I've been looking into refrigerated chillers as a possibility for sub-zero cooling instead of using Dice or LN[sub]2[/sub]. Most are meant to recirculate water and can go down to -10°C, not that great. There are also cryo-coolers that can go down to as low as -110°C. At work, we have one that goes down to -90°C. It would require construction of a heat exchanger coil to tightly fit the thimble and encased in a chamber pumped under vacuum to prevent ice formation. That would be used to run a line to the processor and possibly the video card(s). I'll probably have to use alcohol as the coolant and a cryo-pump to handle the circulation; also need to make sure the tubes and fittings can handle it. Since we have very low dew point (well below -40°C) compressed air and a dry N[sub]2[/sub] generator, I could flood a specially modified or built chassis with them to greatly reduce or eliminate condensation. I would, of course, route the dry N[sub]2[/sub] with tubing to concentrate it over the coolest parts. The motherboard would probably be the GB OC one and I'm trying to decide if I should use the i7 950 or the 990x. Will most likely SLI at least 2 GTX580's. If I manage to pull this off, I'll try to post pics and vids with results. Not expecting to break any world records or anything, just hoping to maybe break the 7GHz barrier and post respectable X and P scores with this thing... we'll see what happens.
Refrigerated Chillers
- Thread starter jayawms
- Start date
SparkleDJackson
New member
if the system was submerged in oil and was piped for the mentioned setup would that eradicate condensation or any icing n such. as if there was any moisture on the system after submerging it the liquid would just chill ontop of the oil to make it the ultimate risk free cooling solution lol. i aint gonna lie. that would rock hard
you need to come join the bench team mate
acetone would work for the coolant too i think
Acetone could work, unfortunately it likes to eat a lot of plastics and rubbers. I would probably use anhydrous Methanol and silicone or teflon tubing with metal fittings; would also have to replace the block seals with silicone as well.
It would be cool to join the bench team if I lived in the UK.
Well, here's a thought; as long as the fluid is (and remains) non-conductive, is non-flamable and stays relatively low viscosity at the target temperatures, how about submerging the entire rig in it and just use that fluid as the coolant. Whatever is used could be piped directly over the cpu and gpu's, and just recirculate the fluid in general to the chiller. The entire rig would be cooled. Be interesting to see how that would work. Maybe that's another project.
anybody is free to join the oc3d benchteam mate, we have people from the uk, mainland europe, north america and africa i think 1 guy is from, we are a uk team as its where oc3d is but there are no limitations on where members come from.
thats true about acetone, not something i thought about tbh. if you could use it to chill a submerged system i think you would be onto a winner as keeping the liquid (oil normally) cold in those systems is the hardest thing about them.
thats true about acetone, not something i thought about tbh. if you could use it to chill a submerged system i think you would be onto a winner as keeping the liquid (oil normally) cold in those systems is the hardest thing about them.
Did a little more research and it looks like acetone and methanol have freezing points too close to -90°C. Oddly enough, it looks like the freezing point of methanol drops a bit with a small amount of water (>30%) added, but I think it would still be a little too risky to use. So now I'm looking at Novec 7000 as an alternative; it's non-conductive and relatively safe. It's freezing point is around -120°C so as long as the viscosity doesn't increase too much at -90°C, it would be better for both projects (unfortunately it's about US$250 per gal). For the immersed system, I'd have to set up 2 isolated zones, a -80 to -90°C zone for the cpu/gpu's and a -10 to -20°C zone for everything else since many of the on-board components (caps, resistors, and such) don't like to be too cold. Also needs a vapor condenser and I need to decide what materials to use for the tank and reservoirs. Would take a little more work but can be done. Still in the early planning and design stages but will see how it goes.
What exactly is involved in becoming a member of the OC3D benchteam anyway? I've seen a few talking about HWbot and boints and such. Never really looked into it all before.
What exactly is involved in becoming a member of the OC3D benchteam anyway? I've seen a few talking about HWbot and boints and such. Never really looked into it all before.
