First: there are two horizons!!! One under Display -> Horizon & Atmosphere Options... and a second one under Telescope -> Telescope Limits. And it turns out that the TPoint Automated Calibration uses the first one!!!
I tried to draw the horizon manually, but that was WAY too difficult. Then I tried to use a similar mechanism as APCC and SkyTools use: point the scope at the border of the horizon and record it. But in TSX, you apparently can't point your scope at individual points and then connect them, but TSX is constantly recording! But I found it impossible to move the scope along a straight line. Weird!
I then turned to the method of capturing a photo, using it in Horizon & Atmosphere Options and draw the horizon based on that photo. First, I took a panoramic photo from where my scope is standing.
I needed to make the sky transparent. Loaded the photo into Photoshop, selected all the sky (and some of the cables in between)
When everything was selected (including all the small details), I inverted the selection and copy/pasted this into a new image with transparent background:
I copied this picture into the Horizons subdirectory in the TSX path and then I could select it in TSX:
Had to move and rotate the image to fit. Now comes the confusing part: just setting the horizon image isn't enough!!! I had to select "Custom Drawn" from the "Horizon Type" drop down. Then click "Create From Current Horizon Photo". Only then would the horizon be used as a restriction in Automated Calibration. This is the small part of the sky that I'm left with :-(
This should help with the Automated Calibration!!!
But I would like to see my photo again! So, I had to go back to "Horizon & Atmosphere Options" and select "Photograph" again. That would show my photo again - but leave the horizon restrictions for "Automated Calibration". Very confusing!
When it got dark, I first tried to redo the Polar Alignment. First, I did a rough PA: slewed to Polaris and then adjusted Azimuth and Altitude until it was centered. Next, I took a 40 point model and adjusted my Polar Alignment according to that model.
Now, I wanted to record the Periodic Error to create a PEC curve. Boy, that process is COMPLICATED in TSX!!! Have to disable everything (guiding, PEC, ProTrack...), then choose my main camera as the Guiding Camera, disable the Relays in the Autoguiding settings, enable logging the autoguiding data and finally select Autoguiding... This recorded the star movement. And the resulting file could finally be used for PE analysis. The PE is under Telescope Tab -> Tools -> Bisque TCS -> Periodic Error Correction tab -> Compute PEC Curve! Load the logfile and click "Fit":
and then "Curve fit to tracking data". And finally "Save to Mount".
Afterwards, I recorded another log, now with PEC enabled to measure the remaining error:
Hmmmmmmmmmm, this isn't too much better (2.3 arcsec vs. 2.8 arcsec).
I then wanted to use PEMPro to record and calculate a PE curve. But PEMPro didn't seem to record the correct frequency:
<screenshot of PEMPro>
So, back to TSX. I guess I need to record PEC from a better location and/or longer then 20 minutes!