Finalizing the PCBs


Now all I can picture is Wally's avatar frenching a gnome.

God damn you internet. You warmed the cockles of my heart, on this day of subzero windchills.
 
I would probably be interested in an aluminum case for the Pyra. Just throwing that out there.
 
I'm not sure how much you know about mold production, but just adding holes at the backside for heat dissipation needs another tool (which is a couple thousand EUR) and you'll see a small recession where that tool is applied.
We're not printing 3D cases here. With molds, you have a top and a bottom. Think about how they're made: They're being drilled with CNC machines from the top and bottom.
If you want to add any holes SIDEWAYS in the case, you can't drill that in the mold, it's physically not possible.

Therefore, for every hole, clip, or whatever that can't be milled, you need an inlay too which needs to be produced additionally and put into the mould each time for production.

No, with the Pandora, compromises had to be made as well.
It's mostly built around the OMAP3: It uses features that the OMAP3 supports (SD Card slots, etc.)
Everything else (touchscreen, Wifi, analog sticks) are standard features for a mobile device as well.

We've added a lot more things in the Pyra (mobile internet), even things that the OMAP5 doesn't support out of the box (like three SD Cards and an eMMC).
The Pandora has nothing like that, AFAIK.

The Pandora was amazing, but there were a lot of compromises that had to be made as well.

I was about to produce the aluminum case in the school I worked in so yes, I know a bit about mold production.
What I understand is that it's too late to modify the cases as it was my suggestion to get an external sided heat dissipator, not drilling holes.

Misinterpretation strikes back. I never said that the OP is perfect for everyone (it is for my needs, also I often use its CPU at 100% overclocked for days).
What I said is that the effort was made towards this goal.
I feel that we still have a margin for optimization with the Pyra, and for the next-gen CPU boards, and the next etc., I express my opinion.
 
  • Like
Reactions: ___
Because they're in a race to be the thinnest for some glob forsaken reason. They pack everything so tight there's no room for a heatsink or even air, so they design chips with fast-finish in mind so they can throttle back down, then market them at their top speed. That IS their "handling" it.
Exactly. They could produce better devices that stay cool but the "race" to have the thinnest device makes everything bollocks.
Good example, I STILL search for a 15" budget gaming laptop that stays cool and especialy quiet. Hard to find because companies combine less powerful components with even thinner cases. I had hopes for the recent Acer Spin 5, which is one of the rare Notebooks that combine a Core ix 8xxx CPU with an GTX1050 gfx card. CPU TDP 15W, GTX150 more but nothing impossible to cool. But in reality, they again put that in a to thin case instead of having a decent case with bigger cooling solutions. Which you have in more powerful laptops but then again with more powerful components and you again have a heat/noise problem.
Why can't someone build a Laptop with "to big" cooling solutions that cool not that big components? Like my Desktop MSI GTX 1050TI Gaming X, gigantic twin fan cooler, completely overdone for a 1050Ti but stays cool and quiet and passive until it reaches 60°C! :D
 
Would it be possible to drill one or two holes next to heat sink? So the hot air with higher pressure could flow outside the case? Or would it not make a difference? I am not talking about production but hardware mods by users.
 
I was about to produce the aluminum case in the school I worked in so yes, I know a bit about mold production.
What I understand is that it's too late to modify the cases as it was my suggestion to get an external sided heat dissipator, not drilling holes.

Well, any holes that are on the SIDE of the case (regardless how big they are) can't be done directly in the mold, except for if you have half of that cutout in the top and half on the bottom part of the case (which is the case for the USB ports, for example).

Everything else that has plastic surrounded needs to be done with an inlay, and it's never too late to do something like that as that doesn't need a mould change but a new inlay.

Where did you want to have an opening and do you have a heatsink that would perfectly fit in there?

Misinterpretation strikes back. I never said that the OP is perfect for everyone (it is for my needs, also I often use its CPU at 100% overclocked for days).

