It's all white!

[Black Sliver]

Well, I thought I was already cautious by adding "current", so don't know why you brought up the Raspberry Pi but apparently I also need to add "viable" to limit it to any cooling that would actually fit in the Pyra.

Because its one of the few you can "hack" and touch. The power budget of a single ARM core stayed the same; it's a few companies that squeeze way to many cores into the chip, or overclock it (without a proper cooling solution), that screw their power design.
No idea what you have, but my current 4.5" phone with 4x1.21GHz A53 does NOT throttle at 400% Load, without direct sunlight, when GPU is at 0%

Edit: giving a crazy clock*cores value and throttling without giving a typical "design average" is just bad power design. Please compare to better design choices like Intel's turbo or Intel's Core M.

 

[Lightkey]

The reason why SoCs often have eight cores is so that four are more efficient at light load and take over whenever you don't need higher performance, that's called big.LITTLE. I presume you know that, just pointing it out because the A53 you mentioned is the model used for light load in that design, it's half the speed of the old A15 in the OMAP5 on paper, I'm not counting that as current.

 

[the_marshal]

Only the company's announcement: https://www.sepa-europe.com/en/news/new-sepa-micro-fans

That look pretty nice ! So small...

I still think the idea having a fan in a small mobile handheld is weird but why not. Especially if it's a good quality fan which is quiet.
And really the fan is likely only needed in continuous high load or other special cases. And it can always be turned off.

Another concern with fan is dust and the fact it can fail (moving part).
Unlike most other device the Pyra is easy to open / fix so that helps.

 

[netcat]

Personnaly I would take one with a 2G limitation any day instead of having to wait 6 more months.
The java-people might feel different ...

ironically,
one of the original java use cases was for resource constrained embedded devices, eg: set top boxes
the idea was that the classloader could unload classes no longer needed and free up memory

but [you got me started]
i just spent a few hours trying to get the smallest possible java hello world
experimented with oraclejdk, jamvm, oraclejre/embedded
smallest i could get was 49M (with default settings its 255M!)
process virtual size
this is ridiculous
for completeness python requires 1.8M, perl 608k, c 328k (all default settings)

 

[Black Sliver]

@netcat isn't that just the acquired memory POOL? (Since java does it's own memory management)
[doublepost=1500109293,1500108005][/doublepost]@Lightkey I still stand on the point that a CPU core that can't be cooled is a bad power design and that (a lot of) recent SoCs do have cores that can be cooled; mine IS fairly recent (as in the age).
The mode where you have a heatsink that eats bursts of a more powerful core, like loading websites, should be advertised as such and the only company that does this is Intel. It also requires fine-tuning, to not have lag-spikes (like not switching between min/max but from "turbo" to "sustainable frequency"), and the "sustainable frequency" should be written on the datasheet.

Given the above, please also note that having a faster core with the same FLOPS/Watt, that needs throttling within a second, makes no sense to use in a device (stock heatsink (e.g. display and case for a phone)) unless it's for the numbers for marketing; e.g. comparing the "sustainable frequency" would yield that flagship phones are not necessarily (that much) faster than cheap ones, unless the heatsink (display) is bigger.

 

[netcat]

@netcat isn't that just the acquired memory POOL? (Since java does it's own memory management)

yeah - all those memory regions are allocated up front, and will have empty space

Heap Usage:
PS Young Generation
Eden Space:
capacity = 524288 (0.5MB)
used = 393296 (0.3750762939453125MB)
free = 130992 (0.1249237060546875MB)
75.0152587890625% used
From Space:
capacity = 524288 (0.5MB)
used = 0 (0.0MB)
free = 524288 (0.5MB)
0.0% used
To Space:
capacity = 524288 (0.5MB)
used = 0 (0.0MB)
free = 524288 (0.5MB)
0.0% used
PS Old Generation
capacity = 524288 (0.5MB)
used = 0 (0.0MB)
free = 524288 (0.5MB)
0.0% used

one culprit is metaspace 20.796875MB
and then there's off-heap jvm internals and actual jvm code, shared libs which add up
understandable, since the jvm is (supposed to be) crammed with smarts
but common .so code segments are shared across processes, so there won't always be huge overhead

anyway
i don't plan to do much java on the pyra
i'm kindof interested in julia...

 

[comradekingu]

Yay, a german company, nice!
The other one I have is pretty loud, so that one seems neat

So is http://www.ebmpapst.com/ The bigger the better, in terms of noise to airflow ratio. A bigger fan spinning more slowly pushes the same amount of air at less noise.

