Overclock.net banner

1 - 8 of 8 Posts

·
Registered
Joined
·
1,290 Posts
Discussion Starter · #1 ·
About to pull the trigger on buying parts, and before I shell out $$$, just want some double checking from fresh eyes (sorry for the long post, but work with me and ye-shall-be-rewarded with a build log).

Basically I want to have my RX480 submerged in chilled water. I figure it's a lot more efficient to do it this way rather than trying to chill the loop water directly (which is what everyone who uses a chiller seems to do), because I need to have a few gallons of water to build up a thermal reservoir so the compressor isn't cycling on and off constantly because it's so powerful vs the heatload of a desktop. (And I don't want to buy 5 gallons of coolant, which would be very expensive).

===============================

The parts I've selected are:
Hailea HC-500A
Submersible Pump - 600 GPH
10 gallon container

The container is cheap and would fit the 480 rad, so I'm going with that over an aquarium. The plan is to put a lid on that container, and then drill 5 holes in it. 2 for the 3/4OD tubing to the radiator, 1 for the power cord to the subbed pump, and 2 more for the 3/4OD tube going to the chiller. The chiller has 1" fittings but they'll be stepped down to G1/4->1/2"ID using adapters. The pump will come with barbs. I'd run the tubes and then seal them in the holes with silicone. The tubes going to my desktop box are a typical watercooling loop...MCP655, res, rad, waterblocks, rinse repeat

I've asked XSPC and the RX480 is copper and brass (this is confirmed by their crappy paint job already flaking to reveal brass), so I believe it should be OK being submerged in tap water. I'm fairly certain that because the rad will be submerged that it will have a much better heat transfer ability than it would if it was in air (water transfers heat better than air). So I think in air I would need an RX480 but if it was submerged in room temperature water, I would only need a single 120 rad to accomplish the same thing?

I can change out the chilled water fairly easily if it becomes grimey (or just treat it like an aquarium and use various anti-growth solutions). Just have to remove the rad and subbed pump and dump it outside in the grass.

===============================

Now for some math:

It takes 4.18 seconds to raise 1 gram of water by 1 degree C, if there is 1W of heat applied to it. Assuming 500W heat load and 8 gallons of water (which = 30.32 kg), it would take 4.18 * 30320 / 500 = 253 seconds for the desktop to heat the water up by 1 degree C, assuming the chiller is turned off. If we set the buffer of turning the compressor on to be 5C rise above target temperature, we get 20 minutes before it needs to cycle, which seems reasonable.

Concerns:

I don't have any good way of stirring the chilled water through the radiator. I'm relying on heat transfer alone between water and itself to bring the container to a consistent temperature. This is my biggest issue that I can think of.

I'm not worried about condensation. Being an engineer, I've already built a controller that will sample ambient temperature and humidity and cycle the chiller manually (after I bypass the built in thermostat) to keep the liquid above the dew point.
 

·
Aint da way it used to be
Joined
·
2,837 Posts
The best passive way to "stir" your 10 gal container would be to put it in the bottom middle of your tank and allow convection to do the stirring.
 

·
Registered
Joined
·
277 Posts
I agree Allow convection to do the stirring.
 

·
Registered
Joined
·
1,290 Posts
Discussion Starter · #4 ·
Do you guys think I should go with a smaller rad to give me more room in the container? It seems you need big rads and fans when you're pushing air through the fins, but with water there's a lot better heat tranfer going on. Would a single 120mm rad be able to dump out the heat coming back from the desktop? It should as long as the surface area of the fins is greater than the surface area of the waterblocks being used? (In other words, if you only have 1 front door (waterblock fin), there's no reason to have 3 backdoors (rad fin)).
 

·
Aint da way it used to be
Joined
·
2,837 Posts
Yeah I agree. It makes sense. In theory it wont be able to dump more heat than its taking in. But outside of a Physics text book situation the application may be different. May want to look into this a bit more. As long as your rad is significantly larger in surface area, say like 150%+ larger, you should be just fine. How many blocks are in your system?
 

·
Registered
Joined
·
233 Posts
This is a really cool idea (providing it works as expected). Unfortunately I don't have input to help your decisions, but if you do end up doing it please make a build log and link it here I would love to see how it would work out.
thumb.gif
 

·
Premium Member
Joined
·
16,904 Posts
Did you ever consider buying a cheap used air con unit and using that to cool a loop? Lower temps (easily below zero), cheaper and most likely easier to setup. Takes about the same amount of space as your proposed setup.

Just throwing out ideas, if you're not interested just never mind then
tongue.gif
 

·
Registered
Joined
·
1,290 Posts
Discussion Starter · #8 ·
Quote:
Originally Posted by Jolly Roger View Post

Yeah I agree. It makes sense. In theory it wont be able to dump more heat than its taking in. But outside of a Physics text book situation the application may be different. May want to look into this a bit more. As long as your rad is significantly larger in surface area, say like 150%+ larger, you should be just fine. How many blocks are in your system?
CPU (EK Supremacy nickel), and 2 680 full cover's, and the mobo block that comes stock on the ASRock OC Formula. It's hard to estimate the surface area of those blocks though, because in the case of the cpu block they also use channels/fins to increase surface are. But I think even the casing of a radiator alone (not looking at the fins) is probably more surface area than the cpu waterblock. I think an all-copper 240 rad would have to be sufficient....but it's just estimation
Quote:
Originally Posted by Jj333 33 View Post

This is a really cool idea (providing it works as expected). Unfortunately I don't have input to help your decisions, but if you do end up doing it please make a build log and link it here I would love to see how it would work out.
thumb.gif
Yep. Maybe someone can then look at what I did and make it even better (as I did for other people's builds using chillers)
Quote:
Originally Posted by Alatar View Post

Did you ever consider buying a cheap used air con unit and using that to cool a loop? Lower temps (easily below zero), cheaper and most likely easier to setup. Takes about the same amount of space as your proposed setup.

Just throwing out ideas, if you're not interested just never mind then
tongue.gif
Yes, I considered it. But an aircon unit is supposedly the exact same thing as a water chiller, just done a little bit bigger and with a fan put on it to blow cold air. I also don't want to mess with hacking up the evaporator coil and potentially leaking refrigerant everywhere
smile.gif
I also don't want below zero, since I don't want to have to deal with any condensation proofing (although I already have a solution for this in the form of placing my desktop in an airtight box filled with dry air from a compressed bottle
biggrin.gif
)

=====

And that reminds me, I also have to looking into building a controller to pre-run the chiller before my desktop actually switches on. Wouldn't want to try booting a 5+ Ghz overclock with room temperature water.

Keep going on the suggestions and advice guys. I made this post for you guys to challenge my ideas and present new ones. I've been thinking of doing this build for a couple of years now, and am fortunate enough to now have the funds to do it. Living the dream
smile.gif
 
1 - 8 of 8 Posts
Top