THe Sandia Cooler - Breakthrough in Air Cooling design

[Seero]

Because they have to thermally couple a spinning impeller to a stationary base. They are going to need some kind of liquid lubricated thermally conductive bearing between the two thermal masses. If such a device were feasible they would be ubiquitous by now. Yet out of a billion air cooling systems around the world, precisely 0.000% of them use non-stationary thermal masses coupled by bearings. Probably because it is cheaper to do liquid cooling at that point. The thermal mass is always decoupled from the moving parts.

In this design, the weight of the rotating blade is lifted by air. When the blade turns, air pressure between the blade and the base increase to an extend that it lifts the blade, but when that happens, the area increases and therefore the pressure decreases. Eventually, it will float in mid air, approximately 0.3mm away from the base. They called this effect "the air bearing." No grease, lubricant, or thermally conductive bearings.
Correction, not .3mm, but .03mm. Sorry.

The rotating portion also have a flat bottom surface. So the air bearing acts on 2 flat surfaces, the top surface is rotating, and the bottom is stationary.

The air which acts as air bearing does not come directly from the rotating blade, but are supplied via a different machanic which are generated by the same motor. It was described as follows:

The prototype device is configured as a static (externally pressurized) thrust bearing. In realworld thermal management applications such an externally pressurized air bearing would be replaced by a hydrodynamic (self-pressurizing) air bearing, which uses a minute fraction of the mechanical power supplied by the brushless motor to generate the required lifting force.

Because the rotating blade floats in mid-air, the only physical contact, if there is any, is the nod from the rotor to prevent the rotating blade block to fly away. This design features a motor called a sensorless blushless motor, which means there exist minimal physical contact.

That means, in simple terms, this design only consists of 3 parts, the base, which can be a circular aluminum plate; the rotating blade, which is no harder to make compare to the Intel stock HS (the phototype was made from a single block of aluminum); and the sensorless blushless motor, which is used by lots of cooling fans.

 

[snarfbot]

In this design, the weight of the rotating blade is lifted by air. When the blade turns, air pressure between the blade and the base increase to an extend that it lifts the blade, but when that happens, the area increases and therefore the pressure decreases. Eventually, it will float in mid air, approximately 0.3mm away from the base. They called this effect "the air bearing." No grease, lubricant, or thermally conductive bearings.

well in the provided marketing image, its indicated that the blades spin clockwise which is fine as it would bring air to the middle, but then its also indicated that air moves downwards in the center, which doesnt make sense. it isnt going to centrifugally pump anything in its present form, at least depicted by the prototype image.

all this makes the rest of their claims suspect imo.

 

[drizek]

Then you have non-stationary interconnects. And they are 1000 times more failure prone than non-stationary thermal masses. For example, take the ribbon cables that wrap around screen hinges. Few of those will last more than 10 years. Probably 10% dont last 2 years. If you can see the folly of flexible ribbon cables, then I suspect you can probably see the folly of this cooling device.

They could just make it wireless...

 

[Seero]

well in the provided marketing image, its indicated that the blades spin clockwise which is fine as it would bring air to the middle, but then its also indicated that air moves downwards in the center, which doesnt make sense. it isnt going to centrifugally pump anything in its present form, at least depicted by the prototype image.

all this makes the rest of their claims suspect imo.

From the OP, it turns CCW, meaning air exhaust from the sides.

It isn't my claim, I simply read the document. I was trying to clear out confusions.

Check post 18, Patrick Wolf linked the doc.

 

[dx11101]

Cool Product but i wonder what happens when it gets dust bunnies.

 

[Bill Brasky]

Cool Product but i wonder what happens when it gets dust bunnies.

This. I'd like to know how it's actually immune to dust...

