I had to be sure the res itself was insulated vs condensation and be sure what did drip was caught by a pigmat to control the mess - Had the res sitting on one of those for that purpose. Of course be sure your coolant lines are also insulated vs condensation and be sure any water that runs along /drips from them doesn't cause trouble.

[Witchdoctor]

Wow bones, that is insane bro, nice write up ........ :thumbsup:



It may just be easier to strap some pots on them and have at it ........ :)

[MaadDaawg]

My main concern would be condensation on the gpu water blocks. I'd think you'd need to eraser underneath and all around them to prevent water damage to the pcbs

[Bones]

I agree it's probrably best to go with a DICE pot and be done with it - quick, easy and not as much to do plus you'd get better temps anyway.
Problems with condensation at the blocks are a concern and you'd have to do some prepping to solve that problem.

However if wanting to experiment with supercooled liquids this is just one way you could try it.

[punx223]

I agree it's probrably best to go with a DICE pot and be done with it - quick, easy and not as much to do plus you'd get better temps anyway.
Problems with condensation at the blocks are a concern and you'd have to do some prepping to solve that problem.

However if wanting to experiment with supercooled liquids this is just one way you could try it.

^^ what he said.... I have dune subzero liquid already and you gotta have the tubes wrapped and also socket/block prepped just like LN2 but even more so because the condensation will be condensation and NOT ice so that means alot more loose water around....

[Kal-EL]

What they said.

[MaadDaawg]

too much work the return I'm thinkin :blink:

[crustytheclown]

stuwi did it back in the days... http://www.xtremesystems.org/forums/showthread.php?235312-Sub-6.7-s-1M-Sub-6-10-min-32M-48.8-K-3D05-single&p=4034671&viewfull=1#post4034671

[MaadDaawg]

Thanks Crusty, maybe wasn't such a stupid idea after all huh :D

Was he using glycol as a coolant?

Looks like a used a rad in front of the SS head/block? do you know what kind of pump he had on it?

[kikicoco1334]

why don't you instead of cooling like they way y'all been talking about why don't you guys just dock/cut a hole on a cheap ras and stick the SS head in the ras and have the ss cool it

cuz i was using TEC like that and it's not as good as docking the sucker on the die but it's still pretty cool

[kikicoco1334]

here is kinda a better idea of what i am talking about

http://i18.photobucket.com/albums/b102/ghostloveff/cooler.png

[crustytheclown]

boris had just bolted an evap on an existing waterloop.he used some car coolant if i remember right... he did it for his 4870x2 which is a pain in the @s$ to ln2 or phase it.
it's very easy to do it.A nice start for testing without much trouble insulating

[Witchdoctor]

Well like I said earlier it is not a matter of this working, Physics dictates that it has too... simple math .........

it is how long it will take to get to desired temprature and if it can in fact hold it's ground with the rediculous amount of heat energy being disapated by 4x 580's with in the surface area of the evaporator.

Would be fun to try though

[MaadDaawg]

then again.. a single Feser quad at ambient keeps them running nice and cool, so....

[MaadDaawg]

why don't you instead of cooling like they way y'all been talking about why don't you guys just dock/cut a hole on a cheap ras and stick the SS head in the ras and have the ss cool it

cuz i was using TEC like that and it's not as good as docking the sucker on the die but it's still pretty cool

don't own a cheap ras :Dizzy: :laughing:

[Splave]

here ya go

http://i.imgur.com/UNAZH.jpg


http://i.imgur.com/o69Az.jpg


using a tuniq tower air cooler as a cooling radiator attached to cascade

Source: PT1T

[Witchdoctor]

That is a great idea ............ :thumbsup:

[Kal-EL]

Thats an interesting application of an hs, innovative :)

[MaadDaawg]

I like it :thumbsup:

Is that Eheim tubing? looks like it

[kikicoco1334]

don't own a cheap ras :Dizzy: :laughing:
>.>


what kind of water is that?

[Mr.Scott]

>.>


what kind of water is that?
Wet. ;):D

[MaadDaawg]

Blue :D

[kikicoco1334]

you guys are funny >.>

[GFDuke]

Looks like windshield de-icer fluid.

[MaadDaawg]

Looks like windshield de-icer fluid.

That's what I was thinking too Duke, but that's not the question Kiki asked :D

[punx223]

Hate to rain on the parade again here but I have mounted a SS to a air cooler actually same damned one.... Tuniq Tower and the heat pipes gave out after about 12 minutes of course I was just screwing around and maybe the fluid it was dipped in helped keep it from happening but just FYI.

[MaadDaawg]

wasn't gonna try it anyway :D :laughing:

[Mr.Scott]

Hate to rain on the parade again here but I have mounted a SS to a air cooler actually same damned one.... Tuniq Tower and the heat pipes gave out after about 12 minutes of course I was just screwing around and maybe the fluid it was dipped in helped keep it from happening but just FYI.
I was kinda wondering about that, given that the heatpipes have liquid and wicking in them. When they freeze, they expand, and eventually pop goes the weasel. In that pic you can already see the base blowing apart.

[MaadDaawg]

I was kinda wondering about that, given that the heatpipes have liquid and wicking in them. When they freeze, they expand, and eventually pop goes the weasel. In that pic you can already see the base blowing apart.

All around the mulberry bush .... :laughing:

[Splave]

this is probably why he used an old ass tuniq tower and not a nice TRU Copper L33t edition ;)

[Witchdoctor]

Basic principles of conduction can be acheived with just about anything. The idea is sound, using a conducter with heat pipes obviously is not

This would be nive if it had longer rods ..........

LOL .......... an old scholl swiffy FTW

[Neuromancer]

Looks like windshield de-icer fluid.

the low temp stuff is yellow iirc. (EDIT: nevermind that is the de-cer stuff)

Yeah any solid CPU block dipped in liquid will work but as was mentioned much earlier. Load will overtake the capability of the SS. It wont hurt the SS, I wouldnt think since there is a medium between them that wont change temp quickly, but still the water will be getting warmer than it does cooler.