My students didn’t want to buy 217-4252, so we did this instead.
We made the attached version of the same part in 2014 (when we were putting entirely too much load on our ball shifter-as-winch). It’s nice because it’s only one piece, but you do need a special t-slot cutter:
I’m not sure I’m reading this drawing 100% correctly, but what is keeping the bearing from sliding out of its slot?
Nothing, but it doesn’t need to be kept there. Both the cylinder rod and the shifter shaft are already constrained in their assemblies so they will stay collinear.
This is a good way to make this with the small bearing. If you are willing to use a bigger piece of stock, you can use some snap rings to make this sort of mechanism in just one piece. Look up the old aluminum shifter cones that WCP used to sell (they might still sell them?)
Ooh I really like this - especially how much easier it would make assembly/disassembly. Pity about the t-slot cutter, that’s probably the easiest way to make it though.
I assume your ‘high load’ reference is why it’s steel? 
Putting my “janky robotics ideas” cap on… here’s a concept for a hacksaw-and-hand-drill version. Take 1/2" square tubing, run a slot on one side to collar the bearing and make a hole on the other for the piston. Use a pair of jam nuts to mount the piston, using the tip of the piston to locate the end of the shifter shaft? The packaging is kind of a nightmare and the whole thing will end up sweeping through a much wider area due to the square profile.
…but you can make it with a drill and a hacksaw…
Mods, maybe these threads should merge https://www.chiefdelphi.com/forums/showthread.php?threadid=160053
Yep, that would work in a pinch. The only issue is the bearing wouldn’t contact the coupling on the push stroke, so you’d be rubbing a spinning shaft against a stationary one. If you could add a couterbore or washer on the inside of the tube, it might work better.
These are still used on at least one of their products. The Part number WCP-0126.
These are a two piece design, with a back disk which is retained by three 4-40 bolts. There are two bearings (R2 Sized) located in the cone. The cone has three rear bolt threads, and a single forward thread to mate with the pneumatic cylinder.
We machined the pneumatic actuator threads on these cones last year to match our actuator.
CAD for the entire gearbox with the shifter is located at:
http://www.wcproducts.net/cad/WCP-0070-Rev1(WCD)-STEPP.zip
Maybe I am missing something here, but what is to stop me from using a woodruff key cutter and an endmill to do the operation of the expensive (comparatively) t-slot cutter?** It could all be done in 1 setup on the mill (save swapping out the tapdrill, tap, endmill & woodruff from the spindle). If you used the TIG wire technique for squaring the stock in the mill vise you wouldn’t even need to part off in a lathe and could just use a hacksaw to rough to length.
** for that matter I could also use a small boring head… but there are always 100 ways to hold and machine most common sized stock.
You could definitely use a 3/8 woodruff key cutter instead