Hacker News new | past | comments | ask | show | jobs | submit login
AirJet – Solid-state chip for active device cooling (froresystems.com)
274 points by peter_d_sherman on Jan 22, 2023 | hide | past | favorite | 110 comments



I am willing to pay for it assuming it works without some pitfalls. They actually have a working prototype [1] to showcase in CES 2023.

To put this in perspective, with this and TSMC matching towards 14A / 1.4nm we could have machine that is roughly 2X the performance of Macbook Pro with M2 Pro at the thickness and size of MacBook Air by the end of this decade.

I am also hoping this will be more reliable than current Fanless cooling solution which tends to degrade after a few years.

[1] https://www.youtube.com/watch?v=YGxTnGEAx3E


>current Fanless cooling solution which tends to degrade after a few years.

What do you mean by that? I've had both PCs and laptops that were completely passively cooled and there was never any degradation.

Lots of different cooling solutions, the issue is there isn't a perfect one that works for all wattage while having no drawbacks. Either it's excessive noise, limited performance, large size, high price, complicated maintanance and so on.

The lofty claims they make are suspicious, I want to see this in action on a high performance chip without any other cooling.


> fanless cooling solutions which tends to degrade after a few years.

Thermal expansion/contraction between cooler and chip itself "pumps" out the thermal paste which over time degrades cooling performance. If the system is running 24/7 its probably not as noticeable because you don't get the frequent heating/cooling you'd see on a device that gets power cycled often.

Nothing a repaste job wont fix.


Are there designs that don't use thermal paste? As I understand it regardless of solution you need paste or live with significantly reduced conductivity. That goes for any system - fan cooled, liquid cooled, passive,...


Graphite thermal pads are available for consumers (e.g: Thermal Grizzly Carbonaut, IC Diamond Graphite) which offer similar performance (within 5°C) to paste without this issue.


Does Apple use paste or pads in the MacBook air?

Edit: to answer my own question it looks like Apple uses thermal paste but the cooling in the M2 MacBook Air looks extremely flimsey to the point of being almost non-existent. I doubt it will be affected much at all by this.


You're looking for phase change materials (PCM), which soften at the desired operating temperature. Prevents pumping out because it's still mostly-solid.

E.g. https://thermalmanagement.honeywell.com/us/en/products/therm...


>> Are there designs that don't use thermal paste?

Yes. Glue. Paste is for when you might want to one day separate part from heatsink. If you don't care about repairs, it is perfectly reasonable to just glue them together with thermal adhesive.


is it theoretically possible to have cooling paste be jelly like? so it just returns to previous position after cooling down? or its just to irrelevat problem to invest in it?


>is it theoretically possible to have cooling paste be jelly like?

That is already available and used in some Laptops. Look up Honeywell thermal pad.

The problem is as always, cost.


I thought thermal pads are generally thicker and less performant


There are thermal pads that perform equally or better than thermal pastes, due to their design. In particular, something like https://thermalmanagement.honeywell.com/us/en/products/therm... can be better than a good thermal paste like https://www.thermal-grizzly.com/en/products/16-kryonaut-en. The thermal conductivity is improved due to the use of a phase shifting material. Reliability is improved due to the fact that it will naturally return to solid without cracking during heat cycling.


It is, and thermal paste manufacturers need to balance performance against thermal cycling effects.


They seem to be targeting certain devices (laptops and gaming handhelds), not top of the line chips.


I want to see it survive 6-12 months of use in a typical home/office environment with dust.

The immediate thought looking at the design was "this probably has lots of tiny little channels, and those channels are going to get clogged with dust."


In a video from PCWorld they mention that their system has more backpressure than fans, which means you can put filters in the device.

Time will tell, I'm happy someone is pursuing this either way.


> roughly 2X the performance of Macbook Pro with M2 Pro at the thickness and size of MacBook Air by the end of this decade.

besides compiling code, training models, or rendering videos, what else really benefits from this level of performance increase? like what "need" is there at consumer/professional levels? obviously better efficiency is always good for a ton of reasons. just curious if there's a "need" where we are wishing "man, wish CPUs/GPUs were 2x as fast!"

what doors does this open that aren't already open? i feel like i wouldn't really notice if my CPU doubled performance wise tomorrow.


All existing air cooled laptops and desktops would benefit because they can keep the same performance and operate much quieter. A colleague's laptop was nicknamed the hoverbook, because it always felt and sounded like it was going to take off whenever he typed 'make'. And heat isn't comfortable on laptops. Plenty of models you need to be careful to not burn yourself if you actually try to use it on your lap.