 

[sm625]

It's not immune to dust. Because the impeller is moving and vibrating, it is less prone to gathering dust. But it will gather dust. And when it does... it is going to wobble, and that "air cushion" they are trumpeting will be quickly be turned into a liability since it allows the impeller's relatively high mass to gain velocity before it makes contact with the baseplate. So you'll get a knock. Which will throw it out of balance even more. Then you'll get oscillations and the device will quickly destroy itself and/or whatever it is attempting to cool.

By the way, if the impeller is lifted up and away from the base plate while it is spinning, then how can they maintain the same thermal conductivity? Oops... When the impeller starts vibrating, the thermal conductivity will decrease in an exponential fashion.

 

[snarfbot]

hmm, alright i read the paper they still make some dubious claims such as the following:

For example, a survey of commercial CPU coolers indicates that a conventional fan-plus heat-sink device equal in size to the version 1 prototype device in Figure 6 has a typical thermal resistance of 0.6 to 0.8 C/W. Our version 1 prototype device on the other hand has a measured thermal resistance of 0.2 C/W. This represents a huge advance in a field that has long seen only incremental progress in cooling performance. Moreover, as discussed later, we have reason to believe that a 2nd generation prototype could readily achieve a thermal resistance of 0.1 C/W in a device of the same size.

that seems pretty revolutionary, but they fail to mention the heatsinks they are comparing their product to are 10 years old, like the aerocool glacier a socket 370 hsf.

a modern low profile heatsink like the evercool hpl-815 does around .2 c/w, so its pretty misleading.

a stock intel box cooler from a core 2 for instance can outperform those old socket 370 coolers by a large margin.

so pretty flawed methodology if you ask me, and just another of the many things that make me skeptical.

 

[snarfbot]

in fact, the actual image they use of the "commercial heatsinks" they compared their product to comes directly from this page:

http://www.dansdata.com/coolercomp.htm

its a socket a heatsink roundup from 2001. unbelievable, you can find it on page 22 of their pdf. its literally a screenshot of his page, embedded in the pdf.

they didnt cite dansdata in their sources either, so i sent him an email.

 

[Seero]

This. I'd like to know how it's actually immune to dust...

In conventional HS, dust mostly gets stuck at the surface (stationary) where the fan pushes at, while the fan blade itself isn't all that dusty.

In this design, air doesn't travel between 2 surfaces, but one, the rotating blade, which is constantly spinning. Like an airplane, the front portion of the wing is very clean as particles in the air polishes the front portion of the wing as it changes the direction of the particle which it travels. The top body of the wing is dirty as even a slight pump will cause drag which accumulate dust.

In other words, unless there are particles can has the length of the space between 2 blades and strong enough to with stand the wind force, then dust simply can't be accumulated as it has 2 forces acting on it, centrifugal force and impact force from the travelling air/particle.

This theory applies on all centrifugal fan, which is used by lots of high end video cards. I have seen dust pile formed on the surface of the rotating blades(i have pets), meaning that the same thing will occur on this design eventhough the theory is sounding.

I have added some correction/information to my previous post.

 

[KingFatty]

I haven't read the PDF yet, but aren't air bearings a known technology proven to work reliably? It seems like here they are taking a proven technology and applying it to the field of heatsinks.

I don't see how the proven air bearing technology would suddenly fall apart just because it's applied to use with a heatsink. I think they have held up under more demanding usages, right? Sorry if I'm misunderstanding as I haven't read the PDF yet.

 

[Markfw]

hmm, alright i read the paper they still make some dubious claims such as the following:



that seems pretty revolutionary, but they fail to mention the heatsinks they are comparing their product to are 10 years old, like the aerocool glacier a socket 370 hsf.

a modern low profile heatsink like the evercool hpl-815 does around .2 c/w, so its pretty misleading.

a stock intel box cooler from a core 2 for instance can outperform those old socket 370 coolers by a large margin.

so pretty flawed methodology if you ask me, and just another of the many things that make me skeptical.

