Comments by Mike Gray from Yahoo forum:
Well, this is the method I used. That doesn't mean it's right or that it's not complete overkill but it worked for me. Be prepared to devote an evening or at least a good part of one to the PEC training. Do a calibrate and train before every PEC training session. It helps to match the motor performance to the state of your battery pack. Use at least a 2x barlow to jump the focal length up. It gives you better accuracy. You have to have excellent polar alignment. Close or pretty good won't cut it. The star will drift off and you will end up trying to train the drift instead of the PE. That's what causes the star to do it's Tanya Harding impression. It took me about an hour to do a drift alignment on both axis so there was absolutely no DEC drift on the eastern equator or the meridian. You can see that I got a little sloppy when I did the last PEC measurement because the PE line is sloped. I left it like that intentionally because the line hid under the 0 grid line of the chart. I used 5 arc-seconds of drift over a 10 minute period as my cut off point. It's difficult to get to that point but it is worth it. Use a 2 circle reticule overlay on K3CCD. Orient the camera so the star travels exactly along the x-axis for the entire width of the screen. Now that you've got the hard part done, center the star in crosshairs and let it sit for 10 minutes (1 worm revolution). It will end up either right or left of the y axis. That shows you how much you need to adjust your custom tracking rate. If you have to add Right arrow to get the star back to it's original position then the motors are running too slow and vice-versa. Don't try to rush this part because wrong tracking speed looks exactly like PE over the short run. You have to let it track for at least one full revolution of the worm to know where the star will end up and how much to adjust the tracking speed. Remember the tracking rate adjustment for future use. It doesn't change much once you get it nailed down. I'm sure that some mathematical genius out there can figure out a formula for how much tracking speed correction to make depending on how much the star drifts taking into account the focal length and pixel size of the camera. I just took a SWAG and narrowed it down from there. When you enter the PEC, zoom the screen up to 200% and only make corrections at Guide speed (speed 1). Let the star move about 5 pixels or so off of center before you try to move it back. Use small corrections or you will end up overshooting the center. If you do overshoot, don't reverse directions you're just compounding the error. Let the star drift back to the center on it's own or use that spot as your new home point. When you get done with the initial training, do at least 1 or 2 more PEC Updates. It will help average out any mistakes you made on your initial pass. Like I said before. This will probably eat an entire evening or close to it so don't waste an excellent night watching a computer screen when you should be viewing or imaging.