Oh, and gaming of course. Which has driven most consumer level performance improvements. Laptops haven't gotten 4k gaming sorted yet and already people talking about 8k.... and have you seen the size of the newest NVidia cards? Literally the size of a brick and often in need of scaffolding due to the weight.


It would enable more background processing of files (image processing, etc) without seeing an impact on performance from the users perspective.


M2 Pro performance in ~7 years?

I don't think, Apple will sit idle until then.


Apple can make advancements, i don't mind. I'll just be happy to get some of the Battery:Performance ratio of Apple in an actual PC.


I'm a little confused by these two quotes.

"The first ever solid state thermal solution" "Inside AirJet are tiny membranes that vibrate at ultrasonic frequency"

How is it solid state if it has moving parts?


Yeah it's fluffy terminology. They use that to distinguish from usual mechanically assembled devices. With rotating parts, friction, etc.

This is more like a one part cast/deposited with some degree of elastic deformation.

Supposedly a lot more reliable.


Likely a piezo or mems device. Both of those use material flexture or deformation to achieve some output.


It's a mems device, not piezo.


Do piezoelectric fans count as being solid state too, then? If so, this one wouldn't be the first.


"Do piezoelectric fans count as being solid state too, then?"

Not in my book. The first thing that comes to my mind when I hear "solid-state" and "cooling" in a sentence is "peltier element". The name "AirJet" gave it already away though, that this is likely something different.


> The name "AirJet" gave it already away though, that this is likely something different.

They mention that those are MEMS membranes in the presentation [1]. But it also sounds like Airjet developers (or marketers) count MEMS membranes as solid-state, and piezoelectric fans as not, while I wonder where the line is.

[1] https://youtu.be/YGxTnGEAx3E?t=146


> "They mention that those are MEMS membranes in the presentation [1]."

This video was really insightful, thanks. Especially interesting was, that they use an effect called jet impingement that is also used in jet engines.

> "But it also sounds like Airjet developers (or marketers) count MEMS membranes as solid-state, and piezoelectric fans as not, while I wonder where the line is."

I think I can imagine where this distinction comes from. MEMS devices can usually be manufactured in facilities and with materials and processes very similar to those used for producing semiconductor chips like processors. Piezoelectric elements, on the other hand, require special materials and their production doesn't fit very well into the common semiconductor manufacturing process. Now, the step to call everything traditional semiconductor industry related "solid-state" is not that big.


It's quite normal to be confused what solid state means. For example vacuum tubes don't have moving parts and aren't considered solid state.


I always interpreted "solid state" to mean "neither clanky nor sloshy"


Well for electronic components your basic three chioces for states of matter are solid, liquid, and gas.

Silicon crystals are naturally solid-state, car batteries and most can capacitors are liquid-state with their electrolyte, leaving many vacuum tubes no other choice but to be gaseous-state.

Now these air chips don't have any electrolyte so that rules out liquid, and all the components are solid materials so the component itself is actually solid-state, and so are other MEMS devices, even though they have moving parts.

Like an electric motor which is just iron, copper, and steel, a key solid-state electrical component.

It's still just a motorized fan.


> "Well for electronic components your basic three chioces [sic] for states of matter are solid, liquid, and gas."*

Three states of matter is basically a fairy-tale we use as a first, easily understandable basic working model of matter. In reality there are so many states that Wikipedia has a dedicated page to list them: "List of states of matter" [1]

This is also the reason why every physicist cringes when the usual popular science mags announce the discovery of "a fourth state of matter" about once every year.

> "Silicon crystals are naturally solid-state"

Solid room temperature silicon can be in a crystalline state or an amorphous state. While amorphous silicon may appear solid it has many properties of a liquid and thats where the word "liquid" in LCD (Liquid Crystal Display) comes from. So, its not that easy.

[1] https://en.m.wikipedia.org/wiki/List_of_states_of_matter


There are probably transition states that are still to be identified.

I'm trying to narrow it down to the obvious choices ;)

Physicists are the experts on this kind of thing and I just go along with the choice they made when they started calling electronic components "solid-state" to begin with.

My apologies for being so controversial.


That sounds completely wrong? We don't consider hard drives or fans "solid state", even though they're not liquid or gas. The difference between what's considered "solid state" and what's not seems to mostly be in whether the component has major moving parts.


SSD means the data itself is stored on semiconductor (memory) chips, instead of a physical magnetic pattern on the mechanical recorder of a HDD.

Other than electrolytic capacitors, both peripherals are electrically solid-state.

Even the helium in some HDDs is for physical purposes, not as a gaseous electrical element.


