I don't think polar alignment (PA) is the problem... as long as the autoguiding can overcome the bias from any PA error.
What happens when PA is off is that there is a constant error (think of it as "force" if you like) that is tugging the mount in some direction. There is no real force, of course, just that the plate is moving in the RA direction at a rate that is not precisely the sidereal rate, but the plate also moves in declination.
If the PA error is large enough, the RA and/or declination error will be too large for autoguiding to compensate for. That is where you need to bump up the loop gain - which can be problematical if the gain is anywhere close to or greater than 1, at which point, the loop will oscillate.
If the autoguiding system is "strong" enough (again, using the force analogy), you will still be able to keep the guide star perfectly steady. What then happens is that over a long exposure, the plate will appear to rotate around the guide star. That's right; the rotation will not be around the center of the plate, but around the plate where the guide star is located.
But even a 1 arc minute error of deviation of the polar axis from the north celestial pole should not be a big problem with your scope's focal length, camera pixel size, and exposures of under 300 seconds. However, if you compare a plate taken an hour later, you will definitely see that the heavens have rotated with a 1 arc minute type polar error. Pretty much all auto stacking software will de-rotate the images when it stack images.
Right now, there is a bug which I believe ZWO is still trying to track down that was introduced sometime I believe between ASIAIR v1.3 and ASIAIR v1.5.0. My guiding error when using ASIAIR has pretty much doubled between last Summer and this Winter. I can imagine that with some other mounts and autoguiding settings, this doubling could cause the kind of runaway that you observed (and others have reported for quite some time now, both on this forum and on the ASIAIR Facebook page). It could be precipitated by PA error but I suspect the bug is somewhere else and polar alignment is not the root cause.
One reason because that is a lot of air traffic in my area, between Boston, Hartford, and New York ( I live in central CT ) and I have lost far too many 5 minute subs to that problem.
Hah, that's funny. I am right under the SFO-SEA air lane and when I am close to Meridian, I get a lot of airplane tracks. They go away when stacked with appropriate stacking setting (check up Google on "3-sigma clip"; yeah, that 3 sigma that we learned in college :-). The default setting in AstroPixelProcessor gets rid of them as long as you have more than a dozen subframes or so.
Chen
