17 Nov
2005
17 Nov
'05
2:08 p.m.
While I don't use a water immersion system, my main system has a water
cooled processor, motherboard, hard drive and graphics card. Actually it's
automotive windshield washer fluid, 30% methanol, with an added surface
tension reducer. A methanol mixture has a lower viscosity and better wetting
properties than water alone, so it runs better through the hoses and
transfers the heat better. The drawback of low viscosity is that a methanol
mixture will leak even when water doesn't.
I like a quiet machine, so I keep the pump and reservior in the garage,
where I allow things to be a bit noisier. My next step in my quest for quiet
will be to attempt to build a sealed oil filled power supply with a heat
exchanger to transfer the heat to the water. I'm also thinking of soldering
copper tube to the SDRAM heat sinks.
It would probably be cheaper just to buy a noise cancelling headset.
Yes, this does violate the 10 year rule.
Eric
On 11/16/05, Allison <ajp166 at bellatlantic.net> wrote:
The real problem with water is not it's
conductivity. High power
tube transmitters have used distilled water in the past as it's a
really poor conductor if kept clean.
DEC experimented with water cooling too. The Aquarius project was
a water cooled VAX. Too many headaches with leakage, heat transfer
to the environment and installation issues. Systems like that use
a chiller and heat exchanger to cool the closed loop water system.
Those are costly and difficult to install. Murder if it should leak
in a computer room. Then there is an efficientcy problem as you end
up using power to move heat which adds heat..
With all that, the circuits they were trying to cool were getting more
power efficient. So by time they worked out wet cooling air cooling
was again attractive or at least far easier.
It's still packaging. ;)
Allison