Do you mean the reported polar alignment error after running only five points of a model? If that i[s] the case, you need to ignore that kind of thing. Watching the polar alignment or RMS numbers as things go along is not helpful and generally leads to worrying about things that will be completely different at the completion of a "full" model. If you watch the reported RMS of each additional point as the model runs, you will begin to think that the RMS is terrible when in fact, by the end of the model run, the RMS is just fine. When you are having an actual problem with something, then watching the numbers may be of some limited help, otherwise just avert your eyes and let the software and mount do its thing.
6.9 is not a bad RMS. Per Frajvall, an expert at modeling, always said that anything below and RMS of about 10 was fine in the majority of cases. Spending time getting even lower ends up being a matter of either over pruning the model or simply diminishing returns.
Something else. I wrote a post in response to airscottdenning's earlier post and I am seeing the same thing here. When you see a linear trend in the graphic representation of the model points, there is most likely something moving in one plane on the mount and might be impacting your model results. I experience this last year when working with a TAK 106 that should have modeled very well but wasn't. When I finally noticed the trend in the data, I realized that something must have been moving back and forth rather than just randomly. I went out to the observatory and started looking for something moving back and finally found that the draw tube on the TAK focuser was able to move back and forth in one plane. When the focuser was locked down, the problem went away. Of course, it wasn't possible to use auto focusing then.
While mccomiskey's points are not as bad as airscottdenning's there is still a fairly evident trend visible on both graphs. When a system is very rigid, the points should be gather around the center point of the graphs in a somewhat random pattern (perfect rigidity would place everything at the center point). There will be some thinning of points near the poles since model points should be avoided in those areas, but the general trend should be points centered around the central point of the graphs. In the end, that can or will be something that will raise the final RMS numbers.
Edited by EFT, 06 December 2020 - 11:01 PM.