Sandia-designed CPU cooler lacks fan, rotates heatsink instead

CPU coolers aren’t rocket science, but the Sandia National Laboratories has produced one anyway. As one might expect, it’s rather special. The patent-pending "Air Bearing Heat Exchanger" promises to substantially reduce the amount of energy required to cool microprocessors while simultaneously cutting noise levels. Fans are the only source of energy consumption for common air coolers, and they happen to be the sole generators of noise. Sandia’s cooler doesn’t have one. Instead, it rotates the heatsink itself. Here’s a diagram pulled from the official press release:

According to Sandia, spinning the specially designed heatsink at "a few thousand RPM" generates a centrifugal pumping effect that shrinks the boundary layer between the cooling fins and the air by a factor of 10. A lengthy and equation-filled whitepaper (PDF) on the technology describes this layer of dead air as the "primary physical limitation to performance" of traditional air coolers. That same whitepaper talks up the noise-reducing potential of this new approach to cooling, and it points out that the spinning fins won’t accumulate dust. Sweet!

Although Sandia’s press release focuses more on potential applications for data centers, the technology is said to be suitable for laptops and PCs. In fact, high-performance gaming rigs are mentioned specifically. It’s even more encouraging to see the design described as "simple, rugged, and cost-competitive."

From what I understand, the so-called "Sandia Cooler" should have a much lower profile than traditional heatsink/fan combos that offer similar performance and noise levels. If that’s indeed the case, the world of PC cooling could be on the verge of an interesting transition. Sandia is currently looking to license the technology to partners who will bring products to market.

Comments closed
    • etana
    • 8 years ago

    Now they just need to power it with the heat from the CPU.

    • My Johnson
    • 8 years ago

    Read the PDF before making an inane comment about the concept, People:

    [quote<] During operation, these two flat surfaces are a separated by a thin (~0.03 mm) air gap, much like the bottom surface of an air hockey puck and the top surface of an air hockey table. This air gap is a hydrodynamic gas bearing, analogous to those used to support the read/write head of computer disk drive (but with many orders of magnitude looser mechanical tolerances). Heat flows from the stationary aluminum base plate to the rotating heat-sink-impeller through this 0.03-mm-thick circular disk of air. As shown later in Figure 18, this air-filled thermal interface has very low thermal resistance and is in no way a limiting factor to device performance; [b<]its cross sectional area is large relative to its thickness, and because the air that occupies the gap region is violently sheared between the lower surface (stationary) and the upper surface (rotating at several thousand rpm). The convective mixing provided by this shearing effect provides a several-fold increase in thermal conductivity of the air in the gap region.[/b<] One important point about the air bearing is that the ~0.03 mm air gap is not maintained by using extremely tight mechanical tolerances. Much like an air hockey puck on an air hockey table, or a hard disk read/write head, the air gap distance is self-regulating. If the air gap distance increases, the air pressure in the gap region drops, which causes the air gap distance to decrease. This built in negative feedback provides excellent mechanical stability and an extremely stiff effective spring constant (important for ruggedness). Unlike an air hockey table, which relies on gravity to counter-balance the pressure force acting on the puck, the air-bearing cooler can be mounted in an arbitrary orientation (e.g., up-side-down, sideways, etc.). And unlike a computer disk drive, incidental mechanical contact between the two air bearing surfaces does not damage either surface[/quote<] [quote<]Having said that, in small diameter high-speed devices, such as those that might be used for direct cooling of a CPU IC package, this thermal conductivity enhancement factor may be considerably larger. This effect is potentially very important because the reduction in air gap area associated with small-diameter devices will itself raise the air gap thermal resistance considerably. [b<]It should also be noted that in future devices the incorporation of a hydrodynamic bearing, which typically entails the addition of air gap surface features such as grooves, will also likely enhance convective heat transfer in the air gap region. Lastly, in some cases it may also be worthwhile to incorporate additional surface features, texture, etc. for the sole purpose of further enhancing convective heat transfer in the air gap region.[/b<][/quote<] Emphasis mine.

