The Sandia Cooler (2016) | Hacker News Reader

The Sandia Cooler (2016)

ip.sandia.gov

82 pointsRichardHeart40 comments

ThermalTake has actually released one based on the Sandia design (and Cooler Master flirted with this one), but it's not really living on its promises. It works, but there's a lot of non-Sandia designs with comparable size that works as good or even better than ThermalTake's Sandia-like design.

Reviews:

Gamers Nexus: https://www.youtube.com/watch?v=u2tCnjb6lp8 (video), https://www.gamersnexus.net/hwreviews/2806-thermaltake-engin... (text)

Quote from Steve: "Ultimately, the Engine 27 isn’t a bad cooler – it performs about the same as similarly sized products, so it’s not some crime against humanity. That said, it’s priced significantly out of its performance bracket, and the high-pitched whine at max RPM can get a bit irritating. You’d want to run this at a lower RPM to account for that."

Linus Tech Tips: https://www.youtube.com/watch?v=oCghRn2Zae4

Quote from Linus: "The Thermaltake Engine 27 gets my "Better than Nothing" award for working better than I expected given its size, making it a great option if you don’t have anything else that will fit."

I have one of those I used in an SFF build, and I like it - it's pretty as well as low-profile.

But it's no revolution, and only rated up to 70W.

(UK Product page - https://uk.thermaltake.com/engine-27.html)

The rotating mass of the heat sink would need an enclosure for safety reasons, negating most of the airflow advantages. "Transfer heat through an axle" sounds like one of those "draw the rest of the owl" steps too; no idea how that's supposed to work.

*edit: Maybe a hydraulic motor where the fluid is turning a big impeller? Torque converter with an open side, kinda thing...

Perhaps the CPU is rotating too? Wait, perhaps I should patent this idea first.

The amazing part of this is that each individual core makes contact with each region of the (obviously stationary) memory drum once per revolution, providing dynamic cache adjacency.

Mel nods in approval: http://www.catb.org/jargon/html/story-of-mel.html

Make sure to incorporate an anti-twister mechanism for extra weirdness points: https://en.m.wikipedia.org/wiki/Anti-twister_mechanism

Right. High thermal gradients through ball bearings or roller bearings are tricky, and you don't get much contact area for thermal conduction in the bearing (meaning most bearing applications need to be mindful of cooling). (Pro tip from MIT 2.72: make sure to position bearings such that thermal stress from the shaft undergoing more thermal expansion than the rest of the machine doesn't jam the bearings.)

I guess you'd want a bearing with sodium-potassium alloy (hazmat!) or mercury (hazmat!) or something. Or... make the casing for the shaft into a heat pipe with ammonia or alcohol vapor.

What about old style Babbitt bearings

You're right. Plain bearings provide plenty of contact surface for thermal conduction. Though, thermal expansion will tend to increase clearance and decrease contact area (ignoring differences in thermal expansion between dissimilar materials).

A linked video claims that the commercial version attempts to transfer heat from one rotating metal surface to another (the the part that connects to the CPU and the bottom of the spinning thingy) separated by an extremely thin layer of air. This is "supposed to" transfer heat as if the metal is touching but "doesn't see to work very well in practice"

As described by the lab, the Sandia cooler predominantly transfers heat across an air-gap (maintained via self-generated fluid bearing) between the base plate and rotating impeller.

As described, they claim that if they can maintain the air-gap to 10um, then they get a net win since the specific (per surface area) thermal resistance is sufficiently low (due to only being 10um), and the transfer surface area is comparatively enormous. Nevertheless, the air-gap still forms the dominant component of their overall thermal resistance.

Reading through their posted development report [0], it's apparent that they were having trouble hitting that 10um target with their designs.

Not exactly a surprise that this is the most difficult part. Not surprising that the "commercialized versions" (see other posts) don't use this mechanism.

[0] https://ip.sandia.gov/techpdfs/Development%20of%20Sandia%20C...

I was working on a so called finless cooler around the time this first rose to prominence. The idea was that a toroidal vortex would give better heat transfer than a lot of fins subject to viscous shear.

Our cooler was simply a folded aluminium sheet with a fan placed into a hole in the top. We had a somewhat similar couette type flow under our motor hub directly over the heat source. We had a modest 15% gain over a reference GPU cooler which was almost entirely due to the better thermal conductivity of the aluminium sheet over the cast heat sink in the reference design.

