[One more thing to report to you: First I went to a star in Gemini and found the scope is roughly tracking without guiding. Then I went to M42. Now stars on the MetaGuide screen were slipping to EAST as if it was not tracking. So I checked the turning of the RA pulley; sure enough it was turning to correct direction. At this time I calibrated and guided as I described above.]
Did you set the backlash values in OnStep to match your mount? The slipping behavior described above is typical of not having the RA backlash value set except it happens when doing a goto from west to east. There are values for the amount of RA/Dec backlash (arc-sec) in the ASCOM driver setup window (and in the Android App.)
[I just did angular analysis on the size of elongation to measure angular movement in RA and DEC direction during the 30 sec of exposure.
The best sub frame shows RA~30" and DEC~30" and the worst sub shows RD~30" and DC~60". These are very crude estimate of angular movements during the worm period. I will measure again when the autoguiding improves.]
Is that guided or unguided?