 

[FBnil]

I'm mostly concerned about losing the Pyra adapter, or losing track of which adapter it is. I have so many indistinguishable AC adapters; I don't know what half of them go to. With USB cables at least they're mostly interchangeable.

Can we get the Pyra logo on the adapter? So when it inevitably ends up in a big box of power adapters in my next move, I can at least remember which one it is.

Although it is a nice idea, the Pandora charger had enough identification on the backside. And you can always buy a label machine, or if lazy, use a piece of paper and some transparent tape. (which I use to mark my stuff at work)

 

[Eight Bit]

How did the key mat test work out with backlight in a real case @EvilDragon , mission accomplished?

 

[Yerffej]

We will also add a few venting holes below the USB port.

With this in mind, I think a fan in a dock would be good idea.

 

[oskda]

... aaaand I'm not sure a display loves being heated to 120°C

I suppose you are joking

1/ An LCD display can resist much more heat than a lithium-ion battery without damage. Let's look Pyra BOE LCD display: http://www.lcdsolar.com/Products/ProductDetail.aspx/BOE/BTL507212-W677L/37940?S=IL#.WW0qPCc3VoA
Operating Temperature -20 ~ 70 °C
Storage Temperature -30 ~ 80 °C

So it can function even with 70 °C, and it doesn't damage even at 80 °C.

What lithium-ion battery can resist without damage 70 °C working (charging/discharging) and 80 °C store ?

Usually a Li-ion battery is charged at maximum 45 °C. Li-ion battery degradation is strongly temperature-dependent.

So well, we have LCD can resist much much more heat.

2/ Most modern smartphones have all components in a block, and LCD displays doesn't suffer for that. Why should it suffer in Pyra?

3/ Let's look at a OMAP5 board:

CM-T54 board (http://processors.wiki.ti.com/index.php/OMAP5_Boards): «Temperature range 0°C to 70°C (optional from - 40°C to 85°C)». Interesting: that OMAP 5 board can function at 70°C, like BOE LCD screen. I don't know exact temperature Pyra board can resist, but I suppose it won't be far from CM-T54 board.

And obviously battery can't resist 70°C function (at least without exploding).

¿¿120°C?? ¿Do you spect Pyra will reach 120°C?

Can Pyra battery, in the lower case part as CPU board, resist 120°C? Obviously it can't, nor 70 °C (a temperature Pyra LCD display can resist). Pyra battery isn't glued to CPU, as LCD isn't glued to CPU.

LCD in upper case part as I think about it: aluminum in the external part connected to CPU/SOC to act as a very big heatsink (it could have external surface corrugated with small ridges to increase surface); LCD bezel like actual design in plastic (or another material with low thermal conductivity), and LCD separated from CPU-display board by insulator. I suppose heat will go for radiator more than for insulator and LCD after insulator LCD doesn't touch aluminum heatsink.

And LCD can resist 70 °C, a temperature battery in lower part case and my hands (holding Pyra) can't resit.

I assume actual Pyra design, as now it can't be changed. And I suppose Pyra 2 (Pyra with CPU update) would have same design, but it may be Pyra 3 or 4 could get a some different update. I think CPU sitting in upper part with external upper aluminum case acting as a big heatsink is a very interesting idea. And a clamshell design (like Pyra) can have it, where a smarthphone with one-block design can't have this advantage.
[doublepost=1500334215,1500332682][/doublepost]

You'd have to move all signals (USB, etc.) over the flex between top and lower case, which is just crazy (and maybe not even possible because USB, SATA, etc. need matched impedance for the source/traces/sink).

It is the other way: if you move CPU board to upper case part, you only need a USB cable to connect with lower part, and a power cable (for energy).

How do you connect keyboard/gamepads/etc in a modern computer? By USB. In Pyra lower case part we would have that: keyboard, gamepad, wifi/4G, BT, SD... all that can be connected by USB.

So we would need a USB cable, something STANDARD, while with actual design we need a complex custom and expensive cable.

Modern (expensive) battery management controllers should avoid this automatically by having a better-tuned tickle charge phase, at least in notebooks. Engineers learned from their mistakes

I am not talking only about charging: simply leaving battery charged to 100% (or near full) degrades battery life.

I think what you are saying about tickle charge phase doesn't apply to modern Li-ion batteries. When a Li-ion battery is full the power input is cut off to protect it (there is no trickle charging, and battery is only charged other time when below a certain %). The problem is, as I said, that even leaving the battery at 100% or near, degrades its life. It would be better to charge only at up to 80% and don't discharge below 40%. Obviously in smartphones we need 100% battery (even with this usually the don't last a complete day) so we charge it fully.
[doublepost=1500336399][/doublepost]

I've bought a JXD S5800 years ago for pretty cheap because the screen was messed up from the previous owner leaving it inside a very hot car. So yeah, you're right about that.