      • just brew it!
      • 8 years ago

      I’d still be concerned that the mechanical tolerances required to make this work (both the bearing surfaces and rotor balance) are going to be problematic. The air gap is only 1/1000th of an inch. So the surfaces are likely going to need to be machined to tolerances significantly better than that.

        • MrDigi
        • 8 years ago

        Maybe it could double as a HD storage device.

      • Peldor
      • 8 years ago

      Does the PDF have some corrupted text for everyone or just me? Installed font issue maybe?

      Page 23 for example:

      [quote<]The objectives of the heat transfer measurements were: , * , *                                                                   "     % -*                                                             $                 ,2% . *                                                          % / *                                                 "                              % 0 *                 "                )  $ &* [/quote<]

    • Bensam123
    • 8 years ago

    I’m rather curious as to what it uses inbetween the rotating metal heatsink and the actual processor. Unless the processor socket itself rotates with the processor there has to be some sort of interface between the two. As such whatever interfaces the two has to be fucking magic or this is a terrible idea.

    The best I can think of is a heatpipe structure or perhaps liquid with a coupler of sorts that can spin, sort of like a union.

    Either way this doesn’t sound like a great idea… especially considering that it has to use more energy spinning a large piece of copper over plastic fins.

      • just brew it!
      • 8 years ago

      Looks like the interface is a microscopically thin air gap. Make the gap thin enough, and the thermal resistance will be small.

      This is the part that makes me skeptical of the whole concept. Maintaining a gap small enough to still allow effective heat transfer will require extremely tight manufacturing tolerances.

        • Bensam123
        • 8 years ago

        Interesting…

        Hard to imagine whatever fixes the fins to the base, which obviously will need to rotate, will be capable of maintaining that airspace and margin of error (like ball bearings).

          • zzatz
          • 8 years ago

          I don’t understand your question. The fins aren’t attached to the base.

          There are two pieces separated by an air gap. The base is one piece, and attached only to the CPU. The impeller is another piece, with a flat bottom and fins above, made from a single piece of metal. The impeller spins, creating an air cushion bearing between the base and the impeller. Due to the large area, close spacing, and turbulence created by shear forces between the stationary base and rotating impeller, there is very good heat transfer across the air gap.

          Fluid bearings are common. Hard drives use them. Some turbochargers use foil bearings, which work the same way. When stopped, the parts touch. When spinning, the rotating part drags fluid – in this case air, in others cases liquid – into the gap, and the fluid maintains the spacing. What maintains the spacing in air hockey? In that case, it’s a pump blowing air through the holes in the table. If you could make the pucks spin fast enough, they could create their own air cushion and wouldn’t need a separate pump.

      • Saribro
      • 8 years ago

      Look at the picture: “Stationary base plate”. Just like any heatsink base mounts to a processor now.
      The fins spin above that plate on a very narrow airlayer, kind of like the heads of a hard drive hovering the plates on a tiny airbubble.

    • dukerjames
    • 8 years ago

    I got a better idea, a turn-table style stand for ATX cases!
    Just place your PC on the stand, your whole PC will be spinning at few K rpms.

    Pros: Easy configuration, compatible with all existing hardware.

    Cons: Wires??!!? It’s best if your PC is powered by a big battery.

      • yogibbear
      • 8 years ago

      We could introduce an electric coupling arrangement…. totally not going to end in tears and many darwin awards.

    • Kent_dieGo
    • 8 years ago

    It’s a neat idea but not practical due to the many cost, longevity and safety issues already brought up. It offers nothing that existing proven products do not already provide.

      • shank15217
      • 8 years ago

      Gee, how did you figure?

      • just brew it!
      • 8 years ago

      Safety?

      What, you’re expecting that thing to fly off and cut your face or something?

    • SpotTheCat
    • 8 years ago

    The machining work done to make this work was not small. Normal heatsinks are slapped together and lapped flat. This thing must have tight tolerances all around. They can’t just lap one side flat and expect it to work, it has to be balanced and fit just right in all aspects of its enclosure and maintain both static and dynamic balance. Impressive that they got it to work, now can they figure out how to do it cost-competitively to heat pipes on a large scale?

