A bunch of pictures

[Linux-SWAT]

The unit being in hands even 50° will be shockingly hot.

 

[WizardStan]

The unit being in hands even 50° will be shockingly hot.

Plastic transmits heat slower than flesh, your own body will heatsink it away from contact before any damage can occur; you'll likely only notice mild discomfort even at high temperatures.Source: I have a 3D printer and have no problem holding onto a fresh part at near 100°; it feels quite warm, but no where near the kind of hot you would get from something that transfer heat well, like very hot water or metal of similar temperature.

 

[Grench]

The unit being in hands even 50° will be shockingly hot.

Plastic transmits heat slower than flesh, your own body will heatsink it away from contact before any damage can occur; you'll likely only notice mild discomfort even at high temperatures.
Source: I have a 3D printer and have no problem holding onto a fresh part at near 100°; it feels quite warm, but no where near the kind of hot you would get from something that transfer heat well, like very hot water or metal of similar temperature.

Actually - your body/flesh/hands act more like a water block heat exchanger than a heatsink. The liquid (blood) captures some small amount of heat and moves on.


The case materials themselves are effective insulators.


Wish I had some cad skills. What is the easiest Linux based CAD package?

Obviously not to scale:


Base of Pyra:

/------------------------------------------\


|                                          |

|                                          |

|             |--------------------|       |

| SoC board ->|                    |       |

|             |                    |       |

|             |--------------------|       |

|              |------| |------|           |

|              |      | |      |           |

|              |      | |      |           |

\______________|______|_|______|___________/

Base of Pyra with unbent metal heat sink:

/------------------------------------------\


|                                          |

|                                          |

|             |-##################-|       |

| SoC board ->| ################## |       |

|             | ################## |       |

|             |-##################-|       |

|              |##################         |

|              |##################         |

|              |##################         |

\______________|##################_________/

        @@@@@@@@@@@@@@@@@@@@@@@@@@@@

        @@@@@@@@@@@@@@@@@@@@@@@@@@@@

        @@@@@@@@@@@@@@@@@@@@@@@@@@@@

        @@@@@@@@@@@@@@@@@@@@@@@@@@@@

The # and @ are all one thin sheet of aluminum.

# is all inside the case UNDER the SoC board and OVER the battery.

A thermal conductive pad is between the # aluminum and the SoC.

@ is the portion of the aluminum sheet that comes out the front of the Pyra UNDER the SD slots.  Not shown here is that it folds down, then back UNDER the front lower edge of the pyra and back to the battery door

Cross section:

  <- Back  ------   Front Edge

PPPPPPPPPPPKey DeckPPPPPPPP

===========Key Mat========P

===========Main Board=====P

 ====SoC====  ==SD Slots==P

----30ga Aluminum sheet---| <- fold                    

 ========Battery===      P|

PPPBattery DoorPPPPPPPPPPP| 

End of aluminum -> -------| <- fold

P means plastic outer shell.  Note that the aluminum folds around the lower plastic shell outside of the case plastics.

Ideally the leading edge (to the right in crude drawing) would be corrugated (many folds) for better dissipation.                         

Assembly:

Clip heat sink / radiator on lower case plastics.  It is a C that should go around that leading edge of the lower plastics piece.

Attach heat transfer pad to heat sink.

Assemble case - pad should contact and compress directly against the SoC.

Of course molds would have to be adjusted for this.

The aluminum would need to be stamped (cut) to shape and run through a brake (bent/folded).

Each piece of aluminum in quantities of 5K should be somewhere in the $0.15 to $0.50 range depending on who's doing it and gauge thickness of the aluminum.

 

[TrashyMG]

Something to think about Kapton film/tape is a good electrical insulator, but fairly decent as a thermal conductor.. Could be used to insulate a thin copper/aluminum plate in areas that shouldn't be touching metal.

 

[Grench]

Something to think about Kapton film/tape is a good electrical insulator, but fairly decent as a thermal conductor.. Could be used to insulate a thin copper/aluminum plate in areas that shouldn't be touching metal.

