what is the coldest thing you can cool a cpu with?

[adarw]

ive seen people use liquid nitrogen but is that the coldest you can get?

[svmlegacy]

Liquid nitrogen is often too cold, causing some CPU's to glitch out with a "cold bug"

 

Anything at absolute zero would work, but it wouldn't be there for long.

[Chronified]

Liquid nitrogen is about as cold as you can get because modern hardware glitches out of it gets too cold, LN2 is borderline too cold, you can't just fill the pot with it, you need to trickle it in gradually and keep an eye on temps. 

As for long-term you need to stay above freezing or figure out a way to get rid of moisture. I have an Aircon pointed at my Mining rack and a dehumidifier in the room at 30% ambient. Works great for keeping the rack cool lol

[adarw]

5 minutes ago, svmlegacy said:

Liquid nitrogen is often too cold, causing some CPU's to glitch out with a "cold bug"

 

Anything at absolute zero would work, but it wouldn't be there for long.

 

2 minutes ago, Chronified said:

Liquid nitrogen is about as cold as you can get because modern hardware glitches out of it gets too cold, LN2 is borderline too cold, you can't just fill the pot with it, you need to trickle it in gradually and keep an eye on temps. 

what about dry ice?

[Chronified]

1 minute ago, adarw said:

 

what about dry ice?

Dry ice is solid making it hard to use for cooling and it's actually "warmer" than LN2. 

LN2 sitting stable in a container is about -195c / -320f 

Dry ice is only -78c / -109f

 

I've theorized that someone should make an arduino trickle system for LN2 that slowly releases a container down into the CPU block instead of pouring it manually, have it hooked up to a temp probe that automatically restricts/adds more based on current CPU temps. 

[BondiBlue]

2 minutes ago, adarw said:

what about dry ice?

That's far warmer than LN2, and it wouldn't last long. Besides, it's a solid, so it would be difficult to keep it situated properly. 

[freeagent]

I'm no pro but I think Helium is a bit colder. I know a few guys who could answer this with confidence, I'm sure they will see this soon. You could buy a phase change system, its not as cold as LN2, but it can get fairly close I think.. depending on how the gas is mixed.

[For Science!]

7 minutes ago, freeagent said:

I'm no pro but I think Helium is a bit colder. I know a few guys who could answer this with confidence, I'm sure they will see this soon. You could buy a phase change system, its not as cold as LN2, but it can get fairly close I think.. depending on how the gas is mixed.

yes, but its significantly more expensive, and for it to not get explosive you need to keep the helium cooled with nitrogen.

[StDragon]

If the CPU gets too hot, then the Leidenfrost effect takes hold and you can't get rid of the heat fast enough.

[svmlegacy]

On 9/27/2021 at 6:29 PM, For Science! said:

yes, but its significantly more expensive, and for it to not get explosive you need to keep the helium cooled with nitrogen.

Are you thinking of Hydrogen? Helium is inert, but expands a lot when it boils.

 

The coldest fluid typically used for cooling is in fact helium. Liquid oxygen systems exist too, but generally those are a fire hazard due to the oxidzing nature of oxygen.

[Guest]

Dry Ice, Liquid nitrogen, Helium are the 3 most common CPU chilling approaches.

 

However Single Stage and Cascade Units also get pretty darn chilly. Comparable to Dry Ice temps actually. But a lot less popular these days I think. 

 

TEC chilling, had my 2700X down to -30c at idle. That was fun.

Water chillers, also can get into negative temps.

 

Seen mention of running into Cold bugs. Sometimes the board will CBB cold boot bug, or a cpu CB being cold bugged. I can pretty much tell you which generations of AMD chips will cold bug on you. Boards are actually a toss up, be surprised what a bios flash on a different revision can do.....

 

Anyhow, before I go onto a full 3 page rant about sub zero temps, there's a WHOLE LOT MORE to it than just slap a pot on it. 

 

Any questions, I'd be happy to answer them. I've done all of the above cooling types with the exception of Helium. 

 

[For Science!]

5 hours ago, svmlegacy said:

Are you thinking of Hydrogen? Helium is inert, but expands a lot when it boils.

 

The coldest fluid typically used for cooling is in fact helium. Liquid oxygen systems exist too, but generally those are a fire hazard due to the oxidzing nature of oxygen.

No, explosion does not have to involve a flame/flash. an explosion by definition is a rapid expansion of gas, so yes, helium despite it being inert, can explode by virtue of massive expansion from liquid to gas. Another obvious risk is asphyxiation, not as bad as argon, but it will still be a risk worth considering.

[IPD]

Liquid helium can get you down to about 3 Kelvin, iirc.  Although I can't speak to the efficacy of a semiconductor at that temperature.  Usually that's exclusive to superconductors.