      • just brew it!
      • 8 years ago

      Yup, that’s the main reason I doubt whether this will ever be competitive. Even if it is technically feasible, the tight tolerances needed to manufacture it will likely prevent it from ever being cost-effective, except possibly for applications where space is at a premium (e.g. high-end laptops).

        • zzatz
        • 8 years ago

        OK, the current approach requires two flat surfaces: the spreader on the CPU package, and the base of the heat sink. The new approach requires three flat surfaces; both sides of the base, and the bottom of the impeller. Milling flat surfaces is pretty cheap.

        The prototype impeller has all surfaces machined, because that’s cheaper when you’re only making one. Production units could be die cast. That’s more expensive than extruding, but not by much. Plenty of cheap objects are die cast.

    • flip-mode
    • 8 years ago

    It amazes me I’ve had the Ninja Rev B for so many years – since my socket 754 build in 2004 or 2005. Amazing heatsink.

    • SonicSilicon
    • 8 years ago

    Can this work vertically?

      • Peldor
      • 8 years ago

      They explicitly say it will in the paper.

    • Coulda
    • 8 years ago

    Isn’t rotating hot metal what creates magnetic field?

      • Meadows
      • 8 years ago

      A magnetic field is the movement of electrons, not the movement of metal.

        • Meadows
        • 8 years ago

        Really now, why thumb down facts?
        Edit: oh I get it, you rascal. Someone thumbed down everyone here, including UberGerbil, who’s never wrong. I saved him from the terrible fate of [i<]minusness[/i<], though.

      • yogibbear
      • 8 years ago

      Potentially yes, though it’d be pretty weak.

        • yogibbear
        • 8 years ago

        Okay you guys give me negative votes? If one part of the same piece of metal is hotter than another part (i.e. there’s heat transfer going on) then basically you’ve got electrons vibrating away. Now you go and spin that object. You get a weak magnetic field.

          • just brew it!
          • 8 years ago

          I thought that in order for the dynamo effect to work the metal had to be molten, and undergoing convective flow. I could be wrong.

          Regardless, the magnetic field will be so weak that you’d need very sophisticated instruments to measure it.

            • yogibbear
            • 8 years ago

            Yep, you’re correct. All I was irked about was that my “potentially yes, but pretty weak” statement copped negatives when it is more correct than simply saying no.

      • UberGerbil
      • 8 years ago

      In general heat disrupts magnetic fields (in permanent magnets), it doesn’t create or strengthen them.

        • ClickClick5
        • 8 years ago

        So wise. (OoO)

    • Arclight
    • 8 years ago

    Idk, a hunk of metal spinning at 6k? I doubt anything that is shapped like a fan and spins at 6k does’nt create a high pitched noise. Imo it’s safer to go water cooling if you hate tall heatsinks, instead of this contraption. Untill i see a video with this thing in action i won’t believe the hype, sorry.

      • Meadows
      • 8 years ago

      It’s not shaped like a fan.

        • Arclight
        • 8 years ago

        Maybe not the fans you know. Have you seen turbine inspired fans?

        If this the rotating part is not shaped like the fan how can it move air efficiently to dissipate the heat? Do you think this will work in vacuum chamber? If not then it’s just an air cooler and has to move the air just like any other fan. Even if for the sake of conversation it’s not designed to move air, it will still create noise at that speed. I wanna hear it in action, don’t you?

        Come to think of it, it looks just like the fans they use to cool industrial pump motors.

          • Meadows
          • 8 years ago

          I’ve seen blowers, yes, in fact I think I had a 7600 GT that used one. This isn’t [i<]quite[/i<] like them, however. In fact, just for the sake of authenticity, let's use Sandia's own words and call it an "impeller structure".

    • just brew it!
    • 8 years ago

    Very interesting physics at work here, but like others who have already posted I have doubts regarding whether it will be practical in real-world scenarios. Heatpipe coolers have gotten quite affordable, and perform well enough from a cooling and noise standpoint. (As an aside, I just replaced the stock HSFs on a couple of my systems with CoolerMaster Hyper TX3 heatpipe coolers. I’m quite happy with the performance, especially given that they only cost $25 — dropped my CPU temps by around 10 degrees C, and they are noticeably quieter than the stock HSFs.)

