Whew - finally, I find the time to sit down and update you on what's going on with the Pyra devlopment / production.
Time to start with the case.
So, what has happened in Greece during the last few weeks?
While it seems at first not much has happened, they have made some important progress.
According to my contact, they have fixed the remaining two issues (shoulder buttons and stylus) and he personally went there regularly to check the shoulder buttons and told them if it needs more changes or not.
I hope he's as picky as me, as that would mean they should work 100% in all cases now
After he approved them everything, they made a small test run of 50 parts to test the tolerances during production.
Producing ONE case with working shoulder buttons doesn't mean that they will all work with 50 different mass produced parts!
So they had to produce them... and then test the functionality of the shoulder buttons with all those 50 case.
Quite a lengthy but very important process if we don't want to have any bad surprises when we finally start the mass production!
The final result is on the way to me - but I'm on a short holiday trip from Tuesday to Sunday next week, so I won't be able to test it before I get back.
You can be sure it'll be one of the first things I do though!
Unfortunately, I didn't receive an answer to my question what exactly is included in the package so far (well, it's weekend and my contact also wants to have some free time).
I hope that it will have the logo glued in as well and the aluminium plate for cooling, as that would mean I can build a full Pyra!
Next up, the PCBs:
We're still waiting for the mainboards, but they shouldn't be too far away now.
As you know, we received the improved 4GB RAM CPU boards and you're all eagerly awaiting results (as am I!)
Well, unfortunately, a quick test by Nikolaus didn't show much improvement, it was still running unstable.
You can probably imagine how much that demotivated me... however, after sleeping over it for one night, I was able to think properly again.
First, Nikolaus test doesn't say much. We played around with a LOT of parameters trying to get the 4GB RAM to work with the previous PCB version - including some that noted "Do not use for in live mode" in the datasheet.
Nikolaus just put in his testing card from back then, he didn't change or review any parameters, so it could be the setup is completely wrong and therefore the RAM is unstable.
For further testing and debug, some of the CPU boards have been sent to Tony Lindgren (an expert in OMAP things). He received the boards but didn't have time yet to work on that yet, but I can't wait for results!
We also wanted to send some of the PCBs to zmatt, but he hasn't replied to any of our emails yet. I sure hope he's doing fine and just incredibly busy - hopefully, nothing has happened to him... you never know on the internet. Though he's still idling around in IRC, which is a good sign.
This brings up another important point:
Kernel developers!
There's still quite a few things which should be done before the release. We don't have proper powersaving, the TILER implementation needs to be tidied up, 3D is not yet implemented, Audio needs a better setup, etc.
It seems there are less and less kernel developers having the time to work on such things in their spare time.
That's why I decided to hire freelancers to help out as well!
We already have one who has knowledge of the OMAP series and is working on the AM-series from TI at the moment. He's a friend of Nikolaus and has already done work on the GTA04.
He agreed to help us out, but he doesn't have too much time - still, any help is good
Now, here's a question for you:
I know we've got quite a lot of OpenSource fans around here. Maybe some know some good kernel developers, who are able to include and improve hardware support and fix various issues.
We can provide a test unit as well as the needed datasheets - but it needs someone who is capable of debugging and fixing low-level things.
I am fine paying for the work as well.
Any ideas or contacts?
Okay, back to our next steps:
As I don't know yet whether the 4GB RAM will work properly or not, I'm also thinking of how to continue in case they DON'T work yet (let's hope they do though!)
I always like to have a plan B ready upfront
Well, let's assume everything is working and ready except for the 4GB RAM version.
In this case, I'll start the mass production - with 2GB RAM boards. No need to wait further if everything is ready!
Quite a few of you have signalled they will be happy with 2GB RAM for a start. And 2GB RAM will eat a bit less battery as well.
Of course, we will continue to work on the 4GB RAM boards so we'll have them available some time later. We already own the RAM chips - it would be a waste of 30k EUR not to use them!
There's a company in Munich (so round the corner) who claims they can get ANY RAM chip to work (and so far, they did, according to our memory distributor).
They have measurement devices that go inbetween the RAM chip and PCB and monitors EVERY signal.
So if something fails, they will EXACTLY know what.
We'll get that to work, even if we start with 2GB RAM first.
So much for now. I really had hoped to post better news than that (you can imagine how happy I would be with better news!) but that's life. And everything Pandora-related ALWAYS had been hard work instead of good luck.
