Fujitsu Could Help Smartphone Chips Run Cooler

angry tapir writes: If parts of your phone are sometimes too hot to handle, Fujitsu may have the answer: a thin heat pipe that can spread heat around mobile devices, reducing extremes of temperature. Fujitsu Laboratories created a heat pipe in the form of a loop that's less than 1mm thick. The device can transfer about 20W, about five times more heat than current thin heat pipes or thermal materials, the company said.

Comments

Waste of time

[Anonymous Coward]

Instead of finding clever ways to dissipate heat, wouldn't it be better to avoid conversion of that precious battery power into heat in the first place?

Re:Waste of time

[BronsCon]

It certainly would be; however, that's not something Fujitsu specializes in, so they're sticking to what they're good at. This will also save power, in a less than obvious way; as chips get hotter, their internal resistance increases, causing them to draw more current, causing them to get hotter still, pushing the internal resistance up higher, causing them to draw yet more current. Keeping chips cooler actually causes them to draw less power, overall. Even with cooler-running chips that barely sip power, something like this is useful to keep them cool and efficient.

Re:Waste of time

[Neil Boekend]

Normally current decreases as resistance increases. Why is chip power management different?

(no snark intended. I really want to know.)

Re:Waste of time

[Anonymous Coward]

Because the parent got one word wrong. Typically resistance decreases as temperature increases in semiconductors, not the other way around. Metallic conductors behave in the opposite way.

Re:Waste of time

[BronsCon]

Preface: I'm not qualified to discuss this in more detail than layman's terms will allow. I'm sure someone more qualified will step in and clear things up (and correct any inaccuracies in my information), and welcome them to do so.

You're thinking of resistance as a current-limiting mechanism, and you're absolutely correct in that respect. What happens, though, as the resistance of the signal path through a CPU increases, the switching current of the gates of the individual transistors in that data path also increases. This increase in switching current is greater than the current-limiting effect of the added resistance, increasing the over-all current draw of the chip.

That explanation is surely chock full of WTF-level inaccuracies, so don't quote me on that; standing by for correction.

Re:Waste of time

[BronsCon]

Sign in and pick up some karma. I just posted an explanation that you can correct in more detail than what you posted here. Undo the damage I've done, please? :)

Re:Waste of time

[Anonymous Coward]

The how would you explain super-conduction (very low resistance)
that is usually achieved in very low temperature environments ?

Re:Waste of time

[viperidaenz]

Rds increases as temperature increases in MOSFET's.
It's BJT's that decrease as temperature increases.

Re:Waste of time

[hcs_$reboot]

Just do that and go directly to Stockholm (Sweden) to immediately receive a Nobel prize.

Re:Waste of time

[Anonymous Coward]

The efficiency at UHF depends on other factors. The temperature coefficient of Rds or Vce(sat) isn't the issue.

What matters here is THERMAL resistance, or if you prefer, call it by the inverse, thermal conductivity. It's just about moving the heat around to get it out. Semiconductors fail when junction temperatures are too high.

Re:Waste of time

[thegarbz]

Some of that is actually pretty okay, but the bit you're missing is that CMOS used in most microprocessors effectively draws no gate current. Sure you need to charge up the gate but that quantity is not appreciably related to the switching resistance beyond the fact that you want to switch a gate as fast as possible to minimise time in the active region.

Re:Waste of time

[Anonymous Coward]

An entirely different principle.

Re:Waste of time

[MattskEE]

In undoped semiconductor resistance will usually decrease as temperature is raised because a higher temperature excites more electrons into the conduction band where they can carry current.

But for cases where there is already a lot of charge the opposite usually applies. In something like a MOSFET the electrons are supplied by the source contact or in a doped bulk semiconductor there will be lots of charge from dopants. In these cases increasing the temperature doesn't significantly increase the charge. However raising the temperature does increase the electron scattering rate which reduces the electron mobility and slows electrons down, so the resistance actually increases.

Re:Waste of time

[BronsCon]

Thank you for that, I actually wasn't sure of the mechanism by which what I said was made correct, if that makes any sense. It's been decades since it's been explained to me, so I only remembered the end result (increased resistance) and not why that, paradoxically, was the case.

2nd post

[Anonymous Coward]

2nd!

Do that for the laptops as well

[hcs_$reboot]

And desktops. That would reduce the fans work (and their own heat!) and the ugly noise that comes out of it.

Re: Do that for the laptops as well

[Anonymous Coward]

Desktops and laptops adjust use heat pipes. This is about getting the technology to fit in a phone.

Re:Do that for the laptops as well

[Anonymous Coward]

They already do this for laptops. The tech used is just too large to fit in a smartphone. Fujitsu's claiming they've made one that's small enough to be of use in a smartphone.

