Hi Jared,
Those all sound like motor tuning issues. The first step would be to make Plots on the Step Response Screen while making a reasonable sized move of maybe 1000 counts or whatever size seems to have the most problems. Increase the P gain in steps of around 50% until the system starts to oscillate or goes unstable (increasing violent oscillation). You may want to set a Max Following Error limit such that as soon as the servo starts to oscillate it disables. Then decrease the gain back down by to about 1/2 where it should be stable. When adjusting gains only use simple numbers as tiny relative numbers won't have a significant effect (1.2 and 1.20034 will respond the same).
If you would then like to post plots on cnczone then we can go from there.
Regards
TK