It turned out that all that matters is cost, nothing else.

Immediately popping into my mind: what rotates it to facilitate pumping, and also, if the top rotates, how is the thermal transmission between the baseplate and the rotating element being facilitated? It's either have to be through the axle, or you'd need some sort of lubricious, yet thermally conductive compound for the rotating element to flow through, which, not gonna lie, sounds pretty magic to me.

Also, I'd need to look into the mechanics of this cylindrical impeller bit more. Boundary layers don't go away magically in laminar flow conditions. They might shrink, but they don't disappear. I also look at the center of their prototype, and all I see is a debris accumulation point that will become more and more obstructed over time in high debris concentration air. There isn't that much preventing dust build up on the top too, which I think may contribute to further build up.

Noise, no comment, except I know that if you've got spinning parts you've got harmonics and vibration, audible or not.

The burning question for me though, is how does it pan out in test designs. If it keeps stuff cooler under operating conditions, with better MTBF than what we're traditionally using, screw it, it's better.

Especially since in a sense you're combining two distinct parts into a single one, which would in theory simplify fabrication. However, that looks to be all metal, so it may not be cheaper than a fan static heat sink combo.

Be a fun thing to test and put through it's paces to be sure.

> what rotates it to facilitate pumping

A motor.

> how is the thermal transmission between the baseplate and the rotating element being facilitated

Through the micro-meter thick air bearing.

> The “Sandia Cooler” architecture simultaneously eliminates all three of the drawbacks of conventional air-cooled heat exchanger technology

... and replaces them with brand new drawbacks that we aren’t sure how to mitigate

Since we're talking about CPU coolers, I hope my comment is not OT: what's with the recent popularity of tower coolers? They seem to pop up a lot in PC build discussions, lately. Am I in some echo chamber? Do they provide any obvious advantages over "traditional" coolers?

Without really knowing what I'm talking about: It seems logical to me that a tower cooler can be more efficient because of less restriction in air flow (air blows straight through + can have one fan on each side) and possibly more efficient heat transportation through several heat pipes into a generally larger surface area.

This, in theory. A heat pipe is much more effective at moving heat than conduction. It's also what laptops use.

Big issue is that heat pipes usually work best upright and effectiveness (efficiency?) suffers in other orientations.

I don't think it's recent. The first tower form factor cooler I bought was in 2011 or thereabouts.

Part of the reason I bought that one when I did was because the stock cooler wasn't effective and the fan was incredibly loud (poor heat dissipation, small fan, high RPM). The tower coolers have a larger surface area and ship with larger/slower fans, reducing noise. They're probably still popular because they work, and they're the only air coolers I buy now.

Top-blowers have been relegated to the low-cost / bundled category for ~15 years or so. Tower coolers are much better and fit better into the overall airflow of almost all cases.

There are no high-performance top-blowers.

Pretty happy with my Noctua CPU cooler - I don't think it gets better than that with performance/noise for the price point (sub $100!) - has anyone found better coolers than those? Interested because I wanna build another PC soon!

But they don’t fit into the space requirements this cooler is designed against.

I went with an AIO (all-in-one) water cooler for the first time in my latest PC build. Many PC cases today have an area in the front or at the top for a radiator and its fans. Mine is in the front and has 2x120mm fans that suck air in. I have the same amount of fans blowing air out (back+top). All four fans are running on very low RPM so it's very quiet.

I really recommend this setup. I think 2x120mm is a sweet-spot for price/performance when it comes to radiator size. Smaller will require a louder fan. Bigger is obviously better but more expensive and might not fit your case.

I used to have a big air cooler. One benefit that I didn't plan for is that I'm no longer worried about the stress on the motherboard mount from the heavy piece of metal hanging from it. Another is less stuff in the way since the bulk of the cooler is the radiator and fans off to the side.

I never understood why one would like to place the radiator at the inlet, thus heating the air inside the case. The hottest part is still the GPU (even if Intel tries their hardest to change that), and by placing the radiator at the inlet you are heating the air to cool the gpu with an additional 50-150 watt.

I actually placed it up top first, blowing out. But then I looked up some tests and recommendations and apparently front blowing inward is better. Can't remember the exact reasoning but I remember being convinced enough to move the whole thing. Temps are great now but I never did a comparison.

