I remember reading an article on this subject in some model engineering publication and at the time I thought that this seemed like far too much work to be reasonable when applied to a generic 4in 3 jaw chuck. The article contained some detailed information on the best way to calculate the optimal radii for the back of the jaws which engaged the scroll.
Recently I came across the problem on an unusual very small (1.75") high quality 3 jaw chuck, but of course I have no idea where the article was to be found.
Google didn't locate the article although I found
Alan