Sandia intros revolutionary heatsink with rotating fins

By on July 11, 2011, 3:22 PM

Although most of our staffers use aftermarket heatsinks, we rarely cover such devices. It's not because we don't find them useful, they're just boring. The heatsink market has been relatively unchanged for years: you have a static, thermally conductive block with an array of heatpipes and fins alongside a fan. Naturally, that configuration varies depending on the noise level or cooling capacity desired, but most heatsink manufacturers are essentially reinventing the wheel.

Hoping to shake things up a bit, Sandia National Laboratories has unveiled a new technology that vows to "dramatically alter the air-cooling landscape." Referred to as the "Sandia Cooler" or the "Air Bearing Heat Exchanger," the contraption relies on rotating fins instead of a fan to dispel heat. The organization says conventional heatsink and fan combinations have a ton of "dead air" against the fins where there isn't a lot of airflow, and Sandia's design solves this limitation.

"In a conventional CPU cooler, the heat transfer bottleneck is the boundary layer of 'dead air' that clings to the cooling fins. With the Sandia Cooler, heat is efficiently transferred across a narrow air gap from a stationary base to a rotating structure. The normally stagnant boundary layer of air enveloping the cooling fins is subjected to a powerful centrifugal pumping effect, causing the boundary layer thickness to be reduced to ten times thinner than normal," the firm explains.

The Sandia Cooler is said to dramatically increase cooling performance while occupying up to ten times less space than current state-of-the-art CPU coolers. It's also quieter, immune to dust clogging, and more power efficient. In fact, Sandia claims its technology could cut US power consumption by up to 7% if widely adopted. The group believes its technology is suitable for most computers and electronics, as well as household appliances like air conditioners. 




User Comments: 43

Got something to say? Post a comment
insect said:

The simplest ideas...

Whatever genius was sitting, drinking a beer, and and said, "Hey, what if we spin the fins instead!" is sitting pretty now... or will be (if the technology is cheap enough to adopt into all new electronics).

albertjkruger said:

At the risk of sounding nerdy: this has got to be one of the most interesting/fascinating subjects that I've read of today, although I hardly understand how exactly it works.

mailpup mailpup said:

With the Sandia Cooler, heat is efficiently transferred across a narrow air gap from a stationary base to a rotating structure.
This is the part I don't get. How can air "efficiently" transfer heat. This gap sounds like insulating dead air to me. Conventional heatsinks need a thermal compound to transfer heat. With no thermal compound and with direct heatsink to CPU contact you still have a partial air gap and the CPU will overheat. How does air now become such a good, efficient transfer medium?

Staff
Jesse Jesse said:

When can I buy one?

Technochicken Technochicken, TechSpot Paladin, said:

How is heat transferred to the rotating fins? It must use some sort of layer of thermal compound, which certainly won't help the cooling efficiency. Also, how does this cut power consumption?

mario mario, Ex-TS Developer, said:

Sounds very interesting and it is an area where you don't see much innovation. I would love to see an animation of how it would actually work since it's not 100% clear to me how it works.

Guest said:

There is a stationary baseplate that contacts the processor to transfer the heat. The fans are .001" above this surface and rotating and it pulls the heat from the baseplate

Phil D.

red1776 red1776, Omnipotent Ruler of the Universe, said:

I wonder if these guys said at one point ..."I can't help but think we are missing something"

[link]

. Also, how does this cut power consumption?

I wonder if it's just because cooler components lower the resistance, increase efficiency, and need less voltage?

Sounds very interesting and it is an area where you don't see much innovation. I would love to see an animation of how it would actually work since it's not 100% clear to me how it works.

As would I. I am interested about what exactly is happening right at the point of heat transfer. It sounds like its different than standard molecular conductivity.....I'm sure that sentence was scientifically incorrect, but you get the idea:p

TekGun TekGun said:

I wonder how loud this will be, if it outperforms current top end air coolers and is quieter then awesome.

1977TA said:

Wow, these guys are going to make a killing, awesome product!

Puiu Puiu said:

mailpup said:

With the Sandia Cooler, heat is efficiently transferred across a narrow air gap from a stationary base to a rotating structure.
This is the part I don't get. How can air "efficiently" transfer heat. This gap sounds like insulating dead air to me. Conventional heatsinks need a thermal compound to transfer heat. With no thermal compound and with direct heatsink to CPU contact you still have a partial air gap and the CPU will overheat. How does air now become such a good, efficient transfer medium?

Where did you read that they'll replace the thermal compound? In the diagram you can clearly read "stationary base plate" --> this will come in direct contact with the CPU (with the paste in between). They just replaced the fan+big heatsink with a more efficient way of cooling the heatsink --> they rotate the upper part of it (the biggest part).

dividebyzero dividebyzero, trainee n00b, said:

Sweet.