Thermal conductivity of polyimide film 0.37W/m*K

http://www2.dupont.com/Kapton/en_US/products/MT/index.html

Aluminum is ~ 205W/m*K

205/0.37 ~ Aluminum is 554 times more thermally conductive than polyimide film.

Copper is about twice as efficient or 1000+ times more thermally conductive than polyimide (Kapton) film.

 

[TrashyMG]

205/0.37 ~ Aluminum is 554 times more thermally conductive than polyimide film.

And? You like Aluminum touching bare circuit board? if some heat still could be dissipated through Kapton to the PCB perhaps then it may be worth using it over a regular sheet plastic for an insulator.  

 

[comradekingu]

Aluminium is more weight-effective, but since we are short on space, i would use copper.

Copling the SoC to the hinges could be one idea.

Snap the case together with a resistor dissipating 1W and see what happens. My guess is a flat shim copled well to the SoC is enough.

 

[Grench]

205/0.37 ~ Aluminum is 554 times more thermally conductive than polyimide film.

And? You like Aluminum touching bare circuit board? if some heat still could be dissipated through Kapton to the PCB perhaps then it may be worth using it over a regular sheet plastic for an insulator.  

Please re-read my message.  In that model there is a thermal transfer pad between the aluminum and the SoC board.

Ex:  http://www.amazon.com/StarTech-com-Heatsink-Thermal-Pads-HSFPHASECM/dp/B0009B0K2I

Note:  Electrically non-conductive, thermally conductive 0.7W/m*K (not great, but way better than tape)

http://www.startech.com/Computer-Parts/Fans/Thermal-Pad-Heatsink-Paste-Alternative-Package-of-5~HSFPHASECM

 

[Grench]

205/0.37 ~ Aluminum is 554 times more thermally conductive than polyimide film.

And? You like Aluminum touching bare circuit board? if some heat still could be dissipated through Kapton to the PCB perhaps then it may be worth using it over a regular sheet plastic for an insulator.  

Please re-read my message.  In that model there is a thermal transfer pad between the aluminum and the SoC board.

Ex:  http://www.amazon.com/StarTech-com-Heatsink-Thermal-Pads-HSFPHASECM/dp/B0009B0K2I

Note:  Electrically non-conductive, thermally conductive 0.7W/m*K (not great, but way better than tape)

http://www.startech.com/Computer-Parts/Fans/Thermal-Pad-Heatsink-Paste-Alternative-Package-of-5~HSFPHASECM

Found a better example 3W/m*K:

http://www.timtronics.com/PDF/timpad/TIM-PAD%201001-1002-1003.pdf

 

[TrashyMG]

Please re-read my message.

re-read what? I didn't even see your message until after I posted my suggestion for Kapton film/tape.


Kapton is what I'm familiar with at work, I'm open to better solutions.. That first Amazon link, reviews suggests it could be a horrible product..  The TimTronics solution looks promising.

 

[EvilDragon]

Thanks for sharing this. How is ED's setup different? Didn't he say we wouldn't need much of a heatsink at all?

  Well without airflow the Heatsink on the devboard is pretty useless, It may drop the temperature 5-10C.. It runs hot, but not over the critical mark from my observation.. Another thing to consider is there isn't any CPU frequency throttling setup like the Pandora yet in the OS.. I can change it manually between 500MHz, 1GHz and 1.5GHz but nothing in between..

aTc noticed some similar heating situations as I have.

Yes, the heatsink on the OMAP5 drops the temperature by 9 - 10°C (according to the internal sensor).


I've tested that already.


The critical temperature is a standard setting in Linux. You can define it in a config file.


I doubt our Linux test distribution has temperatures setup for the OMAP5, it's probably the standard x86 stuff. 

From an OMAP5 tech sheet: "Industrial temperature qualification (-40 to +85°C)" at least the SoC has some decent tolerances. ^^" http://newscenter.ti...ustrial-designs

Well It's running a bit hotter than that.

That's the temperature of the SoCs surroundings, not the SoCs temperature.


