The fanless heatsink: Silent, dust-immune, and almost ready for prime time

For what it's worth, laptops with platter drives used to have a noticeable movement effect. If the drive was spinning pretty hard and you tilt the laptop, you can feel it want to go back to flat.

Duoae wrote:

This is not a gryoscope.

It's not a gyroscope, but it is a gyro. You spin anything, and it will behave that way, to conserve angular momentum.

There will be a "correcting" centripetal effect but that should not materially affect the distance of the air pocket (or bearing as the guy in the video above calls it) below the spinning heat sink portion because it is, as is mentioned at around 2:21 in the video, the spinning itself that causes the separation of the two surfaces.

I don't think we know that. Hard drive heads lift themselves off their platter in the exact same way, using the streaming air to maintain a gap, and they most emphatically can and do crash very hard if they're jostled, gouging out huge grooves in their media. Laptop drives take special measures to avoid hard knocks like that, but I don't think these coolers will be able to do anything similar. And if you change which direction gravity is pulling, then the only scenario where I can see that not mattering is if the cooler is actually balancing tension between magnets and air movement.... the forces involved would need to be much stronger than gravity for gravity not to be a factor.

The cooler doesn't appear to have any physical contact with the mechanism underneath. There doesn't seem to be a spindle holding it in place, it looks like it's spun purely with magnets. They might be able to get it working at any angle, but that seems problematic to me. Unless the forces involved are very, very strong, moving it around or holding it vertically is likely to result in fairly poor outcomes. Magnets can be very strong indeed, but being strong enough to ignore gravity, while also being weak enough to allow mere air pressure to pry them apart, seems unlikely at first glance.

I am not, however, a physicist, and this is not physical advice.

The other thing I'd be thinking about for a laptop is how's the airflow going to work for it. From what I could gather it sucks air in from the top and exhausts out the sides, which I would thin is a very big ask in a laptop where you've got half an inch if you're lucky. This is perhaps not ideal in most 'wind tunnel' PC cases either.

The other thing to consider about cramming one of these in a laptop is vertical clearance; you've got the blade unit that spins sitting on top of a brushless motor. Is it even feasible to build one of these thin enough to fit inside a current gen laptop case?

Hypatian wrote:
Duoae wrote:

This is not a gryoscope.

It's not a gyroscope, but it is a gyro. You spin anything, and it will behave that way, to conserve angular momentum.

Hence my second paragraph >:/

Malor wrote:

The cooler doesn't appear to have any physical contact with the mechanism underneath. There doesn't seem to be a spindle holding it in place, it looks like it's spun purely with magnets.

From the article:
IMAGE(http://www.extremetech.com/wp-content/uploads/2012/06/sandia-heatsink-impeller-cooler-cutaway-640x300.jpg)

See that thing in the centre?

Also from the link to his Q&A discussion in that article:

b) The other point is that the air gap distance is not maintained by holding tight mechanical tolerances. There is a negative feedback in air bearings that provides a passive stabilizing effect. For example, imagine I pushed down on the air bearing, such that the air gap distance was reduced by 10%, and then released it. The smaller the gap distance, the higher the pressure generated by air bearing (this applies to both hydrostatic and hydrodynamic gas bearings). This increases the upward force acting on the heat-sink-impeller, which acts to return it to its original height. The air bearing basically behave like a very thin, very stiff, compression spring.

c) The latter point might sound counter-intuitive, but indeed air bearings are extremely mechanically stiff, rugged and reliable. This is also discussed the report. Briefly, the effective spring constant (change in force per unit change in length), k, is given by k = dF/dh, the first derivative of force (F) with respect to height (h). Force is just pressure (P) times area (A), so for the stiffness of the compression spring we have k = A (dP/dh). It turns out dP/dh is a very large number; the pressure in the air gap region is very sensitive to a change in air gap distance. That for instance is why an air hockey puck that’s floating 0.001” above the surface of an air hockey table will very reproducibly return to an “altitude” of 0.001” if you take it off the table and then put back on the table. And if we increase the temperature of the room in which the air hockey table resides, it expands slightly, increasing the height of the air hockey table surface with respect to the floor it’s sitting on. But this doesn’t throw off the gap distance; the air hockey puck simply goes along for the ride and remain separated by 0.001” from the surface of the air hockey table.

