I've somehow managed to kill one of my BBBW boards. It now displays the dim pulse of the power led when plugged into either the 5V barrel jack supply or USB and nothing further after that. From digging around on the web it appears that this is a sign that I've somehow managed to damage the 3358 SoC.
I've been going over my last usage of the board to try and establish what killed it and would be grateful for any advice people can provide.
I had the BBBW connected to a motor driver board via PWM out on P9_16, and had two QEP inputs on P8_33 and P8_35. I was powering the motor encoders from the VDD_5V on P9_05 and was using the 3.3V rail from P9_03 to pull up some external resistors on the encoder inputs. The FETs that driver the motor are powered from a separate 12V battery, however the grounds from the battery input and BBBW are tied together. The schematic for the motor driver is attached if anyone is interested.
This was working initially, however I think I may have done the damage when I pulled the power from the BBBW instead of powering it down gracefully. I am working on developing some additional drivers for the RTEMS operating system on the Beaglebone and so don't have the ability to gracefully power the board down like you can with Linux.
I have exactly the same configuration running on another BBBW on a mobile robot without problems, however that is entirely powered from battery, (BBBW and motors) so there is no split supply issue there.
After doing some reading it appears I may have made a couple of mistakes but would appreciate your thoughts.
1. I shouldn't have pulled the power on the BBBW without disconnecting the motor battery first. In theory all inputs should have been at zero volts after power was removed from the BBBW, but from the reading I have done any residual input voltage on header pins after power has been removed can damage the board.
2. I should have either used SYS_5V (P9_07 or P9_08) to power the encoders rather than VDD_5V. This point is a little less clear as there seems to be conflicting advice as to whether you can use the VDD_5V header to power external circuitry.
One final observation I have is that looking at the schematic for the BBBW it doesn't seem to use a separate external PMIC but does that all inside the SoC. Does this make this version of the BBB less robust?
Any and all advice or feedback would be most welcome.