GPD Win (x86 Computer / Palmtop)


As mentioned a couple of times during the last year already, when someone asked for air vents on the Pyra:
A fan actively replaces the heat with fresh air.
Any heatsink will just give you more time until the unit overheats, the fan is the needed component to get it out of the device.

Are there any thermal analysts or physicists or the like that can comment on this?

You seem to be suggesting that if constant power is applied to the heatsink it'll just get infinitely hot. In reality it's always going to reach some equilibrium temperature somehow, even if the paths for convection or conduction are poor. And the larger heatsink with more surface area should have a lower equilibrium temperature. Every test I've ever seen shows passive cooling w/o airflow to still get more of a reduction in equilibrium temperature than no heatsink.
 
Nikolaus has a heat camera and we did a few tests. The difference in a closed case with or without any copper or heatspreader is not that much.
With heatsink, it took a lot longer to reach the higher temperatures, but in the end, it got very close (about 3 - 4°C colder).
 
Hmm, not sure about the position of the fan.
It looks like the openings at the side are exactly where you place your hands...
Also my thoughts. However, they found an actual working cooling solution for such an tiny Handheld, respect. It seems they don't give up and even find solutions quickly when problems appear. Hire them, ED! ;)

The aluminum case back would actually assist as both a heat sink and a heat transference interface - to the users hands - which in turn means blood - which in turn makes the device liquid cooled too.
Since heat goes upwards, should't the cooling metal case part actualy be sitting on the top side of the base instead of the bottom?
 
Nikolaus has a heat camera and we did a few tests. The difference in a closed case with or without any copper or heatspreader is not that much.
With heatsink, it took a lot longer to reach the higher temperatures, but in the end, it got very close (about 3 - 4°C colder).

So as you observed the equilibrium temperature is lower. The actual amount it'll vary be is going to depend on a bunch of other variables, you can't really generalize it to say that passive cooling design will always have a minimal effect on equilibrium temperature.
 
Have you ever bled from a burn? If yes, you should see a doctor. Oo
Nobody said anything about denaturing protein. The aluminum back's temperature increases. It interfaces with hands. From a materials perspective a hand is a bag full of circulating fluid (blood). Blood will move the heat away from the hands - and thus away from the aluminum. No crispy flesh or bleeding out needed for this example.

If something "feels hot", what you're actually sensing is the transfer of heat from the object. Cold does not exist. Cold is our perception of a device or object that has less heat or more capacity for additional heat than the relative perspective of the measuring person or device. Even a Bose-Einstein condensate at so close to 0K that atomic structure collapses does not, "have cold". It simply has less heat.

Air conditioners do not produce cold. They capture heat from one place, transport it to another then shed it (and the heat of operation) to the 2nd environment.

ED & Exophase are oth on the right track about a heat sink only haing a set capacity before it looses effectiveness unless the heat is then somehow transported away. We call this movement of heat, "Heat Exchange". The device used is a "Heat Exchanger". A solid to ambient air heat exchanger is often called a radiator.

The GPD Win solution here is to make it's heat exchanger more efficient at transporting heat from the device materials to the surrounding air by moving more air past the hot parts (radiator) via convection (moving air). The aluminum back also allows for some heat to be transported from the device to the person using it via conduction (touch) to an object (hand) with circulating fluid (blood). The aluminum back is then a heat sink (mass of aluminum) that also acts as a conduction heat exchanger (touching a surface with less heat). Their intent appears to be to allow more compute cycles prior to throttling or a higher average cycle rate at equilibrium.

The Pyra solution as I understand it is to absorb short spike-loads in heat via a heat sink (chunk of metal) and dissipate system heat load relatively slowly through the plastics to the surface it is sitting on or to the user. Pollycarbonate is a good insulator - which might not be ideal. This is the reason for my posts in other threads here about ideas for options in case the internal components generate more heat than the device can shed through it's pollycarbonate shell.

The net, though, is that a device like these either needs to shed heat or lower heat generation or melt. The heat has to go somewhere, stop being generated or allowed to build. I hope that the transmission through the pollycarbonate shell to the surrounding air/desk/hands is adequate to maintain enough headroom to do the processing that the unit is supposed to do.

All of that out of the way... I have an idea - IF heat is an issue. For a 'gamer optimized' version of the Pyra, nix the 4G radio & antenna and put in a tiny fan to move air ducted from the SoC out a vent hole. It could maintain a higher clock for a longer time at the expense of battery life and loosing the 4G capabilities. The power lead circuits for the mobile module could then be repurposed into power for the fan in this scenario.

