I totally agree with you Paul and Ray.
I image with 3nm filters so light is focused really well, thus stars are much tighter, even with my doublet, as opposed to RGB imaging. Here is an example of stars that have not been altered by deconvolution, sharpening nor any other tool:
http://www.astrobin.com/full/227959/0/
And perhaps because of stars being fairly tight, tracking errors become more noticable, in my case RA axis is the worse one of the two.
I use an integrated OAG so flex is insignificant if not non-existent (it was a long and bumpy journey to get to that point!).
RA graph being on one side is not a problem at all; it is a result of lower aggressiveness and min error settings - increasing these settings in my case will usually lead to over-corrections (I use relatively short 1.5 s guide camera exposures with an OAG) and making star shapes worse. If the graph stayed on one side of the horizontal axis and at the same distance for the duration of a sub, it would mean nice smooth tracking and round stars (assuming DEC was excellent too and no flex).
Anyway, from what fellow astrophotographers have kindly explained, I gather that my AZ-EQ6 is not substandard and expecting consistent tracking as in the attached graph night after night is a bit unrealistic.