What about active proccessor cooling?

Posted:
in Future Apple Hardware edited January 2014
Specifically what is the smallest miniature passive refrigeration compressor/condenser ever made or invented? For those of you who don't understand; an active compressor has an electric motor which drives the refrigerant through its hot cold cycle on most of our home refrigerators. Now I've seen passive motorless frigdes of less than 1 cubic foot and they will make a case of pop or beer mighty cold! Are there micro sized passive refrigerstion units out there that would fit in a desktop machine? Since power consumption on a desktop machine is not important and I'm quite certain the thermo mechanical interface/active heatsink would be fairly simple, why can't you devise a G4 system running at 3 GHz? Hell the fan in my current G4 makes so much racket I can hardly stand it! A tiny passive refrigeration unit say the size of a couple ATA drives and a micro tank of refrigerant would certainly fit in the G4 tower, especially if you remove that 4"x 1.5" cooling fan in there now! If refrigerant failure occured or temps reached critical levels the chip could power down to 900 mHz or so and no damage would be done? Is this a totally insane concept?

Comments

  • Reply 1 of 9
    neutrino23neutrino23 Posts: 1,517member
    The biggest problem with any kind of active cooling is cost. You would add hundreds of dollars to the cost (at a minimum). [edit: I know of scientific detectors using active cooling which cost $10,000 or $15,000 more than the same detector without active cooling. Just an example.]



    The device you are referring to is probably a Peltier cooler. If you run an electric current through the right combination of materials then one junction absorbs some heat and another will emit heat. In effect, this is a solid state heat pump. The problem is that it is not very efficient. It is good for making something cold (like the inside of an insulated box) where not very much heat is being added. It is not very good for removing heat in large amounts. Remember, it is a heat pump. You still need a fan or something to remove heat from the hot end of the device.



    There is another solid state cooler technology on the horizon called Anti-Stokes cooling. In this you shine a powerful laser on a special crystal and it gets cold. Neat idea but not practical yet.



    There are a handful of mechanical chillers (Sterling engines, JT coolers and such). They are all much too expensive for this application as well as being finicky and not lasting very long.



    The best device to cool a CPU in a consumer product is a well designed heat sink and a fan.



    If you want to try to run a CPU faster than normal then you would need some kind of mechanical chiller pumping antifreeze of some sort over the chip to try to remove huge amounts of heat. Fun exercise but not practical for a consumer product.



    [ 03-24-2002: Message edited by: neutrino23 ]</p>
  • Reply 2 of 9
    kd5mdkkd5mdk Posts: 81member
    Apple just needs to get some PAL8045 heatsinks. Would help keep them fanless.



    Those monsters are awesome.
  • Reply 3 of 9
    Neutrino,

    The cooled area would or could definately be sealed and the the heat exchange would be direct metal to metal contact with the heat sink. Say the cooling fins or vanes of the heat exchange device were actually sealed or contained in the refrigerated space, then indeed that refrigerated cubic area occupied by the heat sink would be isolated from the outside in all ways except the thermocoupling of the metal heat sink to the refrigerated area just by the very immediate proximity (actually inside the enclosed refrigerated area) to the heat sink's total cooled surface area. The way the cooling fins protrude out into the space of the computer's enclosure a good alloy heat sink is as much a cube or box anyway! So whay not a sealed box. Definatly the temperature in the heat sink's refrigerated enclosure would not be cold under full opperation but it would be a lot cooler than a fan only blowing ambient air across standard heat sink surfaces! The enclosure might be a 2"x 3" cube. Also a a small & quieter 2" fan could be used to pull the cooled air through the enclsure. I'm sure there are any number of ways to interface the two.

    I'm simply asking if there is even such a miniature passive refrigeration compressor that even exists today right here and now? Mind you I'm not after super cooling the CPU but just giving it a fairly good boost of cooler weather to live at higher than normal clock speeds. If I can buy a tiny refrigerator for under a hundred bucks then it is no great leap to suggest that that same item can't be shrunk for a hundred or two more! And devising an efficient heat exchange of the size I mention above is not rocket science either. I concure that proprietary measuring devices you mention are extremely expensive and that is due to their limited market of specialize users. Superconductivity and the other things you mentioned aside I just tend to think that logically applied "low tech" refrigeration could give unheard of speed results for a given CPU architecture.