Don't use it in a Sauna that has a higher temperature than 85°C.

 

[EvilDragon]

It won't work like that, that's the problem.


Heat evenly distributes on a heat sink.


Without a heat sink, the SoC might have 110C.


With a 5x5cm heatsink, it might go down to 80C evenly distributed on the heatsink.


With a 10x10cm heatsink, it might go down to 50C evenly distributed on the heatsink.


...stuff about holes in a case and a fan...

Heat distributes evenly into a heat sink - and if it has nowhere to go, does not stop increasing in temp.


The SoC might have 110C at 5 minutes run time.


Using the heat sink it might be 50C at 5 minutes run time - but unless it's transported to the air or hands, it isn't going to stop there. A heat sink alone only draws the problem out longer.

But it DOES go somewhere.


Plastic doesn't fully stop the heat.


I've run the SoC at 110°C for more than half an hour with a plastic cap, and after 10 - 12 minutes of full load, it reached the maximum temperature (both the SoC and the plastic) and didn't change anymore for the last 20 minutes.


Plastic is NOT a heat stopper.


If the surface is big enough to cool distribute the heat evenly, the device will get warm but that's it.

Can you show me a picture of a PC which has a passive heatsink with holes in the case to let out the heat?


The only passive ones I know are simply heatsinks that are fully inside of the device.

Not holes, but I can show you several with heat sinks and external radiators.http://www.cappuccinopc.com/slimpro-sp675fp_fanless_mini_pc.asp

http://www.logicsupply.com/computers/feature/fanless/


The Shuttle box in this example appears to use vent holes, but likely an aluminum case too:

http://www.makeuseof.com/tag/5-silent-fanless-mini-pcs-that-will-save-you-money/

Yes, and you see how these work?


They're using the case as heatsink, with lamellas to get as much surface as possible.


How is that different from me telling that we need as much surface as possible?


Again:


Please show me a single computer, tablet or smartphone that uses a few tiny holes above huge heatsinks without any fan to prove that tiny holes will help get a lot of heat out of the case.

Um... well, the metal case is irrelevant here, since it uses the case as huge heatsink and the other heatsink doesn't do anything to lead the heat out of the case...?

The case is plastic or plastic with reinforcing fibers right?

With all due respect, I think you're completely missing the ideas being presented.


We're talking about using a heat sink that exits the case to be a combined heat sink AND radiator.

And where would you put such huge holes into the case?


Or do you really think the small holes in the hinge will make a big difference, when 99% of the heatsink is still inside of the case?


Do you think the heat will all magically move exactly to those holes because they get fresh air there?

My idea would be to have a thin aluminum heat sink/transport sandwiched between the SoC and the battery that comes out the front of the device under the SD cards and wraps down and back to the battery door. It then has around 10CM^2 of surface area outside of the case to dissipate the heat from the SoC and the battery.

So you want to put the heat where your hands are holding the device?

Either idea uses a substance with high thermal conductivity to draw heat into itself and an exposed surface outside of the case to radiate it to the surroundings.

Yes, but that will only work if there's a huge enough area going outside of the case, and I can't see where that should be.

From the mini-table above, you'll notice that aluminum is approximately 800 times better at transporting heat than the case plastics can be. Put another way, 1cm^2 of exposed (out of case) aluminum surface can do a similar job of dissipating heat from the SoC as 800cm^2 of plastic surface area.

Please explain how the dissipation would be different outside the case than inside of the case.

A heat sink grants time - if there is nowhere to let the heat out, that time is limited.

Yes and no. If the surface is big enough to evenly distribute the heat inside of the case and the plastics lets enough heat through (which I think it will), there will be a maximum temperature.

Those holes won't help much, as mentioned before!


The surface outside of the case is too small to have a real effect!


It will probably cool down everything by 1 or 2°C.

Again, I think you're far overestimating the thermal conductivity of case plastics and far underestimating the thermal conductivity of copper and aluminum.


It's orders of magnitude in difference.

That doesn't have anything to do with each other.


I never said we shouldn't use large areas of copper to cool everything (in fact, I said the larger the surface of the heatsink is, the better it will be).


