I've thought long and hard on why using the expander mandrel might be a problem. Honestly, I'm not sure I've came up with anything that was really very insightful.
The best I have is this: it's a neck tension issue. It's definitely not a primer issue. When you size a neck down from the outside in a bushing die, springback will cause the neck to open up slightly beyond the point to which the bushing sized it down. When opening up necks with a mandrel, springback does exactly the opposite. Although I'm not using a force measurement seating die system, I can actually feel the difference in seating depth force for some cases when seating bullets. Given the leverage inherent to a full-sized Rockchucker press handle, if you can feel the difference, it's probably pretty large. What I can tell you for certain is that when I prepped brass using the .222 carbide mandrel as the final sizing step, out of the clear blue I started seeing ES values of 50, 60, or even 80 fps, whereas I could usually keep the ES under 20 fps during several years and many thousands of rounds reloading heavy bullets for F-TR competition in the .223. Going back to my original brass prep method alleviated the issue, whereas different Lots of primers had essentially no effect.
The only other thing that comes to mind is that when you use an expander mandrel to size the neck, you're sizing the neck from the inside all the way to the shoulder. A bushing die will typically only size the outermost 75-80% of the neck. So that is another difference, although I really don't have any clear idea why that should send ES values so high in my specific loads. As I mentioned above, I have used the same expander mandrel approach with .308 Win cases, and it seems to work just fine. So it's possible there's something specific about the inherently smaller .223 Rem case that lends itself to the erratic ES issue. I can't really say with any certainty. What I do know is that I quit using the mandrel approach with .223 Rem because the ES values were simply unacceptable for competition loads. Further, my original brass prep approach using a bushing die was never broken and worked just fine to begin with. Like the OP, I simply thought opening up the necks from the inside with a mandrel as the last step would produce even more uniform neck tension and was therefore worth a try. Because I know I can generate acceptable ES values using my bushing dies, I'm not really interested in wasting any more time trying to make the expander mandrel approach work, or for that matter, trying to figure out why it didn't work so well in my hands.