But as usual, I'll keep going forward! No way to give up!
Time to start with the case.
So, what has happened in Greece during the last few weeks?
While it seems at first not much has happened, they have made some important progress.
According to my contact, they have fixed the remaining two issues (shoulder buttons and stylus) and he personally went there regularly to check the shoulder buttons and told them if it needs more changes or not.
I hope he's as picky as me, as that would mean they should work 100% in all cases now
After he approved them everything, they made a small test run of 50 parts to test the tolerances during production.
Producing ONE case with working shoulder buttons doesn't mean that they will all work with 50 different mass produced parts!
So they had to produce them... and then test the functionality of the shoulder buttons with all those 50 case.
Quite a lengthy but very important process if we don't want to have any bad surprises when we finally start the mass production!
The final result is on the way to me - but I'm on a short holiday trip from Tuesday to Sunday next week, so I won't be able to test it before I get back.
You can be sure it'll be one of the first things I do though!
Unfortunately, I didn't receive an answer to my question what exactly is included in the package so far (well, it's weekend and my contact also wants to have some free time
).I hope that it will have the logo glued in as well and the aluminium plate for cooling, as that would mean I can build a full Pyra!
Next up, the PCBs:
We're still waiting for the mainboards, but they shouldn't be too far away now.
As you know, we received the improved 4GB RAM CPU boards and you're all eagerly awaiting results (as am I!)
Well, unfortunately, a quick test by Nikolaus didn't show much improvement, it was still running unstable.
You can probably imagine how much that demotivated me... however, after sleeping over it for one night, I was able to think properly again.
First, Nikolaus test doesn't say much. We played around with a LOT of parameters trying to get the 4GB RAM to work with the previous PCB version - including some that noted "Do not use for in live mode" in the datasheet.
Nikolaus just put in his testing card from back then, he didn't change or review any parameters, so it could be the setup is completely wrong and therefore the RAM is unstable.
For further testing and debug, some of the CPU boards have been sent to Tony Lindgren (an expert in OMAP things). He received the boards but didn't have time yet to work on that yet, but I can't wait for results!
We also wanted to send some of the PCBs to zmatt, but he hasn't replied to any of our emails yet. I sure hope he's doing fine and just incredibly busy - hopefully, nothing has happened to him... you never know on the internet. Though he's still idling around in IRC, which is a good sign.
This brings up another important point:
Kernel developers!
There's still quite a few things which should be done before the release. We don't have proper powersaving, the TILER implementation needs to be tidied up, 3D is not yet implemented, Audio needs a better setup, etc.
It seems there are less and less kernel developers having the time to work on such things in their spare time.
That's why I decided to hire freelancers to help out as well!
We already have one who has knowledge of the OMAP series and is working on the AM-series from TI at the moment. He's a friend of Nikolaus and has already done work on the GTA04.
He agreed to help us out, but he doesn't have too much time - still, any help is good
Now, here's a question for you:
I know we've got quite a lot of OpenSource fans around here. Maybe some know some good kernel developers, who are able to include and improve hardware support and fix various issues.
We can provide a test unit as well as the needed datasheets - but it needs someone who is capable of debugging and fixing low-level things.
I am fine paying for the work as well.
Any ideas or contacts?
Okay, back to our next steps:
As I don't know yet whether the 4GB RAM will work properly or not, I'm also thinking of how to continue in case they DON'T work yet (let's hope they do though!)
I always like to have a plan B ready upfront
Well, let's assume everything is working and ready except for the 4GB RAM version.
In this case, I'll start the mass production - with 2GB RAM boards. No need to wait further if everything is ready!
Quite a few of you have signalled they will be happy with 2GB RAM for a start. And 2GB RAM will eat a bit less battery as well.
Of course, we will continue to work on the 4GB RAM boards so we'll have them available some time later. We already own the RAM chips - it would be a waste of 30k EUR not to use them!
There's a company in Munich (so round the corner) who claims they can get ANY RAM chip to work (and so far, they did, according to our memory distributor).
They have measurement devices that go inbetween the RAM chip and PCB and monitors EVERY signal.
So if something fails, they will EXACTLY know what.
We'll get that to work, even if we start with 2GB RAM first.
So much for now. I really had hoped to post better news than that (you can imagine how happy I would be with better news!) but that's life. And everything Pandora-related ALWAYS had been hard work instead of good luck.
But as usual, I'll keep going forward! No way to give up!