I only said that those small holes inside the plastic won't change much.


Again:


I need to do more testings, but from what I've seen so far, a large surface area to distribute the heat should be more than enough.


The only proper way to do this is with simulations - but as long as there's no expert here who really knows how to do such simulations or knows what he's talking about, we're all doing guessing games here.

 

[ekianjo]

Again: I need to do more testings, but from what I've seen so far, a large surface area to distribute the heat should be more than enough.

Yeah, ED is right here. Effectively most heat transfers occur through surfaces - you can't direct the heat to go into little holes - heat dissipates itself in all directions, and heatsinks, fans, are just tools to direct heat in one direction rather than the others, but do not solve the fact that heat is dissipated anyway.  

 

[Grench]

It won't work like that, that's the problem.


Heat evenly distributes on a heat sink.


Without a heat sink, the SoC might have 110C.


With a 5x5cm heatsink, it might go down to 80C evenly distributed on the heatsink.


With a 10x10cm heatsink, it might go down to 50C evenly distributed on the heatsink.


...stuff about holes in a case and a fan...

Heat distributes evenly into a heat sink - and if it has nowhere to go, does not stop increasing in temp.


The SoC might have 110C at 5 minutes run time.


Using the heat sink it might be 50C at 5 minutes run time - but unless it's transported to the air or hands, it isn't going to stop there. A heat sink alone only draws the problem out longer.

But it DOES go somewhere.


Plastic doesn't fully stop the heat.


I've run the SoC at 110°C for more than half an hour with a plastic cap, and after 10 - 12 minutes of full load, it reached the maximum temperature (both the SoC and the plastic) and didn't change anymore for the last 20 minutes.


Plastic is NOT a heat stopper.


If the surface is big enough to cool distribute the heat evenly, the device will get warm but that's it.

Can you show me a picture of a PC which has a passive heatsink with holes in the case to let out the heat?


The only passive ones I know are simply heatsinks that are fully inside of the device.

Not holes, but I can show you several with heat sinks and external radiators.http://www.cappuccinopc.com/slimpro-sp675fp_fanless_mini_pc.asp

http://www.logicsupply.com/computers/feature/fanless/


The Shuttle box in this example appears to use vent holes, but likely an aluminum case too:

http://www.makeuseof.com/tag/5-silent-fanless-mini-pcs-that-will-save-you-money/

Yes, and you see how these work?


They're using the case as heatsink, with lamellas to get as much surface as possible.


How is that different from me telling that we need as much surface as possible?


Again:


Please show me a single computer, tablet or smartphone that uses a few tiny holes above huge heatsinks without any fan to prove that tiny holes will help get a lot of heat out of the case.

Um... well, the metal case is irrelevant here, since it uses the case as huge heatsink and the other heatsink doesn't do anything to lead the heat out of the case...?

The case is plastic or plastic with reinforcing fibers right?

With all due respect, I think you're completely missing the ideas being presented.


We're talking about using a heat sink that exits the case to be a combined heat sink AND radiator.

And where would you put such huge holes into the case?


Or do you really think the small holes in the hinge will make a big difference, when 99% of the heatsink is still inside of the case?


Do you think the heat will all magically move exactly to those holes because they get fresh air there?

My idea would be to have a thin aluminum heat sink/transport sandwiched between the SoC and the battery that comes out the front of the device under the SD cards and wraps down and back to the battery door. It then has around 10CM^2 of surface area outside of the case to dissipate the heat from the SoC and the battery.

So you want to put the heat where your hands are holding the device?

Either idea uses a substance with high thermal conductivity to draw heat into itself and an exposed surface outside of the case to radiate it to the surroundings.

Yes, but that will only work if there's a huge enough area going outside of the case, and I can't see where that should be.

From the mini-table above, you'll notice that aluminum is approximately 800 times better at transporting heat than the case plastics can be. Put another way, 1cm^2 of exposed (out of case) aluminum surface can do a similar job of dissipating heat from the SoC as 800cm^2 of plastic surface area.

