Two tales to tell


EvilDragon

Administrator
Staff member
Joined
Mar 4, 2003
Messages
29,986
Age
46
Location
Ingolstadt
Phewww, sorry for the long time without any updates.
I catched the flu last week and while I could continue to work, I was a bit dizzy - so not in the best condition to post an update :)

Many of you have asked "so, with the last known issue fixed, what's left now before we get our Pyras?"

You shall get an answer in the next post (tomorrow or the day after).

But first, let me tell about two more small issues we found last week, which we already fixed (most probably :))


1. The tale of the battery that was overheating without getting hot

No, the tale doesn't have anything to do with a korean company ;)

As you know, we fixed the power stability issues last week, so we hoped that now battery charging, etc. should work properly as well...
Umm, well, nope sadly, it didn't.

But the reason for that was very, very simple. While testing the battery circuit, Nikolaus saw that the charger chip reports the battery is overheating... even if it's still cold.
There's something wrong, that shouldn't be the case.

It didn't take long to find what's wrong: If Ground and the middle Test pin of the battery have a connection, the charger chip thinks it's overheating.

Well, I made some pictures...

IMG_20161014_170146.jpg IMG_20161014_170036.jpg

You can already see here that the battery pads aren't sitting in the middle of the case but the connectors are!

It can be seen even better with the overlay picture I created:

Akku.jpg

Aaah, yes. Pretty clear this can lead to a short circuit.

But good news: Unlike in the Pandora, the battery in the Pyra sits rock-steady in the case and doesn't move around.
I was only able to move it around by 0,1 - 0,2mm.

And the battery connector on the PCB will also always be in the exact same position.
Aaaand we got about 1,8mm tolerance (so the battery can move 0,9mm to each side without losing contact or causing shortcircuits).

So there's only one thing wrong here: The position of the battery connector. We measured the exact position and ordered a testing PCB already which should arrive within short time.
That should be the final tweak needed for the battery.

And as we're still waiting for some parts and the cases for the prototypes, we don't even lose any time with that.

But more importantly:
No more contact issues as we have them on the Pandora. Yay :)


2. The tale of the prototypes that resisted to boot

With the power supply issue fixed, I wanted to build some more prototypes to ship them to some of our kernel developers.
However, I couldn't get them to work properly, whatever combination of CPU Board, Mainboard and Display-Board I tried.

BUT they worked find without a display board attached to... eh?
Still some power issues?
But my prototype runs rock steady now, off a normal USB AC Adaptor without any battery.

That's something I had to investigate...

First thing I tried was plugging all the CPU boards into my working prototype. If they would all boot up fine, it would show that the CPU boards are intact.
And indeed, they booted up fine.
So the CPU boards were okay.

Hm. But I couldn't imagine that ALL other mainboards and display boards were broken... so what could it be?
The LCD cable...?

And yes, it seemed like it:
I tried reseating the LCD Cable and even tried different ones.

The interesting part: The reasons the unit didn't boot up were different, but consistent each time I reseated the cable.

That means:
As long as I didn't touch the cable, the reason was always the same one. And reseating it changed the issue.

Sometimes it didn't find the display, sometimes not even the LEDs on the mainboard could be found - and sometimes, it worked!

The mainboard LED one was interesting: The LEDs are on the same i2c bus as the LEDs in the display. But if they don't work, it's something more serious - like a shortcircuit in the cable somewhere.

And THAT was the clue:
The cable and connector we're using have two pin rows: One at the edge of the cable and one a bit further up.
You can probably see better what I mean with this picture:

lcdcable.png

The pink/purple things are the connectors. By using these two rows, the connector can be made less wide, so you safe space.

But what happens, when the cable is not fully inserted? Right, the lower row makes a shortcircuit, connecting both rows together.

Hm, but why should that happen now, we've tested the cables and they worked without any issues... well, except that so far we only used the FIRST version of our cables - and for the prototypes, I used the SECOND version of them!

And yes, it seems we accidentally moved the EARS a bit (those two things at the side of the flex cable that make it easier to insert the cable into the connector).
And due to that accidental movement, the cable only goes in by 2,2mm into the connector before it's blocked by the ears - and according to the spec sheet of the connector, it needs to be inserted 2,5mm!

Geez! That's the reason it caused various issues and sometimes even worked when I pushed it in real hard.

