'ghetto' watercooling - possible?

mb2 · Post by **mb2** » Sun May 01, 2005 9:09 pm

just kicking ideas about here..

would it be possible to run this sort of WC setup..
old HS epoxied into waterblock, cheap tubing and cheap pond/fish tank pump.. with a few of litres of water on a deep tray (ie large surface area) w/ no radiator.. perhaps have the water after coming out of the system run down a sheet of aluminium for some more SA + (possibly?) better radiating of the heat before running into the resevoir.

i know this clearly wont kick it with a proper WC setup.. but how much could it cool?

..i mean it would take several hours to warm such a large quantity of water, even if it never cooled down atall.. and i dont imagine using this for a 24/7 computer.. 99% of the time i turn it off at night..
even less strain if i have some speedstep/cool'n'quiet type thing going on.

so how possible is this and what sort of wattage do u think i could cool?

Straker · Post by **Straker** » Sun May 01, 2005 9:17 pm

if you're just doing it for benching or something, sure - tap water is like 4C. a lot (relatively speaking) of people just use a block, pump and a plastic garbage can or such full of water, easier than phase change for short-term cooling.

wim · Post by **wim** » Sun May 01, 2005 11:45 pm

keep in mind that in an open system the water will get gross stuff growing in it and you'll have to cleanup everything once in a while..
if that doesn't scare you, go ahead

Elixer · Post by **Elixer** » Mon May 02, 2005 12:43 am

You might get some ideas from this:
http://www.silentpcreview.com/article171-page1.html

mb2 · Post by **mb2** » Mon May 02, 2005 7:42 pm

(i'm sure i replied to this earlier today?.. o well..)

yes.. a garbage can full of water is a good idea i hadn't considered.. huge capacity.. and wouldn't look too conspicous (the kitchen bin which is due for replacement i have in mind..)
does the surface area of the water make much difference?

would it not work for a full time system? (in the way that i use it ie not 24/7) .. i dont really want to do it for 'benching' only

how close does the CPU temperature stay to the water temperature? eg if u had water at 40*C what would u expect the CPU to be at? [educated guess i know]
how much does the temperature of the water affect overclocking? does it need to be cold to get a decent OC?

are there any other widely available cheap liquids which conduct heat well that could u used instead of water.. eg some kind of oil?

i guess there are things u can put in the water to stop the bacteria having a chance of growing?

"easier than phase change for short-term cooling" isn't phase change rather expensive too?

i could get some car radiator from a junkyard (or wherever) like that guy (thanks for the link.. i read it a while back it probably inspired this idea somewhat) .. but it seems really big and ugly.. if i could think of a place to hide it i guess perhaps..

Katharsis · Post by **Katharsis** » Tue May 03, 2005 3:48 am

from reading many similar posts on various overclocking forums where people have tried this, the general answer is unless you're a mad scinetist type, "don't do it". the water has to be distilled and have anti corrosion / bacteria protectors. other oils and such were usually too thick or expensive (flourinet). any open water source will get dust, hair, bugs, etc in it. the only difference having a larger tank of water makes, is it will take the water longer to change state. longer to heat up and longer to cool off.

the best fanless solution i've seen is a guy dug a trench in his backyard below the frost line and ran a long copper pipe down and back. you can use a large radiator, with a large low speed fan, just remember to make sure the radiator is rather low pressure to avoid killing the pump.

now for the waterblock, theres been a few where they took an old 386 or 486 heatsink (big square thing with no fan) and surrounded it with acrylic sheets. they worked amazingly well as long as the fins are spaced widly enough for proper flow

if you got your labcoat and plenty of spare time tho, don't let me stop you

mb2 · Post by **mb2** » Tue May 03, 2005 10:18 am

"any open water source will get dust, hair, bugs, etc in it"
well, i would put the lid on the bin so hair/bugs etc wouldn't be an issue.. just anything that would grow in the water itself

"the only difference having a larger tank of water makes, is it will take the water longer... to heat up and longer to cool off."
yea.. my hope being that it would take all day to heat up, and then it would cool down during the night?

[other unanswered questions still stand!^^^]

[..if i do decide to go crazy and try this then it will be like july before i have the time anyway..]

jamesavery22 · Post by **jamesavery22** » Tue May 03, 2005 12:25 pm

mb2 wrote:"any open water source will get dust, hair, bugs, etc in it"
well, i would put the lid on the bin so hair/bugs etc wouldn't be an issue.. just anything that would grow in the water itself

"the only difference having a larger tank of water makes, is it will take the water longer... to heat up and longer to cool off."
yea.. my hope being that it would take all day to heat up, and then it would cool down during the night?

[other unanswered questions still stand!^^^]

[..if i do decide to go crazy and try this then it will be like july before i have the time anyway..]

Sealing a large bucket isnt going to go well for heat dissipation. Even a 50gallon trashcan will heat up relatively quickly. When you do passive, surface area is pretty key. 50 gallons of water will hold a lot of heat, but if it cant dissipate any of it (namely sealing it in a large rubber drum) its just going to get hotter and hotter.

Maybe you want to look into bong cooling.

One guy put a 50 gallon garbage can inline with his houses main water line then submerged a heatercore in it. The water in the house was used enough to the point were the water held in the garbage can was refreshed before it could heat up too much. Thats obviously not practical for most people though.

