Topic: Banding problem: my experience and how i completely got rid of it
I spent the last couple of weeks with my solidoodle doing all possible tests and research in order to attempt to solve the infamous banding problem. I am happy to announce that I managed to solve it very effectively and I would like to share my experience.
My initial problem could be rated as "pretty bad", meaning that it was not terrible but I often had delamination in large prints and the parts were not smooth at touch. I managed to break down the problem into three parts and found respective solutions.
1) Hardware: this is the most important source of banding. The fix for it is now well known and documented: see the wiki http://wiki.solidoodle.com/z-axis-anti-backlash. This solution fixes the Z backlash and considerably reduces the overall wobbling of the Z axis. This wobbling makes the bed shift nonuniformly in Z, making some layers thinner and some layers thicker. Prints change noticeably with this fix.
2) Slicer software: some other users reported problems with Slic3r (version > 0.9.7). I can confirm these problems. Specifically, I found that the feature "randomize starting points" generates shifts between one layer and the next. Such shifts resemble banding because apparently the randomization function has some periodicity. The shifts are probably due to backlash and can be compensated with Neil Martin's hysteresis fix, but unfortunately I did not have success in using it, noticing a constant shift in the direction of the compensation. My suggestion at the moment is to turn off the "randomize starting points" feature.
At this point, your prints will hopefully look quite smooth, with possibly a tiny banding with a period of 1.41mm, due to a residual wobbling of the Z axis. In my case, I could see the banding only by looking at the part with the right angle... However, I could still feel it by touching the part. To remove this last wobbling, I implemented the following:
3) Firmware: Z wobble correction (as described in this other post). I am very satisfied with the final result (top: all corrections applied minus the firmware, bottom: all corrections including firmware. Layer thickness: .3mm):
The final firmware correction only compensates for the bed level: other users reported/hypothesized an XY coupling to the Z wobble. This could probably be the case if the rod wobbling is very bad. My firmware mod is not designed to take care of this, but it could be easily inserted. Manually finding the right parameters would be really painful though.
Hope this helps, and if someone has different experience, please share it!



