Improv Water Chiller: SS Phase & CPU WC Loop

[MattNo5ss]

It's been almost a year since since my Venomous X and 8800GTX became more than friends, after eying each other across the PCB of a EVGA E760 Classified for about 4 months. Some of you may remember when they finally tied the knot (or zipped the tie) not so long ago.

Now the story of the rise, fall, and rebirth of a SS. Shortly after the union of the VX and 8800GTX, a newcomer appeared with promises of enough cold for everyone. As the CPUs grew, it wasn't long (~5 months) before they noticed his promises of cold starting to dwidle. The SS realized this, and knew he had bit off more than he could chew in coming to this climate in the southeast US. So, the CPUs called out for help from the oppression of the inadequate SS. A hero emerged from the hot and humid land of Atlanta, his name, Cascade. He came in and with nearly 3x the cold, easily overthrew SS from his throne. The SS has been in hiding ever since, embrassed to show his face amongst Cascade and the CPUs he let down. So, what's SS to do when always cast into the shadow of Cascade?

.
.
.

Alright, since you have the background, it's time to begin.

My plan is to try to combine the SS and my CPU loop to make some sort of water chiller. I have no idea if it'll work. It's just an idea I had a while back, and thought it would be fun to test.

Components

• EK Supreme HF
• DTek FuZion v1
• MCP355 + XSPC Res Top
• Black Ice Xtreme 360
• Single Stage Phase Unit
• Distilled Water (~50%)
• Anti-Freeze (~50%)

I have the components set up like this:

Here's the actual setup while filling and bleeding:



Distilled + Anti-Freeze looks like my loop is running on Moutain Dew



Next, I hooked the SS up to the DTek block and put a temp probe in the res:





Now, a couple shots of the setup before the SS was turned on and after about 3 hours of the SS being on:





So, one observation is that it takes FOREVER for the water to cool down. The only reason I stopped when I did was because 1) I needed sleep 2) a lot of condensation. I'm not sure how cold the water would have gotten if I left it on for longer. I didn't insulate because I wanted to see just how much condensation would form. See pics below for condensation:









Here's a couple of evap pics when I was removing everything:





Next steps:

• Insulate the EK block and tubing coming out of it, since that will be on the motherboard.
• Insulate the motherboard socket area.
• Put a heat load into the loop.

I'm not sure if the condensation will be as bad with a heat load in the loop increases temp, or if the temps will get colder if left on longer. So, better to be safe than sorry.

Any help or suggestions would be appreciated

 

[Bobnova]

Gotta get rid of the rad, it's doing it's absolute best to maintain the water temps at the ambient air temp, so the moment the water goes sub-ambient the radiator becomes a heatload. A big one, at that!

 

[Janus67]

Very cool

I wonder how much heat (or in this case cold) was lost through the tubing, would it be good if you wrapped it in something to help contain the cold?

Also I'm not sure if the radiator is helping it would just be pushing ambient temps through the coils instead of keeping the cold in it.
edit: bob beat me to it.

Maybe you will need to use a different ratio of distilled/antifreeze?

 

[Bobnova]

I would use 50/50, that is good to -30c or so.

 

[MattNo5ss]

Could I just insulate the rad somehow by wrapping it in frostking or something? How much difference do you think would there be in insulating the rad and completely removing it?

I was planning on wrapping tubing in that "tape insulation" or whatever it's called for outdoor pipes.

Yeah, seeing as how the water temp didn't get very cold relative to the evap, I think I could reduce the anti-freeze by a lot. Maybe even completely remove anti-freeze if a heat load is on the CPU block. I used 50/50 b/c it said on the bottle it would be good to -34F or -36C.

 

[Bobnova]

Wrapping the rad in insulation will help a lot. I'd do that before you do anything else and see what happens.

Keep in mind that while the water temp may only be at -10, the SS's cpu block is much colder and water will freeze to it even when the rest of the loop isn't that cold.

 

[EarthDog]

Been waiting for this thread! Nice!

I figured due to the amount of heat/cold transfer through the block wasnt going to be as efficient as a direct contact it would take a while to cool down. We were pretty close to freezing so I cant imagine with a heatload it would freeze.

Looking forward to more results!

Gotta get rid of the rad, it's doing it's absolute best to maintain the water temps at the ambient air temp, so the moment the water goes sub-ambient the radiator becomes a heatload. A big one, at that!

exactly what I was thinking when I mentioned stuff in the other thread.

Keep in mind that while the water temp may only be at -10, the SS's cpu block is much colder and water will freeze to it even when the rest of the loop isn't that cold.

You dont think that the nomalzized temps of the water and tis flow over that cold spot will prevent that from happening?

 

[Dolk]

More!

 

[Janus67]

Been waiting for this thread! Nice!

You dont think that the nomalzized temps of the water and tis flow over that cold spot will prevent that from happening?

Maybe because of how the waterblock is small channels it could begin to crystallize on part of it and then slowly crystallize more and more (think of how a river/stream freezes).

 

[MattNo5ss]

I forgot to take a pic of my FC5V2 and other temp readings... Notice the inline fittings between components, that's where the temp probes are located, plus the one in the res. All the readings were within 1C of each other, so it seems the temps do normalize even though there is an evap on a block. That makes since because it's the same with a CPU on a block. Although, I'm curious whether this will change when there is both a very high heat load (CPU) and very low heat load (SS) in the loop at the same time. I expect it will level out, but we'll see.

First, I'll insulate the rad alone and take readings, then I'll insulate the tubing and take reading. Hopefully, to see how much heat is added to the loop by each piece.

 

[EarthDog]

Yep.. thought about that as well.

 

[Bobnova]

The loop will normalize, but the boundary layer of water in the block will be much colder.

Check out TiN's AC unit chiller and you'll see what I mean, he had plenty of flow and it froze anyway.

 

[Devil_Dog]

Never thought of that. When I pick up an SS I might give this a try if your end results are fruitful.

 

[ChanceCoats123]

If you have a dehumidifier, definitely run that... You're getting a lot of condensation and you weren't even below zero because of the dew point. Less moisture in the air = lower dew point = less condensation.

 

[MattNo5ss]

That's just how it is in the southeast: 24C ambient, 80-85% relative humidity, and a dewpoint of ~21C is not a good combo. Yes, that's a 3C difference between ambient and dewpoint

Good insulation will be a must

 

[ChanceCoats123]

Oh snap... That sucks. And I thought it was humid in the Midwest...

 

[bennoculus]

Pshh... Try living in Florida, Chance. I'm sure Sebastian can vouch for me (since he's lived there longer than I ever did). The summers there SUCK!

 

[xsuperbgx]

I agree that you need to remove the radiator. As soon as your water temp going into the radiator drops below ambient temperature then it is hurting your temps. Even if load temps are slightly above ambient you would probably be better off removing it, as a restriction in the loop.

 

[hokiealumnus]

I say insulate the radiator.

Part of the benefit to this cooling combo will be the very thing that causes the loop to take forever to get cold - water likes to keep its temperature. If you have a radiator full of water at the preferred temperature (and insulated to keep it from being a detriment), you'll keep the loaded temps down longer.

This is a great way to take a SS that doesn't behave itself under high loads and make a hybrid cooling solution that will keep its temp under load. Excellent execution, it just needs insulation is all. I'd also leave the antifreeze in the loop. It's not hurting anything and you'll be grateful when your rad doesn't burst a pipe from freezing up.

 

[PenguinsPal]

South Carolina is terrible for humidity...