From the calibration graphs, it's not just the RA that has steps that are too large but the DEC calibration steps are almost all backlash clearance (the ones labeled B).? There are just a few small steps South, then North finally acts half way normal.
So, major DEC backlash problems during the calibration at least.? The large RA steps indicate there's something wrong in the settings, and PHD2 will therefore use a gain that's way off.? That could explain it.
Now, if I could only fix my own guiding problems...