Sign in with
Sign up | Sign in
Your question

Add this post to the "curious idiot" files

Tags:
Last response: in CPUs
Share
February 23, 2002 7:57:07 PM

OK, I got an outrageous idea that I think I can make work, but I need some help from all you metal/metalloid and EE experts out there...

I am a big fan of Northwood and like taking it to new personal heights with aircooling. I built a watercooler that really seems to do the trick, so here's my next idea (and probably one of the dumbest ones I've had in awhile).

In my lab, I have copious amounts of liquid Nitrogen. You can see where this is going already. I thought it would be fun to see just what my 2.0 GHz Northwood can do. I'd like to run a closed liquid nitrogen system with bleed valves, but here's my concern:

At -195 degrees, I don't think I have to say how cold this stuff is. While I can make a cooling system that handles this temperature, what about the CPU and its casing? Will it shock-cool the hardware and cause serious cracking and destroy my CPU and board?

I'll let you know how this completele waste of time and resources goes...
February 23, 2002 8:27:04 PM

Some of the best inventions were discovered by someone willing to persue their dream, regardless of controversy and criticism. I wish you the best.

<b><font color=red>Cast your vote with your $,</b></font color=red> <b><font color=blue>shed your pride with your opinion.</b></font color=blue>
a b à CPUs
February 23, 2002 9:20:44 PM

this is a serious possibility. I have seen on the japanese sites where this kind of thing is done. The biggest worry is cold shock. Changing the temp too fast and cracking the core. The second is condensation. At -195 degrees condensation will FREEZE on everything.
I am in no way trying to talk you out of this, just stating the known pitfalls in doing this sort of overclocking. Would like to know how it goes, so keep us posted. Good luck and have fun doing it.

I aint signing nothing!!!
Related resources
February 23, 2002 9:25:23 PM

Maybe a tiny pack of liquid nitrogen could be attatched to the top of a heatsink, it would work better than a fan, and the heat sink would prevent the Cpu from being shock cooled! Atleast in theory... :smile: I wish I had a lab.

<i><font color=blue>If wishes were fishes we'd all cast nets.</font color=blue></i>
February 23, 2002 9:34:07 PM

Why not create a hollow inside the heatsink, leave room for expansion of the chemical, and pop in a drop of the Nitro. Close up the hole (welding wouldn't work cause the surface would be too cold, I think) and you have a miniture chill cored heatsink.

Fanless!!!

Better yet (and a lot safer and cheaper) have some Nitro in place behind the main intake. That would cool the air without creating an environment for condensation. It would be through the system before it would condense.

:cool: <b><font color=blue>The Cisco Kid</font color=blue></b> :cool:
February 23, 2002 11:28:55 PM

Quote:
Why not create a hollow inside the heatsink, leave room for expansion of the chemical, and pop in a drop of the Nitro. Close up the hole (welding wouldn't work cause the surface would be too cold, I think) and you have a miniture chill cored heatsink.



So you want to seal an amount of ln2 into a hsf and you think it would magically stay cold forever, lol.




"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
February 23, 2002 11:42:32 PM

Well, it certainly wouldn't stay there, but I don't plan on my chip running on Nitrogen forever, either. :)  My biggest concern with trying something like that just for a single speed test would be being able to cover that hole up with something strong enough to withstand the massive gas pressure that would build up (There is about a 1:1500 volume ration between liquid and gaseous N2 at atmospheric pressures; ie, the heatsink wouldn't have enough internal volume to contain it, even if you did hollow out most of it).

I keep visualizing the hole plug shooting out at insane pressure and velocity, putting a hole in some other component, with the remaining liquid nitrogen shooting out and shock cooling all those tiny wires on the PCBs!

No one said this was going to be easy (or practical) I guess!

<P ID="edit"><FONT SIZE=-1><EM>Edited by RIchardJSampson on 02/23/02 08:45 PM.</EM></FONT></P>
February 24, 2002 12:09:44 AM

Sounds fun.

...
February 24, 2002 12:19:55 AM

Hmm.

Well, to prevent shock-cold cracking the core, you will need a way of gradually dropping the temp on the core instead of blasting it down to -190... you haven't got a way of trickling the liquid nitro through the system (the cooling system, not the computer system :p ) in a gradually increasing rate, have you? You would be looking at minute ammounts of liquid nitro to begin with, and slowly ramping the volume in the cooling system up until it is full and at it lowest temp.

As for condensation: closed system dehumidifier. The entire case would have to be enclosed in a closed system with a dehumidifier that has an external condensation outlet for extraction of water vapour. You would need to run this a while before you start running the cooling.

Just an idea.

-

I plugged my ram into my motherboard, but unplugged it when I smelled cooked mutton.
Anonymous
a b à CPUs
February 24, 2002 12:22:05 AM

Maybe a two stage heat exchanger would be best. Choose a medium for your water cooling system that remains liquid at a reasonably low temperature [automotive antifreeze ?], then metered liquid nitrogen to cool the heat exchanger for the 'antifreeze'. Play till you solve condensation/chill shock problems.
PS Let me know who is willing to keep a supply of Northwoods coming so I can approach them the next time I get bored.

Working on computers is what you do when you get tired of glowing in the dark.
a b à CPUs
February 24, 2002 1:21:14 AM

Get a water block. Run a funnel to it. Slowley add liquid nitrogen, letting it run out the open side. When the system is chilled, turn the flow on full and start the thing.

What's the frequency, Kenneth?
February 24, 2002 1:45:38 AM

I would try and get the white papers or other Docs on Northwood and see what its temperature tolarances are.
The other guy had to right idea to try and SLOWLY freeze the chip. I did see a site in Russia where a guy tried liquid nitro on his board. The nitro condensed water vapor from the air and made a huge block of ice form on his mobo. So yeah, humdidity free is the key.
And you DONT NEED A HEATSINK. at those temps, its pointless to have one. The cpu will sit naked in the socket. But you have to isolate the chip from the rest
of the mobo. Look into how vapo-chill isolates sockets, and maybe a similar setup will work.

