Elementi taggati con: handheld
Is there really a light at the end of the journey...?
Go to the original news post
Well, well, well.
Are we really getting there?
Yes, it could be the case.
4GB RAM CPU Boards are working!
Finally, after a long, long search for the reason, Tony Lindgren found the real issue for 4GB RAM crashes.
It was a hardware and software fix. Even the previously produced boards work better now - and maybe a lot better with some changed timings :)e.
That does mean... well, the CPU boards should now be 100% finished.
And we might even be able to use the 500 bare PCBs we already have - if they can still be populated (as they have been sitting around for quite a while now).
A memory stress test still crashes after 10 minutes, but that is due to overheating, as it doesn't do this if a heatsink is added to the CPU board.
10 minutes without a heatsink is quite good though, as a memory stress test both fully stresses the CPU and memory, and doesn't happen in real life.
Nikolaus ran a 4GB Board without heatsink and 1,5GHz and did an apt-get upgrade which installed 400 new packages... without any issues.
After a few hours, he shut it down again. Without any crash or overheating!
Once the mainboards are finished and the new case has arrived, I'll assemble a full unit including our cooling solution, and then we'll no for sure.
This issue together with the rotation issue we had are most probably due to TI mostly having abandoned the OMAP5.
There are not many devices out there using the OMAP5 and therefore, no one has run into that issue before.
So yes, we might have some more low-level software work ahead of us compared to the Pandora, but it surely will be fun seeing the system evolve :)
Well, with the CPU boards and the DIsplay boards final now, the last thing missing are the mainboards.
What's up with those, you might ask?
Well, the have arrived at Global Components - but there was an issue. Some of the silk layer (that's the white print on the PCBs) are too close to some of the BGA pads. And as we don't populate single PCBs but multiple ones at once, the tolerance is an issue - and some BGAs might not be populated properly.
Unfortunately, the chinese manufacturer hasn't checked that thoroughly (last time, they fixed that issue themselves without telling us upfront). It would be possible to populate them by doing some VERY accurate measurements - but that would cost over 3000 EUR.
That was pretty annoying - but luckily, producing another set of bare PCBs WITHOUT this layer is a lot faster now, as all the tools have already been created and the software for the machines has been written.
The new PCBs will arrive June 18th (so in around a week) and will be populated shortly afterwards.
Well, according to my contact, a (hopefully) final case is already on the way to me.
It has been shipped a while ago via DHL but has not arrived yet.
He'll try to get a tracking number for me, and hopefully, it'll arrive this week.
This case should have the logo and everything glued in. But I won't assure that until we have it here.
... and then?
When the case is fine, 500 transparent cases will be produced, which will be used for the prototype preorders.
After that, the moulds will immediately be moved to the new company so we can produce the remaining cases in proper color. As most companies close down in August in Greece, we have to be fast here. Hopefully, that'll work out - then we could have a small release party at the GamesCom this year and who knows, maybe some users can already pick up their units there.
But yeah, that would be a miracle, as we also need to get the pre-cut material for the heat dissipation and I don't know yet how long the production will take. That said, it's existing material that just needs to be cut - so it shouldn't take long.
So who knows - one can dream, right? :)
With all the boards and the case finished, it's also time for CE / FCC testing, which we're planning for early July.
So yay.. we got another few steps closer to release :)
Final round... Fight!
#pyra #pyra-handheld #umpc #pc #hardware #debian #linux #console #dragonbox #handheld #computer #dragonbox-pyra
With full steam into the new year!
The original news post and discussion can be found here.
Whew, a LOT has happened during the last 10 days!
Mostly good news... so let's start right away!
1. CPU Board Redesign
As you know, Nikolaus worked on improving the vias and decoupling of the CPU Board to finally fix the 1,5GHz issue.
Well, and thanks to a community member, we were able to simulate the outcome so we wouldn't need to produce a new batch just for testing that (which would cost 9000 EUR and take 6 weeks of time).
The community member offered us to do the simulations... and in case you're asking why we couldn't do that ourselves:
The software used does cost about 90k USD... so... yeah :)
So we're really lucky. Where would we be without the community?
He simulated and compared the old revision and the new one with Nikolaus changes and checked that it fulfills everything that's needed according to the datasheets.
