Granted that your mount's PE curve looks terrible, but you should be able to do better than 0.7" total RMS through the whole night.
The next time it happens, take a look at the correction pulses to see if the correction pulses have become a constant comb, like I posted earlier (a horizontally very long thin graph that you have to download to your destop to see). If so, there is some spot in the PE curve of your mount where the slope is worse that what you have assumed it to be, and you will need to increase your max duration to match it. This is why you cannot use someone else's numbers.
Increase the max duration until the comb goes away.
Now, that location is actually dependent somewhat on the Hour Angle of the mount (and the Pier side). The motors and gears etc come to the same pisition for the same hour angle. So, you can actually go back to the same hour angle to re-test. If you are using ASIAIR which does not allow you to input Hour Angle, just remember that RA = LST - HA. You can then impute the RA from the Local Sidereal Time.
This is the problem with the piece of paper ZWO gives you. It is only useful for visual (where PE is important) and of zero value for autoguiding (where the slope of PE is important). They show the expanded view of where PE is max, but not where slope of PE is max, and that is where you actually need the expanded view to dial in the proper guiding parameters.
The bigger problem is that I don't think ZWO understands this.
Chen