Re:Do that for the laptops as well

[Anonymous Coward]

There are plenty of cases which do this for desktops as well. Zalman had a really overpriced one a long time ago, and there are a lot of newer better ones available these days (most targeted at HTPCs). As you say, the new thing is the size.

Re:Do that for the laptops as well

[bloodhawk]

laptops and desktops don't have the space restrictions and hence heatpipes and many other cooling solutions are readily available and already in use.

Re:Do that for the laptops as well

[Anonymous Coward]

Nobody uses laptops and desktops! If you do, you're a dinosaur. Please die off already. A young hipster deserves to have your job, which you're so selfishly occupying.

Re:Do that for the laptops as well

[hankwang]

"That would reduce the fans work (and their own heat!)"

Years ago, I bought a Shuttle barebones Pentium 4 with a heat pipe, hoping that it would be as quiet as the Mac Minis of the time. It was disappointing. It saved having separate CPU and PSU fans and it was a bit less noisy than the average beige box pc, but still very noticeable.

A problem with heat pipes is that they are not flexible, so the motherboard, case-mounted fan, and heat pipe must match exactly. Not so suitable for do-it-yourself pc building. And in order to reduce noise, you still need to spread the heat over a large surface, i.e. attach a lot of lamellas to the heat pipe, that you could just as well have attached directly onto the cpu cooler.

Re:Do that for the laptops as well

[hcs_$reboot]

I bought a Shuttle barebones Pentium 4 with a heat pipe, hoping that it would be as quiet as the Mac Minis of the time. It was disappointing.

So I'm inclined to ask, what's the Mac Minis secret to be so silent?

Re:Do that for the laptops as well

[hcs_$reboot]

Seemingly I have a job you have no idea of ... ;-)

Re:Do that for the laptops as well

[dunkelfalke]

Not using a Pentium 4, I guess.
My core i5-3350p has passive cooling.

Re:Do that for the laptops as well

[jones_supa]

So I'm inclined to ask, what's the Mac Minis secret to be so silent?

It's probably explained in some video with white background and soft piano music: "I saw my friend's PC. It was okay, but it wasn't exactly quiet. I asked myself, how could that be improved. I wanted it to be whisper quiet. I knew it was okay to ask more. And that's where the story of the new cooling system of Mac Mini begins. The iCool."

In practice: some buzzy 40 mm radial fan with the text "O.E.M." silkscreened onto it, accompanied with thick layer of the most crusty thermal compound found in the market.

Re:Do that for the laptops as well

[ChunderDownunder]

"Mac Minis of the time".

Which in the P4 era (mid 2000s), would have been powered by motorola PowerPC chips as used in mac laptops.

Re:Do that for the laptops as well

[Anonymous Coward]

Please don't end a sentence in of.

doh.

Re:Do that for the laptops as well

[Anonymous Coward]

Well, for my mac mini, they put a defective temperature controller in the g4 which shut it down permenantly. No noise at all! Last POS apple crap i bought. Nowdays, they probably ship empty shells and a picture of steve jobs taped to the monitor. How long does it take to realize the display model isn't working at the phone counter for most people?

Re:Do that for the laptops as well

[drinkypoo]

Years ago, I bought a Shuttle barebones Pentium 4 with a heat pipe, hoping that it would be as quiet as the Mac Minis of the time. It was disappointing. It saved having separate CPU and PSU fans and it was a bit less noisy than the average beige box pc, but still very noticeable.

Nah, there are other reasons to use a heat pipe. I've got a $20 cooler master cooler with three heat pipes in it, and a big upright heat sink with a big fan on it. The heat sinks carry the heat up to the big fan where it can be efficiently removed.

I may also try a peltier cooler, I just broke down three peltier refrigerator coolers. THAT would allow me to eliminate a fan, so long as I redesign case airflow to go over the heat sink.

Re:Do that for the laptops as well

[Andy Dodd]

They DO have space restrictions, just not as severe as those in a smartphone.

Interestingly, heatpipes in a laptop are a way to deal with the space restrictions - they allow a laptop to dissipate MUCH more heat in a smaller space.

With smartphones, they simply had to "dissipate less heat".

Although I question how much of a benefit this will really be. As it is, even without heatpipes, smartphone thermal throttles are usually set WELL below the CPU's junction temperature limit - the reason is that it's to prevent other components from getting too hot (like the battery). I remember talking to some Sony engineers, and IIRC, the CPU thermal throttle in most Xperia Z family units is not set to protect any of the internal components, but to protect the user's hand. Fujitsu's tricks might actually reduce the junction temperature at which a CPU can operate without burning the user.

