Polar scope calibration is one issue. But when I look at the star drift, it's more complex than I had thought.
This is stacking of 50 subframes 2 mins each without star alignment. The first image is plate solved with Unimap to visualize the grid.
I could not get my heads around the diagonal drift in 100min since any mis-alignment should be in DEC direction unless the polar axis is way off the NCP. Right now I'm forced to use 2 minutes sub due to this issue.
Horizontal direction is RA. The small bump in the trail happens when I checked the exposure.