Please explain how the dissipation would be different outside the case than inside of the case.

A heat sink grants time - if there is nowhere to let the heat out, that time is limited.

Yes and no. If the surface is big enough to evenly distribute the heat inside of the case and the plastics lets enough heat through (which I think it will), there will be a maximum temperature.

Those holes won't help much, as mentioned before!


The surface outside of the case is too small to have a real effect!


It will probably cool down everything by 1 or 2°C.

Again, I think you're far overestimating the thermal conductivity of case plastics and far underestimating the thermal conductivity of copper and aluminum.


It's orders of magnitude in difference.

That doesn't have anything to do with each other.


I never said we shouldn't use large areas of copper to cool everything (in fact, I said the larger the surface of the heatsink is, the better it will be).


I only said that those small holes inside the plastic won't change much.


Again:


I need to do more testings, but from what I've seen so far, a large surface area to distribute the heat should be more than enough.


The only proper way to do this is with simulations - but as long as there's no expert here who really knows how to do such simulations or knows what he's talking about, we're all doing guessing games here.

Wow - I really don't want to get into a pissing match with you over thermal conductivity, so I'll try to keep this brief.

You're stuck on thinking of tiny holes.  Forget the tiny holes.  We've all moved on to the idea of having an exposed metal piece to act as a radiator.

Stop thinking of case plastics as a radiator.  They are an -insulator- when compared to copper or aluminum.

The case plastics on the Pandora at 140x83x29mm yield approximately 140*83*2+83*29*2+140*29*2 = 0.36m^2 of surface area that can transport heat at a rate of 0.5W/m*K.  We know that it can handle, but barely, the heat output of the Pandora.  So, drawing roughly 500mAh converting to heat at 5VDC W=V*A gives a maximum heat envelope for a Pandora of approximately 2.5W/hr.

Theory:  The Pyra WILL consume more power per hour and WILL produce more heat than the Pandora.  It will produce more heat than can be comfortably dissipated through the case plastics to the surrounding air.

Basis for theory:  The new SoC consumes more power.  More power draw is essentially 1:1 with heat in computing devices.  The new battery will be about 40% larger and will dissipate heat proportionally.  My back of envelope numbers say that the Pyra will need to shed heat at a rate of approximately 4W/hr or more.

We are giving you realistic options of how to transport this heat out of the case and radiate it to the air using materials that conduct heat by factors of 500+ times better than the case plastics.

For the Askaurus version of the heat to air radiator, picture the space between the hinges cut out of the base and in it's place a copper to air radiator block with fins or some such.  No tiny holes.  It's a copper block 4cm long and 1cm wide with textures giving it an exposed surface area of >= 10cm^2.  This would have the ability to dissipate heat at a rate similar to a sheet of reinforced plastics 1.1m square.  This little bumpy strip of copper can transport and dissipate heat at roughly 3 times the rate of the entire Pandora's plastics.  It may feel hot to the touch - but is out of the way.  This is in addition to the plastics dissipation - so, the net is that by adding his radiator in the hinge the Pyra can stay relatively cool with 4x the dissipation rate of the Pandora.  I believe this option to be more expensive in both materials (molded copper) and application (connecting the block to the SoC - no direct path, requires an additional layer on the main board of copper).

For my version the sheet exiting the unit would fold over and down and back leaving a couple of cm on either side for the palms - no burns AND at 15cm^2 of exposed aluminum surface would have the ability to dissipate heat at a rate similar to a sheet of reinforced plastics 1m square.  This little 5cm wide, 3cm folded over the leading edge sheet of aluminum can transport and dissipate heat at roughly 2.8 times the rate of the entire Pandora's plastics.  It may feel warm to the touch, but is out of the way.  This is in addition to the plastics dissipation - so, the net is that it will have ~3.8 times the rate of the Pandora.  I believe this option to be less expensive in both materials (stamped sheet aluminum) and application (exits the case where there is already a seam at SD card slots).

Either way, I suspect that the Pyra is going to need some form of non-plastic external radiator with thermal conductivity of > 100W/m*K.

