It's time for another lengthy newspost, full of fresh and juicy stuff!
1. The cases
Probably the most anticipated topic by everyone here, so let's start with that one first.
As expected, nothing has happened anymore at the old company, so it's time to move.
Originally, my contact planned to move the moulds himself, but it's safer (and insured!) to let a logistics company do that.
The moulds will be picked up on Monday and they'll be delivered the same day.
My contact will also be there on Monday to plan a test run with multiple materials (they got 15 different ones in stock for testing).
After that, I'll fly over to Greece to test them out and check the results.
If everything is fine and we'll have proper material in the selection, we can finally plan the mass production.
All of that is planned to happen in November.
They also have WAY better contacts for various other materials. We can basically get ANY material out there within 15 days if need be (compared to 10 weeks that the previous company needed.)
Additionally, they already produced an SNES shell with their own custom made material which has already been shipped to me.
And apologies to you, as the material is NOT a mixture of carbon with plastic, but rather glass fibre with plastic. It's a LOT more robust.
But I'll make a video for you where I torture the poor SNES shell
Okay, so much for the cases, let's move on to...
2. PCB work
... as Nikolaus has been very productive this week as well.
He finished modifying a mainboard to test all the remaining issues we had and tested them.
Here's a list of most of the fixes:
This fixes all known issues the mainboard had. The last thing is tweak the trace to the Wifi antenna (which will be done with a quick dummy PCB).
That's good news, I think
Next up: The CPU Board (4GB issue)
Some final tests have been made, with 4GB RAM from a different manufacturer, and they mostly confirm what we already expected:
The noise / groundplane on the CPU board needs to be improved.
How is that confirmed?
Well, regardless of the manufacturer, CPU Boards equipped with 4GB RAM show the same issue: They are pretty unstable and freeze easily.
However, some of the CPU boards work fine for hours (with memtester), some work but show errors (with memtester) and some won't even boot.
This shows that this is no software issue, but rather a hardware issue. And as there are fluctuations (tolerances) between the PCBs and we KNOW our groundplane is off the allowed tolerances, we can be 99% that's our problem.
So Nikolaus has already started to improve the layout of the CPU board.
Regarding the already produced bare PCBs which would work fine with 2GB:
We have some plans what to do with them, but nothing is set in stone here, so I won't tell anything yet
But we might also offer to downgrade to 2GB RAM if you like to receive the Pyra a bit earlier. This depends on how fast we get the cases done and the final mainboard.
Please don't email me about that now - we'll let you know in time, no worries
1. The cases
Probably the most anticipated topic by everyone here, so let's start with that one first.
As expected, nothing has happened anymore at the old company, so it's time to move.
Originally, my contact planned to move the moulds himself, but it's safer (and insured!) to let a logistics company do that.
The moulds will be picked up on Monday and they'll be delivered the same day.
My contact will also be there on Monday to plan a test run with multiple materials (they got 15 different ones in stock for testing).
After that, I'll fly over to Greece to test them out and check the results.
If everything is fine and we'll have proper material in the selection, we can finally plan the mass production.
All of that is planned to happen in November.
They also have WAY better contacts for various other materials. We can basically get ANY material out there within 15 days if need be (compared to 10 weeks that the previous company needed.)
Additionally, they already produced an SNES shell with their own custom made material which has already been shipped to me.
And apologies to you, as the material is NOT a mixture of carbon with plastic, but rather glass fibre with plastic. It's a LOT more robust.
But I'll make a video for you where I torture the poor SNES shell
Okay, so much for the cases, let's move on to...
2. PCB work
... as Nikolaus has been very productive this week as well.
He finished modifying a mainboard to test all the remaining issues we had and tested them.
Here's a list of most of the fixes:
- Removed an unneded connection between OMAP/BT UART and Modem
- TCA6424A replaced by PCAL6524HE (which should provide better interrupt handling)
- New Wifi antenna has been chosen: Ethertronics M830520.
- EMU switch (NX3L2267) controlled by NRESPOWERON has been added to CPU schematics. This isolates the shoulder button and LID sensor for the first 10 ms after powering on the OMAP5. This fixes an issue where the system would not boot up when either the lid is closed or a shoulder button is pressed while powering up or rebooting.
- Added fixes for OTG mode
- Found the capacitors that are responsible for the sirring noise. (This doesn't need a new board layout, just different capacitors (various will be tested soon)
This fixes all known issues the mainboard had. The last thing is tweak the trace to the Wifi antenna (which will be done with a quick dummy PCB).
That's good news, I think
Next up: The CPU Board (4GB issue)
Some final tests have been made, with 4GB RAM from a different manufacturer, and they mostly confirm what we already expected:
The noise / groundplane on the CPU board needs to be improved.
How is that confirmed?
Well, regardless of the manufacturer, CPU Boards equipped with 4GB RAM show the same issue: They are pretty unstable and freeze easily.
However, some of the CPU boards work fine for hours (with memtester), some work but show errors (with memtester) and some won't even boot.
This shows that this is no software issue, but rather a hardware issue. And as there are fluctuations (tolerances) between the PCBs and we KNOW our groundplane is off the allowed tolerances, we can be 99% that's our problem.
So Nikolaus has already started to improve the layout of the CPU board.
Regarding the already produced bare PCBs which would work fine with 2GB:
We have some plans what to do with them, but nothing is set in stone here, so I won't tell anything yet
But we might also offer to downgrade to 2GB RAM if you like to receive the Pyra a bit earlier. This depends on how fast we get the cases done and the final mainboard.
Please don't email me about that now - we'll let you know in time, no worries
