Originally posted by: AdamK47
Anyone know how to make a 24/7 LN2 cooling setup?
This might seem like a throw-away comment but I've given more thought to this question than a sane, rational individual should.
The simple answer is: yes. If you think about it, LN2 cooling is nothing more than phase change cooling, even when it's just a guy pouring some of it into a copper pot + insulation + anti-condensation grease mounted on a CPU. LN2 is going to be at its boiling point when exposed to room-temperature conditions, so most (if not all) of the heat it absorbs will go towards change-of-phase from LN2 to gaseous N2, with the gas itself absorbing very little heat post-change-of-phase due to its comparatively lower density and its tendency to drift away from the contact surface.
So, it's phase change cooling.
They already make double and triple-stage cascade phase change coolers out there that can hit temperatures as low as ~-130C (~143K). Liquid Nitrogen can exist at temperatures as high as ~-196C (~77K), so it is somewhat reasonable to conclude that with more stages, a lot more power, a lot more noise, and much bulkier equipment, that cascade phase change coolers could get down to that temperature range which would effectively be 24/7 LN2 cooling. Whether or not that would be the most efficient or effective way to go about things is unclear.
Another option would be to do what some administrators of large, hot computers have done by keeping a huge reservoir if LN2 and simply dripping it out onto hot computer surfaces (CPU, GPU, PWMs, RAM, etc) as needed. As long as the reservoir is well-constructed and there is sufficient ventilation, there should be no risk. The downside is that you'd need a huge tank of LN2 in your house, and you'd need to refill it every so often which would require handling several dewars and might or might not require hazardous material permits that average citizens can not easily obtain. Even if it were legal to purchase and use such a large quantity of LN2, it would probably raise some eyebrows in the wrong places and otherwise be a big pain in the arse. Anybody hardcore enough to want to use LN2 to cool any part or all parts of their system will probably be running a machine chewing up 1kw or more in power which would require quite a bit of LN2 for cooling, at least by the standards of a home user. The main advantage to this approach would be that industrial production of LN2 is presumably more efficient than small-scale produciton of LN2 utilizing a device mounted to your PC. At best, I believe the industrial producers operate refrigeration facilities capable of 50% carnot efficiency. You could probably not match that at home. After absorbing the cost of purchasing and rigging up your own reservoir, your own dewars, and locating a supplier willing to sell to you (plus applicable permits/licenses where required), the overall cost per litre of LN2 would be lower by just snapping it up from industrial suppliers. A nice supplier might even let you ferry the stuff to your reservoirs in borrowed dewars at no extra cost. Of course, you'd have to take care not to spill it and then succumb to the resulting gas . . .
Anyway, 24/7 LN2 cooling for any or all parts of the system requiring cooling would be very interesting and potentially worthwhile for the right power-user, but it would be eccentric and, more likely than not, outrageously expensive. Due to the sheer inefficiency of LN2 cooling (at least when compared to air and water cooling), the costs involved would shoot up very quickly as the overall system power draw increased. LN2 cooling will become much more feasible, and potentially necessary, as overall power draw on computers decreases from process improvements and other technological breakthroughs. For example, there has been a discussion recently about superconducting transistors. If they could ever get superconducting transistors to work at temperatures of 77K or higher, LN2 cooling would produce amazing opportunities. The power draw from such transistors would likely be very low, so the overall power cost of chilling processors based on such transistors might be correspondingly low, potentially even within the reach of the average home user. Of course, that's just taking "normal" operation of such a theoretical CPU into account; most of us want to use LN2 to make processors run far out-of-spec. Having off-the-shelf technology with amazing capabilities that required LN2 might be better, but it would be a lot less exciting than a heavily-overclocked Phenom II, even though cooling a Phenom II at 6ghz+ with LN2 24/7 would be a lot less practical.
edit: According to at least
this site, the US Department of Transportation classifies LN2 as a hazardous material except when shipped in a dry shipper. So, when transporting standard dewars, you'd need training and certification. Since most (if not all) dry shippers seem to be designed to transport something other than LN2 (and just use LN2 to keep the contents cold), it would likely be impractical to move large quantities of LN2 utilizing dry shippers. I guess you could get around most of that trouble by having your LN2 delivered, but then costs go up.
Someone got keyboard happy.
Facebook
Twitter