MSI cooler powers its fan using… heat

Fans are hardly the most power-hungry components in today’s PCs, but MSI has nonetheless found a way to reduce their power utilization to zero—at least as far as motherboard chipsets are concerned. As TweakTown reports, MSI has a chipset cooler prototype kicking around in its labs that employs a Stirling engine to power its fan using the heat produced by the chipset.

The Stirling engine concept works a little like a car engine, using pistons to deliver mechanical energy. However, instead of pushing fuel through a valve and burning it to power the pistons, the Stirling engine uses a working gas—usually air, hydrogen, or helium—that’s trapped in a closed loop. The gas expands when heated and contracts again when cooled, thereby moving the pistons and powering the engine.

MSI’s prototype uses heat from a chipset’s north bridge to power the very fan that cools it. An audacious move for sure, but one that apparently works, since TweakTown says MSI had a working concept at its headquarters in Taiwan. The site also says it believes the technology is “not far off from becoming a reality.” The cooler will reportedly debut on one of MSI’s forthcoming Nvidia-based motherboards.

Comments closed
    • blitzy
    • 11 years ago

    seems kind of dumb since in order for the fan to work the chip has to get hot, im no expert but i thought heat reduces the lifespan of microchips. I suppose if cards can get away with passive heatsinks for cooling then this may work, but it just seems to me that for the fan to work the chip would be getting too hot.

      • green
      • 11 years ago

      i’m not a chemist so i don’t know how various gases/liquids expand with heat
      but if the temp at which the motor ‘kicks-in’ is around 35-45C, that should be okay shouldn’t it?

      after kick-in the system acts to lose heat and drops the temp so the motor slows/stops
      at which point there’s a heat build up again resulting in the motor kicking in again
      i’m assuming this eventually levels off so you get a point where it’s running speed varies only slightly

      so it’s sort of like a lava lamp with a water wheel in the middle…

    • Spyrano
    • 11 years ago

    Sterling engines are indeed, cool, but why have moving parts if you can help it? Just stick with passive cooling, perhaps using carbon nanotubes.

      • DASQ
      • 11 years ago

      It’s more efficient in cost and motherboard realestate.

      Though, I want the two combined, gigantic heatsinks, AND a sterling engine fan.

    • just brew it!
    • 11 years ago

    Unless they can make it /[

    • willyolio
    • 11 years ago

    that thing might work even better if the heatpipe/sink assembly was attached to the upper part of the engine. since the engine works by temperature difference, cooling the upper area would let it turn more of the heat energy into useful work.

      • gerryg
      • 11 years ago

      Hopefully you can beat them to market with your version, eh? Since they’re experimenting, I’m going to assume they’ve tried out several different configurations, and the one shown may not be the last. I too have to remind myself not to be an armchair engineer when I don’t have all the info.

      • Majiir Paktu
      • 11 years ago

      I agree, I posted that in my own comment; however, I think the reason they didn’t is because the engine itself cannot transfer heat quickly enough. It does seem a bit strange, though; you’d think they’d use the fan to cool the ‘cold’ end of the engine, and let all the heat be used as part of it. Go figure.

      • ludi
      • 11 years ago

      If the job can be done effectively without the extra complexity and associated expense, technical elegance loses to simplicity every time.

    • leor
    • 11 years ago

    I’ve always wanted something like this to come out. We have all this energy coming off our components in the form of heat, and we throw more energy at it to dissipate it. It’s a brute force inelegant solution, so I’m glad to see someone coming up with something that takes some of that wasted energy and puts it to work.

    • donkeycrock
    • 11 years ago

    I thought we were trying to move away from moving parts in a computer!

      • gerryg
      • 11 years ago

      Blame the lack of progress on Crysis. 😉

      • Meadows
      • 11 years ago

      I thought air coolers are highly popular.

    • gat0rjay
    • 11 years ago

    Third party cooler indeed. I’ve had 2 MSI video cards.

    1) Burst into flames while booting up computer (NO OC on it, I swear)

    2) Was playing light weight graphics RTS and the fan literally melted apart and dropped onto my sound card. (Again, No OCing)

    So MSI can’t really do worse in my eyes…

      • Mourmain
      • 11 years ago

      Erm…

      *looks worriedly at his passively-cooled MSI card*

      I guess my reluctance to leave the computer on its own for too long as it may decide to burn down the house is not *[

        • Flying Fox
        • 11 years ago

        The guy was talking about the fans on his video card?