Yes, which works because the OMAP3 and all ARM SoCs from that time and earlier were designed to run fullspeed without cooling.

This has changed with later CPUs like the OMAP5, which is designed to run at about 800MHz most of the time without overheating but also offers more speed for short periods of time (i.e. to speed up website rendering, which only needs more CPU power for a short time frame).

We can't change that design. It's not also just the heat, but if you're trying to run the OMAP5 with 1,5GHz full power, it will drain the battery within 1 - 2 hours!
Heat always means battery power goes poof!

What I said is that the effort was made towards this goal.
I feel that we still have a margin for optimization with the Pyra, and for the next-gen CPU boards, and the next etc., I express my opinion.

Of course, you can always optimize things.
Give me 500k EUR just for the development and I'd hire expert teams just to figure out how to create a design that offers the best cooling solutions with the smallest case possible.

The more parts you add to the Pyra, the more expensive it will be, and we're already on a FAR higher budget than planned.

Just a custom made heatsink will easily add 10k EUR at least (for the mold), and that's without production.

Ideas are fine, but they can also become very expensive quickly.
If you can find enough people to donate the additional money to me, I'll be happy to create a custom solution here.

Would it be possible to drill one or two holes next to heat sink? So the hot air with higher pressure could flow outside the case? Or would it not make a difference? I am not talking about production but hardware mods by users.

Sure, drilling holes into the case as self-modification would probably take five minutes.
 
Sounds like its best to settle for the 1ghz and look to more power on a future upgrade board.
 
Sounds like its best to settle for the 1ghz and look to more power on a future upgrade board.
That's the smart option.

What I'm more concerned about is the fact that any new CPU upgrade is going to be a fairly different CPU/SoC. OMAP5 is already several years old and TI clearly isn't going to continue the line. Not to mention a new SoC will have new memory demands, new driver demands, and so on.

Kinda curious, ED, is there any sort of "futureproofing" within the Pyra in that sense? Obviously the CPU board was made to be upgraded, but stuff like the memory is not. If this new SoC can't run on DDR3L (I think what's what's being used, anyway), then those of us who got Pyras are kind of hung out to dry - a form of early adopter tax, as it were. We'd need a motherboard plus a CPU board swap. Even with the potential for a loyalty discount, ED still needs to turn a profit - he can't merely sell the boards at-cost.

Not to mention that the OS has to then check for several hardware configurations, but at least those should be pretty consistent to test for.
 
I always thought that the memory was on the CPU board as well. Is that not correct?

<edit>
Yes it is. From the main site:
* The CPU-Board (CPU, RAM and storage)
* The Mainboard (all ports, Wi-Fi and Bluetooth)
* The Display-Board (for the screen)
</edit>
 
What I'm more concerned about is the fact that any new CPU upgrade is going to be a fairly different CPU/SoC. OMAP5 is already several years old and TI clearly isn't going to continue the line. Not to mention a new SoC will have new memory demands, new driver demands, and so on.

It should mostly need a new kernel supporting the SoC, everything else should be similar.

Kinda curious, ED, is there any sort of "futureproofing" within the Pyra in that sense? Obviously the CPU board was made to be upgraded, but stuff like the memory is not. If this new SoC can't run on DDR3L (I think what's what's being used, anyway), then those of us who got Pyras are kind of hung out to dry - a form of early adopter tax, as it were. We'd need a motherboard plus a CPU board swap. Even with the potential for a loyalty discount, ED still needs to turn a profit - he can't merely sell the boards at-cost.

No, the RAM is also on the CPU board, as is the eMMC.
Everything that's tied together. A new CPU board will almost ALWAYS require a different memory, so it wouldn't have made sense to put that onto the mainboard (besides the technical difficulty that the memory should have as short traces to the SoC as possible).

Not to mention that the OS has to then check for several hardware configurations, but at least those should be pretty consistent to test for.