^^^^^ THIS

And the fact that modern coolers are simple. For example, my son's computer has an X3350 CPU and a Thermaltake 120 extreme. The fan died, and it still run's fine, light load 50c, idle 40c, NO FAN. Under full load, is 80c, so we stopped folding until I get a chance to replace the fan.

 

[Seero]

I haven't read the PDF yet, but aren't air bearings a known technology proven to work reliably? It seems like here they are taking a proven technology and applying it to the field of heatsinks.

I don't see how the proven air bearing technology would suddenly fall apart just because it's applied to use with a heatsink. I think they have held up under more demanding usages, right? Sorry if I'm misunderstanding as I haven't read the PDF yet.

In this design, air which is used as air bearing also act as a heat transfer agent to transfer heat from the base to the top rotating block.

 

[CTho9305]

I haven't read the PDF yet, but aren't air bearings a known technology proven to work reliably? It seems like here they are taking a proven technology and applying it to the field of heatsinks.

I don't see how the proven air bearing technology would suddenly fall apart just because it's applied to use with a heatsink. I think they have held up under more demanding usages, right? Sorry if I'm misunderstanding as I haven't read the PDF yet.

I get the impression people just like to think they're smarter than the researchers at Sandia Labs who built an actual working prototype and measured its performance, dust buildup, etc.

 

[Markfw]

In this design, air which is used as air bearing also act as a heat transfer agent to transfer heat from the base to the top rotating block.

Yes, and air is 400 times less dense than solids (water or transfer fluids used in heat pipes)

I still say its a crap design

 

[Idontcare]

Did the Sandia link change or something? I can't find the 48pg pdf. All they offer in the link is a 1pg overview.

 

[MustangSVT]

Group A: . This is revolutionary and they have "scientific data" and "sample", thus they MUST be doing something right and they are "scientists". This will be the next BIG thing.

Group B: Air (which being one of the best INSULATORS) is being used as heat transfer unit. ARE YOU NUTS? Complicated mechanics and iffy at best data within lab environment. Not to mention PRICING concerns.

I would love to see this coming out as it does sound a bit interesting but I just can't be blind enthusiastic when it just sounds flawed. It reminded me of CAI ads promising 50 WHP gain .

Lets see how my basic basic physics & engineering "beliefs" turn out.

 

[Seero]

Did the Sandia link change or something? I can't find the 48pg pdf. All they offer in the link is a 1pg overview.

They did. From page 11 of the PDF.
"..can be mounted in an arbitrary orientation (e.g., up-side-down, sideways, etc.)."

http://prod.sandia.gov/techlib/access-control.cgi/2010/100258.pdf

Patrick's link still works.

 

[Idontcare]

Patrick's link still works.

:thumbsup: thanks for that! I did not catch the link in Patrick's earlier post.

So any bets on how soon before we see cheapo versions of this thing coming out of China?

The transition to heatpipes was nearly binary, I would not be surprised if we see these things being sold on Newegg within 6 months.

 

[beginner99]

I wonder why they did not run a straight up comparison with an i7 and several modern Heat sinks. I mean that would be much better from a marketing standpoint...if it works.

I'm not dumb but I have better things to do than look at a ton of equations but a nice comparisons table hell yeah. Takes you 5 sec. to see if it works or not.

According to:

http://www.ninjalane.com/reviews/cooling/noctua_nh-d14/page3.aspx

a NH-D14 has about 0.15 C/W.

OK, it also is a little bit bigger than this prototype but it does show, that it ain't that revolutionary.

 

[Idontcare]

I wonder why they did not run a straight up comparison with an i7 and several modern Heat sinks. I mean that would be much better from a marketing standpoint...if it works.

I'm not dumb but I have better things to do than look at a ton of equations but a nice comparisons table hell yeah. Takes you 5 sec. to see if it works or not.

According to:

http://www.ninjalane.com/reviews/cooling/noctua_nh-d14/page3.aspx

a NH-D14 has about 0.15 C/W.

OK, it also is a little bit bigger than this prototype but it does show, that it ain't that revolutionary.