The fix is easy:
For the prototypes, cut the ears and the cables work.
For the final units, move the ears further up again.

So yay, also fixed, and also not any time lost as we found that all out while waiting for other parts :)


Wondering what's left, what will happen next and what the status of everything else is?
You need to be a bit patient. I'll do a comprehensive update post soon. Hopefully tomorrow, but that depends on when I can find the time to do that.
 
Is the battery some standard size and configuration, or is it custom designed for the Pyra? If the latter, what is/was the process of getting them made?
 
Is the battery some standard size and configuration, or is it custom designed for the Pyra? If the latter, what is/was the process of getting them made?

Custom designed, by the same company that produced the Pandora battery as well (one of the largest battery manufacturers worldwide).
 
The battery uses fairly standard +/-/sense pins and a pretty normal internal chemistry, but the exact dimensions are custom designed for this unit. AFAIK ED talked to the same battery company that's been successfully producing custom batteries for the Pandora to make one for the Pyra.

Weird happenings with the LCD cables. Have you tracked the issue back to the layout files that have changed, or was it some manufacturing variance that will be harder to predict that caused this?

FWIW, I also had an infection this week, but it was only a head cold with a slight fever, and symptoms were mostly resolved after the first day.
 
I've been checking 4 or 5 times a day for an update! Release! It is good to hear you're getting everything ironed out, can't wait for the next update!
 
Haven't checked out how pyra is in a while, and surprised it still is into prototyping. I thought they became available in ~july and i'm too slow to this.

I have suggestions.

1. Wouldn't it be better to solve short-circuiting issue geometrically? Like, if you'll make contacts on the battery a bit slimmer, there would be no short-circuiting (instead this issue would cause power outages, if the charger chip doesn't do them already after seeing overheat)

2. Maybe overheating should be fixed a straightforward way, through some sort of metal cooling chunk that goes into the frame (similar to the way laptops do this) which could be then made out of aluminium alloys. That would definitely allow for great SoC replacements in the future. There are ones that make a frying pan out of classical smartphones. Might not be physically possible and can be too significant change at this stage of pyra, but since it's not to the pcb - no bugs to test out. May worth considering.
While i'm at it i'd also suggest going for BGA that has more SoC options for future replacement. I think BGA are what board designers struggle with mainly, if i'm not wrong. There have to be thousands of very different BGAs.
 
Haven't checked out how pyra is in a while, and surprised it still is into prototyping. I thought they became available in ~july and i'm too slow to this.

I have suggestions.

1. Wouldn't it be better to solve short-circuiting issue geometrically? Like, if you'll make contacts on the battery a bit slimmer, there would be no short-circuiting (instead this issue would cause power outages, if the charger chip doesn't do them already after seeing overheat)

2. Maybe overheating should be fixed a straightforward way, through some sort of metal cooling chunk that goes into the frame (similar to the way laptops do this) which could be then made out of aluminium alloys. That would definitely allow for great SoC replacements in the future. There are ones that make a frying pan out of classical smartphones. Might not be physically possible and can be too significant change at this stage of pyra, but since it's not to the pcb - no bugs to test out. May worth considering.
While i'm at it i'd also suggest going for BGA that has more SoC options for future replacement. I think BGA are what board designers struggle with mainly, if i'm not wrong. There have to be thousands of very different BGAs.


you really should read the entire post. The battery does not overheat. The chip says that it would even if its cold.
 
So there's only one thing wrong here: The position of the battery connector. We measured the exact position and ordered a testing PCB already which should arrive within short time.
That should be the final tweak needed for the battery.
Does this mean that the 30 boards with an estimated arrival of October 20th for the 2nd prototype run will need to be re-started?
Do you have a new estimate on when the 2nd prototype board production run can happen?

And as we're still waiting for some parts and the cases for the prototypes, we don't even lose any time with that.
I'm glad to see it all coming together. We're at the 1 year anniversary of the pre-pre-order and prototype pre-orders aren't we?

Thanks for the update!
 
I really don't know what to do
pre-order the Pyra & hope i get it in time for Christmas for my son (& myself don't tell him )
or buy a GPD Win which is not as good as the Pyra and you cant upgrade it
if I knew for sure it would be ready for Christmas I would pre-order right now
oh the dilemma
and before you said it I am not buying both
money is an issue for some people like me
 
Back
Top