Here are some nice pictures that demonstrate the improvement:
Before someone cries in panic: RED does not mean BAD here.
The target voltage is 1,05V and the color shows how much voltage would be where.
So RED is great, green is okay and the more blue we have, the higher the voltage drop is (which is bad).
Here's a nice summary:
As you can see, while we're not reaching TIs ideal specifications 100%, the boards have been improved A LOT!
Here you can see the changed impedance of the old and new revision:
Quite a bit of improvement here!
The software can also simulate what happens when you add more capacitors.
We know it's not possible to achieve ideal conditions with that little space we have on our PCB, but that's not needed.
Ideal conditions are just the conditions under which the manufacturer guarantees that the SoC will work for sure.
This is rarely being achieved with such tight PCB designs, but they still work.
Heck, our current CPU Boards work with up to 1,2GHz, and they're MUCH worse than our improved layout.
This improved layout should easily enough to reach the 1,5GHz, but we can probably still squeeze one more capacitor to improve it even further.
So yes, our redesign worked, the possibility to simulate the behaviour was the best that could've happened to us!
Time to produce a few more CPU boards with that new layout - which should definitely be the final one!
2. The keymat
This monday I also received 30 samples of the tweaked keymat.
So, are we there yet?
YES - and no! But that no is not really an issue.
YES: The action buttons and keyboard feels GREAT! They are really nice to press, have a good pressure point.
So yes, the keymat is fine. No need to change it anymore, I really like it.
So what's the no?
You might remember that I decided that the plastic part of the DPad should also be produced by the keymat manufacturer and be glued on top by them.
Well, that plastic part is the first version and is feels great, it has just one minor issue:
The pivot goes down a bit too much. Which means: When you place the keymat onto the table, the pivot already touches the table without you even touching the DPad.
This, of course, makes it feel awkward.
Just using a bit of sandpaper fixed that issue, and for the mass production, it's a minor fix as well.
The most important thing is that the keymat is FINALLY finished! That was a pretty complex part, with all the layers, transparency, etc.
And now it's done - I'm really relieved. One thing less to worry about.
The DPad is a VERY simple part that can easily be changed and even printed with a 3D printer.
3. The mainboards
GC has populated the new prototype mainboards and display boards this Tuesday.
The last thing needed to assemble them to final units are some more cases (I've only got five here) - everything else is here.
So we can assemble multiple units now - and once I receive the next batch of cases, we can assemble 28 prototype units!
Now, some parts are lying around at Nikolaus' place, some at GC, some here... so we need to meet up to assemble everything, I hope we'll find the time for that next week!
4. The displays
800 displays have arrived at Global Components - 230 have been killed by the delivery service (by crushing a box against a wall...)
These will be replaced, of course - and I've asked them whether I could order 500 more... so maybe it's good that this happens ;)
The will be here in time, no worries. We'll need quite a bit of time to assemble the units (a new mass production will always take some time to speed up), and until then, the displays will be here.
5. Quick summary
Pheeww. So, CPU Board should be fixed now. Motherboard and Display board are done as well.
Keymat is finished, just the DPad needs a tiny fix.
Cases are finished except for the wiggling stylus and the too-hard-to-open battery cover.
Displays are finished, touchscreens are perfect.
All electronic parts needed for the mass production are already lying around at Global Components.
We've gotten a LOT closer to the finished Pyra now.
Now I'm hoping for the next case revision and the prototypes... yay!
#pyra #pyra-handheld #handheld #computer #gaming #hardware #linux
Please help DebianMobile!
If you care about open hardware and free software, if you care about a trustworthy operating system in your pocket, please consider to help getting Debian run on mobile devices. Join the effort and let's see, what can be achieved. Currently, this is a hackers/developers party, there is not much for end users, yet, so please be warned and don't hold your breath. #debian #debianmobile #openhardware #freesoftware #mobile #handheld #pyra #neo900 #zerophone
Back from the Dead!
Go to the original news post
Well, I'm back! Not from the dead, but as I didn't really visit the boards or IRC a lot lately. I might've felt to some like I was dead.
I'm doing fine, no worries. But the shop was swamped with orders until two days ago (I was packing and shipping 40 - 100 orders per day!) that I didn't find the time for much else.
