From what you have said it seems you are using some kind of position control scheme. I use a custom position control scheme (see here for details) which is designed to move as fast as possible while ensuring stability and reducing oscillations as much as possible.
Usually the position control scheme needs to be tuned with the load attached, so it is normal to get oscillations if you are running the motor outside the robot leg.
It would still be helpful to see the waveforms for the new scenario where you were able to reproduce the restarts, which should still provide a good insight into the potential root cause of the shutdown issue.