Slashdot Mirror
Sandia's Floating, Dust-Free, Spinning Heatsink
An anonymous reader writes "Sandia Research Laboratory believes it has come up with a much more efficient solution than heatsink-fan cooling a CPU that simply combines the heatsink and fan components into a single unit. What you effectively get is a spinning heatsink. The new design is called the Sandia Cooler. It spins at just 2,000 RPM and sits a thousandth of an inch above the processor. Sandia claim this setup is extremely efficient at drawing heat away from the chip, in the order of 30x more efficient than your typical heatsink-fan setup. The Sandia Cooler works by using a hydrodynamic air bearing. What that means is when it spins up the cooler actually becomes self supporting and floats above the chip (hence the thousandth of an inch clearance). Cool air is drawn down the center of the cooler and then ejected at the edges of the fins taking the heat with it. And as the whole unit spins, you aren't going to get dust build up (ever)."
http://hardware.slashdot.org/story/11/07/12/1348243/the-fanless-spinning-heatsink
Can we get some new editors??
I think I've read about this magic heatsink before... somewhere....
http://hardware.slashdot.org/story/11/07/12/1348243/The-Fanless-Spinning-Heatsink
Actually read the article, the spinning heatsink is attached to a base plate. It DOES NOT sit directly on a CPU die.
Maybe I just didn't get the message, but what draws heat away from the die itself? This setup probably does away with thermal paste and similar junctions...
From the video... there's a normal heatsink, and the fan draws the heat from the heatsink through the air bearing.
The other thing is that hydrodynamic bearings are only self-supporting and quasi-frictionless after a threshold RPM is reached. Before the whole setup is spinning fast enough for hydrodynamic effects to take over, it's going to grind against the chip die, and unless they came up with something good, it's going to destroy it on startup...
It's Sandia... I'm sure they've thought of that.
There's a video too. It has 45 likes and 117 dislikes. Apparently Sandia sucks at making videos. https://www.youtube.com/watch?v=uGpV_VPUn8g&feature=youtu.be
It's OK. Not even the editor read the article, or they would have seen it was from 9 months ago.
"National Security is the chief cause of national insecurity." - Celine's First Law
It spins at just 2,000 RPM and sits a thousandth of an inch above the processor
What could possibly go wrong? Seems like a pretty tight tolerance with all the vibration that could occur in a server room.
Read up a little on the science of Hard Disc Drives - heads usually rode on air, just above the platter surface. Same effect could be employed here.
A feeling of having made the same mistake before: Deja Foobar
But...all my fans get a layer of dust on each fan blade. What are they doing differently that will stop this?
If you watch the video, one of the heatsink's designers specifically says that when the device is spinning quickly (at 2,000 RPM), any dust particles that land on the device get flung off by centrifugal force.
Ask someone who's worked in a USA machine-shop: it's called a thousandth (the "of an inch" part is implied).
Machinists are not programmers, so beyond about 1/64" they switch to thousandths.
Below that, tenths (ten-thousandths of an inch).
Below that, millionths.
Actually it was almost a year ago. Consider it a dupe
“He’s not deformed, he’s just drunk!”
I suggest that this is how we managed to put a very expensive and blurry space telescope into orbit.
Not in this case; there was an extra washer installed on one side of the arm mount for the mirror grinder, meaning that the arm was skewed. I agree with your general sentiment of reducing areas of potential confusion, though.
Interesting... in electronics design, the 'mil' is a common unit of measure. E.g. a trace might be 6 mils wide. A 'mil' is 1/1000th of an inch. Checking wikipedia, it seems some call it a 'thou' - can't say I have encountered that.
I would suggest that one of the major reasons that US still uses Standard measurements in engineering has to do with "network effects" that date to the two world wars. During the second world war, European factories were heavily bombed and after the war they needed to be re-tooled. In contrast, American industry tooled up for the war, (using standard measurements) but was never bombed, leaving a surplus of high quality tools, many of which are still serviceable to this day. When you are making a new mill or lathe, it doesn't really matter whether it is calibrated in standard or metric, but re-calibrating an existing machine for a different system of units is very costly.
On a typical manual mill, for example, turning the traverse handwheel a complete revolution moves the table by an integer number of thousandths of an inch (usually 100 or 200, which are 2.54 and 5.08 mm). To operate the mill in metric units requires either that the operator remember that a revolution is 2540 micrometers (awkward) or rebuild a significant precision part of the machine (the leadscrews and leadscrew nuts). You might think that this wouldn't be a problem with CNC mills, but many use stepper motors to turn the leadscrews. Those stepper motors might have only 200 or 400 steps per revolution (giving a resolution of 1 to 0.25 mils, or 0.0254 mm to 0.00635 mm) which can make it inconvenient to use metric units.
If that weren't bad enough, collets (basically an adapter to hold the "bit" in the mill) come in standard sizes to hold mills (what you call a mill "bit" used on a milling machine. yes, it is confusing) of standard sizes, which are typically fractions of an inch on US equipment. When you are machining a piece of metal, the finite diameter of the mill it usually important. The accessories that go with a milling machine can easily add up to more than the cost of the machine itself. So, to really operate a mill in metric units in a convenient way, you'd also need re-purchase all the little parts that go with the mill.
Someone is probably going to reply that these issues don't apply to modern CNC tools. I'm not familiar with those, but the point is that there are a significant number inexpensive and serviceable tools in the US that can only work with metric units in a very awkward way (or at great expense).
See The fanless heatsink: Silent, dust-immune, and almost ready for prime time, and an interview with the inventor.
Disbelief of the dust-immune property of this cooler is addressed in the first question of the interview:
Jeff Koplow: I did not mean to imply that there is literally no dust fouling; some dust accumulation eventually becomes visible to the naked eye on the very leading edge of the blades. The point is that dust fouling is reduced to such a large extent that we are unable to detect any degradation of cooling performance operating the device in a relatively dirty environment over an extended period of time. Thus for all intents and purposes the dust fouling problem has been taken off the table. In contrast, with conventional CPU coolers, eventually the entire heat exchanger surface becomes entombed in dust. I suppose there are some applications in which computers are operated in extremely dusty environments that might be too much for the heat-sink-impeller. This is common sense. In trying to figure out a way around the longstanding problem of CPU cooler dust fouling, I was thinking in terms of residential and commercial environments where the vast majority of PCs are found.
Once again, it is disappointing how many people so yearn for the status quo, when presented with clearly superior technologies. Not that they always pan out, but it is disheartening to see such hostility toward progress.