My friend a closed car in the sun can reach very very high temperatures

Pyra LCD can function at 70°C, and resist up to 80°C for storage. And nobody is talking about glueing LCD to CPU, but the opposite: to glue CPU to external upper aluminum case to act as heatsink.

If you look at it you will see most moderns smartphones have LCD in the same body than CPU, and there should no problem for that because they aren't glued together.

 

[Letalis Sonus]

0 to 70°C is the lowest common limit being used in the industry, it's standard for commercial grade electronics. Besides, it only guarantees that the display won't fail up to this temperature, it does not tell you how much it accelerates aging.

Batteries are products where on the one hand aging directly affects their performance noticeably, on the other hand thermal damage means that it might explode in your face - hence manufacturers are a lot more careful about providing operation temperature levels, especially since those things can get pretty hot on their own.

 

[oskda]

0 to 70°C is the lowest common limit being used in the industry, it's standard for commercial grade electronics. Besides, it only guarantees that the display won't fail up to this temperature, it does not tell you how much it accelerates aging.

Batteries are products where on the one hand aging directly affects their performance noticeably, on the other hand thermal damage means that it might explode in your face - hence manufacturers are a lot more careful about providing operation temperature levels, especially since those things can get pretty hot on their own.

Some people are saying you can't put CPU board in upper case part, because of heat damaging LCD, and I am showing near all modern smartphones have LCD in the same body than CPU without problem for LCD, and even that battery (in a smartphone in the same body than CPU and in Pyra actually in lower case part where CPU resides) is more sensible to heat than LCD. So they say you can't do a thing that near all smartphones do (some of them have even aluminum case).

 

[Binky]

if you move CPU board to upper case part, you only need a USB cable to connect with lower part, and a power cable (for energy).

Surely You'd need many USB connections to even approach the combined throughput of the Pyra's various I/O? (Which includes USB3.0, SATA, HDMI-out)

So we would need a USB cable, something STANDARD, while with actual design we need a complex custom and expensive cable.

Who says a standard USB cable can withstand being twisted back and forth in the same place that many times?

 

[Djoga'Ro]

You could move all those ports up, too. Or instead, only move the battery up.

 

[mclien]

As for the heat sink: We could think about to use silver for that. It has the highest heat conductivity of all metalls. I just est. the material price for that (my usual supplier only has sheetmetal up to 2mm thickness). That would be about 25,- EUR (material only)
But once we have the geometry of the required heatsink, I should be able to make some (depending on the shape, it could be either just cutted and bend sheetmetall or it might be better to make a little cats mold for it..

 

[kuru]

Not much better than copper (399 vs. 427 W/mK), but a lot more expensive and not readily available.

Maybe an aftermarket option for possible enthusiast overclockers

 

[EvilDragon]

I suppose you are joking

1/ An LCD display can resist much more heat than a lithium-ion battery without damage. Let's look Pyra BOE LCD display: http://www.lcdsolar.com/Products/ProductDetail.aspx/BOE/BTL507212-W677L/37940?S=IL#.WW0qPCc3VoA
Operating Temperature -20 ~ 70 °C
Storage Temperature -30 ~ 80 °C

So it can function even with 70 °C, and it doesn't damage even at 80 °C.

What lithium-ion battery can resist without damage 70 °C working (charging/discharging) and 80 °C store ?

Usually a Li-ion battery is charged at maximum 45 °C. Li-ion battery degradation is strongly temperature-dependent.

So well, we have LCD can resist much much more heat.

Except for that you either need to make the lid extra thick to put a layer of plastic cooling inbetween the SoC and the display or you would put it directly behind the display, whereas on the bottom, we've got space for that to shield the battery from the heat a bit.

2/ Most modern smartphones have all components in a block, and LCD displays doesn't suffer for that. Why should it suffer in Pyra?

It does. An app in my Droid4 crashed at some time, causing it to run at full CPU speed for over an hour.
I've got a nice visible square in my display since then.

Also, smartphones usually rarely run with full power, so the SoC doesn't warm up. And it's also throttled as well.

3/ Let's look at a OMAP5 board:

CM-T54 board (http://processors.wiki.ti.com/index.php/OMAP5_Boards): «Temperature range 0°C to 70°C (optional from - 40°C to 85°C)». Interesting: that OMAP 5 board can function at 70°C, like BOE LCD screen. I don't know exact temperature Pyra board can resist, but I suppose it won't be far from CM-T54 board.