Q: With such a heavy, metal fan, are there problems with vibration and gyroscopic effects?

JK: The brushless motor spindle is what keeps the heat-sink-impeller centered on the base plate. We don’t observe problems with vibration because the air bearing automatically provides very uniform mechanical support of the heat-sink-impeller. Gyroscopic effects are not a problem because unless you really go out of your way to spin your computer rapidly about an axis perpendicular to the air bearing axis (say, by mounting it on some kind of high-speed turn table), the forces generated are very small. Slow rotation generates very little gyroscopic torque because the magnitude and direction of the torque vector is given by the first derivative of the angular momentum vector with respect to time. The heat-sink-impeller is a monolithic piece of aluminum operated at stress levels far below its yield strength, so we don’t have to worry about it flying apart. There’s nothing special about the aluminum alloy we used. We used 7075 aluminum because that’s what we had on hand. In the future we will use 6063 aluminum because it provides somewhat better thermal conductivity (I did not know that at the time).

AnimeJ wrote:

The other thing to consider about cramming one of these in a laptop is vertical clearance; you've got the blade unit that spins sitting on top of a brushless motor. Is it even feasible to build one of these thin enough to fit inside a current gen laptop case?

Tough question but really there's no reason to stick one of these directly onto any chipset - they could easily stick it on a copper pipe connected to the chipset in an analogous manner to this:

IMAGE(http://www.ixbt.com/mainboard/gigabyte/m59sli-s5/heat-pipes.jpg)

For a laptop you could have it (or multiple fans) "vertically" sticking towards the back sucking air inward through a grille.

Anyone else got anything that needs clarifying?

[edit]

Just for thoroughness:

Q: Does the air bearing heat exchanger only work in a horizontal orientation? Or are other angles possible?

JK: As discussed in the white paper, a downward restoring force many times that of the gravitational force acting on the mass of the heat-sink-impeller is generated by attractive interaction of the permanent magnet rotor and the high magnetic permeability stator. For this reason the device can operate in any orientation and the air gap varies little as a function of orientation angle.

Duoae wrote:
AnimeJ wrote:

The other thing to consider about cramming one of these in a laptop is vertical clearance; you've got the blade unit that spins sitting on top of a brushless motor. Is it even feasible to build one of these thin enough to fit inside a current gen laptop case?

Tough question but really there's no reason to stick one of these directly onto any chipset - they could easily stick it on a copper pipe connected to the chipset in an analogous manner to this:

heat-pipes.jpg

For a laptop you could have it (or multiple fans) "vertically" sticking towards the back sucking air inward through a grille.

Anyone else got anything that needs clarifying?

Right, heatpipes let you move heat to a more convenient position to exchange it with the air, but if you look back at the video, there's this frame when they're talking about fluid dynamics (I'd guess air is the fluid in this case):
IMAGE(http://i.imgur.com/rG8Vn.jpg)
In most laptops, you're still pushing air through that heatsink, so it works in a flat configuration, while the sandia cooler draws air itself from above itself. It would seem like because of the compromises you make with laptops in exchange for portability/size, this is one of those tradeoffs that it's unworkable for that form factor unless you have it outside the casing.

This is the sort of implementation I'm talking about. You have a side/top and rear view of the "laptop" (I apologise for the crudeness of the drawing!). The heat pipe goes to one or several small coolers with their axis on the horizontal (so not vertically like in that picture above and just to avoid the confusion - I said vertically in my other post because we had been describing a vertical orientated axis as being horizontal throughout the thread). This sucks in air from the outside of the casing through those holes or grates on the back (depending on which works best). By having multiple small coolers you can make them smaller and so keep the thinness of the laptop. Note that also really thin laptops would never use a fan like this anyway and instead it'd be better for the mid-sized laptops in that respect chipset closer to the fans, reducing the size of the heatpipe required - this was just drawn as easily as possible, it is not an intention of how the layout would work on an accurate scale (I feel like I'd get criticised for that if I didn't state it explicitly!!).

You could also situate the

IMAGE(http://img6.imageshack.us/img6/8173/heatsinkdesign.jpg)