If this is an 8530, then welcome to the club of the dodgy Z. I think David Steele gets around this with autofocus, but I cannot. JEOL are aware that there is a problem and are "working on it". The issue for me is that ca 3 hours after changing samples, the stage mysteriously and suddenly (we have seen it twitch) moves ca 30 microns in the + direction and this value is not recorded by the software. This means that each point recorded is now 30 microns out and of course plays havoc with totals. There are ways around this:
1) put your samples in, focus and go away. Come back 3 hours later.
2) put your samples in, peak and calibrate then start recording your unknowns. By the time the calibration is done (for me anyway) sufficient time has elapsed that the "little hop" has been and gone.
3) record your unknowns, but confirm these a couple of times at intervals. Also make sure if you are doing linear traverses that the option "check all positions in a sample" under "automate" is selected.
4) put your samples in and come back the next day!!
Unfortunately, the JEOL system is a bit like the slow country cousin requiring patience. One has to be very patient and allow the stage movement to settle down before confirming.
I've had a loss less problems by using one or more of the above approaches.
Cheers
Malc.