          • Mourmain
          • 11 years ago

          Who, gat0rjay? I think so. (So was I). Why?…

        • gat0rjay
        • 11 years ago

        Yeah, It was just a quick flash really. It only lasted a few seconds. I imagine it would’ve done no more harm if I wasn’t around.

      • odizzido
      • 11 years ago

      I’ve had bad luck with MSI myself. I stay away from them.

      • Flying Fox
      • 11 years ago

      The fan on the MSI Ti4200 is still going strong.

      • moritzgedig
      • 11 years ago

      got stuck

      the fan on my MSI GC didn’t serve the cards lifetime too.
      It was also to loud.

      A melting fan is a clear sign of a mechanical blocking, did it get stuck?

        • just brew it!
        • 11 years ago

        It is not at all uncommon for the bearings on the GPU fan to seize up. I’ve had it happen a couple of times; it’s why I now favor video cards with passive cooling. Since I’m not much of a gamer any more, reasonably priced cards with passive cooling are actually a possibility — the current card in my main rig is a passively cooled 7600GS.

    • alex666
    • 11 years ago

    Very interesting physics and engineering challenge. My late father-in-law, who was quite the electrical/communication engineer and very interested in PCs, would have just loved this. Kudos to MSI for trying something like this. It will be interesting to see if and how it pans out. If it does, maybe the same be applied to CPUs and GPUs.

    • Usacomp2k3
    • 11 years ago

    Starting with the chipset makes sense to me. Then maybe market it for gpu’s, and lastly cpu’s. At least that’s how I’d do it.

    I wonder if you could use this to cool a hard drive. They’ve got lots of surface area that gets pretty hot.

    • dlenmn
    • 11 years ago

    I’m an idiot… meant to be a reply

    • provoko
    • 11 years ago

    Third party cooler please.

    • Flying Fox
    • 11 years ago

    Can’t they just produce cooler running chipsets in the first place? Is that such an outrageous idea?

      • DASQ
      • 11 years ago

      Yeah, but that’s more expensive. A lot cheaper to just throw on a slightly larger heatsink and a $2 fan.

      Intel uses all it’s old processor fabs to create chipsets (when it moves to a smaller process), so they’ll always be one generation behind the CPU’s. And hotter than they could be too.

      • UberGerbil
      • 11 years ago

      Well, MSI doesn’t make chipsets. You play the hand you’re given.

      And if you’re Intel, no you can’t — at least not if you want to maximize the return on your capital investment (their whole fab investment model involves moving a fab over to chipsets once CPUs move on to the next node). They probably could do more to tune chipsets for lower power (as they have done for Centrino), but for Quickpath at least that may no longer be necessary, since a lot of hottest bits will be on the CPU where they should get finer control anyway.

        • Flying Fox
        • 11 years ago

        Oh, it’s definitely not MSI’s fault, and I applaud their efforts in trying to apply innovative engineering to the problem.

        Intel chipsets are actually cool enough, even DAAMIT manages to put out chipsets don’t really require active cooling.

        I’m just ranting against Nvidia. 🙂

          • gerryg
          • 11 years ago

          Trying out a stirling engine HS/Fan combo on chipsets is the right place to start. If the engine is cheap enough (it doesn’t look particularly expensive) to compete with a standard HS/Fan combo, it’s great news! The chipset is the best place to experiment since I’m assuming it’s more reliably within a certain temperature range compared to a CPU (or GPU? harder to imagine given space/orientation constraints), which is likely their end goal once they have reliable longer-term data on failure rates. Considering the trend toward green technology and the desire for quieter systems, there could be a lot of appeal here. Of course, in an ideal world all chipsets and processors would only need passive cooling, but there’s enough need out there in various applications that this sounds like a positive thing to me.

    • Majiir Paktu
    • 11 years ago

    My chipset fan runs at 5,000-10,000 RPM. I wonder how reliable a Stirling engine fan will be?

    I would assume there would be no conventional heatsink, and that instead the hot end of the Stirling engine would be all that is in contact with the chipset. Hopefully the Stirling engine is effective in transferring heat to the heatsink? (Where I presume the fan would blow cool air over the ‘cold end’ of the engine.)

      • DASQ
      • 11 years ago

      If you looked at the link, you’d see there IS a heatsink. How else would the fan work?

      But then later you reference a heatsink, so I think you’re just confused.

        • Majiir Paktu
        • 11 years ago

        Yeah, I looked at the pictures they provided.

