All forms of backlash need to be tracked down and eliminated. This includes both ballnut backlash and ballscrew thrust bearing support and adjustments, as well as drive coupling problems, or incorrect adjustment of software backlash settings.
To test your motor tuning, you can set up a straightedge, at an 45 degree angle to X and Y axis. Then program an interpolated table move along that edge, and true up the position of the straightedge until you see no progressive indicator movement during the entire traverse of that edge.
Then use the same dial indicator to track movement along that edge, but watching in particular, the indicator needle during the very beginning and ending of the movement. If you see a small pulse of movement (like a few thousandths) at the beginning of the move, but when it ends, there is an opposite direction pulse which brings the indicator back to zero, then motor tuning can help with that. This test would demonstrate that one motor is starting late and ending late. This could be because of softer motor tuning settings, or stiffer frictional resistance on one particular slideway.
First you get good, then you get fast. Then grouchiness sets in.
(Note: The opinions expressed in this post are my own and are not necessarily those of CNCzone and its management)