Hi,
Maybe I can add some stage shifting issues from my experience here at the geological survey of Finland (mineralogical lab). I had issues on both instruments, our CAMECA SX100 and our new JEOL iHP200F:
1.CAMECA
After programming many spots and returning to the first one it was shiftet some tens of microns in x and y direction, but in the lower part of the holder in the opposite x direction. Long story short, after a few investigations I figured out the stage was "rotating" when moved. This is due to the worn out shuttle on which the sample holder is mounted. You can see the worn out groves where the "fork" of the rod is holding the shuttle when inserting into the sample chamber.
Solution: I replaced the shuttle with a less worn out one and reduced a bit the speed of the stage. No shift after that.
2. JEOL
The new iHP200F has a large shuttle, on which the sample holder is mounted. We use the split holder from Boerder and have our standards permanently inside the sample chamber. When moving in x direction, I got a shift in y direction always towards the gate. The longer the x distance traveled, the bigger the shift in y direction up to several 10s of microns. With the JEOL engineer we took a look at it and, long story short, two metal springs on the side of the stage (with the wheels at the end) are too soft for the massive shuttle, so the faster it moves, the more the whole shuttle "wobbles" off the stage towards the gate.
Solution: (Temporary) We stiffened the metal springs by glueing a small metal pin between the spring and the stage, thus making it stiffer. Long term solution would be to use either a thicker metal plate or a more stiffer material for this part. The JEOL engineer is in discussion with Japan about what to do. Since we were the "first" ones to report that, they are waiting to confirm from another instrument if this happens on any instrument or if it is only in our instrument the case.
In my experience if there is some shift in stage movement it is a mechanical issue and not a software issue (since, like you say John, software is often still the same as before the problem occurs). Maybe mapping the shift (like we did in our CAMECA, testing how the shift appeares throughout the whole movement range) might help to realize what the problem is.
Greetings from the cold North (Finland)!
Radek