        What I meant was that the heatsink would be connected only to the ‘cold’ end of the Stirling engine, rather than a smaller one to the engine and a much larger one to the actual chipset, as it appears in the picture. Their design is somewhat inefficient it seems, since they are taking heat away from their own source of power by doing this; but I suppose they found out that the engine cannot transfer enough heat on its own.

      • derFunkenstein
      • 11 years ago

      10k RPM? That would generate more heat than it dissipates, I’d think.

        • Majiir Paktu
        • 11 years ago

        PWM shows mostly 5k RPM, with occasional bits up at 10k RPM.

          • Forge
          • 11 years ago

          I’m almost certain that your fan is always at 5K RPM and the monitoring agent sucks. If your fan suddenly ramped from 5K to 10K, it would almost certainly self destruct, and I can guarantee you’d hear it.

            • Majiir Paktu
            • 11 years ago

            I assumed as much, just wasn’t sure. nTune does seem a bit screwy at times.

      • BoBzeBuilder
      • 11 years ago

      Assuming its a nForce4 chipset, unplug the damn fan. That’s what I did, and the temps have stayed pretty much the same, no overheating whatsoever. My system’s much quieter.

        • Majiir Paktu
        • 11 years ago

        I’m not too concerned with noise, my GPU fan is much louder; and it’s an nForce 680i. I’m overclocking the FSB, so I’d rather keep things as cool as possible. I may try out the passive solution sometime, though.

      • Kaleid
      • 11 years ago

      Sheesh. I refuse to have anything that rotates over 1000 rpm in my system (excluding the harddrives of course).

    • mortifiedPenguin
    • 11 years ago

    Question: what happens when the fan cools its heat source low enough that the fan itself stops working? Does that mean that the chipset is cool enough not to worry about?

      • Jive
      • 11 years ago

      I would guess its still connected to the motherboard power for those just in case moments.

        • Lord.Blue
        • 11 years ago

        Or it just responds to the heat when the heat gets back up to a temp that triggers the piston.

      • UberGerbil
      • 11 years ago

      Well, if they designed it right that should be the case. Of course, there’s the minimum temperature at which the Stirling engine will begin to operate, and the maximum temperature you don’t want the chip to exceed. As long as the former is below the latter, and there isn’t too much inertia so the engine begins spinning the fan quickly enough that the temperature doesn’t ramp too high before it gets going, you should be fine. However, electronics generally prefer to remain at a steady temp (due to mechanical fatigue from thermal expansion, etc) so you may actually want it to never cool the chip that well (and perhaps it never does).

      There are significant issues of cost and reliability before this starts showing up for sale, however.

      • emi25
      • 11 years ago

      First, let them show the thing, then somebody will test it. Then we will talk.

      • DASQ
      • 11 years ago

      I assume the calibration of the device will lean toward the fan spinning faster as it gets hotter (as it should, I think), and as it cools enough it’ll slow or stop spinning.

      It’s entirely automated, as long as a) The fan mechanically works and b) The heatsink is working properly.

      • RHITee05
      • 11 years ago

      The system forms a very effective negative-feedback loop. Higher chip temp will cause the fan to spin faster, cooling the chip down. Lower temp will slow the fan, causing the chip to warm up. The system will find it’s own equilibrium point determined by how efficient it is.

        • UberGerbil
        • 11 years ago

        The only problem: the natural equilibrium point may not be at the ideal speed for fan noise nor an ideal temp for the chip. Well, actually as a first approximation you want the chip as cool as possible, so you don’t really want it to slow down at all unless noise is a concern.

          • dlenmn
          • 11 years ago

          q[

            • bthylafh
            • 11 years ago

            q[

            • dlenmn
            • 11 years ago

            The absolute zero bit wasn’t to be taken seriously

            q[

            • Mithent
            • 11 years ago

            Provided that it will last the lifetime of the chip, then it doesn’t matter, of course; maybe it will only last 10 years at 30C, but if you’re likely to replace it within 5 then it makes little difference. Unless you’re worried about your computer’s usefulness as a museum piece, 15-20 years should be more than enough for most users.

            • green
            • 11 years ago

            if the cooling solution results in a running temp of 60C, the chips’ life is 1.25 years
            if your warranty is only 1 year then guess what… new chip required
            there’s a reason why warranty didn’t cover overclocked chips

      • willyolio
      • 11 years ago

      if it’s that low, it’s probably near ambient temperature. stirling engines don’t need much heat difference- i’ve seen ones that operate on the palm of your hand at room temperature.

Pin It on Pinterest

Share This