I find the polar alignment procedure a bit peculiar (compared to the 3.28 procedure), especially the bit where it tells you to adjust the azimuth screws to bring the star as close as possible to where it ended up during the previous alt alignment step (you have to memorise that position in your reticle). Be sure to read the updated 3.35 manual.
Version 3.35 doesn't appear to have the glaring bug where it would double the polar alignment error with every alignment step rather than reducing it. I have gotten to sub-30-minute alignment errors and didn't go any further (I don't do imaging). There is still the occasional odd result with 3-star alignments, where the alignment was declared successful yet goto accuracy was not good enough to put the target in the field of view at 150x. Other times it works well and goto is reliable at 240x. It seems the alignment algorithm still isn't as robust as it should be and depends much on the set of alignment stars on offer.
Looking forward to further refinement
Cheers
Steffen.