I'm interested in water chillers. They seem to offer the best of all worlds, below ambient temperatures, low noise and relatively cheap. But i'm having trouble finding info about ways to build chillers and just general information about them. I dont want to use TECs just because their inneffecient, expensive and risky. I've seen people using the guts from drinking fountains, air con units and dehumidifiers (that means u pug ) but any articles i find dont cover stripping down the units to their guts or even how they work. I know theres a compressor and a rad and stuff but really i have no idea how they work or how best to utilize their cooling capabilities. Any info you could provide me with would be much appreciated.
Hehe, ok olv, I'll try and dig it out sometime this week and chill the kids' celeron machine with it for "testing purposes". I did gather a wealth of info on it at one stage but I'm not sure if it got lost or whether I've "archived it" (same thing really) but I'll have a look.
Well after a lot of searching a reading things dont seem to be as complicated as i imagined. A dehumidifier simply seems to consist of a compressor with a hot coil (or radiator) on one side and a cold coil on the other. Submerge the cold coil in your resevoir and voila there you have some nice cold water. Add some insulation to all your cold parts to fight the losing battle against condensation and there you have it. Is it really that simple?
Pretty much. The cold bit is the evaporator, the condenser is the one you cool with a fan. We had a thread on this recently somewhere - "phase change" is a good search term to use (if you haven't already been doing so). You'll want antifreeze in your coolant - use a hydrometer (sp?) to get the right balance of viscosity at your target temps, otherwise it will ice up and/or go sludgy on you. One thing I did with mine that I hadn't seen people doing at the time wasto rip off the aluminium vanes from the evap side, prior to submerging them. They're designed to channel air close to the tubes, water has different characteristics, so you're probably better off without them (although I should have really tested before and after to be fully methodical). I'll start insulating the bits later (as long as the chiller pipes weren't damaged in the house-move).
Right. Just confirmed, it still works. Here's a pic mid temp-drop (after running for a couple of hours) In the time it's taken to upload the pics, it's dropped another full degree. I didn't put enough antifreeze in at first, because I'm just using tap water for the test and don't want to use my whole supply of it. You can see ice forming on the tubes here. This will insulate the evaporator and make it less efficient - a higher concentration of antifreeze will be required... The antifreeze I'm using is Glycol based - non-toxic (I have kids). Because I'm running open bath, I don't want anything that will release noxious fumes. Still. 30°C below ambient ain't a bad start for the £3 investment I made for the dehumidifier originally (although I'll probably spend more than that for antifreeze).
you mention no nasty fumes which is good as you've got kids but im sure that dirty great big fan will cause some problems lol
Heh, kids they might be, stupid they ain't... I can make a fan basket out of modder's mesh for the final product if need be, anyway. I've just thrown this setup together to show the vague principle of the design and how it works. here's a pic of how it looked after I originally stripped the casing off it The radiator-like thing nearest the camera is the evaporator section which I've since cut free and stripped of its vanes. That side is now face down in the water, you can see the edges of the underside of the white plastic shroud to give you the idea how I've just bent the whole assembly down 90° and spaced the condensor rad back from it a bit. I've now half-insulated the slocket I'm using for test purposes. I basically painted a dielectric coating over the exposed contacts on the back of the card and cut a small square of the neoprene on the right to fit inside the socket on the other side. I need to get some Vaseline to smear over the pin-holes of the socket so that I can still make contact without leaving air space around the cpu pins for moisture to develop. I could probably use silicon grease but vaseline is thicker and can be built up elsewhere (like inside and around the cpu socket). I turned the chiller off another hour after I last posted and it had dropped to -12.5° and had about an inch thick of ice on the pipes, even with a load of antifreeze in there.
Okay, I went to the chemist and got some vaseline. I took the precaution of adding it where the neoprene is because there was a tiny gap at the side in one place. I also put a layer on top of the chip for safety's sake (although there's nothing to warrant it on the Intel chips really). You might be able to make out that I managed to leave the "Warranty void if removed" sticker intact, too. I haven't got enough neoprene to do the waterblock, so I'm going to just orient the mobo to a WTX-style in a mobo tray for now so that it will just drip onto the floor, instead of any components. I'm wondering if this would be better as a Project Log but I'll think about that when I get a bit further in. Right, should have it running tonight with any luck - the missus is out today, so I've got full use of the kitchen till she's back... muahahahaha!
