How cold can a computer go and remain functional?

[chsh1ca]

I originally posted this to Cases & Cooling, but someone suggested I post this here. I was originally considering doing so, but figured that I might be better off in Cases & Cooling because of the subject.

Original post:
I was discussing cooling with a friend via IRC, and he raised this question, and to be frank, I tried to answer it as best as I could.

The conversation is as follows:
Friend: "so realistically, how cold can a computer go?"
Me: "probably about to the point where the temperature difference between the aluminum/copper traces and the silicon they lay on is big enough for it to break connections. However cold that is. You'd have to find a chart of temperature expansion/contraction rates for the various materials and do some math to figure out how far the differential could go before breakage occurred, and then add say, 10 degrees for variance. That'd be my guess, but I'm not an expert on temperature and how different materials behave under different temperatures."

My answer was based on the basic principals that all chemical combinations have different expansion/contraction rates, but I find the answer pretty lacking.

Can anyone point me to a site that has this sort of information, or maybe just post it here?

 

[Smilin]

I think actually with the sizes involved you could get a PC down to near absolute zero and it would work fine...fast as sh1t actually. Hard ocp recently did one of those massive overclocks using refrigeration and popped the heat spreader off the P4 they were using...mind you at -50 C who needs a heat spreader?? You'd have to bring it down in temp slowly to let the thing flex a bit so the different contraction rates of the materials could be overcome. I don't think you could mathimatically figure out the temp - there are too many miniscule differences from pc to pc to know which would crack at what temp.

There are a couple exceptions to this: The cmos battery, hard drives and fans. Some of those moving parts use lubricants that would fail and the cmos battery would stop putting out enough juice when it got cold enough.

 

[f95toli]

A silicon based semiconducting circuit will stop working when you have cooled it to a point where the carriers "freeze". A current can only flow in doped silicon as long as there is enough thermal energy to excite the carriers (electrons and/or holes) to current carrying states. In other words: If you cool (even highly doped) silicon it will become a very good insulator at some point, the exact temperature depends on the amount of dopants but I guess it should be around 20-30 K (-240C), maybe a bit higher.

GaAs and GaN (and a few other III-Vs) based components can operate at lower temperatures (they operate using a slightly diffrent principle) and they are therefore used in cooled low-noise amplifiers (used in for example radio astronomy).

This unfortunately means that a AMD processor would not work at 4.2K (liquid helium).

 

[Evadman]

There was an article I remember seeing about an experement using LN to cool a system down. @ ~ -150c it stopped working.

 

[Peter]

First thing to leave will be the storage media - HDD and removables. Then, if you watch humidity condensation issues, carefully selected PC hardware will go down to about -60 degrees celsius. "Extended temperature" industrial/aerospace/whatever systems are usually being rated for -40. Those then have solid state storage (FlashROM drives usually).

 

[chsh1ca]

Thanks for the info, I'll pass it on.

 

[Shalmanese]

I think there are some liquidy bits on consumer level motherboards that turn solid at some temperature and dont work. It might be caps or batteries or something.

 

[Peter]

Battery is first concern. That's why Extended Temperature computers don't have one.

 

[grrl]

I think there are some liquidy bits on consumer level motherboards that turn solid at some temperature and dont work. It might be caps or batteries or something.

Yeah, what about the caps?

 

[drag]

From what I understand about compaciters. There are 2 types of capaciters that are commonly used today in small electronics. I'm sure most of you know capaciters work, so I will try not to sound to stupid. One type is the cheap type, It uses a fluid for a dielectic in order acheive it's capacitive amount, these oils or chemicals or whatever tend to evaperate over time making. That's why cheaper motherboards tend get more and more unstable over time because as this stuff evaporates the capacitor losses it's capacitance potential or whatever... If this stuff gets to cold it can crystilize and the motherboard will fail.

I got this from a website were this guy litterally submerged his entire motherboard in liquid nitrogen. Liquid nitrogen is a dielectic fluid so it can't transmit electricy like water can. He was able to get his 300 celeron (could a been a 500 something It's been a while) to clock over 1ghz before the capaciters froze and destroyed themselves.

If you were to remove all those capaciters and replace them with the better solidstate stuff then you could probably get it much cooler.

BTW IF you want to try this it is quite easy to obain liquid nitrogen out in the country. Farmers use the stuff for freezing sperm for studding animals long distance. ANd it ain't too dangorous to handle just as long as you don't pore some on your shoe or something. I've stuck my hand in some before. You hand is so hot compared to the nitrogen that the air it will boil off the nitrogen befor your skin even comes in contact with it! (of course i just stuck it in and pulled it strait out! I couldn't even feel the cold, don't try it for more than a split-second!)

 

[drag]

I found it.. it was at www.octools.com, but I can't find the actual article... and they use flourite for the dielectic, I was mistaken when I said that they submerged it in nitrogen, i guess, but that's were I got the frozen capaciters thing.

 

[dejitaru]

The capacitance of a capacitor increases at low temperatures. This may not cause problems.

 

[RU482]

I used to work in Engineering for a company that made pen-tablet computers

We would run benchmarking tests while the unit was in a temperature chamber.

we tested from -30C to 60C

 

[drag]

It wasn't the capitance of the capacitors that were the problem, it was the fact that the cheap ones would destroy themselves when the oil used to make them froze....
I could be wrong, but then again we are talking about liquid nitrogen cold temps.

 

[dejitaru]

Originally posted by: drag
It wasn't the capitance of the capacitors that were the problem, it was the fact that the cheap ones would destroy themselves when the oil used to make them froze....
I could be wrong, but then again we are talking about liquid nitrogen cold temps.

Well, most electrolytics serve no "critical" function in a computer's operation. The IC and PCB mounted ones are somewhat more important (in general).


What about the processor? Would that operate at ~zero Kelvin?

 

[CTho9305]

Originally posted by: dejitaru

Originally posted by: drag
It wasn't the capitance of the capacitors that were the problem, it was the fact that the cheap ones would destroy themselves when the oil used to make them froze....
I could be wrong, but then again we are talking about liquid nitrogen cold temps.

Well, most electrolytics serve no "critical" function in a computer's operation. The IC and PCB mounted ones are somewhat more important (in general).


What about the processor? Would that operate at ~zero Kelvin?

The CPU's voltage regulator (all those caps around the socket) isn't critical?

 

[dejitaru]

The CPU's voltage regulator (all those caps around the socket) isn't critical?

Nope. That was my first consideration.
As long as you're getting a voltage, you can hack a few workarounds to have your system operating under minimum acceptable conditions again.

 

[MrDudeMan]

Originally posted by: dejitaru

The CPU's voltage regulator (all those caps around the socket) isn't critical?

Nope. That was my first consideration.
As long as you're getting a voltage, you can hack a few workarounds to have your system operating under minimum acceptable conditions again.

we have a winner

 

[Wintermute76]

I'ts not necessarily that electrolyic caps are cheaper to use, more that at the capacitance needed they perform better than an equivalent ceramic cap caramic and electroolytic cap article

Also not all electrolytic caps are the wet type, there are some dry electrolytic caps out there, not sure on their uses tho.

 

[dejitaru]

Originally posted by: MrDudeMan

Originally posted by: dejitaru

The CPU's voltage regulator (all those caps around the socket) isn't critical?

Nope. That was my first consideration.
As long as you're getting a voltage, you can hack a few workarounds to have your system operating under minimum acceptable conditions again.

we have a winner

Oh, indeed.