News - Spinning Sandia CPU Cooler is 30 times more efficient

[IronWarrior]

It's nosy... and it's got a annoying sound which is far more annoying then my Thunderblades which are making a nice hum.

[Noli]

...but i doubt it will ever compete with the likes of liquid cooling on the basis that you are still relying on air to take it away from the cpu, i believe they said it creates dead air between the spinning metal fins and the base ? Which would mean heat transfer isnt brilliant.

1) The mention of dead air between the fins was with *conventional* coolers which don't spin so fast. He was saying that above 2,000rpm the boundary layer effects of this dead air reduce by a factor of 10 (one reason this design might be more effective).

2) The effects he's talking about e.g. >2,000rpm fan and incredibly small distance to the cpu have very large impact to how the air behaves and hence the cooling capability so I think that the comparison changes it from being 'just air' vs liquid to a very different kind of air in the same way that surface tension or capillary effects in water at the droplet size exist but are irrelevant in a pint of water say. Liquid cooling still has to dissipate the heat via radiators ultimately to the air too. Though admittedly the radiators can be outside the case and therefore larger than a heatblock inside. Even so, this sort of fan could also help on a radiator presumably. So yay.

*** CONTROVERSIAL STATEMENT ***

I actually agree with Malphas. There, I said it!
I don't think he was being outright offensive to anyone, just stating that very few of us likely have relevant expertise in this area and that certainly includes myself.

Bottom line is that their results assert 120 W m^-2 K^-1 vs 4 W m^-2 K^-1 for a more conventional CPU cooler. If that's correct, all our waffle about not understanding how it could be effective given this factor or that factor is irrelevant. Unless you want to cast aspersions on their measured results, we have to accept that it's simply an order of magnitude higher efficacy. Period (as the Americans would say). Obviously all they've done is proof on concept level stuff so far. Whether it can be commercialised and seen within our computers is a totally different story.

[Hicks12]

thankyou for that response Noli, it certainly does change my mind set a bit! I assumed being closer to the actual unit would provide masses of improvements in terms of heat dissipation, with the water comment i really meant that for high end systems and generic cooling its all about removing heat from area A (i.e chip), and dumping it where ever, the use of massive radiators provides that huge advantage as you have a reliable means of heat transfer (the water and the copper blocks) and all that heat is getting transfered to large radiators that means the chip is as cold as possible. Obviously if they could make a massive version im sure it would compete, its merely that it would be hard to have such a huge dump site as liquid cooling setups have which is why i love mine haha, although better fans would be nice!.

[aidanjt]

It's nosy... and it's got a annoying sound which is far more annoying then my Thunderblades which are making a nice hum.

Actually the noise is from the prototype motor. If you watch most way through the video you see the part where they turn off the motor and let the top part's own inertia keep it spin, the noise of the assembly itself is practically inaudible. Apparently they're aiming to have a far more silent motor for the final product.

[aidanjt]

Interesting concept, I do wonder why they've left the top open though

Likely to keep down the noise level.

and haven't ducted the exhaust either.

No need. The ATX standard takes care of that for them in desktop PCs. No doubt individual designs for more specific applications might need it though, it's always easy to tweak from there once the big problems are resolved.

The air gap is going to be a limiting factor and will have to stay very small as conduction via air is so low. They wont be able to increase the surface area either.

That's the only part I'm sceptical about. They'd really have to demonstrate significant thermal transfer through this air bearing in practice. I've yet to be convinced that a layer of air, no matter how thin, will do anything other than insulate the rotating heatsink from the heat source. They were also super vague about this point, which is always a bad sign.

I wonder how much of the heat is transferred by radiation (much higher factor with heat transfer through air)

I can't imagine it being a significant quantity. Physical thermal conduction is obviously its core focus.

[Noli]

The air gap is going to be a limiting factor and will have to stay very small as conduction via air is so low. They wont be able to increase the surface area either.

That's the only part I'm sceptical about. They'd really have to demonstrate significant thermal transfer through this air bearing in practice. I've yet to be convinced that a layer of air, no matter how thin, will do anything other than insulate the rotating heatsink from the heat source. They were also super vague about this point, which is always a bad sign.

120 W m^-2 K^-1 for their thin gap air conduction vs 4 W m^-2 K^-1 for direct touch standard headsink. What's the matter? You have been told the effectiveness but just don't believe it?

At the end of the day, rate limiting factor for any heatsink structure (like any reaction) is the slowest point at which heat can transfer (the bottleneck). For a standard heatsink, this might well actually be the point where the metal fins transfer heat to the air (which would keep the metal hotter, which would in turn lower the temperature differential between heatsink and cpu, which would in turn lower the rate of heat transfer since we all know that heat transfer is proportional to delta T). Since, on a conventional heatsink, even the faster, larger fans are still relatively low pressure and large surface area is created by packing fins close together (inhibiting airflow), this may actually be the rate limiting step. Direct touch heat pipe heatsinks hardly thrash metal block ones (they're very close in performance in fact) so that seems to indicate that the rate at which heat is conducted away directly off the chip is not necessarily the bottle neck.