Re:Do that for the laptops as well

[tlhIngan]

Although I question how much of a benefit this will really be. As it is, even without heatpipes, smartphone thermal throttles are usually set WELL below the CPU's junction temperature limit - the reason is that it's to prevent other components from getting too hot (like the battery). I remember talking to some Sony engineers, and IIRC, the CPU thermal throttle in most Xperia Z family units is not set to protect any of the internal components, but to protect the user's hand. Fujitsu's tricks might actually reduce the junction temperature at which a CPU can operate without burning the user.

Actually, thermal control is a compromise. ARMs have traditionally consumed 1mW/MHz, which is great back when everything was 500MHz or less. These days you have "octacore" processors running at 2.5GHz, you're looking at 10-20W in a tiny package with poor cooling (because Package on Package to put RAM on top of the SoC). Add in everything else and going full tilt you could easily have to deal with 15+W.

Given the thermal resistance of the package, you can easily reach junction temperature (125C) stupidly quickly.

One analysis of a chip I saw had 2 cores going 100%, while cores 3 and 4 had to be thermally limited to 50% to keep the junction temperature down. And it needed to start well before the temperature was reached - or you'll overshoot the temperature.

Then there's the system configuration - is the CPU beside the battery? Then you'd want to keep it from getting too hot (you can still get max junction temperature and be only at 45C at the top because of thermal resistance) . Or use the metal in the screen as a heatsink.

A large amount of heat in a SoC is also conducted away through the balls - ground and power planes that double as heatsinks aren't uncommon.

Re: Do that for the laptops as well

[Anonymous Coward]

It better be more than that. Heat pipes tend not to work well when upside down, and people tend to flip thei phones around (sideways, etc.) all the time.

Poorly defined claim

[Anonymous Coward]

The efficiency should be expressed as thermal resistance, expressed in degrees temperature rise per Watt. All thermal resistances in the heat flow path get added up to determine how much hotter the chip is than the air or your hand etc.

I know Fujistu

[Anonymous Coward]

I am a pink belt of the twelf degree.

1 mm is a lot

[Anonymous Coward]

Plenty of phones sub-10mm today. 1mm is not an insignificant piece of the thickness budget.

Wrong reason they'll use it.

[dohzer]

This won't make your phone cooler; the manufacturers will just push their chips harder for the same temperature.

Re:Wrong reason they'll use it.

[DigiShaman]

I'm going to have to disagree with you on that. Of the people I talk to, almost all universally prioritize in the following order of their phone: Battery run-time, screen size/clarity, and performance. Notice how battery run-time is way up front, and CPU performance is in the back. That's because all new mobile phones have a CPU that's "good enough" to be GUI responsive. They're not trying to play PS4 quality games or mine for Bitcoins on the thing.

Re:Wrong reason they'll use it.

[gnasher719]

This won't make your phone cooler; the manufacturers will just push their chips harder for the same temperature.

"Push the chips harder" = "Empty the battery quicker".

Re:Wrong reason they'll use it.

[Andy Dodd]

Yeah. Lots of people rave about spec e-peen, but the truth is, even older chips are MORE than powerful enough to provide a good experience except for the niche things like hardcore gaming.

Interestingly, for hardcore gaming, NVidia gave up on the phone form factor. The SHIELD Portable's form factor allowed it to have active cooling for the Tegra4, and the SHIELD Tablet has a phase change heat spreader (aka heatpipe) over its CPU - a predecessor to this stuff Fujitsu is working on.

For nearly all smartphone users, more basic chips (like the Snapdragon 400) are more than enough, the problem has been that until the Moto G, low-end chips got match with piss-poor software quality. Moto is the first company to release low-end phones with software that didn't suck.

No

[Anonymous Coward]

If you think I'm going to let some stupid smartphone company put a 20W CPU in my smartphone instead of a 4-5W one, you've cracked it.

It's bad enough they keep adding more cores, higher res, and for some ridiculous reason don't add more ram or more battery.

I don't need 8 cores, I don't need a UHD res 6" screen, and I certainly don't need an oven in my pocket.

Transfer the heat to.... where?

[gavron]

In a laptop, the use of similar devices makes sense, as the heat can be transferred
somewhere where it can be dissipated into the air. Unfortunately it's more efficient
to transfer it to the table you have it on, so the bottom gets the heatsink which
makes it horrible to actually put your laptop on your lap-top.

In a smartphone, it's being held in your hand (on the back) and up to your face (on
the front) with fingers on the sides. Where to exactly are they going to move the
heat??? Heat exchanging is nothing new, but the ability to remove heat requires
the device interact with a cooler medium to transfer that heat. Normally that's
your palm, or the air, or both.

