Rotary Tool concentricity/runout issues with the chuck itself

I’m trying to get better precision out of the rotary tool. Has anyone done any experimenting to determine it’s accuracy? Or even better, found ways to correct the issues?

I’ve been having issues with runout (see previous post - Rotary Tool Runout and facing operations? )

After having finally gotten back to it, I turned a “test bar” down at work, and spent a lot of time to ensure it was round, and as parallel as I was able to measure (so within about 0.01mm). Using this round in the rotary tool chuck, I could not get the runout/concentricity less than 0.4mm (yes, almost half a millimetre). The chuck jaws also do not seem to be even vaguely parallel (there is a visible gap between stock and teeth), and due to the tightening mechanism, I could not use some of the tricks that can be used in normal lathe chucks when concentricity is the issue (though that would not have helped with the out-of-parallel teeth).

I kinda already knew, but this confirmed, that the tailstock is next to useless for concentricity. Being held by only 2 screws, these essentially become a hinge point for the tailstock to rotate that then creates more runout.

I’m going to look to design my own tailstock holder (I’ll take the live end out of the existing one), so I should be able to hold runout, but it won’t help with the concentricity.

I have had limited success with using oversized stock and intentionally turning it down to size, but this only works if there is only one operation (easy enough with wood, but not aluminium, due to small steps required when using fine bits). The main reason being that the controller moves the origin 90 degrees between operations! More complicated parts realistically need more than one operation (or 10 days to try to process in a single operation).

Anyone done any similar testing or found any solutions, or can confirm/deny the concentricity and runout issues I’m having? I’m wondering if I just need to take the chuck to a machinery shop and say “hey guys, can you re-face the jaws so they’re concentric?”. Problem is I’m not sure they’d maintain concentricity, as they seem pretty poorly built. But 0.4mm! That’s shocking!

Additional - anyone know why EVERY operation sweeps the length of the work (with assigned stepover) ad the OD diameter? At first I thought it was a safety thing (get rid of anything oversized), but that assumes that people would never put oversized stock in and end up with too deep a cut. Or it was put as a facing operation - but that shouldn’t be compulsory. It just seems to waste a lot of time with no real benefit. Anyone know?

Secondly - I had it do an 8mm diameter and 9mm diameter cuts, done on the same stock without adjusting anything. They came in at 8.4 and 9.6mm. Some may be incorrect Z height, but that should have led to the same oversized (eg. 0.4mm on both). Neither is it the bit slipping in the router, as it did the 9.6 first. if it was slipping (ie. being pushed up into the router), then it should have been a larger error on the second cut, not the first. Anyone else mucked around with repeatability on diameters to see if it just stuffs up?