Takes in ambient (chassis temp) air through the central area, and fires hot air in a 360 spray across the motherboard. Might cool the CPU just fine -at least until the sprayed hot air starts warming up the air volume in the chassis*- but I don't think it's going to be doing the RAM modules, mobo VRM and the back of a graphics card any favours.

*Going to make chassis intake/exhaust airflow placement a challenge

/ Would have liked to see the lovely diagram use a number higher than 40C as it's cooling benchmark.

As would I. I am interested about what exactly is happening right at the point of heat transfer. It sounds like its different than standard molecular conductivity.....I'm sure that sentence was scientifically incorrect, but you get the idea:p

It's not a hell of a lot different than conventional air cooling according to their pdf. Conventional coolers use a low flow (fan) airstream passing over cooling fins and heatpipes to effect the heat exchange. This method uses a very high airstream moving over those same fins- the same convection (advection?) seems to be responsible for transferring heat from the baseplate/heatspreader, to the fin arrangement -kind of like a turbine effect.

Mizzou Mizzou said:

I don't know, if the fans are turned solely by thermals rising off the processor it's hard to see this being as effective as touted. There's also the issue of the .001" gap tolerance, just about any kind of bump or nudge would probably be terminal ... even dust in the case would be an issue. If a motor is involved (not sure) it would presumably be mounted somehow directly above the processor which would seem to create other problems.

In the absence of more information or any credible test results it would seem to me that the jury is still out.

red1776 red1776, Omnipotent Ruler of the Universe, said:

Sweet.

Takes in ambient (chassis temp) air through the central area, and fires hot air in a 360 spray across the motherboard. Might cool the CPU just fine -at least until the sprayed hot air starts warming up the air volume in the chassis*- but I don't think it's going to be doing the RAM modules, mobo VRM and the back of a graphics card any favours.

*Going to make chassis intake/exhaust airflow placement a challenge

/ Would have liked to see the lovely diagram use a number higher than 40C as it's cooling benchmark.

Good point, going to need a vent/pipe to the chassis exhaust fan.

OH:o, I missed the PDF link.

Vrmithrax Vrmithrax, TechSpot Paladin, said:

technochicken said:

How is heat transferred to the rotating fins? It must use some sort of layer of thermal compound, which certainly won't help the cooling efficiency. Also, how does this cut power consumption?

As was mentioned above, there is a tiny gap between the solid mounted baseplate and the rotating fins. The smaller that gap, the better the thermal transfer between the plate and the fins.

It helps to think of it in simplified terms. Conventional heatsinks conduct the heat away from the source, then radiate the heat away. Conduction is typically highly efficient, but the radiant release into air is less efficient - fans are used to circulate the air and help the efficiency, but the flow is turbulent and less than perfectly even. This new method puts a step in between, by basically radiating from the hot plate across the tiny air gap to the moving fins, which then radiate again into the air. In this case, however, the fins are moving, and look to be designed to keep the airflow very uniform across all fins, so the efficiency of the cooling should be stellar.

Which brings the power consumption question... If the fins are much more efficient, a slower RPM can achieve the same goal as the old heatsink/fan combination. In theory, of course. Too bad they cant tap that heat energy to generate the spin too, that would be sweet!

dividebyzero dividebyzero, trainee n00b, said:

Good point, going to need a vent/pipe to the chassis exhaust fan.

Intake hole over the centre of the heatsink/fan, shroud to direct exhaust air away and out.......mmmm....kind of sounds like a graphics card squirrel cage.

Of course the main problem there is that you get a back-pressure buildup on the "sides" that don't vent directly out- which would be an insta-death for a CPU cooler as is described here.

My main query here is that the centifugal cooler is going to need to be spinning at high rpm (and/or have a large diameter) to cool anything above a mainstream CPU stock clocks. It has already been demonstrated that passive air cooling can effect much the same result. In fact, passive heatsinks can do somewhat better than stock clock/voltage . It actually surprises me that more [link]

gwailo247, TechSpot Chancellor, said:

I wonder how those 50 pages of physics equations are going to stand up against 1 benchmark.

Trillionsin Trillionsin said:

So tell me.. .how is this possibly immune to dust clogging anymore than any conventional fans?

If there is airflow, there is a pattern the air will take. If there is air, there is dust. Dust will form on the most susceptible places and then progress just like any other fan/heatsink.

1977TA said:

trillionsin said:

So tell me.. .how is this possibly immune to dust clogging anymore than any conventional fans?