How about a "cold finger" type system with a small vacuum chamber just around the components? Evacuating an entire case will require a pretty extreme pump (well depending on what kind of pressure you want to reach) plus the case will need to be reinforced substantially to deal with the pressure differential. Building a smaller chamber just around what needs to be protected would be easier I would have thought.
Dont know much about full immersion systems though so couldn't compare the two.
Dont know much about full immersion systems though so couldn't compare the two.
YogevSkill
New member
interestin idea .keep us posted. and good luck.
p.s i used to have idea(probebly dumb) to use a mini freezer(-12c) refrigerator as a case with water loop inside filled with alchol so not to freeze.
p.s i used to have idea(probebly dumb) to use a mini freezer(-12c) refrigerator as a case with water loop inside filled with alchol so not to freeze.
Goldskis45
New member
This is incorrect...
Water has a density of 1g/cm[sup]3[/sup]
Mineral oil has a density of 0.8g/cm[sup]3[/sup] (one of the more common submersion oils and most oils do weigh less than water)
[Unless you use something from 3M (their cooling fluid can be extremely heavy) but it runs in the $200/Gallon range]
This means that water SINKS in oil... If you have any pipes protruding out of the oil that can have water condense on them the water will get into the oil and sink to the bottom... If the motherboard is below these pipes the water can make its way down to the motherboard and cause a short the same way it would if the computer was out of oil...
Any pipes protruding from the oil should not be directly above any electrical component
Directly underneath the pipes should be the lowest point of the tank so any water that happens to find its way in will stay away from the motherboard...
Any sub-zero/near-zero system that can have water condense on its pipes should have in my opinion a wrap around the pipes that is porous and will absorb any water that wants to condense on the system
Goldskis45
New member
Have you ever been on Puget Systems site before? I got to see their first submerged rig in action a few years ago they are pretty cool
http://www.pugetsystems.com/submerged.php
fredtrix
Member
Did you manage to see the temps of the system?
Goldskis45
New member
Well the one I saw was the original version so the temps were actually quite warm... but it took 11 hours? for it to get to its max temp of 80[sup]o[/sup]C at full load (Has an amazing heat capacity)
It was just idling while I was there and was chilling at something like 26[sup]o[/sup]
The newer versions with radiators (V1 didn't have one while I was there) under load don't break 35[sup]o[/sup] I think.
Edit: ah, the history page has all the info http://www.pugetsystems.com/mineral-oil-pc.php
fredtrix
Member
I was at the site and read about it, just wanted to know your first hand experiance =] Thanks for the link though.
Scoob
New member
Sounds like an interesting project
I've been looking at simple chillers with a friend for a build he's considering. We'd thought about still using conventional rads in the loop to get temps post-block back to as near-ambient as his current loop is able, before further cooling the fluid via the chiller. The chiller would of course only kick in once a certain coolant temperature is achieved.
The ideal situation of course would be to have the chiller do ALL the cooling work, but that'd see the PC likely drawing 2,000w+ from the wall, which is a bit silly lol. Still, I am impressed that the higher-end chillers can provide up to 1,650w of cooling with only a 550w power draw. I think a lower-capacity chiller with a simple rad to take the edge off the temps first would work nicely.
Anyway, great project, keep us posted!
Scoob.
I've been looking at simple chillers with a friend for a build he's considering. We'd thought about still using conventional rads in the loop to get temps post-block back to as near-ambient as his current loop is able, before further cooling the fluid via the chiller. The chiller would of course only kick in once a certain coolant temperature is achieved.
The ideal situation of course would be to have the chiller do ALL the cooling work, but that'd see the PC likely drawing 2,000w+ from the wall, which is a bit silly lol. Still, I am impressed that the higher-end chillers can provide up to 1,650w of cooling with only a 550w power draw. I think a lower-capacity chiller with a simple rad to take the edge off the temps first would work nicely.
Anyway, great project, keep us posted!
Scoob.
That wont work, a freezer cant dissevolve(?) a constant emission. I thought about it to
wassupdoc
New member
Dissipate
A freezer/fridge is designed to keep cold things cold, not cool things down, with a pc inside it would burn out the motor, not to mention the condensation
