Heat sink breakthrough threatens ventblockers A US government lab has opened for licensing a novel way of improving the cooling technology used in everything from CPU and GPU coolers to air-conditioning units: make the fan the heat sink, and the heat sink the fan. "We describe breakthrough results obtained in a feasibility study of a fundamentally new architecture for air …
That's pretty nice. I wonder if ducting or tubes will be part of it. But, if anyone is thnking of getting patent-crazy, a possible workaround/defeat could be similar to what Dell and others do: put a fan at the chassis opening and a fan near the heatsink, and choose directions of cooling and exhaust according the the final internal arrangement. Maybe tubed, axial fans might work, too...
It's already been thought off! It's called a "flywheel". Take a look at any two cycle weed eater, it has the basic engine that's producing heat, and a flywheel that produces energy to fire the ignition, but at the same time, absorbs heat and cools the engine via forced air. Most are made of cast aluminum. The trick in his quest is to get heat to transfer from the block underneath to the "fan" itself. There is a tradeoff on large fins to absorb heat verses how much air you can push through to cool efficiently. A sweet spot, if I may.
Next will be a magnet inserted on the end of the 'fan" and some type of pickup to either power something internally or monitor RPMs.
...it's one where everyone will say "I could have thought of that".
The really clever bit is in the thin air layer. Sure, air is normally a terrible conductor of heat, but when the layer is so thin it's thermal resistance is much reduced*. Thus, from the the point of view of the heat, the rotating impeller is thermally 'attached' to the base plate (or at least much more so than if it were, say, 1mm away).
I for one hope he/they make a pile of cash out of that. Clever ideas like that need rewarding. And besides, something like that spinning away at several thousand RPM has got to sound just a little bit like a turbine, and that'd be a cool noise for any PC to make.
There must be a pile of kinetic energy built up in that spinner. That could be used as a little energy reserve; lose the mains power, and the spinner becomes a generator providing just enough electricity for a cleanish shut down. Bit like an F1 car's KERS. My idea (unless someone else thought of it first)!!!
*Just like metal loaded epoxy used to connect some flexy circuits; it's a terrible conductor over any sizeable distance, but when used in thin enough layers that doesn't amount to much.
It is not the "thin" or at least not "just the thin".
The heatsink is spinning and there is a constant shear across that micro-gap. So its heat conductivity will _NOT_ be the conductivity of an air gap at rest with the same width.
What is it going to be - god knows, but my educated guess is that it will be more. In fact much more. That is the trick here. The thing works so well because of conductivity across a disturbed boundary layer which is being kept in that condition by the heatsink spin. At a couple of thousand RPMs it is likely to be on par with a lot of heat transfer pastes.
I did not get it on first reading either by the way. I looked at it, thought WTF and only after an hour it hit me: "This is what this guy has done".
Clever, definitely much more clever than it looks. One thing is clear however - this is not Zalman material. At the speed where the layer is disturbed enough for it to work it will produce noise higher than Zalman tends to accept in their gear.
"The thing works so well because of conductivity across a disturbed boundary layer which is being kept in that condition by the heatsink spin. At a couple of thousand RPMs it is likely to be on par with a lot of heat transfer pastes."
Yes, I think that's a far better explanation of what's going on. It's better to move the heat by moving and constraining the air, rather than rely on the heat conducting through the air.
I still wish I'd thought of it!
sorry, been done. Is used on (some? all?) disks so when power's cut the motor becomes a jenny and makes enough elec from disk running down to complete a (typically 4k) sector write, and I guess park the head after.
So I understand anyway.
Indeed, but that powers only the harddisk. I was thinking of the whole machine, or at least enough of it to close a few files properly. A VME chassis has an ACFAIL line, which can be used in embedded applications to do some vital stuff in the dying microseconds of the PSU's capacitor charge; quite a useful notification in some circumstances!
I thought that head parking was achieved through purely mechanical means in that the forces exerted on the head (via the air cushion) by the spinning disk have a tendancy to push the head arm off the disk. But there's clearly enough energy in the disk to do it electronically too.
Just a thought - that's not something that SSDs can really do is it, unless they have a decent amount of capacitance somewhere. So is a power cut a slightly greater problem with an SSD than for a HDD? Whatever volume checking an OS performs after a power loss, it'd be much nicer to find that all the sectors had been correctly written.
