Question AirJet cooling for Laptop chips!

[FlameTail]

Frore secures $100M, collabs with Intel to create a new way to cool processors | TechCrunch

Frore Systems, a startup creating new ways to cool computer hardware, has raised $100 million in a funding round led by Qualcomm and others.

www.google.com

Better Than Fans? New 'AirJet' Chip Promises To Overhaul Laptop Cooling

The San Jose-based Frore Systems claims the new cooling technology can deliver a 50% to 100% increase in performance for laptops without generating much sound.

www.pcmag.com

This is so cool. And Qualcomm apparently invested $100 million so it seems legit.

 

[FlameTail]

Dr.Ian's video about it.

PS: No one else replied to this thread yet? I am surprised. Is this technology not exciting?

 

[soresu]

Dr.Ian's video about it.

PS: No one else replied to this thread yet? I am surprised. Is this technology not exciting?

There's a longer thread talking about it here.

 

[Doug S]

Does it make any noise? If it has some vibrating membrane, and air is moving through those little places with the filters, there must be some noise. But is it like "you have to hold your ear up against it to hear anything" or "wow who thought tiny MEMS membranes vibrating could produce such an annoying high frequency sound!"

 

[coercitiv]

Does it make any noise?

According to Ian Cutress they should be relatively silent. The quoted 21dB value tells me little without a distance spec though.

 

[Doug S]

According to Ian Cutress they should be relatively silent. The quoted 21dB value tells me little without a distance spec though.

A given db level that's a mix of lower frequency sounds that amount to white noise vs the same db level of a single high pitched frequency are also very different in their annoyance level.

 

[coercitiv]

A given db level that's a mix of lower frequency sounds that amount to white noise vs the same db level of a single high pitched frequency are also very different in their annoyance level.

The noise profile is obviously more important. It does lack the turbulence of a high speed fan, but then again it relies on a high speed air jet moving through a relatively narrow space, so I reckon it can go either way.

I should have thought about this earlier today: they have a website and a bit more info than we get from the videos. Their datasheets quote a noise level of 21dBA @ 50cm for the 4W model and 24 dBA @ 50cm for the 9W model. That sounds good... on paper

 

[soresu]

I should have thought about this earlier today: they have a website and a bit more info than we get from the videos. Their datasheets quote a noise level of 21dBA @ 50cm for the 4W model and 24 dBA @ 50cm for the 9W model. That sounds good... on paper

I'm unclear as to why they deduct the power consumption of the device from the dissipation for the net dissipation values?

 

[yottabit]

I'm unclear as to why they deduct the power consumption of the device from the dissipation for the net dissipation values?

Because the majority of the power consumption of the device is turned into heat I’m sure. So the net heat removal rate is the useful metric

 

[soresu]

Because the majority of the power consumption of the device is turned into heat I’m sure. So the net heat removal rate is the useful metric

The figure is exactly what you get if you subtract the stated power consumption from the original dissipation figure - that's under 1% efficiency, which seems a bit unlikely as MEMS generally seem to be pretty efficient devices if early MEMS speakers are any indication.

Unless the cycle frequency of the device needs to be so high to maintain jet impingement air velocity that the perf/watt of the device is at the far end of the bell curve 🤔

Orrr.... the device is so efficient that it is actually transferring heat from the incoming air back to the copper base plate, thus partially negating the cooling effect.

Which makes sense as the boundary layer is broken and the base plate is as free to receive thermal energy from the air as it is the reverse.

I'm sure that I've seen some kind of metamaterial voodoo that could make thermal conduction a one way deal for a surface, but I guess for a 1st gen product they are probably happy to launch with that issue still hanging.

 

[yottabit]

The figure is exactly what you get if you subtract the stated power consumption from the original dissipation figure - that's under 1% efficiency, which seems a bit unlikely as MEMS generally seem to be pretty efficient devices if early MEMS speakers are any indication.

Unless the cycle frequency of the device needs to be so high to maintain jet impingement air velocity that the perf/watt of the device is at the far end of the bell curve 🤔

Orrr.... the device is so efficient that it is actually transferring heat from the incoming air back to the copper base plate, thus partially negating the cooling effect.

