A design error in the original QU-BD OneUp exhibited very strong banding in the Z-direction (I originally attributed it to Z-wobble, but realized a few days later that it was banding, not wobble).
Whosawhatsis (who is now a part of the team at Deezmaker) wrote about this quite some time ago, but I had forgotten about that document until this post on fabric8r reminded me of it. He calls it Z-ribbing, which I think is a far more evocative term.
The ACME Z-screw is a 3/8"-12 having 12 turns per inch, that's 1 turn per 1/12", or 1 turn per 2.116(666...) mm.
With a 200 step motor, that yields .010583(333...) mm per full step [*]. From the repeating 3's, we know that every 3 steps yields a terminating decimal. So layer heights should be in multiples of 0.03175 mm to eliminate the rounding errors that leads to this error.
So, on the OneUp/TwoUp, the desirable layer thicknesses are: 0.03175, 0.0635, 0.09525, 0.127, 0.15875, 0.1905, 0.22225, 0.254, 0.28575, 0.3175, and 0.34925 mm/layer.
Interestingly, when I installed the Z-bearing alignment spacers on my OneUp, the banding went away, leaving behind a faint Z-wobble. It may be that by eliminating the binding of the Z-screw to the Z-nut, the gantry is allowed to move more smoothly, reducing the positioning error per micro-steps.
My prints do show some faint Z-wobble (and I can see the screw wobble when I do long Z-travels) - which can't be eliminated because the thread pitch is far higher than achievable layer thickness.
[*]: BTW, to further extend the calculation, it's 94.48818897637 motor steps/mm. The 16x micro-steps on the controller yields 1511.811024 steps/mm, The value stored in the EEPROM is 1511.81 steps/mm. The remaining .001024 steps/mm is pretty much negligible, as it's only .128 step over the 125 mm range of the OneUp/TwoUp.
UPDATE 1: Ironically, now that I've moved to integral motor steps, the Z-wobble is seems to be more apparent! I need to do a few more tests, but preliminarily, it looks like I might need to split the error between the Z banding and Z-wobble.
UPDATE 2: As I am doing more tests, it's starting to be more clear that I am still seeing Z-banding - much smaller than the original Z-banding before the spacers, but still at the pitch of screw thread. So mechanical issues is the dominant cause. Some calculations I am working on suggests that accumulation of errors still can contribute to the "ribbing" effect at approx. 2 mm intervals, but the effect from that should be smaller. unfortunately, with the interval close to the thread pitch, it's hard to separate the mechanical issues from the mathematical one.