Of course, the question for today is: How did it work? Did you get working boards? Is everything okay?

Well, we got halfway through it, it's not looking bad

There was one problem with the stencil (the holes for the solder paste were too big for the BGAs), so usually, this would've meant: Okay, time to go home until a new stencil will be there.
Well, the new stencil will be there tomorrow morning, but we wanted to try to get a PCB working anyways, so we tried one PCB where the BGA pads had NO solder paste.
This COULD work out - or not.
It seems it worked half.

There are three BGAs on the board (BGAs are the chips with VERY small solder balls that are below it), the most complex one is the OMAP, the second complex one is one that does all the power management.

We populated a PCB and tried to boot it up - and, yay! It booted! U-Boot was clearly seen via serial out and I could access my testing card in Slot 1 via my netbook.
However - I didn't see any picture. Backlight came on, but no picture.

Damn.

So it was time to get in contact with MWeston, who quickly replied to me what I could measure where.
It seems the LCD doesn't get power.
Hmm.

Okay, time to do the remaining tests...

USB... working.
SD Slot 1... working.
SD Slot 2... working (after we found one pin was not soldered properly)
WiFi - no Power.
LCD - no Power.
Audio Amp - no power.
Battery charging - nope, didn't recognize that I connected an AC Adaptor.

Well - there were two solutions:
Either A LOT of stuff is broken on the PCB (but then again, it could only be a lot of missing parts...) OR something with the power management is not working, which would explain why so many parts didn't get powered up.
As mentioned: One BGA does the power management - and the way we tried to get it working was a "could work - or not"-thing.

So yeah, most probably that power chip. It handles every other current than the main 5V one. And exactly these parts are not properly working.

Tomorrow at 8:30am, we'll continue with out adventure. We'll have the proper stencil for that, so population should be fine this time.
In case something still doesn't work, we'll go to do some X-Ray tests tomorrow. This would reveal any missing traces or similar.

Even if the boards work fine, Global Components will do some X-Ray to see if the machines can be further optimized for some parts to minimize the failure rate.

I'm pretty happy that the machine booted up.
Firstly, it's the first time Global Components EVER did a PoP (and it worked on first try), and secondly, debugging would be hard if it didn't even boot.

The GTA04 team had similar issues: 3 out of 3 PCBs they got from their run didn't boot. They managed to get one half working after a day, so for a prototype run, issues like these are normal.
But things are looking good that we'll have fully working PCBs tomorrow, so stay tuned for more updates

This means that actual production run will start early February (unless something else goes really wrong), but I'm happy we got something to boot today


http://boards.openpandora.org/index....ne-2011-12-21/