Now this I could buy because it makes sense in regards to how it would appear to be Y axis variation. At one point I had some flex of the Z axis rotationally around the X axis that was noticeable as I grabbed the spindle and pushed/pulled. I did some tightening of the eccentric spacers to snug the wheels to the X axis C beam and felt as if I had removed the play, but could be wrong. Specifically, how would I go about removing flex of the Z axis around the X beam?
Yeah, that may well be the case - I'm not trying to say otherwise, just trying to eliminate other potential sources and we'll see what's left Even those parameters (assuming the 300mm/min feed in ply, 2 flute cutter) will generate ~0.6kg tangential cutting force, which is enough to bend a 2" long, 1/4" diameter carbide end mill by ~0.01mm (so 0.02mm on overall dimensions); and that's without allowing for the stiffness of the spindle or machine (which is likely to contribute substantially more deflection). By contrast, a 12mm deep cut of 0.25mm with the exact same parameters generates less than 0.1kg tangential cutting force. I'm not saying that this is "the" problem, but I suspect it is a factor.
I grabbed the Z axis assembly and was able to get some rotational play around the X axis C beam. I then went meticulously through the entire assembly and ensured everything was tightened. The eccentric spacers seem to be just about as tight as they will go, and I'm not suffering any motion issues on account of it. Even after all of this, there seems to be some play rotationally, though I was pretty rough with it. Either the carriage is loose around the C beam (wheels not gripping tightly) or the C beam is flexing, from what my eyes told me. I don't know that I believe the C beam would be flexing, so I grabbed it alone (off to the side of the Z axis assembly) and tried twisting it, but didn't see any motion. Likely just my imagination then as far as the C beam twisting. I then re-ran the original 50mm circle to see if I had any improved results, and I did not. I then reworked the G code to do as Misterg suggests by running a rough cut that is over-sized and a finish cut. Again the results were more or less the same: high spot was from 10:30 to 4:30 on the "clock", low spot from 1:30 to 7:30, and roughly .015" difference between the high and low (whereas I saw .017" before). I am now exactly 6 weeks departed from first having the machine built, wiring, and moving. I know this because my wife has reminded me of that fact and the amount of money I spent, as well as the fact that I've yet to "make anything" (her words). I do have a 5 day weekend and what feels like a wealth of knowledge (the Openbuilds team), so I'm hopeful to get to where I need to be.
is your machine actually square? and a little perspective..... you are chasing 15 thousands of an inch in WOOD, tomorrow when the weather changes the wood is going to expand or contract more than that all by itself (-: relax and make something for the kitchen, nothing like actually making a few useful things to teach you what is important and what is not. my own machine is now years old and getting wonkier and wonkier and really needs a rebuild (it is mostly MDF), yet it made a useful part (in plastic) on Sunday and 2 useful parts the week before and 150 useful parts in plywood in the weeks before that. If you watch Edge Precision you will see he has a very large, very expensive, 'precision' machine, yet all cutting processes have a tolerance and if you want to reduce that inherant tolerance you have to do the work required, in Peter's case he is using measuring tools and various offset adjustments to get holes within 0.0005". Just because it is CNC does not mean it is 100% accurate, no matter the cost.