Thankfully, things are back to normal again, so here I am! And I brought some news with me, as things have not stopped in Pyra development.
1. The SATA-Adaptor
Not everyone is interested in one, I guess, but the SATA-Adaptor has mostly been finished. It plugs into the USB-Port of the Pyra and splits the signal to both USB 2.0 and SATA. Keep in mind that it's only for the Pyra.
I'll create a small case for it which you can either print yourself or maybe I'll do an injection mould plastic of it as well (depends on the demand and costs).
There's no direct JTAG connection on the Pyra PCBs (we didn't have enough space), but the SD-Card slot can be used for it! All you need to do is boot a special kernel and use this simple SD-to-JTAG-Adaptor.
This is helpful for everyone who wants to do some low-level hardware debugging.
3. The CPU-Board fixes
Nikolaus has used the last few days to vastly improve the VIA- and capacitor-setups on the CPU-Board. He's rechecking it currently to make sure it's the best we can achieve with the current layout. After that, as mentioned in the last post, it will be sent to a community member who can run a simulation on it which will show us the improvements and make sure it's good enough for the full 1,5GHz speed.
4. And the rest...?
Well, all companies involved in the Pyra production had holidays, and like most companies here, they'll resume work on Monday.
Then I'll bugger them all to move ahead with full speed, and of course I'll keep you informed.
Oh, and by the way: A happy New Year 2017 :) May our journey finally bring us to the finish line :)
#pyra #pyra-handheld #handheld #hardware
For the friends of the decimal system and the believers of the importance of many trailing zeros: Pyra handheld sold already 1000 times - that is 3e8 in hex, an impressive number!
Now let's wait, until they get the thing ready...
#pyra #debian #freesoftware #almostopenhardware #handheld #dragonbox #openpandora
go to the original news post
My normal daytime job is still giving me a lot of work... I had to edit 3 commercials and one cooking show last week, so combined with the daily shippings in my shop, I didn't have much time for anything else.
Of course, that doesn't mean nothing has happened, as there are other people working on the #Pyra :)
1. The mainboards
The bare prototype mainboards (which most likely won't be different from the production run mainboards) will arrive at Global Components December 12th and will be populated shortly after (like the CPU boards, 30 in number).
As soon as we have the cases and the keymats for those, we can assemble them for devs, Pyra preorderers and the remaining ones will be auctioned so anyone can get them (yes, the vouchers for the Pyra can be used to get one of those if you like). These will clearly be marked being prototypes, so they might be of some value to you :)
Speaking of keymats and cases...
2. The keymats
I received some inbetween sample of the keymat, but it was a bit rushed (as they wanted to finish it as fast as possible) and in my opinion, it was way too stiff. I told them to hurry but not rush. It's more important to get perfectly working samples one or two weeks later instead of imperfect ones as fast as possible.
The good thing is, once they're working perfectly, mass production of them won't take long and can be done within a few weeks.
I'll keep you posted. This won't prevent me from building a new prototype though, as the look of the keymat won't change, only the stiffness and feel of it.
3. The cases
According to UPS tracking, a sample of the new case including the sealing for the speakers has been shipped November 28th and will arrive tomorrow until the evening.
Let's hope everything will be fine now - but we won't find that out until tomorrow or Tuesday (depending whether UPS will keep their time :)) Of course you can expect some highres pictures of those - but don't expect them to LOOK perfect, as the moulds won't be hardened before everything is 100% okay physically (and therefore you will probably see a few discolorations in there).
Once the new mainboards are finished, I can finally built a prototype that works perfectly fine using a battery, as these will have the battery connectors in the proper position (running it with battery was a bit of hit and miss before).
UPDATE (did I mention I had too much work? I almost forgot this:)
4. The CPU boards
Nikolaus has been done countless tests (also huge thanks to notaz for a NEON-Stresstest-tool) to rule out different possibilities why the CPU boards could fail running higher than 1,2GHz. It could've been the RAM, the clock itself, power source, software settings, etc.
With these tests we could slowly rule out various issues. The RAM is fine, the power source is fine, so what we're most likely missing are some more decoupling capacitors. So there's a very good change the next CPU boards will work fine. I'll let you know once I get more details.
Exciting times :)
#pyra-handheld #hardware #linux #hardware #handheld #debian #news