Which makes sense as the boundary layer is broken and the base plate is as free to receive thermal energy from the air as it is the reverse.

I'm sure that I've seen some kind of metamaterial voodoo that could make thermal conduction a one way deal for a surface, but I guess for a 1st gen product they are probably happy to launch with that issue still hanging.

The only other place the power consumption could be going is to the kinetic energy of the air. I’m too lazy to do do the math for that now but given the volume the thing moves it would be essentially a rounding error.

So yes I’m sure roughly 99% of the energy the device consumes is going to waste heat, conducted through the vibrating membrane and the housing.

A solid state device like this will be less efficient than a well designed fan or blower at moving air. It makes sense if you think about it - it is oscillating solid material back and forth at high frequency which requires excitation energy. When you move solid material you produce internal stresses which produces heat.

So it’s less efficient than a fan at moving air, but if the entire unit is viewed as a heat pump its efficiency is pretty good. It can remove roughly 10 times the input power in heat

There’s a lot of psuedoscience going on in this thread with misunderstanding of energy balance and boundary layers. There’s really nothing magical about this device except its compactness and potential longevity, which could be pretty big breakthroughs for certain class of device

 

[Khato]

These are certainly an interesting development, but unless battery life is a non-factor then their efficiency will be a problem. 8.75W of cooling for 1.75W of power isn't exactly good. A single fan + heatsink in most current notebooks can dissipate 4x that amount for the same power consumed. Sure the fan + heatsink combination requires more vertical space in particular... but if you make the device thinner as the Frore cooler allows that'd result in less volume available for battery capacity. So what benefit does this really provide on the mobile side?

It definitely does have a place in something like demonstrated in the video - very small form factor devices that don't rely on a battery for power. Could also be used in low power mobile devices to complement an otherwise passive design. (When running on battery device skin cooling is adequate, when plugged in the Frore cooler would activate and allow higher performance.)

 

[Abwx]

10W is not that much, it could be passively cooled with an all aluminium well designed laptop with the complete case removing heat.

This imply of course a minimal added cost but nothing significant when we look at current laptop prices that range at 1000$+.

 

[soresu]

but if you make the device thinner as the Frore cooler allows that'd result in less volume available for battery capacity

Currently this is being aimed at space constrained SKUs anyway - so that point is entirely moot.

 

[soresu]

10W is not that much, it could be passively cooled with an all aluminium well designed laptop with the complete case removing heat.

10.5/8.75W per device with several placed in series can add up to a significantly larger thermal dissipation in the same product height with an adequately sized vapor chamber to accommodate.

This is why they are going on about shifting from a 10-15W SoC to a 28W SoC potentially in the same volume - though I have my doubts that 24 dBA will remain the noise level when 3 are running at full whack.

 

[igor_kavinski]

MacBook Air gets solid-state active cooling in intriguing demo

Proof-of-concept explores alternative to fans for sustained, heavy workloads.

arstechnica.com

Anyone wanna bet that Apple won't care?

 

[yottabit]

MacBook Air gets solid-state active cooling in intriguing demo

Proof-of-concept explores alternative to fans for sustained, heavy workloads.

arstechnica.com

Anyone wanna bet that Apple won't care?

With 5 W power consumption and 0.1 W idle, yeah.

If they can get the power consumption on these down then they will be great for ultralights, but as-is that’s like 30% of the CPU power

With that kind of power draw the only place I can really see these being used currently would be in NUC format

 

[FlameTail]

With 5 W power consumption and 0.1 W idle, yeah.

If they can get the power consumption on these down then they will be great for ultralights, but as-is that’s like 30% of the CPU power

With that kind of power draw the only place I can really see these being used currently would be in NUC format

The interesting thing is these cooling chips are built in fabs like semiconductors are, so future iterations should be more efficient as the fabrication improves.

