I can't say from here what is going on with what I have been told. But I do have some questions and observations.
>>>>>>swings back and forth + and - a considerable amount (+/- 45degrees)
Yup, at first it does.....but then it settles. It can continue to settle for five minutes or more. However, usually within two to three minutes per trial it is close enough to figure how much you need to adjust, or at least which direction to adjust.
>>>>>>it never really seems to stabilize.
What does this mean? On the same run, does it point above the Y axis for a while, then move below, and then back above, then back below, (or vice versa)? Or does it start out on one side, then drift maybe to the other, but generally, after about two and a half minutes stay on one side, with slight variations up and down but generally in the same place? That second behavior is somewhat normal.
>>>>>>moving the scope from the horizon to the equator seems to shift things just enough to throw it off.
No, that is not a "throwing it off." In fact when you get the celestial equator/meridian fix on your azimuth, moving the star to the (20-30 degrees above) horizon and adjusting your altitude actually also means you need to adjust the azimuth again. It is not as if something slipped or something was thrown off. The Azimuth setting was correct only for the original altitude of the mount. Now that you have changed the altitude, you have thrown the azimuth reading off. The altitude change affects the azimuth reading. That is why you have to go back to the equator after the horizon. (And if you are being really picky, back to the equator, and back to the horizon and.......until there is no drift wherever you point.)
>>>>>Seeing conditions were "OK". Not great.
Take longer exposures. Four seconds minimum. It should not matter too much in this application, but if your seeing is not good, generally, go longer in exposures.
>>>>>scope backlash was off
If your scope is perfectly balanced and there is a lot of slop in the dec gear train maybe. But just shove the scope to one side or the other of the dec movement (or drive it for a second or two in one dec direction before starting the test run). It should not be moving from there on, and therefore backlash should have nothing to do with it. YOur dec movement is disabled during a drift run, and your RA is pushing constantly one direction, so backlash has no place to play. (And besides, your RA is not being considered. _)
>>>>>I have not done PEC training.
As Michael pointed out, that should not matter. Of course it is nice to have a mount that is tracking well. But, PHD2 separates north-south drift (what you care about) from east-west.
>>>>>>That sounds like your mount isn't perfectly level.
It is true that mounts do not have to be level to get a good polar alignment. But it is a lot easier to get it done that way, for the reason pointed out (that with a non level mount, an adjustment in Azimuth affects altitude , and vice versa than in a level mount ).
>>>>>it was an incredibly slow process for me.
Yup, it does seem slow. However, although couterintuitive, you actually get quicker with it the more you do it. One would think that since you need a couple of minutes on each trial, there would not be much of a way to speed it up. But, with experience, it gets faster. At first there is a lot of bumbling around (which way to turn the adjuster, how much, how long to let it run before you can decide it is stabilized, and all that). Once you get over that, it moves from incredibly slow to just slower than some other methods. Still, it is probably the most accurate of methods.
>>>>>>being done in combination with a long focal length making it all that much more visible.
Were you doing it visually, I'd grant you that. But PHD2 has sub pixel resolution. Using more focal length will probably not make that much difference.
At any rate, take a moony night and just mess around with it at your leisure. A little practice goes a long way.