I think the major point of this Sandia cooler is that it requires less material to attain the cooling it delivers (that means lower production cost, lower environmental footprint to manufacture) and that the cooling performance of the device over its operational lifetime is vastly improved relative to a traditional HSF.

Meaning sure a pristine NH-D14 will give you 0.15 C/W on day one of use but come back and measure it a year later after its clogged with dirt and grime and you won't have anywhere near the 0.2 C/W you would still be getting from the Sandia Cooler.

If it is just about cooling at any cost then I'd pit my VaporPhase LS against it as well, but the operating costs and initial upfront costs might make a difference in the "value" of the total package

In terms of why the Sandia paper did not take it to completion, this is a prototype intended for scientific exploration (hence the data curves generated with pure Helium versus pure Nitrogen environments).

This wasn't about hyping or selling Sandia coolers, it was about introducing and legitimizing a new concept in approaching HSF applications.

They'll rightly leave the i7 comparisons to FrostyTech and MadShrimps. This is simply about proof of concept and reduction to practice, two things you must do if you want to solidly lock up the patents.

 

[snarfbot]

heres the deal, they compare it to heatsinks that are over 10 years old, and call them simply "commercial heatsinks" of a similar size, which is misleading because you cant buy them anymore and modern heatsinks of the same size perform many times better.

it would be like via marketing its nano dual core as substantially better than other commercial cpu's and comparing it to a pentium 2.

on page 15 they compare it to the state of the art in air cooling:

A second basis for comparison is provided by the performance specifications given in the 2008 DARPA MACE call for proposals on ultra-high-performance air-cooling technology. As shown in the table below, our version 1 prototype device was able to provide the same cooling performance as DARPA’s state-of-the-art device while providing a factor of 4 reduction in size and more than a factor 10 reduction in electrical power consumption. Again, this clearly represents a breakthrough in performance.

they claim that it does .2 c/w while consuming 100 watts, and being much larger. while in reality it performs much better than that .05 c/w @ 33 watts.

http://rfdesign.com/military_defense_electronics/darpa-thermacore-heat-sinks-0411/

the more i scrutinize this research paper the more misleading it appears, if not deliberately deceptive, which is unacceptable in my opinion.

 

[Seero]

heres the deal, they compare it to heatsinks that are over 10 years old, and call them simply "commercial heatsinks" of a similar size, which is misleading because you cant buy them anymore and modern heatsinks of the same size perform many times better.

it would be like via marketing its nano dual core as substantially better than other commercial cpu's and comparing it to a pentium 2.

on page 15 they compare it to the state of the art in air cooling:



they claim that it does .2 c/w while consuming 100 watts, and being much larger. while in reality it performs much better than that .05 c/w @ 33 watts.

http://rfdesign.com/military_defense_electronics/darpa-thermacore-heat-sinks-0411/

the more i scrutinize this research paper the more misleading it appears, if not deliberately deceptive, which is unacceptable in my opinion.

Figure 23

 

[snarfbot]

Figure 23

i fail to see the significance with regard to my previous post.

anyway it did remind me of one other failure with the design, the fact that it consumes between 10-20 times as much power as the average 120mm fan. if you put one of those delta 220cfm fans on a regular tower heatsink it would definitely be under .1 c/w, while consuming about the same amount of power.

regardless im pretty sure the purpose of this paper was simply to justify continuing funding:

The FY09 Tier 1 LDRD and Sandia Royalty funds invested in this proof-of-concept study have resulted in a major breakthrough in air-cooling technology. The arguments put forth in the original proposal in support of the boundary layer thinning effect, negligible air gap thermal resistance, low electrical power consumption, low-noise operation, and immunity to fouling have all been shown to be correct.

and was never intended for any sort of peer review, but thats just speculation on my part.

 

[IlllI]

here is a q & a with the inventor http://www.extremetech.com/computin...tsink-your-questions-answered-by-the-inventor