 

[soresu]

MacBook Air gets solid-state active cooling in intriguing demo

Proof-of-concept explores alternative to fans for sustained, heavy workloads.

arstechnica.com

Anyone wanna bet that Apple won't care?

Probably not for the 1st gen.

Maybe in another years time when Frore start pushing out 2nd gen devices.

 

[soresu]

The interesting thing is these cooling chips are built in fabs like semiconductors are, so future iterations should be more efficient as the fabrication improves.

It's not quite the same thing as you are thinking about a la Moore's Law.

MEMS (Micro Electromechanical Systems) is the tech in question creating the jets, and this does not scale in performance the same way as semiconductor transistor logic or memory devices do.

The devices will almost certainly improve generationally as MEMS microphones did (and MEMS speakers are expected to), but it's likely to be the frequency of the devices mechanical motion causing faster jets that improves their performance, with a much harder ceiling vs electrical switching performance in semiconductors given they are dealing with air resistance as well as electrical/semiconductor problems.

Going by my research into MEMS audio it's also possible that changing the shape of the jet generating components may allow further optimisation of the perf/watt/area, and possibly even the materials used.

For example graphene is hella efficient at acting as a speaker membrane, which is going to be similar to whatever they are using to push the air forward into a jet.

 

[soresu]

The interesting thing is these cooling chips are built in fabs like semiconductors are, so future iterations should be more efficient as the fabrication improves.

Interesting point that the future of lens technology called metalenses will also be fabbed with the same kind of lithography equipment.

The fab industry is on the verge of receiving a huge boost into potential markets with MEMS speakers and metalenses on top of these cooling jet chips.

 

[FlameTail]

Interesting point that the future of lens technology called metalenses will also be fabbed with the same kind of lithography equipment.

The fab industry is on the verge of receiving a huge boost into potential markets with MEMS speakers and metalenses on top of these cooling jet chips.

Metalens is truly amazing stuff, with the promise to completely eliminate the camera bumps in smartphones

 

[soresu]

Metalens is truly amazing stuff, with the promise to completely eliminate the camera bumps in smartphones

True that, and together with a more recent discovery called 'space plates' it could mean lenses for traditional cameras and telescopes that are much, much smaller and lighter.

An optic to replace space and its application towards ultra-thin imaging systems - Nature Communications

The need for space between lenses in optical systems results in a trade-off between potential for miniaturisation and achieved resolution. Here, the authors demonstrate a device that propagates light longer than its thickness, a spaceplate, and can therefore replace space in optical systems.

www.nature.com

You see metalenses can only reduce the volume of the lenses themselves, they cannot eliminate the necessary focal distance inside an optical system that is basically just air - space plates compress that focal distance so that you can shrink the total size of the optical system.

Imagine a cassegrain telescope, except it's just a few centimeters long instead of half a meter or more.

Gonna be some really interesting times for optical engineers in the next 20-30 years.

 

[FlameTail]

True that, and together with a more recent discovery called 'space plates' it could mean lenses for traditional cameras and telescopes that are much, much smaller and lighter.

An optic to replace space and its application towards ultra-thin imaging systems - Nature Communications

The need for space between lenses in optical systems results in a trade-off between potential for miniaturisation and achieved resolution. Here, the authors demonstrate a device that propagates light longer than its thickness, a spaceplate, and can therefore replace space in optical systems.

www.nature.com

You see metalenses can only reduce the volume of the lenses themselves, they cannot eliminate the necessary focal distance inside an optical system that is basically just air - space plates compress that focal distance so that you can shrink the total size of the optical system.

Imagine a cassegrain telescope, except it's just a few centimeters long instead of half a meter or more.

Gonna be some really interesting times for optical engineers in the next 20-30 years.

*_*

 

[FlameTail]

10.5 Watts
AirJet Pro removes 10.5 Watts of heat at a silent 24 dBA noise level, while only consuming a maximum of 1.75 Watts of power. AirJet Pro generates 1750 Pascals of back pressure, 10x higher than a fan, enabling sleek, dustproof devices. A revolution in Active Device Cooling

10.5W cooled using 1.75W.

How does this compare to a fan?