"Solid state" in this context is used to include SSDs but exclude HDDs. The definition you use which considers HDDs "solid state" is not the definition that's relevant to this discussion. The physical state of matter doesn't determine whether a component is "solid state" in this context.


In my lifetime, "solid-state" first referred to "built with transistors rather vaccum tubes", where vacuum tubes are not 'solid' because their insides include a partial vacuum and transistors are 'solid'.

Extending "solid-state" to mass storage devices that have neither a rotational device nor the necessary internal gap (air or partial vacuum) for either rotation or actuators feels natural…to me.


It's built similar to the accelerometers and the compass you have in your phone. MEMS devices are considered solid state.


Wasn't dlp labeled as solid state even though it had spinning mirrors?


they are not mechanical parts, a speaker also vibrates but is solid state


Speakers are not solid state by any definition I’ve run across (except for piezo ones, anyway).


Then a solenoid with a piece of membrane over it would also be solid state.

Why stop there, a servo with a piece of membrane would also be solid state.

Why stop there, a motor with a membrane..

This definition makes no sense to me.

The above are electromechanical parts. Piezos are somewhat an exception because electricity directly leads to a parts deformation without magnetic field moving a component of that part in an electric field. In fact a piezo is just one part with two contacts.


>Then a solenoid with a piece of membrane over it would also be solid state.

>Why stop there, a servo with a piece of membrane would also be solid state.

>Why stop there, a motor with a membrane..

>Piezos

All Correct.

Even when they are electrically powered, hydraulic and pneumatic servos are not usually liquid-state or gaseous-state electrical devices except for any vacuum tubes or electrolytic components.

When the electricity flows only through solid materials like copper, carbon, or silicon, and doesn't (intentionally) flow across things like liquids, air gaps, or evacuated spaces, that's solid-state.


"Solid state" has two meanings:

(1) The gaseous / solid meaning relevant for tubes vs. transistors.

(2) The mechanical / no moving parts meaning.

Two examples of devices that use the second meaning:

https://en.wikipedia.org/wiki/Solid-state_storage

The SS in SSD means it doesn't have moving parts (platter, head) like a hard drive.

https://en.wikipedia.org/wiki/Solid-state_relay

The SS in solid state relay means it's not electro-mechanical like a relay.

In this case (the cooling chip), the second meaning is obviously the correct one because it is being compared to fans, which have moving parts.


> All Correct.

Says who?


Says someone who incorrectly thinks "solid state" here refers to the state of matter and thinks anything that's neither liquid nor gaseous is "solid state".


But they were so confident about it!


Maybe falls in the MEMS category. For example we consider accelerometers to be solid state but they aren't.


The datasheet[0] shows the device moving 8.5 watts at 85 Celsius. That doesn't sound like it's going to make laptop throttling a thing of the past. Maybe I'm not understanding something.

[0] https://uploads-ssl.webflow.com/6387c57559192a648478384e/638...


That’s for a single AirJet. They suggest using 3 of them for a 15” notebook.[1] That is enough for most ultrathins.

1: https://uploads-ssl.webflow.com/6387c57559192a648478384e/638...


So the stack is CPU, heat spreader, array of these products? It'd be nice if the datasheet showed performance curves across temperature and pressure as well.


The examples I've seen don't use a heat spreader, interestingly. A video demo had the CEO saying they depend on the velocity of the air to remove the heat quickly, so no heatsink required.


It’s a bit confusing but from a YT video it seems the device still need a coldplate + heatpipe, and the airjet is stuck on the heatpipes themselves replacing cooling fins. The airjet itself blasts the air into it’s own internal coldplate from what I understand.


That is correct - in response to the question about whether heatpipes could be removed, the CEO mentioned that their requirement is uniform distribution of heat on the copper plating on the airjet and putting it on the heatpipe helps achieve that so their plan is to stick it on the heatpipes.


That makes a lot more sense to better handle fluctuations in load.


Something like an M2 MacBook Air, which currently has no active cooling, would be quite a beast with 8.5W more headroom.


seems iffy, might be useful alongside traditional cooling.

1. high frequency stuff to drive this thing might make noise an issue. 2. wouldn't it have to be integrated into the chip your cooling to optimize the effectiveness? 3. the wording leads me to believe their "performance" is cooling to noise ratio and not just cooling.

IDGAF about noise as long as it doesn't sound like in a server closet, i just want to keep things under 90c with a margin for hot days, crapped out a/c.

this is just a solid state fan. cool for sure, potentially very useful (industrial/embedded will probably love it), but marketing needs to calm down.


