Radiator-less liquid cooling

[Rubycon]

Anyone have such a setup?

Instead of using a radiator your supply/return lines run to a chiller OR (if you're lucky) you have a supply/return chilled water line near the PC that you can tap into.

Such a system could truly be quiet for the exception of hard drives (how ironic that's what makes all the noise on my system because 12 spindles - 10 of them at 10k) and no increase in room temp!

I'm on a ship and there is CW supply available however I've asked and it's off limits.

 

[aigomorla]

Originally posted by: Rubycon
Anyone have such a setup?

Instead of using a radiator your supply/return lines run to a chiller OR (if you're lucky) you have a supply/return chilled water line near the PC that you can tap into.

Such a system could truly be quiet for the exception of hard drives (how ironic that's what makes all the noise on my system because 12 spindles - 10 of them at 10k) and no increase in room temp!

I'm on a ship and there is CW supply available however I've asked and it's off limits.

bah... this kind of process is going to be expensive.

Reason is i only see 2 ways to do it, 1. TEC 2. Compressor.

The chilled water unit will be majorly expensive. Around 700 dollars and be terribly inefficient.

The TEC method will be dangerous because if you overshoot the TEC wattage, your temps will climb climb and climb.


How about just a passive cooling reservoir like a reserator if your only going to use hard drives?


If you can tap into the cold water, as long as your water temps held constant i dont see how this wouldnt work tho. The problem is getting your fresh supply of water.

 

[Rubycon]

I have TEC coolers for laser pump heads that are rated in kW. Hot/cold loop configuration. Chilled water is not a problem. Heck 1000 pc's wouldn't move it much - the main loop is 13,000 liters/minute at 5°C with a running pressure of 110 psig! But installing nipples on the line is not allowed even near fancoil units.

Of course G0 temps (E6600 at 3.6 default vcc) of mid 20's idle mid 40's load is nothing to make fuss over. That's on AIR.

 

[DerwenArtos12]

Originally posted by: Rubycon
I have TEC coolers for laser pump heads that are rated in kW. Hot/cold loop configuration. Chilled water is not a problem. Heck 1000 pc's wouldn't move it much - the main loop is 13,000 liters/minute at 5°C with a running pressure of 110 psig! But installing nipples on the line is not allowed even near fancoil units.

Of course G0 temps (E6600 at 3.6 default vcc) of mid 20's idle mid 40's load is nothing to make fuss over. That's on AIR.

Your ambients have to be 10C! thats not comfortable, not even a little.

 

[aigomorla]

Originally posted by: Rubycon
I have TEC coolers for laser pump heads that are rated in kW. Hot/cold loop configuration. Chilled water is not a problem. Heck 1000 pc's wouldn't move it much - the main loop is 13,000 liters/minute at 5°C with a running pressure of 110 psig! But installing nipples on the line is not allowed even near fancoil units.

Of course G0 temps (E6600 at 3.6 default vcc) of mid 20's idle mid 40's load is nothing to make fuss over. That's on AIR.

LMAO!

Okey then do this:

Its my design with a friend of mine named martin. Only your going to add much bigger TECs. The hotside will need to be cooled down somehow. Why cant you use a radiator? How about passive? A PA120.3 passive could do justice if your ambients are low enough. Heck drop 2 and im sure they would do fine on passive config.

http://www.xtremesystems.org/f...howthread.php?t=172946


My soon to be new addition to my main machine. :X

 

[jagec]

Originally posted by: Rubycon
I have TEC coolers for laser pump heads that are rated in kW. Hot/cold loop configuration. Chilled water is not a problem. Heck 1000 pc's wouldn't move it much - the main loop is 13,000 liters/minute at 5°C with a running pressure of 110 psig! But installing nipples on the line is not allowed even near fancoil units.

Of course G0 temps (E6600 at 3.6 default vcc) of mid 20's idle mid 40's load is nothing to make fuss over. That's on AIR.

Well, instead of using the chilled water directly, you could try something like wrapping some copper tubing AROUND the chilled water line, or some other water supply pipe that isn't stagnant flow. Then run your own system closed-loop. That way you get the cooling you need, without actually tapping into the system and thus breaking The Rules.

Or...you're on a ship...do what ship engines do and cool with seawater. Corrosion will obviously be a concern, so you'd probably need zincs of some kind, but the cool factor would be through the roof. Probably not very realistic on a BIG boat, though...

The TEC coolers would work as well, of course, but I think that icing would be difficult to overcome. You'd have to have a heat exchanger to ensure that the water was cooled evenly to prevent freezing, but at that point you might as well just get a radiator, and use the TEC on the CPU directly. Interesting that you guys have TECs AND a chilled water supply...in our lab all the lasers just use the chilled water.

