Frore Systems has developed a solid-state alternative to a laptop fan. The air flow is generated by a vibrating membrane which resides inside a hollow heatsink. The heatsink is less than 3 mm thick.
Extremetech has an article (link) with some more details.
Interesting find and share. Anybody know the wattage ratings for watts used by the fan, and continuous heat rejection on the Framework’s current cooling solution?
@D.H Currently the FW laptop puts out roughly 30W of heat so multiple Airjets would need to be on the board to allow adequate cooling. Early days but looks interesting for sure.
The Framework Fan has a rating of 30W sustained, the CPUs have a PL1 of 28W-30W. Your results will vary based off of thermal paste and stuff though.
The pro versions of these things have 10W dissipation per chip.
thought of something similar the other day! remembering that piezo fans exist - Forced Convection with Solid State Piezoelectric Fan - YouTube
so interesting, Frore’s website lists a large engineering team, but for now they are only showing expensive CGI, promises and a super basic “datasheet”
would love to see some some thermal engineering math to explore the feasibility of this
interesting as this proposal is, im not convinced its new. I have a prototype device somewhere that I’ve had for well over a decade that I’m pretty sure is a similar idea. That one didnt catch on either.
I knew it exists, learned it in physics class, along the fact that it was very energy-hungry… thus I’m not interested…
Piezo is already used to cool e.g. smartphones.
EDIT: no it’s not piezo, thus I am interested.
Ouch you’re right. That’ll teach me to thoroughly read links before commenting…
Here’s an interview with the CEO by PCWorld that just popped up for me: Solid State Cooling Could Revolutionize Thermals - YouTube. Too bad there’s not that much B-roll, though they do show it blowing a wheel thingy.
They say it’s a type of MEMS chip, the same stuff that powers DLP projectors, or the gyro/accelerometers in your phone. Asianometry has some good videos on MEMS if you’re interested: https://www.youtube.com/results?search_query=mems+asianometry
The proof is in the eating. I guess the Framework laptop would be the ideal test bench for this tech since all of framework’s customers would be able to replace their existing fans with this new mems tech and we could find out immediately if its worthwhile.
In the PC World interview, the CEO says there will be consumer products shipping with this in 2H 2023, and from a recent Techcrunch article, it says that Frore “expects first AirJet chips to ship in Q1 2023” so it won’t be long until we see if they can deliver on their promises.
The article also mentions that they’ve raised $116M and that Intel is collaborating to build AirJet into its Evo platform, hence the mention of the 28W TDP power target) and that they’re already developing next gen versions. Personally, I’d be happy to pay a premium to be able to replace the current Framework cooling system w/ one that was totally silent, but I think the most exciting thing is if Frore is able to move thermal improvement/scaling onto a silicon, vs mechanical rate of change/improvement curve.
BTW, Frore seems to be pretty obviously pitching Apple… (AirJet_Introduction_121422.mp4 on Vimeo and Airjet_Mini_v3_121422.mp4 on Vimeo)
Pretty cool indeed. But in the PC World interview they said they could move/cool about 5W with 1W of input power. That is a very high power consumption for that amount of heat removed. Just to put it into perspective: Noctua quote a 1.68W maximum power draw on a big 120mm 2000rpm fan. Or in other words: Add 20% to your CPUs power draw. Meaning it is 6W extra to cool an intel 30W CPU.
In datasheet Frore claim net 4.25W removed per 1W @ 85C die temperature, 25C ambient. Anyone got resources to compare this power efficiency to conventional fans? compared to @mcz comment, that would suggest comparable efficiency ratios, but with Frore’s chip having lower peak power removal.
I have a pretty standard notebook fan lying around, rated for 3.75W vs Frore’s 1W. I have no idea how much heat a laptop fan removes, would be good to know ballpark in typical operating conditions.
Asides from power/efficiency considerations, as talked about in video the almost unreasonable conduction of vapour chambers could enable crazy conduction of heat via all that copper surface area on their chip, and then a boosted heat transfer coefficient in the convective cooling from a compressed boundary layer! The full heat path once built into a device seems very superior to what is possible with fans.
Since they seem to be working with large OEMs now, it’d be a real shame if their contracts prevented them eventually selling these chips as standardized off the shelf products. At the mention of Apple, recent history isn’t promising and I wouldn’t be surprised if proprietary versions with restricted supply were used by the big manufacturers.
The current fw removed 30w og heat…so it woule consume apparentely an insane 10 watt ish to cool the laptop!
At idle fw consume 5watts.
Best cooling solution…is to not produce the heat. Like arm chips with a good node fab in an apple macbook.
@sunsoy the best info I have on hand is the Steam Deck which uses a 2W fan and APU rated to draw 15W.
I would take the extra peak power draw over noise any day; if I’m running anything full tilt I will be plugged in. Assuming they achieve what they claim, there are other benefits to this tech:
It does seem to drop off kinda hard in terms of low power cooling based on the datasheet, but running it in bursts might mitigate that (and not be an issue since there won’t be distracting noise), or simply having the space for more passive cooling can help too.
Well from looking at the datasheet for the pro version you could remove about 6.5W of heat with around 1W input power. Scaling gets apparently worse at the higher loads which in turn means it does scale nicely to lower heat inputs (aka normal use) which fans do as well.
With fans it really is a bit of a guessing game since power ratings are probably (personal guesstimation) never reached in real life.
There is also a different problem apart from the power consumption: Size.
If they stay as big as they are and you need about 6 of them to cool a framework, they need a lot of space.
Anyone remember when LTT tested something similar? This Bizarre Fan Cost $1100?! - Piezoelectric Fan - YouTube
Piezoelectric is quite different no? Yeah I remember that video though, very cool xD.