3 Axis CNC router

[Neale]

I saw exactly the same problem on my machine. X and Y were fine, but Z looked as if it were losing steps on complex 3D work. In my case, it didn't show up on single test runs; what seemed to happen was that you lost one step every time you changed direction. Fix was simple - as Jazz said, just change step polarity. I think this is more likely to happen with the CSMIO controllers as they have differential outputs - brilliant for noise rejection, but too easy to get the two wires swapped!

My only reservation about this diagnosis is that I wouldn't expect to see any significant error on a single movement, as the effect is cumulative with movement direction changes. But Jazz's test (I did exactly the same thing when I was chasing the error) will soon show the truth!