Some things to consider:
1. the cpu does have a certain depth to it. It does have height. So you HAVE to ensure the ENTIRE chip is the same temp before you turn it on. A small difference in temp across the cpu can cause it to crack.(at those low temps) So let the cpu sit in the nitro for a good day or so.. let the low temp saturate the entire chip.
2. (and im only guessing here) - that extreme low temp may cause the pins to shrink. and they may lose proper contact with the socket. Dont know a way around that
3. you have to isolate the chip from the socket. Ergo, not allow the liquid to contact the socket in any way. The casing they use for liquid cooling might help. Instead of puting water into that casing, ya put nitro. But again, the casing will have to be slowly dropped to the -190 temp.
Of course using that casing wont allow you to freeze the entire chip, just the core.

anyone else got newb advice for the mad scientist?

Benchmarks are like sex, everybody loves doing it, everybody thinks they are good at it.
February 24, 2002 2:25:51 AM

I think your wrong with no heatsink. I think you would Want some thing to chill slow so the chip dont crack. A Box at the end of the heatsink and all you have to do is slowly fill it. At one end the Cpu heating and the other end the box is cooling. Once you know when it stops cooling you can make a drip system on the time. Some thing like every 4 sec it adds More nitro. And there 10 drops in the box.
February 24, 2002 11:49:58 AM

With a heatsink, there is still the gap (microscopic) between the HS and cpu, the temp differences bewteen the HS surface and chip will likely cause condensation. Even slowly dripping the nitro in. Just a guess though.

Benchmarks are like sex, everybody loves doing it, everybody thinks they are good at it.
February 24, 2002 12:19:24 PM

Quote:
I keep visualizing the hole plug shooting out at insane pressure and velocity, putting a hole in some other component, with the remaining liquid nitrogen shooting out and shock cooling all those tiny wires on the PCBs!


Sounds just like one of the pistons of my old car.

I never said it would work Mat. I merely threw an idea at someone who knows the properties of Nitro to see what sort of replies would come back.

I stopped chemistry after age 15, so I don't really know much about Nitro.

:cool: <b><font color=blue>The Cisco Kid</font color=blue></b> :cool:
February 24, 2002 12:33:35 PM

Quote:
2. (and im only guessing here) - that extreme low temp may cause the pins to shrink. and they may lose proper contact with the socket. Dont know a way around that


Some people did the same thing with a tbird a long time ago, when they used a thermal transfer medium they were able to reach very low temps and high operating speeds, however when they put a waterblock on the die and ran pure ln2 through it, the computer would not boot, it however was fine after they went back to the warmer medium. They could not tell what it was, but I would bet that having the chip at ultra cold temps (and not evenly either) would cause the signal timings in the interconntects to be all wrong, thus no operation.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
February 24, 2002 12:34:48 PM

Quote:
I never said it would work Mat. I merely threw an idea at someone who knows the properties of Nitro to see what sort of replies would come back.



It wouldnt work, the nitro would get warm and the pressure would keep it liquid, or it would explode as others pointed out.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
a b à CPUs
February 24, 2002 12:45:49 PM

Allrigt been to a few more sites to see what they were using. The most widely used device for this is a thick copper bowl with a flat bottom.Insulate the sides and bottom down to the core. The copper seems to slowly cool the cpu without ckacking it if you slowly add LN2. I guess you could set up a drip system to replace what turns to gas and escapes .keeping the chip cooled indefinately.

I aint signing nothing!!!
February 24, 2002 7:10:29 PM

I ran a little experiment the other day. I took a graduated cylinder and made a gasketet hole in the bottom, into which I insterted a thermometer. I filled the with powdered silica gel (which is what we use for liquid chromatography -- a process which separates compounds by allowing them to slowly pass through a solid medium). I put some liquid nitro on top, and observed the temperature fall at the bottom of the silica gel. I think you all are correct in that the gradual temperature fall is quite important. So my results were going well. It took about a minute to go from 28 degrees to zero degrees, and another minute and a half to it -35. Everything was going great! Then my thermometer (which is one of those two foot long lab grade ones) broke and then I had to contend with mercury everywhere, which is not a good thing. Of course, a thermometer is a hollow system with changing pressures, unlike a chip. I think that this might be the easiest way for me to start. The N2 flows through the gel powder very slowly, and if I start with just a little N2, I can gradually move it up. Essentially, the silica gel will be my "heatsink" and will most likely directly interface with the die. Now this issue is, how to allow it to interact with the die and minimize heat transfer from the casing.. Any ideas?
February 24, 2002 8:36:47 PM

http://www.octools.com/index.cgi?caller=supercool.html
im not sure of any other LN2 sites that are in english...but theres suppose to be a store in japan that sells books on extreme cooling and all of the parts to do it and thats why most of the LN2 overclocks are from japan

hope this helps
February 24, 2002 8:44:34 PM

http://www.vr-zone.com/guides/a7m266/
heres another, this site might help a little more, but it wouldnt hurt to use both....also something i read and i dont remember where i read it(some japanese site but the guy did speak english) is that at those extreme temperatures only windows 95 would work, nt/98/ME/2000/xp would not work...sorry i dont have the link, maybe someone else can verify this
February 24, 2002 8:50:20 PM

oh yeah, u might also want to do a voltage mod on the motherboard to push your processor to its limits, also you have to have a motherboard that can lock down the agp bus so ur video card actually starts up....if i think of anything else ill let u know
!