Great questions. The hole appears only shifted down and toward the origin after flipping (vs what I’d imagine would be a diameter mismatch or something loose if I saw it on both sides).
This may have all been in vain, as I was doing some back of the envelope calcs and my worst result had ~0.25mm of a ledge between the two halves. If the piece is ~50mm tall, then tan(offset) = 0.25/50; offset = 0.005rad. However, on a ~40in x-beam, that would manifest as tan(0.005)*40in = 0.2in gap on one side. I’m at ~0.012.
So… to your question, I have indeed started wondering if it’s more about my setup. I was using this method to flexibly machine from the top, then flip and find these locating pins between power offs and whatnot. On one run, I watched closely for stepping early on on side to and adjusted my origin manually (0.2mm shift in x). After it was done, I jogged to current origin, put a dowel in, and expected there was no way it would fit in my reference hole… but it did!
So, I think you may be asking the right question. I would never have figured that method should allow 0.2mm slop, but the dowel fit. Even jogging it +/- 0.1mm while in the hole didn’t look that bad at either extreme and I think maybe I’m at the limit of my eye.
I have not been machining the stock other than facing, hoping to rely on these holes. It has occurred to me that I could switch to using a square, and machining two faces so that when I flip I’m locating to a corner against two reference (I’d machine in a pocket for a metal square insert for example).
Anything stand out there? Thanks for taking a look and chiming in! I’ve learned a ton on these pieces with respect to modeling, CAM (really, really tricky not to chip out sharp protruding edges), and trying to get better at multiples via various strategies.