So... I ask again... transfer the heat to where?

E

Re:Transfer the heat to.... where?

[Megol]

In a laptop, the use of similar devices makes sense, as the heat can be transferred
somewhere where it can be dissipated into the air. Unfortunately it's more efficient
to transfer it to the table you have it on, so the bottom gets the heatsink which
makes it horrible to actually put your laptop on your lap-top.

Can you mention one laptop that uses the bottom as a heatsink by design? Hint: there is none.
In real notebook computers the heat is transferred using heatpipes to one or two heatsinks and then transferred to air using fan(s). The heat leakage to the casing is in most cases an unavoidable misfeature - though Apple did/do(?) use the casing as a heatsink in order to avoid spinning up the fans.

In a smartphone, it's being held in your hand (on the back) and up to your face (on
the front) with fingers on the sides. Where to exactly are they going to move the
heat??? Heat exchanging is nothing new, but the ability to remove heat requires
the device interact with a cooler medium to transfer that heat. Normally that's
your palm, or the air, or both.

So... I ask again... transfer the heat to where?

E

The (metal) casing. The power consumption in a phone is much lower than a notebook computer so it is a valid design.
Remember that the heat can't be allowed to build up higher than ~100 degrees celsius (in practice lower) as the hardware would fail, the only valid option unless one adds a heatsink/fan like in the notebook computer is to dump the heat to the casing.

Re:Transfer the heat to.... where?

[gavron]

It depends what you mean by "by design" is :)

Air is a great insulator, but poor conductor of heat.

My familiarity is with generatios of Dell laptops that exchange more heat through the bottom of the case they do they through venting to the air. Their support system even ensures you tell them if you're using your laptop "on a solid hard surface".

FYI 100C is higher than most hardware's failure point.

I know you want links. I'm off to bed. Google is that way --> Lazy is that way ---, and links are found in delis.

E

Re:Transfer the heat to.... where?

[Anonymous Coward]

Transfer the heat from a central point to the rest of the device for starters.

Re:Transfer the heat to.... where?

[N!k0N]

Presumably the entire expanse (limited though it is) of the back of said phone. I know with mine (GS4), it tends to get hot centered at the CPU, and to a lesser degree, the battery. However, the rest of the case is generally cool. So a heat pipe putting the heat "over there" (and thus, equalizing the temperature across the entirety of the unit) wouldn't be the end of the world.

Re:Transfer the heat to.... where?

[Anonymous Coward]

Don't be dense. They tell you to put it on a soild surface rather than a blanket or pillow or something else that will block the grill or fans. Stop inventing clueless "common knowledge" people are stupid enough without your help.

Re:Transfer the heat to.... where?

[DigiShaman]

So... I ask again... transfer the heat to where?

If the case is a solid piece of milled billet aluminum, it serves as the heat-sink as well; dual purpose. Because of the great thermal conductivity that is aluminum, the phone would actually be cooler.

I've actually had my iPhone prompt me that the temperature was too hot and needed to be cooled down. Physically holding it in my hand and pressed to my face felt no hotter than having it exposed to direct sunlight. Ok, so not that hot, but it wasn't blistering either. It was comfortably warm-ish. Had this technology had the chance to dump the heat to the case, I doubt the heat build-up would have ever occurred in the first place.

Re:Transfer the heat to.... where?

[Anonymous Coward]

It depends.

Would you prefer (from a phone performance, lifetime and battery life perspective) to have your heat-energy all concentrated in the center of your device CPU filling 1mm cubic volume of material, all whilst maintaining your phones external casing as close to room temperature as possible for as long as possible.

Or would you prefer to have your heat energy distributed over the entire external surface of your phone, in contact with the "great insulator" as you put it.

When you are a 100 degree computer chip, room temperature Air is an AWESOME thermal material compared to that piece of metal next to you that is 99.8 degrees.

Lower Power Chips

[monkeyxpress]

I seriously doubt this is going to have applications in people's smart phones. The limiting factor to pretty much all mobile performance issues is battery capacity, and being able to put more power through a certain mm^2 of silicon isn't going to help if you can't supply that power in the first place. This looks like a PR stunt by a component maker. These sorts of efficient heat pipes certainly have their place in many types of electronics products, most certainly in laptops and maybe even the next line of crossover tablet devices. But for smartphones I think we will need to wait for a significant improvement in battery tech before we have enough heat to waste that these things become any more useful than just sticking the chip to a big aluminium case like Apple does.

Smartphones

[Anonymous Coward]

It is better, that smartphones have better balance of battery. then better camera, I think