    It's simple really: Make resonably small amounts of cold air "or other cooling medium" and pass it over a redesigned heatsink. I'd pay $400-$500 for a Gigahertz or two on my current chip!!
  • Reply 4 of 9
    airslufairsluf Posts: 1,861member
  • Reply 5 of 9
    eskimoeskimo Posts: 474member
    I think what you are talking about is similar to what Kryotech does/did with it's line of SuperG computers. They were "thermally accelerated" to over 1GHz before normal PC chips reached that level by using a compressor to cool the processors to approximately 40 degrees below zero.

    Picture:





    They have several articles one of which is <a href="http://www6.tomshardware.com/cpu/99q4/991115/index.html"; target="_blank">here</a>



    A review by Anand on the 1.86GHz G2 system (sucessor to one in Tom's review) is <a href="http://www.anandtech.com/cpu/showdoc.html?i=1515&p=1"; target="_blank">here</a>



    They are no longer manufactured as far as I know. I believe that real clockspeed advances along with extraordinary development costs negated their ability to sell many units.



    [ 03-26-2002: Message edited by: Eskimo ]</p>
  • Reply 6 of 9
    xhorxhixhorxhi Posts: 46member
    Eskimo,

    You have a chilly name and you really dig super cool stuff (pun intended) but thats beside the point. You rightly point out that "real" clock speed advances negat the need for such cryogenic tempuratures as 40 below, however that is exactly why "high-end" High cost super cooling is not needed. I'm just saying to use low tech low cost cooling hardware, that lowers the CPU temp by a nominally small percentage to give it a fighting chance at say a 50-70% increase in clock speed over and above the normal range for that proccessor. YES it would get hot but with some much cooler than ambient temperature air running across the heat sink you maybe could keep the unit within it operating range. 40 below temps generated by extremely exotic and expensive cryogenic compressors is obviously not what I had in mind. Maybe I am oversimplifying the electrical and mechanical factors involved here but I know that an extremely inexpensive tiny freezer can make enough coold air to make ice in an hour!
  • Reply 7 of 9
    eskimoeskimo Posts: 474member
    Xhorxhi,

    I also produce semiconductors for a living and I can tell you that the cooling necessary to achieve the type of overclock you speak of on a processor not already capable of doing so is enormous. Forgive me if I lose you in this explanation.



    In order to increase the current drive necessary to switch the logic gates (CMOS transistors) faster you need a higher voltage. As you increase the voltage increased hot carrier effects which will decorate your oxide as well as increase the charge region around the source/drains until short channel effects degrade your transistor performance. At this point your transistors will fail to function properly at all and given too much voltage you will break down your gate oxide causing irreversible damage rendering your processor useless. Of course your processor will error out and crash before this happens but it is possible.



    All this increased voltage drives up power consumption exponentially. Increased power requires increased power dissipation. Such to the point where refrigerant/liquid cooling is required. Simply cooling the ambient air will be insufficient as the chip will heat itself faster than you will be able to dissipate the concentrated heat into the ambient. So you are either going to cool by convential means or by extreme means. There isn't much middle room.
  • Reply 8 of 9
    xhorxhixhorxhi Posts: 46member
    Well now see there you go and get all technical on me!! PPffft!

    But I do understand your explaination of the situation. It appears that at the processor's deepest level of material makeup that the the thermo-conductive ability, or inability I should say is insufficient to disipate heat to the outside no matter how cool the air or heat sink may be! Right? Kind of like when a nuclear reactor reaches critical mass albeit much less disasterous!!



    [ 03-26-2002: Message edited by: Xhorxhi ]</p>
  • Reply 9 of 9
    Regardless of what method you use the amount of energy disipated from the cpu + the amount of energy required to run the cooling system (be it piezoelectric or compressor based) will need to be dumped to atmosphere as heat. A Ti laptop is already pretty warm from what I understand. Cooling the chip further would actually make the case even warmer. So unless you want to carry around a cooling tower no additional cooling is possible on a laptop over what is currently done.



    The same applies to a refridgerator. If you open the fridge in your house, the room it is in will get hotter, not colder.



    The only way this will help is if you put a big fan blowing a ton of air in the system or better yet hook it up to a tap and run water through the system. Regardless you are creating more heat than you are removing, you are just relocating the energy.
Sign In or Register to comment.