      • flip-mode
      • 8 years ago

      The TX3 is excellent for stock or mildly overclocked cooling. I put one on the i5-2500K at work and it keeps it running cool and essentially silent ( of course, the office environment has a much higher noise floor than my basement ).

        • just brew it!
        • 8 years ago

        My only concern with the TX3 is the MTBF of the sleeve bearing fan. I plan to swap the fans out for some Scythe FDB ones at some point.

      • Thrashdog
      • 8 years ago

      I’ve got an old Thermalright XP-120 heatpipe cooler, and while it’s not quite as effective as some of the newer tower-style designs I imagine that it’ll cool any CPU that they’ll ever make — as long as they keep making adapter brackets for it.

        • just brew it!
        • 8 years ago

        I’ve also got some monster solid copper Thermalright heatsinks from the Socket A era. If I could get brackets to mount them on Socket AM2/AM3 I’d probably still be using them. As it is, they’ll probably just gather dust unless I get the urge to resurrect my old dually Athlon MP rig for old times’ sake.

    • mutarasector
    • 8 years ago

    Interesting concept, but one with several problems to overcome. In addition to the ones already mentioned by others here, the two that leap to the front of my mind is 1> orientation seems to be rather limited to horizontal plane usage, at least initially, and 2> and results in limited efficiency due to eddy current air turbulances within increasingly small and tightly packed enclosures. It seems it requires even more form factor compromises to accommodate the required air channel flow/porting to make it work.

    • pragma
    • 8 years ago

    I’m impressed! Most interesting… and scary. That thing looks pretty good but we shall see.

    The future in PC coolers: standard version 7200 rpm, enthusiast niche 10000 rpm, and 4200 rpm for cool and quiet fans.

    The future of $20 walmart fan heaters: selectable 400/800/1600 W for single/dual/quad-CPU operation, but the darn thing WON’T RUN WITHOUT INTERNET (whatever distributed work units it wants). They’ll come with stickers like “Eliminating Cancer!”, or “Helping the World Climate!”, or maybe a simple gray box without stickers – that’s the NSA model. Oh yeah, and the Google heater will come with integrated webcam^H^H^H safety device, that one mustn’t cover, and the sticker reads “Protecting the Children!”

    • CuttinHobo
    • 8 years ago

    So… a spinning heatsink won’t collect dust, but whenever I see the blades of a fan – either a computer fan or a regular house fan – the leading edge of its blades are home to a layer of dust.

      • MrDigi
      • 8 years ago

      don’t question the whitepaper, it can’t be wrong ; ]

      BTW I find fan blade dust doesn’t come off nearly as easy as vacuuming the dust from fins.

    • MrDigi
    • 8 years ago

    Stupid idea, good heat sinks require low thermal resistance to the heat source which is why heat sinks require thermal grease at the interface and mechanical pressure. A rotating heat sink makes no direct contact. Dust will collect the same as on rotating fan blades. What a bunch of BS.

      • sluggo
      • 8 years ago

      Umm … they built one, it works.

      The thermal resistance across the air gap is quite small, less than 1/3 of the overall device resistance.

      Dust does not collect due to the extremely thin boundary layer.

      Reading is fun.

      • tHeNeXuS
      • 8 years ago

      If you just took enough time to read the attached PDF, you would know that the cooler is made of two parts: a rotating part (the one with the fins) and a static one.

      The static part is in direct contact with the surface to be cooled. Heat is transferred from the static part to the rotating one via the the thin air slice in between the two.

        • MrDigi
        • 8 years ago

        The paper misrepresents the thermal barrier issue as a heat removal problem where it is a heat generation problem. The assertion this is a solution to the thermal barrier is a joke. Next stop Popular Science.

          • sluggo
          • 8 years ago

          So you’re saying Sandia Labs put an engineer and a technician on this project for (I’m guessing) six months, built prototypes, completely documented their experimental approach, did all the math, issued a peer reviewed paper, applied for all the patents, and then faked all the data?

          Have you ever taken a science class?

          They completely characterized the thermal resistance of the air gap over a range of fly heights and angular velocities with both H2 and N2. It’s in the pdf. Read, ffs.