You might want to search for people trying to use aquariums as radiators. As others have already said here, unless you want to go madscientist style, dont bother. Results wont be that great and you'll have a lot of issues to overcome.

Unless you live somewhere where its constantly cold and can put that 50gallon sealed whatever outside, its just not worth it.

pddjsteve · Post by **pddjsteve** » Sat May 14, 2005 7:30 pm

jamesavery22 wrote: Sealing a large bucket isnt going to go well for heat dissipation. Even a 50gallon trashcan will heat up relatively quickly. When you do passive, surface area is pretty key. 50 gallons of water will hold a lot of heat, but if it cant dissipate any of it (namely sealing it in a large rubber drum) its just going to get hotter and hotter.

Hi. I actually registered to reply to this thread. I have a 1.6ghz applebred duron that I watercool with a homemade setup. I have a BeCooling slit edge block, and a minijet606 or something pump (I picked it up cheap locally). I use a 30 quart plastic rubbermaid container for a resevior, and no radiator. I have 5 gallons of distilled water with a bottle of water wetter in the system. With a room temperature of 75F, my cpu idles 86F and full load 95F. This has been running about a year now, aside from changing the water twice. Before that, it was cooling a 700mhz p3 for 2 years on a homemade block and everything was room temperature.

jamesavery22 wrote:You might want to search for people trying to use aquariums as radiators. As others have already said here, unless you want to go madscientist style, dont bother. Results wont be that great and you'll have a lot of issues to overcome.

Unless you live somewhere where its constantly cold and can put that 50gallon sealed whatever outside, its just not worth it.

It isn't difficult, you don't need to do much, and it does work. I eventually want to make a custom resevior and increase the capacity, but there is enough water here so that once its temperature stabilizes (I'd say it is only a few degrees over room temp, the resevior is warm to the touch but barely) it pretty much keeps the cpu temp constant. And it is as quiet as your pump is.

As far as cost, well, $5 for water wetter, $3 for the distilled water, $40 for the waterblock, $20 for the pump, $2 for tubing, $5 for the container; about $75.

I've probably spend about $200 total, though, counting its original incarnation for the p3 and with various research and fiddling and tweaking. I actually started out with a $50 setup for my p3. I did make my own waterblock, first epoxying plastic around a heatsink, then making one from copper parts. The overclockers.com article archives have a ton of good stuff, people have already tried a ton. It is good research and inspiration.

When I upgraded to the duron I felt the need to buy a waterblock rather than use homemade because I wasn't sure I could make something that would take the heat (I've had it up to 2.2ghz from stock 1.6 just to see what it could do). It did about 100F/110F idle/load at that point, but I didn't leave it there long enough to really heat up the water.

jamesavery22 · Post by **jamesavery22** » Mon May 16, 2005 9:16 am

pddjsteve,
Thanks for posting! You are the first Ive seen to get a setup like that to work for a 24/7 setup. Got any pics? And how'd you read the temps?

pddjsteve · Post by **pddjsteve** » Wed May 18, 2005 2:35 am

I don't have anything of its current configuration, but after work I'll try to remember to get the old pics up. Hah. In its original form, it was extremely ghetto looking. It looks a little better now, but it still needs some work. I want to replace my rubbermaid container with something acrylic and custom-fit to my needs.

I think it actually is taking on some aspects of an evaporative cooler, too. There is about 3 inches of airspace inside the sealed container, and it is covered in very fine droplets of condensation (almost a mist).

I use MBM5 and compare the socket and on-die temps... they are usually a within a few degrees of one another. I haven't gotten around to calibrating it or anything, though. But the waterblock is barely warm to the touch. I have the cpu running 1.6ghz on a 266mhz bus when I'm not doing anything important, though I plan on modding it to a "mobile" to enable multiplier switching through software. I've used the idle register hack on it.

pddjsteve · Post by **pddjsteve** » Sat May 21, 2005 5:07 am

The oldest setup, my old pc (the pentium 3) you can see the resevior, though.

Side View

Close-up of resevior. The pump is sitting on a piece of plastic that is rubber-banded to a sponge - decoupled from the resevior for quiter operation.

New CPU (duron applebred 1.6ghz - TBredB 1900+ with cache disabled) and waterblock (becooling slitedge).

Straker · Post by **Straker** » Sun May 22, 2005 3:17 pm

mb2 wrote:"easier than phase change for short-term cooling" isn't phase change rather expensive too?

didn't see this til now - i meant short as in an hour or so, so i was comparing it to liquid nitrogen, not a compressor + evaporator etc. n2 is fairly cheap, but er... gets expensive and labor-intensive (and hard to breathe) to use for more than just benchmarking. something like pddjsteve's setup would work fine but that's not really ghetto any more.

fingers · Post by **fingers** » Thu Oct 27, 2005 5:05 am

I run a 10meter coil of copper tubing (about £10 from some plumbing place) and run a sealed unit that cools my XP2500+ @ 2254MHz, NB, Ti4600, Dual HDD Cooler, all from a single CSP750.

The only fan in the unit is the one in my Seasonic S12.

Needless to say this is very, very quiet and being a sealed unit doesn't need the water changing.

Oh and it runs for 10hours on a 35C day at CPU 48C running Prime95 Torture test. It also stayed below 44C on the CPU gaming for 4hours+ following this though the room temp had dropped to about 32C by this time!

Hope this gives you some inspiration