If there is airflow, there is a pattern the air will take. If there is air, there is dust. Dust will form on the most susceptible places and then progress just like any other fan/heatsink.

Magic?

Seriously though, I can't wait to see some benchmarks/comparisons !

red1776 red1776, Omnipotent Ruler of the Universe, said:

This is shaping up to where I think they need to call in James Dyson to solve the logistics of it all.

Guest said:

Couldn't they do this even more efficiently by leaving the heat sink stationary and sealing a fan right above the middle portion of the heat sink then using the fan to push the air through the fins and out the sides?

yorro said:

This is one of those Why-I-Haven't-Thought-About-It Ideas.

Guest said:

Thermal transfer is done through convection (fluids and gases), conduction (solids and fluids) and radiation (waves like emi). Most heatsinks rely on conduction from the processor into a large amount of metal and then a fan to create convection to remove that energy. Radiation is very small.

It seems that what they are saying is that this still leaves a non-fluid air barrier which reduces the efficiency of the convection. And of course, the fan itself does not act as a heatsink because of its separation.

This design seems to be employing convection over a more accurately controlled area and volume of air and that air also acts as a bearing for the fan. The narrow air barrier means there is no dead air so convection is vastly improved (all air has become fluid) and the close proximity of the fan means it can conduct too and becomes part of the heatsink.

It is all about keeping the air gap tiny and making it behave like a fluid in motion, i.e. a continuously refreshed coolant. Make the air move fast enough with no surface adherence (perhaps just a little turbulence is needed to 'scrub' the surfaces) and it is just like a liquid coolant. It's a brilliant bit of engineering.

I'm curious to know more about the air bearing - how is it created initially and then maintained so accurately ? What are the surfaces on the bearing like - ultra smooth, channelled, golf balled ? Is there any other bearing involved ? I think there must be one at the centre....?

Guest said:

ALL I can see from this description is errors in the design based around laminar flow, Reynolds numbers and Bernoulli's principal..

But from a practical pov, the larger mass of the rotating fin assembly will require more torque to overcome the fin assembly's inertia, which means heat output from the motor goes up significantly.

mailpup mailpup said:

Where did you read that they'll replace the thermal compound? In the diagram you can clearly read "stationary base plate" --> this will come in direct contact with the CPU (with the paste in between). They just replaced the fan+big heatsink with a more efficient way of cooling the heatsink --> they rotate the upper part of it (the biggest part).
I mentioned thermal compound as a comparison. Where did you read that there was any thermal compound at all? Immediately above the stationary base plate is an air gap or air bearing as they say. Due to the air gap, there is no direct contact.

I understand what they are saying as to how the system works but the part I don't quite see is, "heat is efficiently transferred across a narrow air gap," that's all.

Guest said:

The heat can be transferred across the air gap because the air in the gap is behaving like a liquid.

The trick is getting it to behave like a fluid - that needs a very fine gap with no still air in the gap and high flow.

That gap is not not going to be easy to maintain. I'm assuming the motor at the centre is what does it so that is going to be an expensive motor and the design overall is not gonna be cheap to produce because of it's fine tolerances.

red1776 red1776, Omnipotent Ruler of the Universe, said:

Anyone remember this?

I wonder how , or if the development is coming.

[link]

Guest said:

The group believes its technology is suitable for most computers and electronics, as well as household appliances like air conditioners.

Most? So much for 5GHz overclocks. If they were confident in surpassing current HSF solutions, they wouldn't say "most". I see this as a low end solution, just better than passive heatsinks.

wcbert said:

You should have more type of these stories here. I found it an interesting read today. Far better the reading about game performance in a video card.

Guest said:

That 0.001" air ain't f..n things up ?

MilwaukeeMike said:

I dont' see how this will be much of an improvement. It definitly appears to be a more efficient fan, but air is still inferior to liquid or metal for conducting heat. This is why 70 degree air is comfortable and 70 degree water would make for a very cold shower.

Maybe couple this fan with a traditional heat sink, but on it's own, I don't see it competing unless it moves a TON of air. I also looked like their experiment used six 1 x 1 square inch tiles that were hot. A CPU's heat is far more concentrated, so we'd need a piece of metal to displace the heat anyway.

Guest said:

Why not just spin the whole computer?

Guest said:

Unfortunately, like most inventions now, it was created by someone working for another company.

THAT company is now sitting pretty and the actual inventor got his beer and hopefully a pat on the back.

NOW, said company will have a patent and be able to lock it away and troll away for decades to come.

YAY innovation!

Row1 said:

I have this, with a low wattage (45watt) AMD AM3 605e processor: GELID Solutions Slim Silence AM2 65mm Ball CPU Cooler. This is very low-profile, and pretty quiet, and has the fan down in the middle of the fins.