HDD heads nowadays auto-park because they're on a spring an voice-coil arangement which -- the power is cut to the positioning voice coil and spring returns the head to the centre.
Wasn't always like this though -- old HDDs were on stepper motors so the head had to be manually "Parked" if hte PC was to be moved, to stop it dragging across the platters. I miss the thunk of my old AT's head parking...
@bazza "I was thinking of the whole machine": Just how much mass do you think this heatsink has? From the picture, it's no where near the mass required to harvest kinetic energy to power a complete rig. Perhaps if it was on the order of Scythe's Mugen R2, then you might be able to help an SSD clear its cache (Intel SSDs and the like, since Sandforce doesn't use DRAM caches). But to help the whole machine, you're looking at several kilos at the least.
What happens when some idiot moves your PC whilst it is turned on so they can get something that has just fallen down the back, that sub-mm air gap will suddenly become much closer to the thing you're trying to cool and is likely to come whizzing out the front like some manic kung fu style star (except a lot hotter)
Angular momentum or some such physiks (or magic, much the same)
... because they are sealed units made in a clean room. This fan will not clog up because - erm err (wave hands a bit) some really clever reason. The close tolerance between the heat sink and fan narrows the boundary layer around the heat sink at the cost of extra air resistance. There will still be a boundary layer around the fan. If they still shift more heat per watt running the fan then cool.
cool air is pulled into the centre of the fan and "comes into contact with the hot base plate" so I think the air gap is exposed to fresh incoming air, full of dust. Hard bits of grit then get into the tiny close-tolerance gap and start grinding away the metal, producing a fine conductive dust that supercharges your RAMs. Ok I don't fully understand the concept yet.
These f**king flash adds are lagggin my interwebs. Why you have such yucky adds Reg? I don't run ABP on your page because I actually think you deserve the revenue but if you keep this up... You know flash adds actually put me off wanting a product? I don't care about boats and don't even know what it's for.
/rant
Im very keen to put a large, presumably heavy and necessarily sharp edge spinining flywheel right next to my graphics cards. on top of my chip. next to my ram.
several thousand RPM, you say? That's... thats astounding. like a chainsaw, then?
--
In all seriousness it sounds like a good idea, but the market is increasingly moving away from fanned systems to either passive-cooled radiators in massive tunnelled racks for business, and sealed-system liquid coolers in home units. personally I still want a total-immersion/submersion cooled beastie at heart. With lights! and bubbles! anything to seal the box, really.
Humans are dirty, and there's no way to avoid crud and the associated thermal problems with crud unless you make the computer a sealed unit (with a rad on top. or machine the entire backplate out of copper, mebbes)
take a look at the latest gfx cards. apart from the ones with liquid cooling blocks on they all have fans. only the nasty cheap cards are fan free. enjoy your low res/detail gaming with those.
liquid cooling is still more expensive than it should be. maybe all PCs should be liquid cooled and the costs would plummet?
there are also instances of when liquid cooling pumps die and your pc fries. im not sure i would trust a £100 to not die and fry £1000 worth of GPU/CPU etc.
The wire safety cage is an eminently sensible component - if only to avoid a Darwin Award following on from some unforseen event.
Good engineers need to survive to complete their good engineering.
It reminds me of one of the best engineering team of the 20th Century - the Wright Brothers. They realised that powered flight could be dangerous so they put the engine along side the pilot so that if there was an accident the pilot would not hit the engine - or vice versa.
There are some good questions needing good answers.
I was sceptical, but many questions I had are answered in the 40+-page paper.
E.g the PC heatsink is 0.2C/W vs three or four times that for a similarly sized classical one.
The question re junk in the bearing remains largely unanswered. The demo unit used lab-standard helium or nitrogen to pressurise the air bearing; p 25 "The use of dry nitrogen, rather than air, is an experimental convenience.". Well some readers will need more convincing than that, though he could well be right (maybe the thing is self cleaning? Really?).
Heatsinks do 2 jobs:
1 provide a conduction path to lose heat to the environment.
2 provide thermal mass to buffer rapid temperature changes
There's a little problem with this design, if the impeller stops spinning, the air gap becomes an insulator and suddenly the thermal mass is no longer connected to the thing it's buffering and cooling....
