Not much text here. Just a video.


And I suspect that's glue on the front, but when talking about cooling the frosting on the top looked a bit like heat damage on the case to me.
Yes, I think it's the unit I commented on in the news thread with all of the prototype build pictures we had. It seems to be glue that goes on clear but dries white (which is entirely backwards if you ask me, but you didn't), and he put slightly too much on when gluing the logo plate in, and tried to clear it up, but ended up smearing a little over a wide area. The release units will use pre-glued logo plates, so this won't be an issue.
 
Hi all,

@EvilDragon : thanks for the very interesting video ! Is that some gray hair I see on your right temple ;)

More seriously though, and I'm sorry if this question has been asked before, but:

Which version of Debian will the Pyra run when it ships ? 8 or 9 (with "Buster" 10 already frozen, and "only" ~130 RC bugs to fix) ?

Cheers, Magic Sam
 
Which version of Debian will the Pyra run when it ships ? 8 or 9 (with "Buster" 10 already frozen, and "only" ~130 RC bugs to fix) ?
Last I knew it was using some flavor of 9, but Knowing aTc he will likely try to push to Buster when he gets time.
 
Last I knew it was using some flavor of 9, but Knowing aTc he will likely try to push to Buster when he gets time.

It will run whatever Debian calls stable. You can of course run anything you want on it, but as something we have to sort of officially support as a default OS, stable should give the least amount of problems.

Then again, if we run into issues where we need lots of newer library versions than what's available in stable, i have no problems to move to something different :)
 
I know that a first-boot wizard is something that you can see by booting up any Linux version and that the Pyra’s one isn’t anything special, but I still appreciate seeing it on the Pyra anyway because it helps me visualize what I’ll be doing in the first few minutes after getting mine.

What struck me with the video is when you were testing the keys. You could test a whole row in less than a second. I immediately thought of a cheap calculator with protruding, sharp-edged rubbery buttons. You couldn’t test buttons like that by rubbing your finger along them, since every single one would try to grab and hold your finger. It gave me a feeling of how nice the Pyra’s keys must be to use.
 
Hmm. Thinking about the boot speed of Gimp, my previous comment about it appearing slow was misplaced as I missed that it's being booted from a standard SD card which is bound to be slower...
 
Thanks ED really cool to see some Pyra action.
Regarding Text or Video i prefer text because i could understand well as my english is not so great (but i do like and watched every video you made).
 
I think the hardcore group of us that been since the beginning of the Pyra development or even through the Pandora development will be fine with just text updates, but I feel perhaps there is a fair amount of people that see these regular posts as more the same, just another post showing it will be a bit longer to get out the door. But a video like this that shows progress to them is likely better than any forum post.
 
I think we all have a little more grey hair than when we started this journey. In my experience, gray hair can come and go when you're under stress (or at least it could when I was young enough to restart my melanin production spontaneously when the stressful situation ended), but in my case at least we're now at the phase where once you go grey you stay.

IIRC ED is a few years younger than me though. Everyone's hair is different, but assuming I'm somewhere near the mean, he's likely to have a good few years before the grey patches start to join up.

As for the Debian version, I assume we'll start on Stretch (9.9), which already has the latest LTS support (so, bug fixes up to 2022). It's the work of an internet connection and a couple of terminal commands to upgrade to buster when it's ready. Of course, this can all change up till the date of the first production, but I guess most of the bug fixing work has gone into stretch so we'll start with that, and once a few of us users have shifted to buster, we can tell the team what works and what doesn't and start moving everyone over to buster. I assume Buster will have LTS support to 2024, but that's not been announced yet.
 
...
This has to be confirmed - but there's nothing else that could possibly cause this.
...

Crazy 'nothing' to consider... What if there were some insulating material on the board or the keymat - or if the keymat were just a bit out of alignment (no alignment pins remember) such that the keymat isn't making correct contact for the non-working button?

How to test... Hmm... Cut the button out of a keymat, skin it down to just the inside contact surface, mount it in the eraser holder for a click pencil, use that on a unit assembled without a keymat in it, poke the button contacts through the empty key holes.

I guess I'd want to be 100% certain that it isn't a contact issue before diving into diodes under microscopes.
 
I guess I'd want to be 100% certain that it isn't a contact issue before diving into diodes under microscopes.
Or check it's not some weird software definition issue before cutting into keymats. A lot of this is up to what is defined in the device tree binary.
 
Last edited:
Yeah, I can't see how a diode can be reversed in a modern production routine, unless it's backwards in the intial schematics. That should be pretty much automatically converted into a layout once you've set the parameters (although there's always a certain amount of hand fettling), and that should be automatically made into the PCB and fed into the pick and place machines. I can't see the tapes in the hoppers having randomly oriented diodes on them; that would pretty much invalidate them for any machine usage. There's always likely to be a certain number of duff units on any tape of sufficient length, but that doesn't explain why the same keys always stop working.

It should be easy to see if the diode has been misplaced, and is only held by one pad or entirely missing now. And even 20EUR multimeters these days have diode testers in them, so if you can physically get at the contacts to the part you can test its orientation fairly easily. I personally doubt the key is missing the contact patch; both the key pad and the contact patch are pretty large on the scale of things, and being right on the edge can be visually inspected easily by lifting the corner of the keymat (with the whole assembly outside of the case of course).

Edit: Or, as Trashy suggests, it could always be a software issue.
 
Or check it's not some weird software definition issue before cutting into keymats. A lot of this is up to what is defined in the device tree binary.

Or just use a paper clip or 'test wire' to jump the connections on the circuit board - which should test to see if the key's contacts are live.

Yeah, my idea with stealing a contact out of a keyboard key to use as a tester was a bit over the top.
 
Crazy 'nothing' to consider... What if there were some insulating material on the board or the keymat - or if the keymat were just a bit out of alignment (no alignment pins remember) such that the keymat isn't making correct contact for the non-working button?

The keys don't even work when the board is removed and I'm manually making the contact with a special tool we built back in Pandora times (basically a tip with a carbon pad on top).
Nope, it's not a contact issue.

Or check it's not some weird software definition issue before cutting into keymats. A lot of this is up to what is defined in the device tree binary.

That would affect all units, not just a few boards with different keys.

Yeah, I can't see how a diode can be reversed in a modern production routine, unless it's backwards in the intial schematics.

Oh, they sometimes are reversed inside the reel even. I had a few Pandoras where some LEDs were reversed and didn't work :)
The more tiny things get, the more stuff goes wrong.

It should be easy to see if the diode has been misplaced, and is only held by one pad or entirely missing now. And even 20EUR multimeters these days have diode testers in them, so if you can physically get at the contacts to the part you can test its orientation fairly easily.

Diode testers only work reliable when the diode are removed from the PCB. You can't see if a diode is reversed, as these are about 0,4mm x 0,2mm parts... you won't find markings.
Yes, if one is missing or misplaced, you can see that, which is what Nikolaus will do. I don't even know where the diodes are on the PCB :)

And no, it can't be a software issue, as all units would behave like this in this case.
 
Yes, diode testers can be tricky to use in situ. I'd tend to think that the current between the pads of a diode in line with either a row or a column should be minimal provided no keys are being pressed, but yeah, if you don't even know where it is you can't test it.
 
Interesting Video, not much to complain about. Except the Phillips-head screws on the prototype, they are butt-ugly! xD Showing screw heads is fine, has some retro-tech-feeling. But then I would choose flat, single slotted screws. ;)

And out of curiosity: how thick would this Pyra-phone be with battery?
 
Back
Top