The difference here is that, although the surface area is lower and the fins are still quite tightly spaced, the *entire* heatsink is itself spinning at over 2,000 rpm which is enough to create high pressure and force airflow through the structure with a *much* smaller dead air (boundary layer) effect than conventional heatsink cooling. If you're worried about insulating effects of air, then it's probably this increased dead air in conventional heatsinks that is limiting their effectiveness. And we probably *think* that they more effective than they are just because intense competition has refined this technology, this mechanism to a certain point. Whereas, it may not be very effective at all, simply that it is just the best mechanism we have so far developed.

I'm just making all that up but the point is that the results should speak for themselves and if accurate, then there must be a good explanation even if it's not my one.

kthxbai

[aidanjt]

120 W m^-2 K^-1 for their thin gap air conduction vs 4 W m^-2 K^-1 for direct touch standard headsink. What's the matter? You have been told the effectiveness but just don't believe it?

I can be told a great many things, including the moon is made of cheese, the west is a haven of democracy and goodness, that pop music is pleasant to the ears, etc. Unless they're backed up with facts they're meaningless bare assertions. And the fact is, a pocket of air tends to act highly thermally insular, *no matter how thin*, that's why micro-air bubbles mixed with thermal interface materials is regarded as very very bad. That's the whole point behind making the heat exchange fins rotate to begin with, to eliminate dead zone air pockets. So I'd need very compelling evidence contrary to current known physics to be convinced that the air bearing isn't creating a bigger problem than it solves.

[Whiternoise]

It's rather silly for unqualified people to be speculating on how effective it is purely on some pictures and brief description in a news article, with no actual in-depth knowledge of it's workings or the level of physics knowledge to accurately comment on its effectiveness supposing they did.

EDIT: Oh wow, this took off about 10 posts ago, but still

[DLUK]

I think the actual relevant points are for are we going to bin H20 or traditional lumps of metal:

*What happens if the machine or desk takes a small knock, how much like a HDD is the new cooler?

*What happens if the motor fails?

*How silent will the production models be and will the air flow in most cases for GPUs be affected as the CPU may be causing a vortex in side the case?

These new coolers may lead to new issues as they replace old ones.

[Lord Midas]

Ok, Malphas don't bit my head off, but... I like the fact that a heatsink and fan have merged into a Fansink, and thus potentially doing the job of two devices more efficiently.

I currently have a Tuniq Tower 120 and at nearly 1kg is stupidly massive (though extremely effective). If I could replace this with a 30 times more efficient tiny little cooler (and without the need for having water in my chassis), then that's good by me. For a resonable price of course.

[coolio]

Am I the only person thinking that the current production price could be quite a big problem with this design?

At $10 per unit, by the time you add the accessories, warranty, packaging, marketing and all that jazz, the price is likely to be a fair bit higher than the price of current CPU coolers... which do their job.

Personally, I think that the biggest effect this design may have is changing CPU designs. They will be allowed to excrete more heat as this concept would be able to remove that heat. This may result in much faster processors or, in a bad case, less efficient processors (which is highly unlikely!).

[KeyboardDemon]

Am I the only person thinking that the current production price could be quite a big problem with this design?

At $10 per unit, by the time you add the accessories, warranty, packaging, marketing and all that jazz, the price is likely to be a fair bit higher than the price of current CPU coolers... which do their job.

Personally, I think that the biggest effect this design may have is changing CPU designs. They will be allowed to excrete more heat as this concept would be able to remove that heat. This may result in much faster processors or, in a bad case, less efficient processors (which is highly unlikely!).

I was wondering about price too. It sounds like a great idea, spinning the heatsink and using the force it creates to expel air out of the sides and suck in air from the top with no room for dead spots and no where for dust to sit on the fins sounds like a good idea. I'd love to see some benchmarks to see it it compares to other well established alternatives.

[mikerr]

The "air gap" is 1/1000th of an inch - i.e. the thickness of a human hair or less, and not static...

[levedy1]

How would that fan work in a desktop PC? I mean most desktops stand up vertically and if this cooler basically 'floats above the CPU' wouldn't gravity be an issue? Also you are going to have to find space in your PC for what seems to be some sort of external motor that powers the whole thing.

[Noli]

I can be told a great many things, including the moon is made of cheese, the west is a haven of democracy and goodness, that pop music is pleasant to the ears, etc. Unless they're backed up with facts they're meaningless bare assertions. So I'd need very compelling evidence contrary to current known physics to be convinced that the air bearing isn't creating a bigger problem than it solves.

Fair enough but all your argument (rightly) boils down to is that every single "fact" in life is based on an assumption or level of trust. I haven't measured the lack of thermal conductivity of air myself - maybe you have - but I take it on very good faith that it is a very poor conductor. The level of faith is enough for me to call it 'fact' and all scientific fact are indeed built on consensus and the weight of evidence with an eternal assumption that what we think we know *might* be shown to be wrong at some point in the future.

In this sense, 'facts' are nothing more than a high enough, albeit arbitrary, level of evidence. Now the people that have developed this have provided figures measured, I can admittedly only assume, in a competently scientific way. They could be rubbish of course but given nothing to the contrary I am willing to accept their published figures. Despite no experimental evidence to the contrary, your doubts, based purely on gut instinct around your theoretical understandings of how these things work, could still be right. Nonetheless, I am going to lend more weight to their figures than to your scepticism.

Please don't be offended by this 'fact'

[Vimeous]

I want one.
However will it work in my tower case?!

Make me wonder what we could do with a similar design suspended by magnets....