The bulky heatsink on modern processors doesn't need fan driven airflow a lot of the time, coupled with the thermal inertia that makes them relatively failsafe. 10s of seconds with a failed fan before dangerous heat levels are reached is not unusual, more than enough time for the CPU, OS or user to take action.
However effective this design is, he'll have to cripple it with old style passive blocks of metal to make it safe to use. That might restrict its market a little, if you've got to put the hefty metal in there either way there's not much point replacing the cheap fan on most systems.
...and I don't believe the dust claims. My desk fan seems to have no trouble covering it's spinning blades in dust... neither do the fast spinning ones in my PC.
Is the lack of thermal contact - the air gap is a significant and inevitable barrier to proper conduction. Those 'fins' on the impeller are *not* equivalent to the stationary fins on a normal heatsink because they are not bonded to the chip with a thermally conductive paste.
All he's really done is get rid of the heatsink entirely (except for the aluminium plate) and put a faster fan right next to the chip. Bravo.
That's a bit of an understatement. The highly turbulent (and thin) air layer should be a *much* better conductor of heat (orders of magnitude over still air). A question would be why does it have to be so *tall*?
The real question is does novel ==better?
I for one welcome our new rapidly spinning, sounding like an idling gas turbine cooled overlords.
The writer's description of how a conventional heat sink / fan combo works is completely backwards.
The article says that the finned block of metal's job is to draw heat out of the passing air.
No, not at all.
The CPU (or other component) generates, relative to its actual mass, immense amounts of heat (seriously: in some cases over 100W from a layer a few microns thick of doped silicon), and this heat is conducted to the hunk of metal. The heat then leaves the hunk of metal *into* the air, which is moved away either by convection, or if the heat burden is high enough, by a fan.
(yes, I know about heat pipes; all they do is move the hunk of metal away from the heat source. the actual heat sink is still cooled by moving air.)
is, I believe, to be read as "fins that dissipate, in the air, the heat contained".
If the author really believed the heat was sucked from the air into the sink then into the CPU, ... I guess we'd have noticed other strange things in the article. Don't assume people are that dumb.
1 its an impeller. we have had impellers before they are not dust and crud proof. the dust and crud collects on the inside centre and blocks the vanes.
2 Impellers generally spin faster and are hence very whiny from the impeller vanes not the bearings! (gigabyte made a cpu cooler a couple of years ago it lasted about 3 days in my rig).
3 This test is mains powered!? of course its going to be a powerful cooler! but will it work as well when scaled back to pc 12v and lower amps?
4 Air gap, all well and good in a lab now air gap meet smokers tar residue and cat maltings.. the end.
5 Air gap (ii) and thermal insulation are two terms that go well together air gap and thermal transfer are not! whilst a sustained heat source will permeate an air gap at a constant and sustainable rate, I believe that a rapidly changing heat source such as a cpu will suffer with excessive lag in the transfer responsiveness. so much so that a rapidly increasing core temperature would spike the base block temperature massively before the thermal transfer over the air gap can wind up to full capacity.
6 Boundary Layer. This device offers nothing new, it might be a carefully designed shape to reduce static air (not boundary layer effects) but this is also possible in traditional systems as sold by companies like Zalman BUT specialist shapes are often more expensive to manufacture hence we still have the cheap "it'll do" extruded aluminium squares, as the cost to effectiveness break point is adequate. is this device offering us cheaper efficiency? I doubt it.
7 High speed airflow, well because of the very high rpm this device generates a very high speed airflow which can reduce the thickness of the boundary layer, but this high speed airflow can be achieved in standard designs also, but at the expense of power (and psu heat) and silence!
These days PC's are going for larger quieter fans pushing big slow volumes of air, for good noise levels, This design is counter to that tack in all ways and has not been shown to offer any advantage on this consideration.
So will these work? well maybe in specialist no expense spared, dust sealed noise isolated kit.. but I don't use much of that.
I was sceptical at first, but reading comments above this looks to be an ingenius idea.
WRT "Several thousand RPM is going to sound like a jet engine"... Most PC fans I have come accross, at full speed, run at over a thousand RPM. Many of those on budget coolers or those which bundled come with CPUs run at 2-3k (I may be a little behind the times, but they did with the last one which came with a CPU I bought). I suspect that it will be able to be slowed, just like a standard fan, although probably not to the same extent (to maintain the air cushion). I guess it's something which would need investigating.