            • MrDigi
            • 8 years ago

            Presumably the author, PHD, performed the project alone, no other credits given.
            I don’t see any reference to this being published in a peer reviewed publication.
            It is not unheard of missing details in seemingly detailed measurements (cold fusion comes to mind).
            Hyping results isn’t unheard of either, like comparing it to a low performance heat sink, claiming “typical of those used in many desktop computers”. This type heat sink was used years ago and maybe recent on very low wattage CPUs. The prototype power consumption figures (up to 20W) also appear much higher than the heat sink they compared it to (3.6W). BTW they don’t reference the source of the heat sinks listed (appears to have been copied from dansdata.com).
            No details are given on how thermal conduction from the back side and mounting of the base plate is accounted for, a common source of errors in thermal resistance measurements. Maybe heat is being carried by the seperately supplied air bearing gas. The real thermal image appears to me to show a greater temperature difference than supported by air gap thermal resistance claimed, though no scales are given. Along with no details on heat input used or actual temperature values, strange given some of the mechanical details given.

            • PrecambrianRabbit
            • 8 years ago

            Clearly you looked at the paper and know something about what you’re talking about. You’re still an asshole.

            • sluggo
            • 8 years ago

            You want to believe they pay a PhD to precision machine aluminum, hand-wind motor armatures and do all the other repetitive data generation, then fine. Whatever.

            I didn’t say it was published in a peer reviewed publication, I said it was a peer reviewed document – do you see the difference? Everything that leaves Sandia, Los Alamos, Livermore, Oak Ridge, Battelle .. all of them, the papers are reviewed internally before they’re approved for publishing.

            If you have an issue with the math or the physics or the experimental procedure, how about you say what it is? Just saying “I don’t believe it because it doesn’t sound right to me” certainly give you lots of room to quibble, but in a technical forum you come off like an idiot.

            The power consumption doesn’t appear much higher, it is much higher. He says so, tells you why, and tells you by exactly how much. What’s your issue here? He’s not designing it to use less power, he’s designing it to remove heat quickly and quietly. Do you understand that those are different goals?

            He doesn’t cite sources for the performance data of the other heatsinks because he’s using commonly published data. Do you have an issue with the actual numbers or are you just looking for something to pick at?

            The heat transfer from the mounting plate to the impeller is fully measured and characterized. What he didn’t do (and he explains why) is mathematically model that transfer. Do you have an issue with that? If so, I’d love to hear what it is. If not, just look at the pages that deal with the thermal path and his measurements. The temperatures of the gases are normalized before they are introduced to the prototype – this is in the text if you can be bothered to read it.

            The heat input is generated by resistive loads and actual temperatures are clearly stated wherever a measurement is referenced. You obviously haven’t read the experimental procedure or you don’t understand it or you’re just trying to bluff your way through a post. In any case, why bother?

          • zzatz
          • 8 years ago

          The paper misrepresents nothing, but you do. It’s fine if you don’t understand how this works, but please don’t call people liars because of your own failure to understand.

          There is a large area with a small air gap, filled with highly turbulent air. That makes the thermal resistance between the base and the impeller much lower than between the impeller and the air flowing through it. The same situation applies with a conventional heat sink; the thermal grease isn’t as good a heat conductor as the IC package or the heat sink, but it’s better than the interface between the fins and the air.

          Think of a garden hose and spray nozzle. A half inch hose will have more resistance than a three quarter inch hose, but neither has much effect on the rate of flow of water. It’s the partially closed nozzle that controls the flow. Thermal grease has some effect on heat flow, but much less than the surface area of the fins and the volume of air blowing over them. This new cooler may have slightly higher resistance than thermal grease, but trades that for lower resistance between the impeller fins and the air. In other words, a half inch hose with the nozzle open more will flow more water than a larger hose and a partially closed nozzle.

      • My Johnson
      • 8 years ago

      There’s an air shear affect that assists in heat transfer.

    • kamikaziechameleon
    • 8 years ago

    cool idea.

    • Mr Bill
    • 8 years ago

    Sort of like blowing a high pressure hose at the base plate. Have not read the white paper yet but I wonder if there is a cooling limit caused by compression heating of the air by the fan and friction as the air flows across the base plate?