Personally, I still want mine with the 4G.
[doublepost=1469216823,1469215022][/doublepost]
Since heat goes upwards, should't the cooling metal case part actualy be sitting on the top side of the base instead of the bottom?
Hot air rises - that is a fact. BUT - it doesn't rise because it is hot. It rises because it decreases in relative density.
For gases, PV=nRT.
For a given number of molecules (n) of gas, if the temperature (T) increases, one or both of the pressure (P) and/or volume (V) must increase proportionally. In open atmosphere, the pressure (P) is a relative constant. So, as the volume (V) expands and the mass/molecules (n) remain constant, it's density (mass per cubic unit) decreases and the warmer, less dense, gas will rise while cooler more dense gas flows in around it to replace it. With a continued source of heat the cycle can repeat or be continuous. This is heat driven convection.

Within a solid object, heat will travel down at pretty much the same speed as it will travel up or sideways. It can actually be measured in it's advancement through solid objects with a common unit being W/mK

http://www.engineeringtoolbox.com/thermal-conductivity-d_429.html

Silica Aerogel ~ 0.02
Wool blankes score ~ 0.04
Polycarbonate transports heat through itself at 0.19 W/mK
Steel is ~ 43
Aluminum is ~ 200
Copper is ~ 370
Silver is ~ 429
Diamond is ~ 1000

Annectdotally, heat moves through butter and coal at roughly the same rate as polycarbonate. Butter simply has a lower melting point and it can be notoriously difficult to get a fire hot enough to push enough heat in to ignite Anthracite coal. The point is, though materials may have similar heat transfer properties there are are other properties that make them better or worse at specific applications.

'nuff physics for one day.
 
So as you observed the equilibrium temperature is lower. The actual amount it'll vary be is going to depend on a bunch of other variables, you can't really generalize it to say that passive cooling design will always have a minimal effect on equilibrium temperature.
I think ED is considering the device to be a sealed system that's airtight. In that case, the only way heat can escape is by heating the air in the case, then heating the case, so the equilibrium temperature is dependent only on the ambient temperature differential and the wind speed.

A handheld computer is quite close to airtight in the scheme of things at least for small temperature gradients. And a metal plate doesn't increase surface area nearly as much as a cast heatsink with a folded shape does. But in cases with more space to play with and a more complex heatsink you'll get a marked difference.
 
Nobody said anything about denaturing protein. The aluminum back's temperature increases. It interfaces with hands. From a materials perspective a hand is a bag full of circulating fluid (blood). Blood will move the heat away from the hands - and thus away from the aluminum. No crispy flesh or bleeding out needed for this example.

If something "feels hot", what you're actually sensing is the transfer of heat from the object. Cold does not exist. Cold is our perception of a device or object that has less heat or more capacity for additional heat than the relative perspective of the measuring person or device. Even a Bose-Einstein condensate at so close to 0K that atomic structure collapses does not, "have cold". It simply has less heat.

Air conditioners do not produce cold. They capture heat from one place, transport it to another then shed it (and the heat of operation) to the 2nd environment.
I went to school. ;)

Since your hands are usually warmer than the surrounding air, your hands have the opposite effect of a cooling solution unless you live in an very warm area or am I missing something?
If you still hold on tight to a device that is so uncomfortably warm that it nearly burns you, you may have a slight cooling effect but I guess if the air is cooler than you it would probably be better not to touch it - cuddling is there to save heat and not for cooling.
 
Last edited:
I think ED is considering the device to be a sealed system that's airtight. In that case, the only way heat can escape is by heating the air in the case, then heating the case, so the equilibrium temperature is dependent only on the ambient temperature differential and the wind speed.

A handheld computer is quite close to airtight in the scheme of things at least for small temperature gradients. And a metal plate doesn't increase surface area nearly as much as a cast heatsink with a folded shape does. But in cases with more space to play with and a more complex heatsink you'll get a marked difference.

You can call a system airtight, but it's not going to be a perfect thermal insulator. Heating the air in the case is not the only way for heat to escape, it will also escape through conduction to whatever the heatsink is attached to and even radiation (which is actually a fairly decent component of heatsinks without active cooling, see here http://www.heatsinkcalculator.com/blog/the-importance-of-radiation-in-heat-sink-design/)

It looks like GPD is at least putting some fins into their heatsink to increase surface area; naturally this has to be balanced against the amount of conduction to the chassis this limits.
 
Since your hands are usually warmer than the surrounding air, your hands have the opposite effect of a cooling solution unless you live in an very warm area or am I missing something?
If you still hold on tight to a device that is so uncomfortably warm that it nearly burns you, you may have a slight cooling effect but I guess if the air is cooler than you it would probably be better not to touch it - cuddling is there to save heat and not for cooling.

As long as the object is cooler than your hands, holding it will have a warming effect. Once it becomes warmer than your hands though your hands will begin to absorb heat and have a cooling effect as Grench described.

-Neelix
 
Yes, I specified airtight to suggest it's not radio wave proof, but to limit direct emission of heat from an internal component. Infra-red radio waves will heat up the case, just as conducted heat, just as convected air passing heat (again by conduction) to different parts of the case. But the only thing actually getting heat out of the system is emission from the case, so when considering the equilibrium temperature, you only need to consider emission from the case (by conduction, convection or direct radiation).

Technically, presumably things that aren't hot enough to produce even infra-red light are still producing radio waves, just at wavelengths a lot longer than visible light. A case is unlikely to be perfectly opaque to all wavelengths (indeed it's probably best not if you want your wifi to work), so heat energy can escape a closed case by broadcasting radio waves out. But presumably this can be ignored in classical thermdynamics, because the amplitudes and energies involved are not worth doing the maths for.

Edit: You're right that I did dismiss radiation and conduction when I said 'the only way for heat to get out it...', so sorry about that.
 
Last edited:
As long as the object is cooler than your hands, holding it will have a warming effect. Once it becomes warmer than your hands though your hands will begin to absorb heat and have a cooling effect as Grench described.
Who would want to hold on to it then?
I like my toys/tech to be cooler than me.
 
I went to school. ;)

Since your hands are usually warmer than the surrounding air, your hands have the opposite effect of a cooling solution unless you live in an very warm area or am I missing something?
If you still hold on tight to a device that is so uncomfortably warm that it nearly burns you, you may have a slight cooling effect but I guess if the air is cooler than you it would probably be better not to touch it - cuddling is there to save heat and not for cooling.

You're getting the general idea - BUT... remember that thermal coeficient?
Stagnant air is 0.024.
Water - the stuff your hands are mostly made of - is 0.58.
That means that you and I are 2400% better at conducting heat than dead air is.

Granted, when turning the device on at ambient room temperatures of ~ 70F or 21C, you (98.6F or 37C) are going to warm the device. However, that SoC is going to start throttling around 75C (167F). Without putting the air in motion over a broad surface area, it isn't really going to do much. The case is going to get warmer than the air AND you. At that point, you, being liquid (mostly), cool it.

No, the Win and/or Pyra's case won't get up to 75C, only the generation point (SoC & stuff) will get to those ranges - the heat sink will spread that out, but will warm up into that range too. The -low- thermalconductivity of the Pyra's case will mean that it will only be able to shed it's heat to you slowly over time. An aluminum back would shed heat faster - but would also feel hotter at the same temperature difference.

It's all trade offs. Sometimes it is better to accept a higher in-device temperature to increase the effectiveness of a radiator then move hotter air slowly than it is to have the fan spinning like mad - generating it's own heat and consuming resources in the process. Getting it 'right' is both science and art.
 
So how long did it take GPD to figure out that heatsinks and fans can help manage the heat?
 
So how long did it take GPD to figure out that heatsinks and fans can help manage the heat?

It doesn't take long to figure out, but you can't test whether you need that before having a prototype.
I also couldn't do all the tests I wanted to find out whether our heat dissipation is enough or not.
However, I would not include a fan.
 
That was a joke based on the PGS update about cooling. Somehow PGS managed to figure out they would need to do something without a prototype.

Should we be concerned about heat with the Pyra? I still remember your news post from 1 April. :p
 
I wouldn't be concerned, as the unit behaves fine in normal usage, but I couldn't test it yet with full power and 3D enabled. It shouldn't be much different from Smartphones with an A15 though, so not ideal, but usable.
 
I'm assuming that pokemon Go is the current pinnacle of gaming technology. After an hour of play my battery is almost drained but my phone is still only noticeably warm, not painfully hot.
 
Back
Top