I can't beleive, though, the number of people who just seem to automatically poo-poo ideas like this. Whether they will work well in real life, or suit your specific needs, or will be economical to produce, they are interesting to read about.
Tolerances on that air gap are very tight, considering it separates two substantial masses - this seems to rely on a degree of precision manufacturing not widespread in consumer devices. Get that even a bit wrong, or just have too high a variation, and these will be difficult to trust.
It gets even harder when you scale it up to bigger things like air conditioners.
The benefit of a heatsink is that you get a period of grace if the fan isn't working to shutdown. With this method the CPU would fry instantly in the event of an impeller fail. This may be OK for larger and (less time critical) failures in air-conditioning etc but I think the metal block on the CPU is here for a while. its a lot cheaper to replace a heatsink fan than a CPU!
Surface area and temperature differential is still all that is involved.
Bringing more hot surface area into contact with cooler air.
Seems to me that this device gets its efficiencies from higher speed airflow which improves the differential.
Be interesting to see the actual for energy usage (with some proper controls).
Noise presumably won't be that much of an issue as this is currently a long way from the domestic market.
I could see a domestic application using a slower rotating heatsink passing through an air flow - the major price there would be space.
Of course with space to trade in a domestic box you could always increase the heat sink surface area by using both sides of the mobo
maybe domestic and SoHo minitower PCs do, but lots of corporate small form factor PCs are horizontal. Some SFFs work either way.
There seem to be a lot of readers who've never checked their existing fan rotation speeds (it's often shown in the BIOS, or HWmonitor or similar will show it if it's a Window box).
"There seem to be a lot of readers who've never checked their existing fan rotation speeds (it's often shown in the BIOS, or HWmonitor or similar will show it if it's a Window box)."
The issue is more than rpm, this system seems to rely on creating turbulence, that will increase acoustic noise. In addition this is a heavy lump of metal doing 2000rpm, rather than a lightweight plastic or ali. 'normal' fan, which means more energy and more destructive power.
Agree with Colin Millar, Peltiers are very much a last resort due to cost, poor efficiency and long term reliability issues.
Just one thing, I think the diagram shows the wrong rotation direction.. if air goes in at center, the fan will rotate anticlock... unless the arrows indicate air flow coming OUT of the disc...
these may have been answered, blame the authorization delay... :(
"heavy and necessarily sharp edge spinining flywheel"? that would be most good HS/fans then.. plus the *production model* has *fan guards* :/
"air gap for thermal transfer" - this is the SAME gap that exists in most fans!
As for the need for fans.. there are enough HUGE heat-sinks about, and if you are NOT doing a massive overclock, a fan may not be needed.. if you do, liquid colling is easier..
Dust? plenty of cases that include fine dust filters, so it stays outside, and 'positive case pressure' will ensure air goes OUT of the cracks, not in.. :)
about the possibility of using a Sterling engine to power a heatsink fan - after all, it'd be drawing it's power from the heat it needs to dissipate and the hotter the CPU got, the faster the engine would work. Unfortunately, it'd probably need a normal motor to get things moving until heat levels are high enough and there's probably several dozen other flaws I haven't thought of yet :)
Back to this device: dust getting into the airgap seems like a potential concern: there's plenty of stuff which can get into a 3-micron gap (http://www.engineeringtoolbox.com/particle-sizes-d_934.html). I'm not sure weight is an issue though - if anything, I'd expect a "Sandia" cooler to be lighter than a standard "metal+fan" heatsink: as it's more efficient, less metal is needed for the same "cooling" capabilities and it also doesn't need the plastic frame and fanblades of the fan (though it still needs the motor). And I'd assume that the centrifugal forces would keep the airgap size consistent, regardless of the heatsink's angle relative to the planet's gravitational pull.
On the other hand, the electric motor has to spin a relatively heavy chunk of metal, rather than a set of lightweight plastic blades. So you'll need a more powerful motor; together with the high-precision needed to mill the two pieces for the airgap, it may be difficult for a Sandia cooler to be competitively priced against a traditional heatsink setup...
As previously mentioned.
A simple case filter, plus positive case pressure = no dust clogging up the internals.
