Realistic repeatability expectations?

[amfsolinc (Ben Ianiro)]

@mikep I rehomed prior to starting the job and setting the zero point and experienced this issue.

[jwhendy (John Hendy)]

You might check my post here. No obvious answers there and it got derailed a bit, but the tl;dr is I think that what I was calling a “repeatability” error was really an “perpendicularity” error between my x and y rails.

You could test this pretty easily. Put some stock down and bore a full depth hole. Power cycle your machine and cut the same hole half depth. Do you see a step?

I was getting what I considered massive steps (on the order of what you see), but it was purely limited to 2-sided flipping. My main takeaways were:

• make sure your rails are perpendicular (my x axis gantry had pretty big uneven gaps between the left and right y axis front plates, maybe 1/8". arctan(0.125/33in) = 0.2deg, which over a 6in work piece happens to produce an offset of 0.02in…
• make sure the dowels are tight. I was using some tiny 4mm dowels and didn’t want to keep remaking the holes in my MDF so I had them pretty easy to insert. I also didn’t like wrenching on my part to get it off of them, so I went on the looser slip fit for the part as well. I think there was far more variability in placement and alignment than I realized there as well. Test that by trying to torque your part when it’s on the table and watch a corner. Think you can move that corner 0.02?

I’m starting to lean toward using a corner locator instead, and when I’m machining my top (side 1), I’ll machine two sides. This way, I know I have two perpendicular references. When flipping (side 2), I’ll reference off the square. Much easier to re-locate to a square (account for any non-repeatability in switches) than to a hole (unless you’re cutting fresh), and less ambiguity in making to edges mate vs. doing the dance of making dowels usable (you have to get the part off/dowels out) vs. super solid.

Hopefully something in there was helpful!

[amfsolinc (Ben Ianiro)]

Just wanted to update the thread, I did a few tests to see if I could narrow down what the issue was and have been unable to replicate any appreciable repeatability issues. I set up an indicator in my router and swept a pins location then moved the machine around and cycled its power and was able to maintain location within 4 thousandths, I appreciate everyone insight and suggestions, for now I am just going to continue to watch for any issues and go from there.