yay thats what i like to see thats proper extreme cooling although it just makes me want to do it more Just a few questions... for now What affect do such low temps have on pumps? I presume your gonna run the pump inline as if it was sub-merged it'd heat the water, so how will you prime it? Approximately how cool would water at -10 keep a heavily overclocked CPU pumping out approximately 100W (good estimate). How cold does the water have to be to keep the cpu temps below freezing? How loud is that fan and is it 'The colder the radiator the colder the condenser? You said it took a couple of hours to get to that temp, of course you wouldn't neccassarily need for it to be that cold before booting up but if your O/C relied on the subzero temps, would it be as simple as you'd have to wait? Think thats it for now, any other useful points you can think of are welcome seems a shame to have to insulate water blocks if they look so nice. BTW how do u plan to insulate the tubing? OK no more q's i promise
Block attached (although I turned it through 90° after this shot) with pump at the ready. I'm using the Hydor in submersible mode - there was a load of crap on the chiller from where it was out in the garage which has ended up in the water (cobwebs and stuff), the grit-screen filter will help to keep it out of the pump internals. The downside of this is the heat the pump adds to the water - it's a lot harder to get below freezing with that in there. I've used metal collar fittings as an added precaution - They'll withstand much lower temps than I'm likely to get but I prefer them anyway. What I've done is to use 5.5 x 8 mm hose sheathed inside some 11 x 13mm (or its imperial equivalent). I managed to get the lengths just right where I could have a perfect seal on the inner tube and an airtight one on the outer by sliding it over the outside of the fittings at each end. I didn't have any pipe lagging to hand, but I thought I'd try it for now and see anyway - the pipe doesn't need to pass over any components, so drips aren't a massive concern at this point. ... good job, really... Panning back so you can get a better view of what's involved... Then back further still for the complete setup and a quick money-shot... (Note the temperature warning set to 30 degrees ) In that shot, the bios is reading room temp at 28°C from the remote thermistor (I've since moved it away from the floor and it's reading 35° ), whereas the cpu temp is reading 20°C. Well those temps are quite frankly, way off. Room temp in the vicinity was actually closer to 22/23°, albeit in the draught from the chiller's fans, the underside of the waterblock was about 3° (although I need to take the block off to relocate the thermocouple wire which I forgot to add during assembly - it's just shoved in right now).
Sorry olv, you posted while I was typing. I would run the pump inline by choice in a final (permanent) setup. I'd prime it by sucking the coolant through (it's clear tube, so just hold it up as high as you can and stop sucking when you see it's through and stick your thumb over the end to lower it back down. Hey, one step at a time, eh? Just kidding - both my "big rigs" are nicely set up right now. I'll wait till I've perfected this side of it before I rip either of them apart to test on. The only other rig here is a 2GHz P4Celeron that should do 3GHz with the right cooling but it's in an ECS board. Louder than all my computers put together (subjectively speaking). I figure I could put four 92mm Papst together with a freshly fabbed shroud but then I need 12V power at the chiller and it's easier to use mains power. As far as colder = colder, I think the main thing is to dissipate all the heat regardless. Someone else might have more insight on this. Sure. A w/c system with that quantity of water can take up to a couple of hours to hit its max temp anyway. Thing is, even if it takes a day to get to mega low temps, you'd just leave it running from then on anyway. With that said - if I can make an insulated res with a cover, it would take a lot less time. I'm probably leeching heat from the floor as well as the surrounding air. If I can cut that out, It'll be much more efficient. Double-skinned, as it is now - then I'll add some pipe lagging over that.
Ooh. That sounds tasty. The rad face is ~205mm x ~185mm, the current fan is ~170mm diameter. I don't know what its CFM rating is but I wouldn't want to drop below it. Saying that - the noise level is mostly from the compressor anyway. If you fancy selling that fan, I might be interested if you have no other plans for it.
This isn't really off topic - but I was considering a similar setup using an old watercooler from my mums work that I found ... the fitting to hold bottles is knackered but the cooling section still works fine ... it appears to just be a spiral of aluminium (i was surprised at that actually, i was expecting copper ... all the rest of the metalwork is...) with a gap in the centre, all surrounded by packing foam as an insulator. The hole in the centre held a plastic bag from which the water would flow from the bottle, and then be directed to the tap ... I was thinking of simply extending the tubes from my watercooling into the bag and using it as a second res. Flaw #1 being that my pump is nowhere near powerful enough, and flaw #2 being that I'd then be stuck with 4 feet of watercooler beside my computer.
sounds good go for it I did see an article the other day where a guy use the cooling loop from a drinking fountain the other day and it worked pretty well, kept CPU temps below ambient.
