Cooling with ferrofluid

[NanoSpin]

Hello:

I am developing a ferrofluid cooling system (for CPUs, right now) at Virginia Tech. This involves a magnetic cold plate, and no pumps are involved (I know, it sounds impossible, but it works).

I want to talk to folks out there who build custom cooling systems for overclocking. If you are willing and have some time now or in the next few weeks, I would really appreciate 10 minutes of your time. We can chat or talk on the phone. I will ask you some questions regarding your main concerns while building your computer, and what influenced your choices.

Your participation will greatly help my research, as I am planning to bring the technology to market soon. PM me or post here, and I would reach out to you ASAP.

Thanks!

 

[aigomorla]

ferrofluid is im assuming nano partical ferric oxide. which is iron?

problem with magnetic fields inside a PC is that it can and will corrupt a lot of data.

And no its not impossible its impractical... There has been liquid gallium setups used by the military for use in desert.

http://www.xtremesystems.org/forums/showthread.php?99240-Can-you-say-Liquid-Gallium-cooled-Xeons

The magnetic field is just your first problems...
Nano particles have been played with in liquid cooling.
The problem with nano particles is the abrasive nature of them.
Its like having a million little razors scratch and etch the surface which the fluid will travel though.

Also its extremely messy, and ferric oxide solution i believe will also be highly conductive.
In the real enterprise world, you not even allowed to use distilled water due to humidity it brings.
Hence why the really expensive servers which are watercooled use a special coolant called Flournet, which clumps up together when it comes in contact with a conductive material.

 

[tracerbullet]

Flournet

Fluorinert, I am thinking, is the spelling of what you're describing. It makes a good die lubricant for punching, too! Funny, I didn't know until now that it was good for anything other than that.

 

[Rubycon]

Why would distilled water bring about humidity [sic]?

All the enterprise systems I've encountered use either deionized water or water-glycol mixture. The latter is used if there is a risk of freezing.

 

[aigomorla]

Why would distilled water bring about humidity [sic]?

All the enterprise systems I've encountered use either deionized water or water-glycol mixture. The latter is used if there is a risk of freezing.

cuz tubing is naturally pourous...

despite the sealed nature of closed loops, water still excapes.

Thats where the problem starts, most rack equipment and server screws are steel.

And well, ruby, you dont need me to tell you what happens to humidity + steel. Also the chance of leaks is too great when distilled becoming conductive after usage. The coolant maybe conductive, but the moment it picks up any dust it will become charged.

And most data centers have used and still use Fluorinert.
Thank you for fixing my error.
As this comes in contact with dust or anything charged, it clumps it up.
Hence it retains its non conductive properties.

Fluorinert, I am thinking, is the spelling of what you're describing. It makes a good die lubricant for punching, too! Funny, I didn't know until now that it was good for anything other than that.

It was used quite common in the early water cooled servers.
The machines which costed like 30,000 dollars hardware wise.. lol...

http://en.wikipedia.org/wiki/Fluorinert

Wiki tells u the original intentions of this stuff... however its really nasty... and it smells really bad... >.<
i vowed never to touch that again after playing with it once...

 

[Rubycon]

The porosity of the tubing does not contribute in any significant way of raising ambient humidity. If it did there would be a significant loss of closed circuit coolant losses!

What does increase ambient humidity is evaporative cooling which is not (formally) used to cool systems directly. Air to secondary coolant exchange systems may be used where a cooling tower is placed outdoors, for example.

These setups use stainless steel (316) or titanium piping and manifolds not tubing!

 

[aigomorla]

These setups use stainless steel (316) or titanium piping and manifolds not tubing!

lolol....

Those servers would also be very difficult to replace a part on... lol.
then again... servers are difficult to break if u setup properly to begin with.

 

[Rubycon]

lolol....

Those servers would also be very difficult to replace a part on... lol.
then again... servers are difficult to break if u setup properly to begin with.

Actually they can be switched off, un-racked and moved to a vacuum station that removes the coolant. The manifold simply unbolts and cpus can be accessed if needed. Very rarely are there failures, a PSU may pop here and there but they are redundant and can be swapped while the server is running.

 

[GoatHornz]

Couldn't just use a heat exchanger an have two separate loops. One for that flows water / antifreeze over CPU and then back to the heat exchanger. And have the second loop with your cold plates and the nano fluid in it and this course the heat exchanger. Just separate the two fluids...koolance even sells one made for water-cooling made of stainless steel. Or you could probably get one made from some other stronger alloy!

 

[Rubycon]

Couldn't just use a heat exchanger an have two separate loops. One for that flows water / antifreeze over CPU and then back to the heat exchanger. And have the second loop with your cold plates and the nano fluid in it and this course the heat exchanger. Just separate the two fluids...koolance even sells one made for water-cooling made of stainless steel. Or you could probably get one made from some other stronger alloy!

That's how the standard systems work. Water to water exchangers so the open side coolant never reaches the sensitive areas of the coolers inside the racks.