    • blorbic5
    • 8 years ago

    TLDR version: The air around current heat sinks doesn’t move enough and ends up acting as an insulator that keeps air from taking heat away from the heatsink. This design has the fan blades so close to the heat sink that there is very little stationary air; this allows the air to better remove heat from the heat sink.

    It sounds like an interesting idea that has more applications then just computers.

      • My Johnson
      • 8 years ago

      Yes, They highly recommend this for air conditioners, heat pumps and refrigeration equipment:

      [quote<] As mentioned earlier, air conditioner power consumption is dominated the compressor, and the problem of low fan efficiency (typically 25% in a 3-ton residential air conditioner unit) is less of a driver than in electronics cooling 15 where fan efficiencies are often just a few percent. While the prospect of an additional 2 to 4% increase in air conditioner efficiency that can be realized by reducing fan losses is certainly welcome [Parker, 2005], this effect will be swamped by direct reductions in heat exchanger thermal resistance (i.e., the boundary layer thinning effect), the indirect effect of drastically reduced fan noise on COP, and the elimination of heat exchanger fouling.[/quote<]

    • smilingcrow
    • 8 years ago

    The guys and gals over at SilentPCReview will have soiled themselves with excitement by now but in such a way that it will be inaudible except to bats.

      • xii
      • 8 years ago

      I smell a PETA action coming up…

    • sluggo
    • 8 years ago

    Holding that .001″ gap between the spinning carousel of death and the stationary baseplate at a consumer price point is not going to be easy, methinks.

    But now that I think of it, people are willing to spend as much or more on heatsinks as they are on HDD’s, so maybe a piece of precision rotating gear can be done at the price.

      • blorbic5
      • 8 years ago

      In the PDF they likened it to an air hockey table. Where it keeps the tolerances by using air pressure and not by mechanical means.

        • sluggo
        • 8 years ago

        It looks like they found a gap height of .001″ requires 6-7k rpm in order to achieve a thermal resistance of the claimed .2 C/W. However they achieve and maintain that fly height, I still think it’s going to be difficult (read: expensive) to manufacture an aluminum impeller with .5″ fins that operates at those speeds. Surface finish, motor concentricity even variations in material density become factors when trying to balance such a massive spinner.

        The prototype in the article was machined out of a block of aluminum and then black anodized. This can’t be done for a consumer part.

        Don’t get me wrong, I love the idea. It’s absolutely a breakthrough for applications that require state-of-the-art air cooling and I already like the commercial air-conditioning application mentioned in the pdf. I’m just not sure we’ll ever see it in a PC due to cost issues.

          • blorbic5
          • 8 years ago

          Manufacturing would be higher but I wonder how much can be saved from not having a behemoth of a heat sink with copper heat pipes. Also the brushless motor and speed control would add cost too. I would think it would start out at around $150-200.

          They also claim that due to time and money constraints they haven’t optimized the design at all.

      • Thrashdog
      • 8 years ago

      It might be possible to widen the gap, stabilize it, and still maintain decent thermal conductivity by replacing the air gap with a fluid film. Light oil has got to be a better thermal conductor than air.

        • My Johnson
        • 8 years ago

        From the PDF:

        [quote<]Having said that, in small diameter high-speed devices, such as those that might be used for direct cooling of a CPU IC package, this thermal conductivity enhancement factor may be considerably larger. This effect is potentially very important because the reduction in air gap area associated with small-diameter devices will itself raise the air gap thermal resistance considerably. It should also be noted that in future devices the incorporation of a hydrodynamic bearing, which typically entails the addition of air gap surface features such as grooves, will also likely enhance convective heat transfer in the air gap region. Lastly, in some cases it may also be worthwhile to incorporate additional surface features, texture, etc. for the sole purpose of further enhancing convective heat transfer in the air gap region.[/quote<] Edit: Formatting FAIL!

    • willyolio
    • 8 years ago

    I don’t see how spinning a heatsink like a fan would make it any quieter or energy-efficient than a heatsink with a fan. then again, i’m not into fluid dynamics or thermodynamics, so if it works… it works.

