You should distinguish clearly between the terms as symptoms versus causes: Z-ribbing (a periodic, axially symmetric, pattern of roughness along the z-axis) may be caused by z-quantization which squashes or stretches layer thickness due to numerical, electrical, or mechanical processes.
Z problem. That’s what it is. Depending on the cause (wobble, rounding error, missed steps) it needs different solutions. But it’s a problem with Z.
If the user can’t differentiate between the types of Z-problem before they start trying solutions, they are going to adjust things that don’t need adjusting and potentially make things worse.
The different causes would have potentially different symptoms. If wobbly Z screws were pulling the print sideways, the roughness pattern along the Z axis would not be axially symmetric, but would still be periodic, and would need to be solved mechanically. A numerical rounding problem would be periodic and axially symmetric, but would need to be solved in software/config. Missed steps wouldn’t be periodic, would be axially symmetric, and the solution could be mechanical, electrical, or software.
I don’t have a suggestion for the best terms to use, but I do think that in choosing the term for a thing, you should start by knowing what the thing is and how it differs from similar things.
I’ve gone with “Z-ribbing” (for now) and finished the first draft of the article. I’d appreciate any feedback people can give.
@Whosa_whatsis Wow! Lots of good info in there. Its a little hard to follow though considering your audience. Images, headings, bullets, etc can help (I assume this is just a manuscript so to speak). And examples of each kind of problem. A little more info on the jump over to using microns and how that translates to steps_per_mm or layer height in mm could be helpful. Also I think you could break down how the reader calculates optimum layer heights with a formula and more examples. Overall great job!
I do plan on adding pictures to demonstrate each issue. If anyone has pics that demonstrate the other issues as well as the one on this post does, let me know!
Thanks g+, i was half a page into my post, clicked a millimeter to far to the left and now it’s gone (even though i didn’t hit cancel or anything like that). Please excuse my compressed writing.
These show wobble more drastically:
- Sells
- 90, Printrbot, Prusa i3…
- anything with the motors on the bottom
These don’t
- Prusa (i2)
- Max v1
The position of the rod constrains differ (top/bottom) - bottom shows wobble in the much-used lower regions of the printer, top has more wobble towards the top (if any)
With the X-axis “pushing” onto the couplers, the couplers themselves will go excentric (saw this on my 90 before i switched to “plum couplers”) - this allows the rods to sag down a bit, creating a state of lower energy, which the system will stay in
With the X-axis pulling, the couplers will center themselves
Think about the coupler as a fairly soft element; take a wire and push on it and you will end up with a bent piece (and the spots where you pushed being misaligned), but pull on it and it will become perfectly straight.
I hope this still makes sense…
@Thomas_Sanladerer I thing I see what you’re saying, but I don’t think that’s necessarily the case. If you use couplers that properly center the two shafts, this should not happen. What you say is certainly true of the wobble of the loose end of the rod, but this will not create wobble in the print unless you constrain that loose end and transfer that wobble to the Z stage instead.
“If you use couplers that properly center the two shafts, this should not happen” Agreed. The only couplers i’ve used that achieve that are my modified plum couplers. What goes along with that, though, is that the Mendel90 leaves almost no space between the coupler and the X-end’s nut when it’s printing the first layer. Just like on the Sells, the wobble will get weaker layer by layer and eventually become so weak that it’s not noticeable on straight walls anymore. 45° overhangs will still show it, though. It takes just about a centimeter on the 90 and more than five for the wobble to become unnoticeable.
The Sells had a different way of introducing off-center movement into the system: It had its Z-rods screwed tight onto the 608 bearing - but because M8 rod is a fair bit smaller than 8mm on the outside, the bearing would not guide the rod dead-center.
I’d think a neat way of absolutely getting rid of Z-wobble would be to have the Z-rods turned down to, say, 5mm on the driven end and guided by a 625 bearing. That way, the coupler could be anything that somehow transfers torque and the rod would be perfectly guided. A bent rod would only show its effects on larger Z-heights, but at those heights would already be flexible enough to be guided by the smooth rods (which imho should be at least 10mm, btw).
I’d use M6x1, M8x1.25, and M5x0.8 throughout, to focus strongly on pitch.
I’d also use ‘runout’ term after talking about the small circles. “Runout”, its definition, measurement, problems, and solutions are well understood in non-additive Manufacturing, and you should leverage that knowledge. http://www.emachineshop.com/machine-shop/Runout/page607.html Using “runout” might make the “sinusoidal ridges along the vertical surfaces, with inverted ridges on the opposite side” more clear in that the overly constrained leadscrews transfer the runout of the leadscrew into runout on the manufactured part.
Maybe add a metric example so folks with metric screws can demonstrate the problem to themselves. Perhaps 1/3mm layers? Or am I misunderstanding things and it fundamentally impossible to get z-ribbing with metric pitch threads?
@David_Forrest it is entirely possible to get ribbing with metric threads. M6 seems like the perfect pitch (1mm) that will work with any reasonable layer height (i’d call 1/3mm unreasonable. Metric and fractions are never mixed.)
However, the typically used M8 (1.25mm pitch) rods will show signs of ribbing with many heights that seem reasonable - like 180µ, for example.
@Thomas_Sanladerer So if 180 micron layer on a M8x1.25 rod with 200 steps shows signs of (5-layer?) ribbing at 28.8 steps/layer, then on an M6x1 rod, a 144 micron layer should show similar ribbing?
The absolute error and the “pitch” of the pattern will be a bit smaller, but generally, yes.
Z Aliasing Patterns or Layer Height Aliasing Problems