I assume my way results in better cooling for the CPU because the radiator gets fresh air instead of GPU-heated air. Maybe it's just about prioritizing CPU over GPU (or having more headroom for a few extra degrees on the GPU).

AIOs don't seem to me like they're worth it, because most CPUs can be cooled by a good tower cooler just fine, and those that don't really only thrive on a custom loop instead of a 240 mm rad. For gaming - well CPU thermal load tends to be modest, so it doesn't matter.

AIOs for GPUs largely don't exist, and that's the component I'd watercool first, not the CPU. So we're back to custom loop there.

> most CPUs can be cooled by a good tower cooler just fine

Water coolers are definitely still an enthusiast part in my opinion. If you're satisfied with "just fine" then no, it's probably not worth it. It will give you better cooling per decibel but probably worse per dollar, at least on stock clocks.

If you want to mess around with overclocking the advantage seems to become a bit bigger for AIOs as you crank up fan speeds.

Every AIO I've ever owned has stopped working not too far into its life. I remember being shocked and dismayed the first time I had one fail, and simply upset when the second one failed. After the second failure, I bought a Noctua NH-D15S and that heatsink has happily survived 4 builds. (Only getting replaced after I switched to Threadripper, which required a different Noctua heatsink. But don't worry, the NH-D15S still lives on in a friend's PC.)

In my opinion, there are two viable cooling solutions: a high-quality heatsink and fan, or a custom water loop. AIOs just don't have the performance to make the failure rate acceptable.

There was this review done a while ago on a product similar to that design, but with some changes made: https://youtu.be/oCghRn2Zae4

I've had some interesting ideas related to heat transfer that I haven't seen anyone try.

First is using suction to remove hot air from deep inside a heatsink. This seems like it would work better than just randomly flinging heated air around.

The second involves actually moving the heatsink. Imagine something like the Sandia cooler, but as a centerless ring about a foot in diameter, and spinning at only 1 rpm. The heat source would be at a spot under the spinning ring. A bushing would transfer heat by touching both the CPU and the ring. Each part of the ring has plenty of time to cool down a bit by the time it comes back around to touch the CPU again. Basically there is always cool metal on the other side of the bushing.

These are both concepts I've thought a bit about, but just don't have time to work on.

I'm not clear on how high speed is supposed to eliminate fouling. Sure, particles are less likely to adhere but you're also increasing the number of particles.

As the speed increases, the percentage of particles that stick declines quicker than the percentage increase in volume of air (and thus particles that may stick). How can you remove dust from a fan? One way is to increase the flow of air to “blow” away the dust. Same principle here.

Just from observing dusty PC fans being turned up to higher speeds, I think the effectiveness would be limited. Most of the cleaning is accomplished from having air flow impact the surface at a different angle than usual, and at a much higher static pressure.

Because it doesn’t work https://youtu.be/oCghRn2Zae4

It would be interesting to see how that would work if you made bigger versions. It was a lot smaller than the others, and I'm curious if it would have to be as big as them to match them or if it would match them somewhere between its current size and their size.

Also, I wonder how much details matter? That certainly looks quite a bit different from the Sandia unit. Sandia was claiming a large increase in efficiency over other units, and usually you only get that kind of increase with either a fundamental breakthrough or by using known methods but getting it all tweaked and tuned just right. I think Sandia's is the second kind.

The conclusion of the video is that it does work, but the performance is not as good as the larger, traditional designs.

The performance is below low profile traditional coolers like the noctua too, it doesn’t work as in live up to it’s promise.

The Noctua was 37mm, the Sandia was 27mm. I don't have performance per unit volume stats, but I suspect the Sandia would perform well there.

Performance per dollar, though... maybe not.

It looks like that design has diverged significantly from the Sandia one, probably because they couldn't manage the tolerances the Sandia design needs. Given that it's a first attempt to bring the concept to market, and that it appears to be significantly compromised, I don't think "it doesn't work" is a fair characterization of the actual Sandia design.

It might well be completely impractical for the commercial CPU cooler market, but it also might be the case that a bit more R&D and tooling up could get us a really effective new cooler design. Improvements could include [perceived] noise reduction as well as better efficacy.

Reply on news.ycombinator.com