Five years on, my PC is still nice a clean on the inside.
Only need to clean the filter once every six months.
Big heatsink with very slow turning fan means very quiet too.
Just need to re-build with newer/faster components.
The rotating bit is predominately just the fan with a fancy impeller.
The mathematical modelling of the gap is questionable: Using pipe flow instead of flow between moving plates is entirely different in fluid mechanics and therefore heat transfer. Further, the graph shown doesn't apply within a boundary layer. And there are two boundary layers in shear between the moving plates in the cooler. That is nothing like pipe flow.
Treating the air gap as a conductive medium, based on the width of the gap is a gross fudge. I'm entirely unconvinced by the calculations.
Issues I see...
1. If the fan fails or begins to there is no direct contact slab of heatsink to act as a buffer.
2. Big heavy chunk of non streamlined metal spinning at high speeds can't be quiet.
3. Big heavy chunk of metal spinning at high speeds likely to cause significant vibration unless very well balanced and fitted - it could rip itself off the motherboard. Also, possible gyroscope effects?
4. Still as likely to get coated in fine dust as a normal fan. If that air gap gets compromised in any way then things start going wrong very quickly.
5. It's being compared with normal HSF combo's with slower running fans. What's the comparison against high speed fans that would blow far more air (and probably be as loud).
Very dubious about ALL of the claimed benefits, but, since the sponsor is after all the US Department of Energy/Sandia National Labs, let's focus on the topical " + Provides increased energy efficiency". Their "drastic improvement in aerodynamic efficiency" supposedly "translates to an extremely quiet operation", no less. The report and patent dangle even more prospects of violent and dramatic upheavals to the state of the art in CPU coolers.
Most CPU coolers use an axial fan for good reason: axial fans give a high flowrate and small pressure rise, which is appropriate to the duty of CPU and casing fans. A typical axial 80 - 120 mm CPU cooler fan absorbs 1.8 - 3.6 W peak power (~0.15 - 0.30A at 12V). Centrifugal (radial) fans are generally used for low flow, high pressure duties.
What we see here is a centrifugal fan with (lots of) thick, cantilevered vanes and no top shroud. It should be good to absorb at least as much power as a similarly sized axial: however, it is very unlikely to function at all effectively in pumping, and nothing like as efficiently, as a properly designed centrifugal. Poor aerodynamics generally also lead to flow-induced noise. And BTW, there will be pesky boundary layers all over the front and back surfaces of the numerous vanes, happily producing friction and absorbing power. For a 100 mm diameter fan, even at 5,000 RPM, these BLs will be nice and thick, helping to block the radial flow in the channels between the vanes.
Then, on rear face of the impeller, there is that magic fluid-filled gap between the rotor backplate and the stationary base plate with the heat source. The flow patterns in this space are complex, and depend on stuff like scale, shape, speed of rotation, fluid properties, surface temperature distributions... Leaving the complexities of the heat transfer aside, there will also be friction acting on the rotor, which requires additional power from the fan motor. Given good data (*) - yes Sandia, they're already out there - you can calculate this windage, and for the Sandia specs (100 mm diameter, 5,000 RPM, 25-micron air gap) it comes out around 2.9 - 3.6W ( air temperatures 20 - 100C; or about 10 - 12W for the 5-micron gap they seem to aspire to for better heat transfer. This ignores sealing issues between the rotor and base plate.
So the motor power requirement of the Sandia Cooler could be roughly at least DOUBLE, maybe FOUR TIMES, that of typical state-of-the-art CPU cooler. There's also the niggling issue of where that friction heat in the air-gap goes - it ADDS to the thermal load that must be transferred out to the impeller and dumped into your room.
That relatively heavy metal impeller (all those fat blades) and high rotation speed, would need balancing to avoid vibration and possible contact or rubbing. Also, both to prevent vibration and hot spots, maintaining a uniform 5-micron cavity gap would be important (and tricky).
"Simple, rugged, cost-competitive" - "Dramatic increase in cooling performance without resorting to exotic methods???" - NOT.
Hard to discern real progress there for CPU cooling, no matter how you paint it, and IMHO apparently not a good technical case for throwing more cash into.
* http://www.esdu.com/graphics/dataitem/di_07004.htm -- Engineers with hard-hats only, please
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
