A bunch of pictures


@ Fusion_Power, Askarus, Levi:

Wow, you guys are really stuck on putting some kind of heatsink in the hing area arent you?

What I don't get is why are you trying to over-engineer what is truely a simple solution?!? there is no need to pipe heat anywhere, besides being EXTREMLY unwise to pipe heat through the mainboard--even if it wouldn't drastically increase the dev time for the system and require a complete retooling of the mainboard, it would lower the longevity of the Pyra. just no--the SoC is milimeters away from the base of the case.

Right below the SoC (relative to the operational orientation of the PCB) there will be a space (Judgeing by the placement of the battery on the pandora) roughly halfway up on the SoC PCB while this does not leave MUCH room, there is still a good eighth inch by 3 inch by half inch space for a heat dissapation mechanism. all that would be required is a mesh or a grouping of slits on the back/top edge of the case between the lower shoulder buttons. based on normal hand hold positions, there is little to no chance of anyone placeing their hand in that space so simply placing a copper plate in that space (attached to the SoC) and then shrouding the area with a high heat plastic skirt would keep out dust,

as for "Moving air" you really shouldn't need to... but if you insist, you can easily add an after-market fan to the unit (something similar to those "laptop coolers")... something that mounts to the pyra using the battery cover (basically replaceing the battery cover and tying into battery power) which houses a fan to blow air through the back vents which in turn forces air out the top vents. in this case, the skirt would keep the air moving through the top as opposed to venting through the unit and over other parts disperceing the SoC heat throughout the system potentially warming everything up and defeating the point of fanned cooling...

I'm not an engineer, but this really seems simpler than heat pipe and thermal routing. and much safer than intentionally transferring heat arround the SD slot, mainboard, and even display...
 
The battery has moved on the Pyra prototype cases that are about - iirc it moves down to cover more of the SD cards, so that leaves more space over the CPU board, if that helps any.
 
Yeah, I also find the idea of spreading the heat through the board a bit weird.
 
Yeah, I also find the idea of spreading the heat through the board a bit weird.
It sounds weird, yes, but that's the way a heatsink works:

The more surface a heatsink has, the lower is the temperature on each spot.

So the 100°C from the chip would be distributed to about 40°C if you have a heatsink with enough surface.
 
Any reduced lifespan to expect from that ?
Umm... no, why should there?

The processor produces the same temperature, whether you cool it or not.

If you DON'T distribute the heat, it's more likely to have a reduced lifespan ;)
 
It may help that the Pyra Processor sits upwards due the placement of the daughter board unlike the uspide down placement of the Pandora SoC.

Not sure how the battery, which also generates some temp, affects the Pyra SoC over time and vice versa.
 
I've never seen the battery temp of my Pandora get above ten degrees more than ambient, so I doubt it's pumping much heat into anything. 40 degrees should be well within the manufacturers guidelines anyway, if the Pyra's SOC can be kept to that.
 
've never seen the battery temp of my Pandora get above ten degrees more than ambient,
It can get hotter in special conditions on the Pandora though. Under normal use, it's very unlikely, but when you are asking the battery to feed more power (such as using a USB hub, plugging more stuff on it, screen with higher backlighting and wifi, and high CPU usage at the same time) it produces a lot more heat. Nothing unbearable, but that's the kind of conditions that should be tested to see if anything goes wrong. 
 
Yeah, I also find the idea of spreading the heat through the board a bit weird.
It sounds weird, yes, but that's the way a heatsink works:

The more surface a heatsink has, the lower is the temperature on each spot.

So the 100°C from the chip would be distributed to about 40°C if you have a heatsink with enough surface.
Mmm... A heatsink WILL lower the overall temperature by dispersing heat through it's material until it's material volume is saturated. If the heat sink is saturated and has no way to (or lacks efficiency to) dissipate the heat faster than it is input, then the system temperature will continue to climb.

Note I said volume, not surface. Heat conductivity from the SoC to the heat sink in this case is a surface to surface transition. The ability of the heat sink to hold heat is limited by volume.

The Pyra case is not going to be a 'perfect insulator', but won't there need to be a way to get the heat from the heat sink to the outside so that it can dissipate into the air, users hands or other?
 