Not the OS, that's the board file that's part of the bootloader, and maybe the kernel.

We already have support for multiple hardware revisions of the Pyra inside the board file. When a new revision is being released, the board file needs to be updated.
The rootfs itself (which is the OS) is not affected. It's a standard Debian Rootfs, basically, that would run on any system.
 
@EvilDragon
->Where did you want to have an opening and do you have a heatsink that would perfectly fit in there?

I pictured that, removing the bottom plastic part under the ports, so no hole, no drilling. A simple aluminium or copper plate rectangle is ok.
It could be pushed to the exterior, but there can be other solutions, I never saw an opened Pyra case.
And I don't _want_, I suggest.

->If you're trying to run the OMAP5 with 1,5GHz full power, it will drain the battery within 1 - 2 hours!

I don't care if the AC adapter is plugged, which I do when I compile mass stuff on the OP during days, and obviously on the devboards as they have no batteries.

->If you can find enough people to donate the additional money to me, I'll be happy to create a custom solution here.

I don't think my suggestion is expensive, but indeed every modification costs.
Maybe once the mass production is secured the kickstarter can be done and add some bucks.
 
I always thought that the memory was on the CPU board as well. Is that not correct?

No, the RAM is also on the CPU board, as is the eMMC.
Everything that's tied together. A new CPU board will almost ALWAYS require a different memory, so it wouldn't have made sense to put that onto the mainboard (besides the technical difficulty that the memory should have as short traces to the SoC as possible).
Derp. That's what happens when you wake up for work at 5:45 AM, kids. For some reason I thought the memory was on the mainboard.

In that case, yeah, it's a lot more futureproofed. Well, short of stuff like not being able to take advances in Wi-Fi tech or native USB Type-C or whatever, but the former is probably not an issue for most and the latter can be handled with cables or whatever.

Though I am kinda bummed it can't support USB-C with Power Delivery for its charge port. I've got a battery bank capable of that (keeps my Switch nice and juicy on the go), and while I can use that same bank to charge the Pyra too, probably won't be as fast since Pyra is still held to the charge current limits that most Micro-B plugs are - 5V@2.1A. It'll still charge fine, but it's probably gonna take awhile if the battery is pretty much empty. My charger on the A-Plugs supports 5V@3A, and the USB-C can output up to 30W at a variety of voltages and amperes.

Or am I wrong on that? I don't remember seeing a ton of what sorts of voltages/tolerances the charging port has for in-current. I can't imagine it'd be much more than standard though.
 
@EvilDragon
->Where did you want to have an opening and do you have a heatsink that would perfectly fit in there?
I pictured that, removing the bottom plastic part under the ports, so no hole, no drilling. A simple aluminium or copper plate rectangle is ok.
It could be pushed to the exterior, but there can be other solutions, I never saw an opened Pyra case.
And I don't _want_, I suggest.

Oh, you mean that small plate you posted here at the thread?
I doubt that would make a huge change, to be honest. It's pretty thin (unless you want to do your own custom thick copper block, so a heatsink connected with the backside, but that would be freaking expensive again), I don't think we'd see a huge change compared to what we have now.

Also, you'd need some way to FIX it into the case, which means at least one more inlay for the mold.
All in all, I'd suspect this would easily add 8 - 10k EUR to the development costs.
 
We have to stay realistic here, if I may provide a real world example :

The GPD Pocket uses an Atom X7 processor (TDP < 4W), active cooling with a fan, proper copper heatpipes and aluminium housing ... it still gets hot and throttles the CPU after a while
 
  1. Produce Pyras using ED's current ideas and plan.
  2. Produce extra cases for purchase.
  3. Sell thousands of Pyras - and several hundred 'extra' cases.
  4. Watch while mod community butchers cases to see who can get the highest clock rate at 'steady state' at room temperature surroundings. Let them compete on it and see what they come up with.
  5. Incorporate those ideas into the next case revision.
Evolution works.
 
Back
Top