How cold is too cold?

[nightfly]

Way back when, a friend and I considered building a computer inside a tiny refridgerator, and just running all the wires through a grommet in the side of the box. However, it was too small, and non forced air, so it turned out not to be as good an idea as we thought. But it brings to mind the possibilities. With a small forced air fridge that automatically dehumidifies the air to keep the problems of crystalization of any humidity out of the equation, what might be the problems of building everything and sitting the MB on a platform in the freezer? What components would be likely to stop working right (with of course a SSD instead of a platter drive)? Of course, as it's in a freezer, I'm assuming all air cooled devices instead of any liquid ones.

 

[cortexodus]

This is a pretty common pop-up topic for machines. Unfortunately, the math behind the thermals just doesn't make it feasible.

 

[nightfly]

Those (chest style deep freezers) are like the little refrigerator thing. The don't have active dehumidifying, or the capacity to cool a 'hot' PC with all it's heat generating parts. Forced air freezers do (commercial freezers where people are constantly opening and closing the door, or the style where you see say, in supermarkets that keep 'open' displays of food ice cold 24/7); there's no ice in my freezer, anywhere for the last what, 15 years, even the ice cubes eventually disappear from sublimation if not taken out and used. But I get the point about having a high performance computer, using up 1000 watts of energy that needs to be dispersed, in the average home freezer.

Ahh, but I'm getting away from my question. I was simply wondering if at some temperature point, the chips on the circuit boards stop working, decrease in performance, or degrade in some way to damage them.
But thanks for the links, it was interesting.

Oh. The 'hair dryer' concept they mention doesn't seem to be valid. A hair dryer is designed to convert as much energy into heat as possible, PC's aren't. We can heat up a room pretty quick with 1500 watt space heaters that turn themselves off when the desired temperature is reached, where as a 1500 watt set of energy consuming PC's or other electronics such as a whole lot of old style CRT TV's using up even more energy (we tried that one in an appliance store during one very cold winter about 40 years ago and gradually froze our asses off) simply won't do the trick, so there's something else going on there.

 

[dragonstongue]

takes a lot of cooling to accomplish low temperature operation vs room or even warm liquid cooling, its just not worth it in so many ways. IT is better to have a constant more or less operating temperature then it is to try and get it below its normal operating range as they are electronics and moving parts with strict tolerances in some cases, way less likely to cause problems keeping them a constant warm then a constant cold, even if the cooler opens up maybe a slight overclocking headroom.

HDD do not at all like cold, they prefer room to warm 30-40c, power supplies like same, cpu prefer cooler, gpu cooler but being hot to being cold is VERY hard on components and you could very well shorten the lifespan if not done very carefully. Just not worth it as many folks have said, though it can be done to a very small point to help wick heat away, to do it primarily is a massive cost for very little gain.

Just as example, HP, IBM and a few others have started doing warm liquid cooling as in temps 40-55c this reduces overall temperature fluctuations drastically, carbon footprint a great deal, and specific cooling costs are magnitudes less. The standard cooling methods being air and/or liquid without trying to force it in abnormal means to below ambient are just the safer better approach IMHO

some chips do not like "cold" though it will be hard to pin down which, I am not good at math but less heat means electrons flow easier which should mean you need less power for the same result, but maybe because of this electro-migration happens sooner hard to say. As far as caps/vregs and such go again they are an electrical component so am sure for the most part as long as the cold is not abrupt it shouldn't be a major lifespan factor as long as they don't get moisture baked into them. I know if chips get to hot it will kill them so I would assume in some cases then yes to cold would slow them down or possibly damage them, not sure, not a chip designer by any means.

 

[NamesLucky]

Ahh, but I'm getting away from my question. I was simply wondering if at some temperature point, the chips on the circuit boards stop working, decrease in performance, or degrade in some way to damage them.
But thanks for the links, it was interesting.

My thoughts on this are that competitive overclockers use liquid nitrogen all the time, and generally over many parts of the board/CPU/GPU. I think if a bad temp "point" did exist they would find it doing that. For 24/7 use it is not super uncommon for some overclockers to attach AC's to their "water" loops, giving -30C idle temps. I have personally ran a computer at this state (not always on though) for over 9 months, and others for over many years.

 

[1337m0dd3r]

I would try posting this in the overclocking and cooling thread in the sub-forum about extreme cooling you'll prolly get more response from there.

 

[Spazturtle]

Watercool and put the radiator and pump into the cooler would be better.

 

[dragonstongue]

for this -30 you refrence phase change no? as I doubt most standard AC would let things get that cold.

 

[lilfiend]

Way back when, a friend and I considered building a computer inside a tiny refridgerator, and just running all the wires through a grommet in the side of the box. However, it was too small, and non forced air, so it turned out not to be as good an idea as we thought. But it brings to mind the possibilities. With a small forced air fridge that automatically dehumidifies the air to keep the problems of crystalization of any humidity out of the equation, what might be the problems of building everything and sitting the MB on a platform in the freezer? What components would be likely to stop working right (with of course a SSD instead of a platter drive)? Of course, as it's in a freezer, I'm assuming all air cooled devices instead of any liquid ones.

I'd recommend reading the phase change section of EOCF if you are actually interested in this type of extreme cooling. The simplest way to do it would be to setup a water cooling loop, but instead of a radiator just have a big cooler chest with the cold side of an AC unit or dehumidifier in it.

If you really get into it then you want 2 or 3 stage cascades with different coolants, all in all its not worth it but damn is it cool

 

[NamesLucky]

for this -30 you refrence phase change no? as I doubt most standard AC would let things get that cold.

5000 btu ac, like a small window ac has capacity for -30 temps, just need to trick the thermostat into thinking its hot, i usually just attach it to the compressor lol. Loaded temps are different though, And depend on oc.