Don't try this at home - watercooling, not as you know it

[porina]

 

I think I discovered LTT early 2016, and after watching pretty much everything on youtube from the old house era onwards. It gave me delusions of becoming a youtuber too. These images are taken from my attempt at making such a video. The video sucked so bad I got a phone call from Dyson asking me to design their next generation of vacuum cleaner. The video never went anywhere. Anyway, I just rediscovered the source files so I thought I might as well post about it here.

 

The concept was, is there another way to do liquid cooling other than the traditional loop approach, or oil immersion? This was my hybrid. You might guess from my avatar I also have a side interest in keeping fish, and I had this old small tank spare. Could I do something with that?

 

The tank is a Fluval Spec with a nominal capacity of 10L (about 2.5 US gal). There is a rear filter compartment which I emptied of the filter material apart from the pump used to circulate water, kept obviously to have water movement.

 

The system is Broadwell based, so i5-5675C, Asus Z97M-plus, 2x4GB Crucial Ballistix Tactical, ancient Intel 40GB SSD and a PSU that would need a new tier to describe how cheap and nasty it is, but happened to be on hand at the time. The heatsink is a CoolerMaster... I think it is a TX3 but not 100% sure on that. I removed the fan, inverted the mobo and balanced it on the lid of the tank. Coincidentally, the lid of the tank has a hole in it to allow the light (removed) to shine in more easily, and it was the right size to pass the heatsink through.

 

 

I had filled the tank already with reverse osmosis + deionised water, so it was as pure as I could get at home. Turn the pump on and I could test the system.

 

Unfortunately I don't have written notes on the performance, and I'm not going through my own videos to get the details out. The test was simple, put a load on the CPU and see how the temperature varies over time. The large volume of water would have significant thermal mass, but the surface area of the tank itself wasn't great so I knew it would likely end up storing the heat in the water. Sure enough, that was what happened. After running for 6 hours, the water got rather warm, with corresponding increase in CPU load temperatures.

 

The thing is, if you have a non-sustained load with long rest periods, say overnight, maybe the heat buildup would be slow enough to allow this to be viable. Or just use a bigger tank to slow the rise?

 

I am thinking if I should revisit this method, as dumping a load of ice in there would quickly allow sub-ambient cooling, for benching timescales anyway. Because of the mount method, there is no condensation risk to the electronics.

 

Oh, there were some minor problems to consider. As the water heated up, gas escaping from the water liked to trap itself within the heatsink. If anyone were to do this seriously, one of the passive air cooling heatsinks may be a better choice with their increased spacing. Also I noticed some potential corrosion effects on the aluminium parts of the cooler. Do note the fins and heatpipes are different material so there is likely some galvanic process going on here. As the system isn't sealed, there will be water loss.

 

The water going in was measured as 0 TDS (total dissolved solids) measured using electrical conductivity. It was significantly non-0 by the end of the experiment.

 

Anyway, it wasn't a serious usage configuration, but was fun at the time. I haven't totally given up on my delusions about being a 'tuber, but I have a much greater appreciation into the efforts involved. It isn't realistic unless I have a lot more time to produce something I can look at and be proud of.

[Technomancer__]

4 minutes ago, porina said:

Because of the mount method, there is no condensation risk to the electronics.

what if the water evaporates and liquefies when it hits the mobo?

[porina]

5 minutes ago, Ethocreeper said:

what if the water evaporates and liquefies when it hits the mobo?

I suppose the potential risk there is if the water temperature is hotter than ambient... wait, I'm dumping heat into it right? Ok, maybe it is not zero risk, but temperature around CPU socket will be comparable to the water since they are linked through the heatsink. Could more distant areas of the mobo be affected? Maybe a bit more spacing to allow air flow would help prevent trapping moist air could help if anyone were to seriously look at it longer term. 

[Tiwaz]

exactly but the better and safer option would be to take an enormous tank to store liquid, at least 20L and use that as your reservoir iwth a watercooling loop without any radiators. Distilled water isnt expensive but its going to if you wanna add Propandiol to keep out unwanted organisms

[SCHISCHKA]

Someone had to try this and document it.

I think the corrosion is the biggest negative. Maybe an all copper cooler if anyone still makes them.

I think this is less practical than evaporative cooling.

[porina]

16 minutes ago, Tiwaz said:

exactly but the better and safer option would be to take an enormous tank to store liquid, at least 20L and use that as your reservoir iwth a watercooling loop without any radiators. Distilled water isnt expensive but its going to if you wanna add Propandiol to keep out unwanted organisms

I had also thought about a "big res" and even ground cooling, which has been done before. But these are less interesting.

 

I have RO/DI water on tap as a part of my fishkeeping setup so the incremental cost to me is insignificant.

16 minutes ago, SCHISCHKA said:

Someone had to try this and document it.

I think the corrosion is the biggest negative. Maybe an all copper cooler if anyone still makes them.

I think this is less practical than evaporative cooling.

It wasn't intended to be practical, just different

 

I had kinda wondered why all-copper coolers fell out of fashion in recent times. I recall they were all the rage on the higher end around the Core 2 era.

[DrMikeNZ]

58 minutes ago, porina said:

I had kinda wondered why all-copper coolers fell out of fashion in recent times.

Copper price increased, cost wasn't worth the benefits and there was a drive towards AIO coolers and improved CPU efficiency. I really enjoyed having 1.5kg of raw copper hanging off the motherboard bending it awkwardly.

[porina]

8 hours ago, DrMikeNZ said:

Copper price increased, cost wasn't worth the benefits and there was a drive towards AIO coolers and improved CPU efficiency. I really enjoyed having 1.5kg of raw copper hanging off the motherboard bending it awkwardly.

1.5kg? That's a lot... until I actually looked it up. Taking Noctua D15 as a high end air cooler, it is listed as 980g without fan. If you were to replace it with equivalent volume of copper that would put it around 3kg. Yikes.