In the PCWorld interview on YouTube they even showed an unbranded Intel laptop cooled by these chips and also mentioned that multiple device manufacturers are planning to put this technology inside their products this year. So I highly doubt this high frequency noise is an issue. They also explained that the traditional arrangement of having a vapor chamber between the CPU and the active cooling to move heat also works with this design without losing efficiency, and this way they also can increase the cooling capacity by putting more chips next to one another.


What's wrong with the marketing? It's marketed as solid state active cooling, where are they suggesting that it's more than that?


It’s not enough to extract all the heat from a processor but it may be just enough to keep it from getting hot enough to throttle.


Their marketing is good and i hope the tech grows to pro laptops and desktops. But the marketing needs to be taught a bit of math.

In the 'how airjet works' video, they say a 50W processor is throttled to 10W and airjet brings it up to 20W so it's a 100% performance improvement for the processor.

Its more like the processor is only 60% worse and not 70% worse


Is there any data on how long this last? Having a perfectly working laptop die because some membrane lose elasticity would suck


Worst that will happen nowadays is that it would throttle heavily.

I broke the fans in my laptop through overzealous cleaning and when I turned them off completely to not hear the rattling the laptop generally still worked - just forget about any heavy tasks. CPU: AMD 4800HS.


"The first ever solid state thermal solution"

Peltier coolers have been around a while, but aren't very efficient. How's this compare?


Are those really considered active cooling? A pettier element just moves heat from one side of a plate to another, right? You still need to dissipate the heat on the hot side.


It's _actively_ moving heat from one location to another. What you do with that heat at the other location is another matter.


Damn it, my phone auto-"corrected" peltier to pettier and I didn't notice before now, after the edit window. Why is phone autocorrect still so broken?


How does dust not get jammed in that slit?


They talk about back pressure in this interview and basically say that it can produce enough pressure to pull air through a filter that would keep dust out. https://www.youtube.com/watch?v=YGxTnGEAx3E


Dumb question, but wouldn't the filter just end up getting clogged instead?


First bigger surface area two you can run it backwards each time you power it on and the filter increases time between automatic cleaning and helps with dusty environments.


I don't think this solid state fan can be run backwards, if I'm understanding how it works correctly


Can you clarify? I'm not sure this addresses the parent poster's concerns.

To add, I'm wondering about the susceptibility of such a system to the use of things like compressed air. High-velocity particulate + tiny spaces = potential for jamming/destruction of the components?


Forget cooling - I want to mount some larger versions on some boots and start flying like Iron Man!


At a smaller scale, it could be a combination VR headset and hairdryer.


I am thinking of all the things that this could change, but all I can think about is my loud 8” ultraportable Chuwi Minibook, with a Intel Core m3-8100Y, and how much I hate that fan.

Never mind that, this might be what’s needed to create a 1” thick Core i9-xx900HX 8” desktop replacement class UMPC that will most certainly be a worthy replacement, assuming they certify it with thunderbolt 4.


> my loud 8” ultraportable Chuwi Minibook, with a Intel Core m3-8100Y, and how much I hate that fan.

Isn't it possible to open it up and replace the fan with a Noctua?


It's a minibook and I don't think Noctua sells fans tiny enough to fit those unique cooling assemblies.


One of Noctua's products is a giant, passively cooled CPU heatsink. It juts out of the case.

They were kind enough to ship the backplate to my motherboard, even though that should have come with the board originally.

I'm glad I could find something that might help me run a quieter home server. Really appreciate their support.


If I understand the mechanism, LTT did a video on similar albeit larger more expensive version of these a few months back. Had some drawbacks IIRC but maybe the chip design addresses that.

https://youtu.be/NY-gA_zA_os


It's not a piezo according to other commenters.


I remember seeing the precursor to this some 10-12 years ago. Can't recall if it was the same company or not. The prototype looked like a block of very densely packed sheets of carbon fiber - imagine a typical car radiator but with finer and thinner fins made of some black plastic-like material - and once energized the sheets would oscillate in a way such that air got sucked through the block at impressive speeds. The prototype wasn't particularly silent.

On the whole this idea doesn't seem like it's going to do much for laptops now that Arm is entering the scene. Maybe it'll make a good solution for noisy graphics cards.


Related:

"Solid State Active Cooling Could Revolutionize Thermals"

https://www.youtube.com/watch?v=YGxTnGEAx3E


Interesting, I had skipped over that video several times because I thought it was going to be talking about peltier coolers. This is very cool.

I can't wait to get a thousand of them and build a silent vacuum cleaner.


I think they're using piezoelectric cooling - https://youtu.be/Rn6qVv9HzHc


In the CES video [0] they are stating it's MEMS.