That's a standard temperature range for industrial used electronics.

¿¿120°C?? ¿Do you spect Pyra will reach 120°C?

The SoC can reach 130°C on the surface when running at full power for 30 minutes and without cooling. I've tested that
Sure, that's the surface, and it will quickly cool off around it, but without cooling and throttling, it'll heat up a lot more.

Can Pyra battery, in the lower case part as CPU board, resist 120°C? Obviously it can't, nor 70 °C (a temperature Pyra LCD display can resist). Pyra battery isn't glued to CPU, as LCD isn't glued to CPU.

It has 1mm plastic and 0,5mm copper (or aluminium) and a 1mm PCB (the CPU Board, as the upper side of the SoC is directed towards the mainboard) between the battery and the SoC.
This would all be missing in the lid - unless you want to make it super thick.

LCD in upper case part as I think about it: aluminum in the external part connected to CPU/SOC to act as a very big heatsink (it could have external surface corrugated with small ridges to increase surface); LCD bezel like actual design in plastic (or another material with low thermal conductivity), and LCD separated from CPU-display board by insulator. I suppose heat will go for radiator more than for insulator and LCD after insulator LCD doesn't touch aluminum heatsink.

Well, that very big heatsink can also be put in the bottom part. The rest is to throttle the CPU so it doesn't get too hot (like all smartphones do).
Don't forget: We cannot let the lid reach 70°C either - that would be dangerous to humans!

Also note: The 70°C is usually the SURROUNDING AIR temperature. A heat source directly behind the LCD is something different - especially as it doesn't evenly heat up the LCD.

I assume actual Pyra design, as now it can't be changed. And I suppose Pyra 2 (Pyra with CPU update) would have same design, but it may be Pyra 3 or 4 could get a some different update. I think CPU sitting in upper part with external upper aluminum case acting as a big heatsink is a very interesting idea. And a clamshell design (like Pyra) can have it, where a smarthphone with one-block design can't have this advantage.
[doublepost=1500334215,1500332682][/doublepost]

I'm not really sure what the difference is between a big heatsink in the lid or in the bottom part.
Except for that the lid would be thicker and more heavy.
And we'd need an LCD cable with 10 times more traces, just to get all signals to the bottom part.

As you're comparing it with smartphones:
All these also have the SoC in the same part as the battery... so...?

It is the other way: if you move CPU board to upper case part, you only need a USB cable to connect with lower part, and a power cable (for energy).

One USB cable... how would that work? For all the connections?

How do you connect keyboard/gamepads/etc in a modern computer? By USB. In Pyra lower case part we would have that: keyboard, gamepad, wifi/4G, BT, SD... all that can be connected by USB.

Except for that this solution needs a lot more power, as these USB connections with hubs have quite a bit of loss here

So we would need a USB cable, something STANDARD, while with actual design we need a complex custom and expensive cable.

You'd like to have a USB cable stick out of the back of the lid and back into the bottom part? :O
That sounds like a Raspberry Pi Hacker solution, but not like a product.

Pyra LCD can function at 70°C, and resist up to 80°C for storage. And nobody is talking about glueing LCD to CPU, but the opposite: to glue CPU to external upper aluminum case to act as heatsink.
If you look at it you will see most moderns smartphones have LCD in the same body than CPU, and there should no problem for that because they aren't glued together.

And they also have the Battery in the same body as the CPU and they're also throttled to never reach 70°C.
So what kind of an argument is that?
[doublepost=1500376014][/doublepost]

How did the key mat test work out with backlight in a real case @EvilDragon , mission accomplished?

Yep. All good here

 

[Eight Bit]

Yep. All good here

This is great news, another item checked from the list. I guess mass production for the key mat can start now?
What else is there to sort hardware wise? Last CPU board check? The 4GB memory timing? More case stuff?

 

[fahrstuhl]

One USB cable... how would that work? For all the connections?
Except for that this solution needs a lot more power, as these USB connections with hubs have quite a bit of loss here

With a different SoC, you could try to route some PCIe lanes through the hinge and connect the bottom half to that.
There seem to exist some SoCs that have PCIe.
But I think PCIe lanes need quite a bit of shielding to keep up the signal quality. On the other hand, a display cable gets a lot of data pushed through it, too.

But yeah, I don't think it makes sense to put the SoC into the display part of the case, especially if a heatsink is required. You really need to keep the top half of the Pyra as light as possible. At least my wrists start hurting from using top-heavy devices for longer periods of time.