BTW, how do you get access to all the cool things you seem to play with on that ship? For some reason I thought it was a cruise ship, not a research vessel.

 

[Rubycon]

Originally posted by: DerwenArtos12

Your ambients have to be 10C! thats not comfortable, not even a little.

About 18°C. But the system is near a supply that is blowing out 8°C air and it helps.

When the PC is not near the supply idle temps skyrocket into the mid 30's and load into the lower 50's. :Q

Originally posted by: jagec

Well, instead of using the chilled water directly, you could try something like wrapping some copper tubing AROUND the chilled water line, or some other water supply pipe that isn't stagnant flow. Then run your own system closed-loop. That way you get the cooling you need, without actually tapping into the system and thus breaking The Rules.

Or...you're on a ship...do what ship engines do and cool with seawater. Corrosion will obviously be a concern, so you'd probably need zincs of some kind, but the cool factor would be through the roof. Probably not very realistic on a BIG boat, though...

The TEC coolers would work as well, of course, but I think that icing would be difficult to overcome. You'd have to have a heat exchanger to ensure that the water was cooled evenly to prevent freezing, but at that point you might as well just get a radiator, and use the TEC on the CPU directly. Interesting that you guys have TECs AND a chilled water supply...in our lab all the lasers just use the chilled water.

BTW, how do you get access to all the cool things you seem to play with on that ship? For some reason I thought it was a cruise ship, not a research vessel.

Yep, cruise ship and there's lots of tech here.

Well it's look but no touch. The laser heads produce up to 10W CW at 532nm for beam shows. I have a 200A FET controller that can regulate the current to the TEC's and it works quite well. Everything has to be approved if it leaves the case. The Enermax Liberty PS that was tripping AFCI protection was really frowned upon so attaching anything external is a no-no.

 

[aigomorla]

rubycon... just run a monster radiator passive.

3 MCR320's on passive should be good for around 600W of heat load.

Each MCR320 you add, add another 200W.

Or you can try this:
http://www.frozencpu.com/produ...tor.html?tl=g30c95s667

Calculate all the blocks your system has. Post a number, and i think i can help you piece out a passive system using large class radiators. Unless the problem is mounting the radiator.


Heres a calculator made by my friend martin on XS. i also use it sometimes to determine what type of pump with radiator i need on such system. If you like to tweek some numbers try it out and see what will work.

http://www.xtremesystems.org/f...howthread.php?t=151627

 

[Rubycon]

The X38 runs so much cooler than the 975X board did I think I'll just forget about water cooling now.

A Q6600 at 400x9 with default volts on air that's Prime/OCCT stable is nothing to sneeze at. I bumped the FSB up to 422 for 3.8G and it boots, gets into windows and runs but as soon as you load up on Prime it will reboot. The chip is of good run. (SLACR) Fitting with the Slacker Stacker 830 hehe. I'm sure it would do 4+ with VCC jacks and water but right now that's just not going to happen. Sub 15S SP 1M times are fine with me.

 

[aigomorla]

Originally posted by: Rubycon
The X38 runs so much cooler than the 975X board did I think I'll just forget about water cooling now.

A Q6600 at 400x9 with default volts on air that's Prime/OCCT stable is nothing to sneeze at. I bumped the FSB up to 422 for 3.8G and it boots, gets into windows and runs but as soon as you load up on Prime it will reboot. The chip is of good run. (SLACR) Fitting with the Slacker Stacker 830 hehe. I'm sure it would do 4+ with VCC jacks and water but right now that's just not going to happen. Sub 15S SP 1M times are fine with me.

lmao!

8C air + passive radiator would be a killer combination. Might even get condensation if conditions were not in your favor.

Also, isnt that air salty on a ship? id be worried about corrosion.

 

[Rubycon]

Air is well filtered. Dust is more of a problem then salt. We keep a strong positive pressure to keep it out.

There are areas exposed to salt and things do corrode fast if they're not protected.

 

[BladeVenom]

Originally posted by: aigomorla

lmao!

8C air + passive radiator would be a killer combination. Might even get condensation if conditions were not in your favor.

Also, isnt that air salty on a ship? id be worried about corrosion.

I also think condensation might be a problem.

 

[Rubycon]

Originally posted by: BladeVenom

I also think condensation might be a problem.

As long as the dewpoint is several degrees below the exposed surface temp condensation is kept at bay. Take anything outside and all bets are off. It's a real problem say with taking a photo when the lens is well below the DP. It will fog immediately. In the Caribbean 30°C dewpoints outdoors are not uncommon.