Row1 said:

Guest said:

Why not just spin the whole computer?

tru dat.

red1776 red1776, Omnipotent Ruler of the Universe, said:

Row1 said:

I have this, with a low wattage (45watt) AMD AM3 605e processor: GELID Solutions Slim Silence AM2 65mm Ball CPU Cooler. This is very low-profile, and pretty quiet, and has the fan down in the middle of the fins.

Ummm, you are kidding right? the GELID Solutions Slim Silence is a inexpensive chunk of aluminum with a fan in the middle. It says it will "dissipate 65W with good case ventilation"

dividebyzero dividebyzero, trainee n00b, said:

Strangely enough, I think the Gelid Solutions etc are Sandia's benchmark (see page 22 of the pdf) . Page 15 of the pdf (and repeated throughout the paper) notes that present coolers in the prototype's size range have a thermal resistance (cooling efficiency) of 0.6-0.8C/Watt...in fact most of the coolers they are comparing their product to are ~10 years old (Socket 370, Pentium III), use 70mm fans or smaller, and most are featured here in a review from May 2001

Something tells me that the Sandia boys don't update their bookmarks very often.

So for an average ~125w quad core CPU Sandia are assuming 75-100C recorded CPU temp using a conventional heatpipe cooler (!), while their own solution would present 25C (I'd assume that the figure is still dependant upon ambient room temp).

Guest said:

i'm guessing by not using a cpu fan. no fan, no extra power.

Guest said:

I'm not sure how this won't get clogged like the fins on a normal heat sink, or how the thermal transfer is going to be as efficient. Even if you view it as a fan have you ever had a fan that didn't have dust on it after a year of use even with cleaning it 4 times a year. All this is doing is making the fan and the heat sink one unit. The fins rotate on a motor which produces heat since it is more weight than a typical plastic fan you will be producing more heat, though in A/C applications that use huge fans made of metal it may be way more efficient in regards to weight. I'm not expert in this but it seems that between the motor adding heat the lack of direct contact with the heated surface the thermal transfer will be not as efficient. Though if the specs were perfectly engineered you could have a tight enough fit, with fast enough spinning blades you would see huge reductions in heat but at a huge cost of production or high failure rate. With the fins being so small failure will be much more likely to result in damage than typical heat sinks on the market that many can still keep a system cool enough with the case fan to last until a replacement fan on the heat sink is purchased.

Guest said:

Why don't they just build the CPUs with heat pipes and fins already part of the shell no need for thermal compound just add your own fan.

Zen Zen, TechSpot Paladin, said:

I have to admit that I drew a lot of pleasure reading all about this so called "product of the future". But to be more honest, I'm enjoying seeing everyones statements about this product. I have to admit this technology appears to be impressive, but the theoretical application behind the product just leaves me boggled. One thing is for sure, in all my years in the computer industry, I've never seen something this innovative in regards to CPU cooling.

I'm going to have to reserve my final thoughts about this product until I see some actual "hard data" and "real life tests" to back up this products claims. This things looks great on paper, but I want to see it in a real computer case, attached to a real motherboard and hovering over a real CPU and the computer being pushed hard, maybe test it by playing Crysis.

I don't know, this all seems like "feel good watered down gospel", little wishy washy to me. I'm an old school computer science major, give me the big fan motor, the big fan blades, the cool slim line heat sinks and a little dab of thermal compound, that's what I trust.

But again, I really have enjoyed seeing everyone's points of view on this product, made for good reading!

SINEKT SINEKT said:

The fans are .001" above this surface

Phil D.

Ouch ! No accidental touching of the spinning stuff here...

Why don't they just build the CPUs with heat pipes and fins already part of the shell no need for thermal compound just add your own fan.

Cause then, if the pipes break or something, you'll need another CPU, not another cooler. Kinda why we have cars built like that. You don't need to replace your car to get a new clean air-filter.

Guest said:

Check out the bitchin' IP theft at: http://www.youtube.com/watch?v=zAf_p-7cTDo

Looks like Sandia's concept, Sandia's design, and hey -- check out 2:45 and 3:55 -- even Sandia's graphics and a photo of Sandia's prototype! Too bad these $230,000 "winners" don't have a license for Sandia's IP. No problem taking credit for it and winning competitions with it, though...

Load all comments...

Add New Comment

TechSpot Members
Login or sign up for free,
it takes about 30 seconds.
You may also...
Get complete access to the TechSpot community. Join thousands of technology enthusiasts that contribute and share knowledge in our forum. Get a private inbox, upload your own photo gallery and more.