Thank you for the question. The "shoulder die" is only for forming 5.56x24 from .22 TCM brass.
When you try to form cut-off .223 brass in that die, in one step, the necks come out far too thick! I tried using it at first and discovered I had to removed the expander ball to keep from yanking the necks back up because the necks form "donuts" inside. I could get some good cases from using the shoulder die without the expander ball, but of course the necks were super undersized. When I tried to expand them, I had a LOT of splits.
I did try annealing but even with annealing, trying to form brass from the diameter of a .223 case body to a .255" 5.56x24 (.22 TCM) neck was just too much.
My solution was to form 5.56x24 cases from .223 brass in stages. I started with a form die in .30-06, then one in .25-06, then one in .244 Remington. Obviously I had to cut off the form/size dies since all that is needed is the neck and shoulder portion. So those dies are cut off just like the .223 Remington Improved die - about 1/4" below the shoulder (inside). Using an RCBS turret press I could mount all the dies at once. The process is basically done by "eye" and "trial and error" so to speak in terms of adjusting the the dies. The dies are screwed down far enough to form the neck but stop short of forming a shoulder less than .720" because that is the final length needed for proper headspace.
With each die adjusted, the process goes pretty fast. Cut off a bunch of .223 brass around 1" long - remember the 5.56x24 case final length is only .935" so you want adjust your cut length so the formed case comes out just a hair too long. It won't hurt if it's longer, but that adds more time trimming to final length.
Lube the cases then pass them into each die in turn using the turret feature which is basically, .30 cal, .25cal, .243 cal, then .223 cal. After passing into the fourth die - the .22 TCM die, the last die is the "shoulder die" which puts the final shoulder angle and neck location on the case. If your pre-cut .223 cases are just right, the final 5.56x24 case will come out just a bit over-length and take only a few turns of the trimmer.
Now some points to consider. Using staged neck forming I get virtually 100% good cases with no rejects, no splits, or other problems. Annealing is NOT necessary using this process. My Lake City once-fired brass came out with necks between .015-.016" thick which is "spec" for the .22 TCM and therefore the 5.56x24.
The other thing is that my progressive case forming was focused on making 5.56x24 brass exclusively, not .22 TCM brass. For this reason I didn't worry about the different shoulder angles between the '06's and the .244, and the only "hard reference line" that matters is the .720" body-to-shoulder length of the 5.56x24! This is of course exactly the same as on a .22 TCM case.
The best way to know you're cases are coming out right is to have a .22 TCM barrel on hand to "plunk test" them. Glock barrels are good because you can see if the base is level with the barrel tenon, headspace is good. If the case is below, headspace is excessive. If the case is above headspace is too little and the slide will not fully close.
Because .223/5.56x45 brass is so thick down where the 5.56x24 shoulder is going to form, some cases will come out of the final forming step with somewhat "rounded" shoulder-to-body transitions on one side, with a more expected" sharper shoulder on the other side. As long as they pass the "plunk test" they're fine to load. 5.56x24 brass made from 5.56x45 brass is made to handle pressures near 70K psi, and from my personal experience it would be difficult to "blow one up" using any powders I've tried. The cases "volume out" meaning you can't put in any more powder. I have taken reformed rifle brass to 11.3 grains (2,067fps) of Enforcer and 10.7 grains (2,063fps) of Lil-gun behind 40 grain Armscor bullets and that's just about all the powder the case will hold and still get a bullet in. Those are hot loads and running about .3-.5 grains below that is still plenty impressive, but the take-away for you is knowing the cartridge and four different guns that I've shot it from handled it without a problem. I post all my load data on
www.ammogide.com which is why I don't tend to post it elsewhere, but as fo today I have 26 published loads on ammoguide which should give anyone interested a good baseline, though of course it's up to each of use to start with a load we feel is safe and work up from there.
My top 5.56x24 loads are equal to Armscor factory .22 TCM fired from 5" barrels which means they're running about 75fps on average over where they should from 4.5" barrels. For creating "stock" I keep my loads somewhat less than the top.