Shorted Tracks on Bare PCBs

[cborn]

After my latest experience, I'd like to suggest adding a step to check blank PCBs for internal shorts before commencing a build.
Following on from my bad experience with the Combo Shield: http://digistump.com/board/index.php/topic,1588.msg7849.html#msg7849
I went on to build my three Pro nRF24L01+ Shields, with the stackable headers.
I then tried the Radio example, but couldn't get communications working, despite swapping the modules around.
I ended up adding debug code to my Rx module, which was running with the OLED stacked, printing out the data received as only 0xffff.
I then added CDC USB serial and enable stdout over it so I could see the full result of the radio.printDetails() call on a terminal program on my PC, and it turned out most of the registers were 0xff as well. Something seemed wrong. I swapped the third radio module in (the 2nd was on the Tx), and got what looked like good readings from printDetails(), but no radio comms. Swapped modules again,and now my OLED module stopped working (permanently  )

Could it really be hardware problems again?

Unfortunately, yes. Removing the nRF modules from the Shields I again ran continuity tests on the bare PCBS.
Of the three, one was good, one was internally shorted from GND to 11 and 12, and the third shorted from GND to Vcc, 5, 9 and 12!

Not a good success rate, not something I've ever come across before. A couple of bad batches, perhaps?
So I STRONGLY suggest that you run a continuity test between all pins on each bare PCB before you start loading it.

[defragster]

Seemed like one of GND/VCC/VIN was commonly involved in the shorts so I tested against those.

I just hit 4 bare OLED Combo PCB's with continuity buzzer/meter and GND/VCC/VIN are only connected as labeled or as might be expected and don't hit any other labeled device ## pins.  I've already built three OLED PCB's with no problem and used a couple of those boards for Rx/Tx and 9DOF as well at the same time.

I got the same GOOD results on my two Bluetooth and one nRF board too - no apparent board shorts with GND/VCC/VIN, but I didn't do any other pin on pin combinations.

These 10 good boards show the flaw isn't in all batches.

[cborn]

Seemed like one of GND/VCC/VIN was commonly involved in the shorts so I tested against those.

I just hit 4 bare OLED Combo PCB's with continuity buzzer/meter and GND/VCC/VIN are only connected as labeled or as might be expected and don't hit any other labeled device ## pins.  I've already built three OLED PCB's with no problem and used a couple of those boards for Rx/Tx and 9DOF as well at the same time.

I got the same GOOD results on my two Bluetooth and one nRF board too - no apparent board shorts with GND/VCC/VIN, but I didn't do any other pin on pin combinations.

These 10 good boards show the flaw isn't in all batches.

Yes, I don't know if it is batch related. As far as I can see, my good and bad boards are identical, including the numbering (BAC-G, 94V-0, 43/14). Resist mask looks good, no excess solder anywhere.
Note that on my nRF boards the consistent short was GND-12, not Vcc.

I sent an email with thermal imaging of the OLED Combo shield board to support, no response yet.
I haven't done one for the nRF boards, although it could be interesting to see if they are consistent. 

[cborn]

Following up on this, I received a reply from Erik and unfortunately they received a bad batch of PCBs.
They are working through it now, hopefully the problem won't be too extensive.
Many thanks to Erik for his open communications on this issue.

[cborn]

Following up on this, I received a reply from Erik and unfortunately they received a bad batch of PCBs.
They are working through it now, hopefully the problem won't be too extensive.
Many thanks to Erik for his open communications on this issue.

Replacement parts arrived today, just about to try them out.