[0] https://youtube.com/watch?v=YGxTnGEAx3E


I design low volume specialist equipment, and would love to try these.

Problem is I can't even talk to manufacturers at my volumes, and this probably won't be on Digi-Key/mouser/RS etc for years.

If I were them I would get these out to us little guys through those distributors so more engineers in general are familiar with them.


I'd be interested to see how shock proof they are, along with how well they stand up to fatigue. I'm assuming they are silicon flappy-bits, which tend to be fairly hard but fragile.

However there are watch people using silicon as balance replacement, so perhaps its more robust than I thought.


This has great potential for the niche sorta between passive cooled laptop and active laptop. 5W even stacked isn't a lot...unless compared to passive then it suddenly looks gigantic



I don't think it's a "dupe" if it's a different link. It's more like a "see also"


How does 1 watt compare to fans? Battery life is usually a priority over noise.


A case fan is around a watt[0] (0.6 - 6 depending on size and rpm) so I imagine it's in the ballpark, might even be a improvement.

[0]https://www.buildcomputers.net/power-consumption-of-pc-compo...


Funny that they clearly use a macbook as their "generic spinning laptop" image in the product video since this is sort of the one category of general computing device in which thermals seem to be more or less a solved problem due to the characteristics of the Apple Silicon chips.


They're solved and not solved. I love my M1 MacBook, but I also know that Apple worries a ton about thermal design. One of the big problems with the iPhone mini was dealing with the thermals in such a tiny package. While The M1 and M2 chips are allowing Apple to create great fanless and fan-lite designs, I'm sure Apple would love to be able to squeeze more CPU power into a MacBook Air or be able to offer even faster processors in a MacBook Pro by using more power.

Intel has been showing off "laptop" chips that are faster than Apple's. I put "laptop" in quotes because Intel often benchmarks their 45W "laptop" processors that would never fit in anything like a MacBook Pro or most other laptops.

Still, I'm sure Apple would love to be able to increase the thermal limit on their designs. Maybe even an iPad Pro could accommodate one of these (it looks like they're around 2mm thick).

Apple has done wonders to create processors that work extremely well within low power/heat constraints. I still think they'd love to be able to remove more heat and give them more room for their designs.


I thought the main difference between the studio and mini was the size of the heatsink allowed higher performance. It seems like if they could get the thermals better, the studio could shrink down to a mini size.


Responding to the (surprisingly passionate?) comments below – this word choice was poor, I should have probably said something along the lines of "this is one of the products on the market that is least subject to major thermal throttling issues, relative to any other laptop that they could have chosen for this demo video."

The main point being, out of all the laptops out there, I thought it was entertaining that they picked the one high-performance laptop brand I am familiar with that does not get hot when I use it.


Thermals are never a solved problem. System designers simply use up what is available (and usually won't stop there).

It's like saying processing speed is a solved problem. Engineers/scientist don't have any problems finding productive ways to use more processing speed.


> thermals seem to be more or less a solved problem due to the characteristics of the Apple Silicon chips.

I am actually at a loss for words. So I'll ask a question instead.

Do you really believe what you just said or do you have no clue how CPUs work and passive cooling works?


this seems like it might be interesting for cooling telescope optics?


Very cool. (Pun unintended)


The only red flag being set off for me is how has Apple, Intel or AMD not bought these guys up already?


If you have the patent on this tech and the ability to manufacture them in bulk why would you sell? There is a huge potential here and selling it to one of the big guys to just then sit on a few billion at home seems quite a boring life.

Apple can easily license or buy from them which is what they do with other tech. Apple does not make OLED panels themselves nor do most that use OLED panels in their products.


> selling it to one of the big guys to just then sit on a few billion at home seems quite a boring life.

Depends for who. If you are inventor, you can sell your invention and then go on working on your next invention without worrying about cash instead of trying to be a businessman doing business.


"Hey Apple, you can have AirJet in your laptops, but we want 30% of your revenue otherwise we sell to Intel"


We've seen the product, now time for the evidence. If they're right about their high pressure air stripping the insulating boundary layer from heat sinks it should be demonstrated with cooling benchmarks.


...the ones they post on site linked?

Either way we'd have to wait till people in industry get their hands on test models


Is this an Indian company? Frore could be a play on crore which would make sense, but otherwise, good lord. Reminds me of that 30 rock joke about the “rural juror”. Feels like the corporate equivalent of 3-yo Timmy dressing himself for the family photo.


The two founders have indian sounding names [0] at least, so it might be the case that this is what they are going for.

[0] https://www.froresystems.com/team




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: