This chart plots the drift in RA during tracking relative to the star position in the first frame. PEC was automatically trained using an autoguider 4 times. Although RA drifts steadily away from the starting position, there are many nice flat plateaus and much smaller variations than the 12 arcsecond swings seen after manual PEC training.

Here's the real payoff. Very small deviations between frames. The chart below demonstrates that deviation between 10 second frames has been reduced to around 2 arcseconds, much improved over manual training. The closer to flatline this chart gets, the better the imaging will be, which is why I'm doing all this in the first place. It's interesting to note that the first 20 frames (200 seconds) were shaky, it takes a few minutes for the mount to settle down after aquiring a target.