Yes, you need to do more testing.  Our options may not even be needed - we have all stated that.  The objective was/is to dream up options that may help you and your project should it need additional cooling, which I'm 80% sure it will.

Your last comment about needing an expert to run the simulation or knows what he's talking about isn't a bad idea.  Maybe someone should find a mechanical engineer to look up the thermal conductivity of the materials involved, compare it to the ballpark energy consumption (heat generation) for the device, compare to a similar device, and estimate whether there is enough heat load dissipation in reinforced plastic to dissipate xW/hr from the estimated surface area of the given material.

Clearly I'm just another idiot in the forum.

Carry on.

Again: I need to do more testings, but from what I've seen so far, a large surface area to distribute the heat should be more than enough.

Yeah, ED is right here. Effectively most heat transfers occur through surfaces - you can't direct the heat to go into little holes - heat dissipates itself in all directions, and heatsinks, fans, are just tools to direct heat in one direction rather than the others, but do not solve the fact that heat is dissipated anyway.  

Since your post is a pile-on troll anyway...

Clearly you can not read.  Please re-examine the solutions that Askarus and I have put forward.  Neither of our current ideas has anything to do with little holes in a case.

 

[ekianjo]

For my version the sheet exiting the unit would fold over and down and back leaving a couple of cm on either side for the palms - no burns AND at 15cm^2 of exposed aluminum surface would have the ability to dissipate heat at a rate similar to a sheet of reinforced plastics 1m square.  This little 5cm wide, 3cm folded over the leading edge sheet of aluminum can transport and dissipate heat at roughly 2.8 times the rate of the entire Pandora's plastics.  It may feel warm to the touch, but is out of the way.  This is in addition to the plastics dissipation - so, the net is that it will have ~3.8 times the rate of the Pandora.  I believe this option to be less expensive in both materials (stamped sheet aluminum) and application (exits the case where there is already a seam at SD card slots).

So where exactly would that 15cm2 surface be on the unit ? below the battery ?

You mention it may get warm, but won't it get more than just warm if it transports all the heat faster than the plastic case ? with a directly exposed hot surface (even if you say it's out of the way, you never know how people will hold the device when they stopped using it and want to put it back) you may incur the risk of injuries or at least incomfort. Any example out there of commercial, portable device that actually has this kind of design ?  

 

[comradekingu]

Can you show me a picture of a PC which has a passive heatsink with holes in the case to let out the heat?

The only passive ones I know are simply heatsinks that are fully inside of the device.

I have built a PC like that. Full case with holes in it. Encasing a fully passive system.

 

[Hồng Thất Công]

I thought Pandora generates more heat than Pyra.  Pyra without Craig should be cooler automatically.  

 

[SNESFAN]

Is it possible to run the CPU board behind the screen and distribute heat through the glass like tablets do? Connect to the main board through a ribbon cable rather than being connected directly. The screen will offer 2 large surfaces to dissipate heat into, one being glass that is more conductive than plastic. And neither of which you will be holding directly in your hands. Also add a copper plate to distribute the heat more evenly to prevent damage to the LCD.

 

[klapse]

From the mini-table above, you'll notice that aluminum is approximately 800 times better at transporting heat than the case plastics can be. Put another way, 1cm^2 of exposed (out of case) aluminum surface can do a similar job of dissipating heat from the SoC as 800cm^2 of plastic surface area.

From what I can recall, there is a difference between a material being able to conduct heat (800 times better) and radiate heat (??? times better).

 

[ekianjo]

Is it possible to run the CPU board behind the screen and distribute heat through the glass like tablets do? Connect to the main board through a ribbon cable rather than being connected directly. The screen will offer 2 large surfaces to dissipate heat into, one being glass that is more conductive than plastic. And neither of which you will be holding directly in your hands. Also add a copper plate to distribute the heat more evenly to prevent damage to the LCD.

Sounds like a reasonable idea, but LCDs already generate some heat on their own, not sure it would not affect their reliability on the long term if they are always exposed to higher temperatures.