That's why ED talked about hooking up the heatsink to the SD card cages, if it turns out to be necessary.  Those don't have a great volume in relation to their surface area, so a small input of heat will raise the temperature more greatly, and proportionally (to the delta between the air around the SD cards and the temperature of the slots) increase conduction of heat to the air and convection of that air away.
 
That's why ED talked about hooking up the heatsink to the SD card cages, if it turns out to be necessary.  Those don't have a great volume in relation to their surface area, so a small input of heat will raise the temperature more greatly, and proportionally (to the delta between the air around the SD cards and the temperature of the slots) increase conduction of heat to the air and convection of that air away.
SD card cages add more sink - but little to no radiator. They're still inside the case - and if they're used as intended (card inserted) then there is little to no airflow in/around them. Case plastics are poor thermal conductors.

I could see coming out the front of the case with a sheet metal flap following the case contours down and around the front lower edge of the case to provide a heat sink WITH an exterior radiating surface.

Of course the SoC in this case may emit so little heat that all of this is unnecessary. I'm mostly dreaming up ideas/options in case they need 'more'.
 
Yeah, I also find the idea of spreading the heat through the board a bit weird.
It sounds weird, yes, but that's the way a heatsink works:


The more surface a heatsink has, the lower is the temperature on each spot.


So the 100°C from the chip would be distributed to about 40°C if you have a heatsink with enough surface.
Mmm... A heatsink WILL lower the overall temperature by dispersing heat through it's material until it's material volume is saturated. If the heat sink is saturated and has no way to (or lacks efficiency to) dissipate the heat faster than it is input, then the system temperature will continue to climb.


Note I said volume, not surface. Heat conductivity from the SoC to the heat sink in this case is a surface to surface transition. The ability of the heat sink to hold heat is limited by volume.


The Pyra case is not going to be a 'perfect insulator', but won't there need to be a way to get the heat from the heat sink to the outside so that it can dissipate into the air, users hands or other?
That's what my bridge solution would do.

I want to get the heat out under the hinge.
 
How about increasing the speaker grill area, adding more holes on the left/right for ventilation. The speakers need to be closed to work properly, but there's still some space beside the speakers which could be used for venting. The extra holes would look like they are part of the speakers.

I thinking passive cooling here, hot air goes up, so it would go around the board and out these vents. Still need some way to allow cool air from outside to go into the case (to get a nice flow going).
 
Last edited by a moderator:
Are there any numbers or past experiences that imply more cooling than planned is needed? Is there any reason to believe the temperature will rise to troublesome numbers? There are plastic cellphones with similar-powered chips in them that can deal with the heat without any additional cooling solutions. Going by what ED just told us it doesn't seem like it's going to be an issue any time soon. I'm sure they'll test the first prototypes properly and then add another delay-inducing design change if need be. For now from my limited point of view I see no reason to believe this will be an issue. That said, if anyone has any actual evidence-based claim (like heat production profile for the chip and the rest of the components) I'd be happy to change my opinion, and so would other skeptics, I'd wager.
 
Last edited by a moderator:
Are there any numbers or past experiences that imply more cooling than planned is needed? Is there any reason to believe the temperature will rise to troublesome numbers? There are plastic cellphones with similar-powered chips in them that can deal with the heat without any additional cooling solutions. Going by what ED just told us it doesn't seem like it's going to be an issue any time soon. I'm sure they'll test the first prototypes properly and then add another delay-inducing design change if need be. For now from my limited point of view I see no reason to believe this will be an issue. That said, if anyone has any actual evidence-based claim (like heat production profile for the chip and the rest of the components) I'd be happy to change my opinion, and so would other skeptics, I'd wager.
I always see the upgrade option.

Build as many cooling as possible just for the case it might be needed in the future.

But that's all.
 
Samsung ARM Chromebook (A15 dual SoC) gets noticeably hot near the area where the SoC is when doing heavy compiling, for example. Pyra will be a bit more restricted in space, which would make heat issues worst.

Also remember that heat is generally bad for lithium batteries.
 
Last edited by a moderator:
Yes, but IIRC your solution requires holes in the case. I strongly suspect most of this isn't necessary while we're talking about a mobile ARM core, but if we're considering what might need to be done for an octacore or inefficient x86 chipset, it's just worth thinking how that would be fitted with minimal retooling.


That's if you want to entertain the idea of using a chip that runs hotter, and thus according to thermodynamics drains the battery even faster.


Edit: @Askarus
 
Last edited by a moderator:
Back
Top