collimating a refractor via diag. (non coll. lens

Hi,
I noticed (using a cheshire ep) that the circles were slightly off so I bought a collimatable diagonal and using a refractor cheshire ep I have bullseye on. The question is two fold
1. If I rotate the diagonal then it is back to slightly off.
2. The circles while they're dead on bullseye to each other the circles themselves are not in the center of the cheshire sight tube.
Is anything wrong or this is the way it should be?
Clear skies
Itz

Of course it is wrong (but it may be tedious to improve on the present setting). If rotating the diagonal shifts the center, then the diagonal is not fully collimated with respect to 90-degrees (the tubes may be mechanically offset). If the circles are not centered, then again the diagonal is not accurately collimated. BTW, I've found an old Edmund Prism diagonal to be far more accurate with respect to center AND more importantly, high-magnification planetary resolution than a modern expensive dialectric mirror diagonal (despite RCs' claim that diagonal mirror surface accuracy matters not). If you cannot align the diagonal any better, THEN do your critical observing with the diagonal in its' best orientation AND place the object of interest in the off-center spot where the alignment is best. In addition to diagonal alignment, objective alignment & focuser alignment are also factors.