Hi @shawn_ricardo
I was thinking a bit more about this and I'm even more confident it's most likely a ground loop issue. When you use a PC, it is plugged into the same wall-ground connection as the 5V power supply, so there is no ground loop between the PC and the Doodle (they share the same common ground from your house/lab/office). However, with our drones, that GND connection is isolated due to the battery. If you are using a bench power supply instead of a battery, then the ground has a (typically) isolated path to push through since most benchtop supplies are isolated from wall power to the load output for replacing batteries (i.e.: you need to NOT have the negative terminal shorted to Earth GND for battery powered electronics replacement supplies).
So, I'm even more confident you should try a power module fed by whatever supply is powering your Starling.
The other smaller thing I thought of could be related is turn-on timing. Some devices, Boson in particular (though I have not heard this with Doodle) are very picky with power up timing and will not enumerate if VBUS is too fast to rise, or if some devices turn on before their host. I'm curious if the turn on timing between the Doodle and the PC is less of a concern (likely the USB port is going into a hub internal to your PC), whereas with us on a Starling, it's direct to the Snapdragon (no Hub) and the timing of it may be more sensitive. The way to check for this is to experiment with turning on the doodle power at various stages (before Starling, after Starling, after Starling for a long wait, and even "with" Starling) and see if that makes a difference. If that changes things, then we have a timing issue that may be addressed by a SW enabled power supply based on when the enumeration is successful....
But, I'd still go with the power module as a priority, since that is what is needed anyway for flight. Might as well test as you will fly.
Keep us posted.
Thanks!