Since the tracking measurements were made on the meridian, and the RA drift is much greater than the DEC drift then I would guess the problem is in the RA axis somewhere.  I just got back from the observatory (remote) and did find a couple of loose bolts on the RA head .. I tightened them down and then did some slewing tests to make sure I didn't break anything.  Maybe this was the problem.  We are scheduled for clear skies tonight so will do a new model and tracking tests.  Skies permitting I'll update this thread tomorrow with the new results.

With a 132 point model the best I could do unguided was about 25 seconds imaging before the stars became elongated.  I have read many are doing several minutes unguided, hence my concern something is wrong.  I wrote this Tracking Analyzer program to assist in evaluating the problem.  What it does is similar to the software program FlexRx where images are plate solved and an drift correction in RA and DEC are calculated.  This information is then passed to the mount as an update to its current corrections.  The idea is to take a couple of test images at the beginning of an imaging session to establish the RA and DEC correction rates and then analyze each subsequent image for correction updates (tweaks) to those rates.  Ideally though you could start with the APPM model correction rates and then pass updates if it is not quite right or there is some other drift present, maybe due to temperature changes on the mechanicals or optics that the model could not account for.  As mentioned above, with the 132 point model I could do 25 seconds before star elongation, but with this method I can do 600 seconds unguided with a 1 pixel or less error.

So far it works very well, and eventually it would be great to have this added to either PEMPRO or PHD2 as an alternative guiding method.  PEMPRO has all the tools already available including plate solve and mount control, so this should be easy to add.

Craig