Cooling with water / coolant in electronic systems. 1

[N9ZN]

I am not a mechanical engineer, my background, for many years, is in Information Technology code design. Please keep you response in easy to understand terms.

This is my first attempt at water cooling a computer, I have no margin for error.

Environment...
I have the option of using 1/4 3/8 or 1/2 water loops and likely will use 3/8 unless someone can build a good case to move to 1/2 inch. This is the first time I have attempted cooling via water and I will be targeting the motherboard voltage regulators, North bridge chip, South bridge chip, and CPU. The loop will consists of 5 water blocks in a serial configuration each of which have a copper transfer plate.

Each block will be mounted on top of each of the area to cool and have a layer of thermal interface material (pads, tape, or paste) between the blocks and the component surface (some metal and some IC chips). The internal are of the water blocs consists of a hollowed area with cool water entering and heated water exiting the block.

Questions...
1. I need a simple method to find the best flow rate for transfer of heat to coolant. My understanding is cooling can be reduced if the flow is to low or to high. Disturbing the system once installed will be difficult.

2. I also need to know what type of thermal interface material would be best for IC chips of a non-metallic material. It must be non conductive, not disperse into the leads of the chip, and have a high rate of heat transfer. Two of the water blocks will be cooling multiple voltage regulator chips per block. For these block it is essential no gaps develop (in the thermal interface material) between the chip surface and water or overheating will occur.

3. Is there a preferred type of coolant to use in this system to enhance heat transfer? My initial thought was distilled water with some type of biocide additive to inhibit organism growth along with a corrosion inhibitor in the fluid. I have no ide what would work best so no corrosion, algae or organisms growth occurs.

4. Tubing material is another area I am uncertain of. Many are available and I have access to most types. Pressure will be no greater than common off shelf cooling pumps Will provide. Fluid temperature is an unknown as it moves through system areas, some components may go as high as 95C while being cooled. Suggestions are appreciated.

5. Maintenance of the system will need to be performed. The system consists of Acetal material, Tubing, copper, brass, and nickel. What is the beat way to clean the inside of the blocks (they can be disassembled), tubing, pump, and radiator? The radiator is my biggest concern due to the many small areas of which I have no direct access to clean. What method would be best for removal of corrosion and any other material build up in the system? Any cleaning solution should be easily removed after cleaning so as not to cause damage to the system while in use.

 

[hacksaw]

start reading everything you can get your hands on, there are no pat answers in building from scratch

 

[IRstuff]

If you don't want to pay for a heat transfer book, you can download one:

http://web.mit.edu/lienhard/

http://www/ahtt.html

As a general rule, going to a liquid cooling system is a bit drastic; I'm unclear why you would design for 95°C case temperature for desktop computer.

1. No, cooling is function of mass flow and conductivity
2. There are companies that specialize in that sort of thing. Berquist comes to mind, but there are a host of them on the web:

http://www.bergquistcompany.com/thermal_materials/index.htm

3. Water is the cheapest, and generally the best

In general, I'll echo Hacksaw, you must get more familiar with HT. Better yet, hire someone who knows what they're doing.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[MintJulep]

Sounds like you have a very clear idea of what the system will look like.

Before you've done any engineering at all related to the heat transfer or performance required.

This is a mistake.

 

[mcgyvr]

http://www.frozencpu.com/

has all the parts/liquid,etc.. you should need.

 

[vpl]

Nice link, mcgyvr, I didn't know there were liquid cooling systems for PCs!

Patricia Lougheed

******

Please see FAQ731-376: Eng-Tips.com Forum Policies for tips on how to make the best use of the Eng-Tips Forums.

 

[IRstuff]

There were some guys, about 15 yrs ago that used LN2 to run their computer at about 3x overclocking, i.e., they took a computer intended to run at something like 1GHz and ran it at 3GHz with LN2 cooling.

If the OP is attempting to massively overclock, then water cooling is not going to be sufficient.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[mcgyvr]

Guys are going into the 8GHz+ range with liquid nitrogen solutions now.. for a short time anyways.

I'm running a corsair H80 liquid cooling setup on my machine (it was a free option when I built it so I thought why not). But its a piece of crap. I was only able to overclock to 3.1Ghz (from 2.6 stock) before temps got out of control. A decent finned forced air heatsink will easily outperform it. So beware.. not all liquid cooling solutions are worth it. There are plenty of benchmarks/reviews out there though.

 

[MiketheEngineer]

And don't forget - water and electricity don't get along too well.

 

[mcgyvr]

Then don't use water... use mineral or transformer oil

http://www.pugetsystems.com/aquarium-computer.php

 

[berkshire]

Another way to go, is refrigerent cooling. Masterflux, and Aspen company make compressors small enough to go into the computor tower and give you a self contained system. It is all very experimental, there is very little information out there that is not proprietary.
B.E.

The good engineer does not need to memorize every formula; he just needs to know where he can find them when he needs them. Old professor

 

[IRstuff]

"only able to overclock to 3.1Ghz (from 2.6 stock) before temps got out of control."

The H80 is air-to-air, ultimately, so you're limited to room temperature air as the baseline, and note that since there must be a positive delta, the coolant must be higher than room temperature, and if the water flow is subpar, then the hot side exchanger could be quite warm.

After market devices are ultimately limited by the packaging of the electronics, the internal heat transfer, and the cold plate temperature. LN2, or even possibly a Freon coolant would get some additional delta T to work with. LN2 gives you roughly 220°C deltaT advantage over room air. That's pretty huge, but obvious, the cost is commensurately higher.

A purpose built processor for overclocking would have the die sitting on a copper block, or something similar, for starters. Anything else is a bit of a Bandaid solution, since design decisions already made compromise the post-facto cooling solution.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[MiketheEngineer]

He said "water cooling"..... I am a structural not a mechanical (but somewhat close by training and experience) and definitely not electrical but I do know about water and electricity BY EXPERIENCE.

 

[IRstuff]

Water, while not necessarily friendly to electronics, is at least removable, whereas PAOs and others, if they leak, are much messier and harder to clean up. If water leakage is a concern, the boards could be conformally coated to minimize the possibly after effects.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[davefitz]

I am not involved in the area of heat transfer from PC's, but there have been many papers published over the last 15 yrs in the ASME heat transfer conferences on this very topic.

My guess is there are "heat pipe" mini coolers that may be epoxied onto the heat sink of the PC CPU, then charged with a small amount of heat transfer fluid ( binary hydrocarbon or one of the banned refrigerants)and the "cold end" of the mini heat pipe would discharge heat either to a finned tube air cooled HX or to a water cooled HX.

One issue with using these non-water HX fluids may be the building code or fire code might restrict the total amount of such fluid that can be in a single room or building.

If water is being used, one can engineer the surface of teh hot end to have nano-particles that ensure CHF critical het flux is not exceeded.