    • Peldor
    • 8 years ago

    It certainly sounds like a beast at 0.2°C/W (with 0.1 likely in a 2nd generation design), the main drawback at this point is that the prototype’s 10cm flat baseplate would be a socket-clearance boondoggle.

    Other than that, woohoo!

    • bdwilcox
    • 8 years ago

    bdwilcox-designed CPU cooler lacks heatsink, rotates PC instead

      • UberGerbil
      • 8 years ago

      Why not move to one of the poles, and have the earth rotate around your PC? Only two per planet, but they’ll be remarkably cool.

        • bdwilcox
        • 8 years ago

        Oh, now that’s just crazy talk!

          • HiggsBoson
          • 8 years ago

          I’m reminded of a ST: TNG episode, where a massive asteroid is threatening to impact a planet and kill the population. The solution as espoused by a depowered-Q is to change the gravitational constant of the universe in order to make the asteroid easier to move.

    • ew
    • 8 years ago

    What happens when dust gets into the air gap?

      • calvindog717
      • 8 years ago

      .001″ is pretty tight, and since air will probably be flowing from the center out, dust will be blown away from the air gap, and not get stuck inside. I haven’t read the papers, but I think that’s what would happen

    • BoBzeBuilder
    • 8 years ago

    In Soviet Russia….

      • dpaus
      • 8 years ago

      This was invented many years ago in glorious Worker’s Paradise by famous-but-mysterious Professor Reguspatoff! (who often stamped his inventions “REG U.S. PAT OFF” on the bottom of them)

    • albundy
    • 8 years ago

    doesnt grinding metal heatsink on metal cpu produce even more heat?

      • Jon
      • 8 years ago

      I’d like a tl;dr on the whitepaper. But my logic reflects yours.

      • UberGerbil
      • 8 years ago

      You missed the call-out for “stationary baseplate” in the diagram? That’s the interface where things get interesting, because you need it to be low-friction but high thermal transmissivity (and also sealed, to keep contaminants out). People have asked a lot of thermal paste in the past, but not that it also be a lubricant. I haven’t had a chance to look at the paper yet, but there’s an interesting engineering problem there and I look forward to seeing how they solved it. The diagram refers to a .001″ air gap, which implies they’re “flying” the spinning parts much like heads are flown over HDs, and of course that’s the “air bearing” in the name, but that doesn’t sound particularly efficient from a thermal standpoint.

      There’s also a potential issue with the motor in the center, since it’s sitting right over the hottest spot on the heatsink and yet has the least facility to get rid of the heat. You could build the motor vertically and move it away from the base plate, but then it potentially impairs airflow (though maybe you could give it a teardrop aerodynamic housing, which also would look kind of cool).

        • Firestarter
        • 8 years ago

        [quote<]People have asked a lot of thermal paste in the past, but not that it also be a lubricant.[/quote<] Many lubricants are involved in cooling though, like engine oil which is vital to engine cooling.

      • Spotpuff
      • 8 years ago

      Look at the diagram. Stationary base plate, air gap, then rotating heatsink part.

      The part touching the CPU doesn’t look like it rotates.

        • Duck
        • 8 years ago

        Yep. Pretty clear I thought.

          • Chrispy_
          • 8 years ago

          You are witnessing the rise of the TL:DR (de)generation.

          They are too damn lazy to learn anthing because they don’t read more than the headline and then just look at the pictures before posting LOLWUT all over facetube and twitbook.

            • eitje
            • 8 years ago

            tl;dr – get off my lawn!

    • Farting Bob
    • 8 years ago

    “Reinventing the wheel” springs to mind, I cant imagine these will be cheaper to make than a heatsink and a bunch of fans in a datacenter where you might have hundreds or thousands of CPU’s to cool. Still, interesting tech.

      • Meadows
      • 8 years ago

      Datacentres don’t complain about noise to begin with. You’re missing the target market by a long shot.

        • DaveSylvia
        • 8 years ago

        Maybe on the noise front but on the lower-profile front, there’s definitely a demand for even smaller foot-print servers.

Pin It on Pinterest

Share This