Vertical Banding Revisited

[KAS]

I noticed something interesting when I created a test piece for this vertical banding issue. This rectangle is 60x10x15mm with no top, It's created with a .5mm boarder and printed as a spiral vase.

It appears the bottom 3 layers matched my large parts vertical banding exactly. Both items were printed the exact same speeds (20mm/s) except for the spiral vase option on the test.
Once it starting making the walls it started looking pretty damn good with almost no noticeable banding.

I'm at a loss why it would band part of a print and not the whole thing...

Click the image and zoom in with the Photobucket controls to get a closer look.
[img]http://i482.photobucket.com/albums/rr189/Onuxis/20150211_191606.jpg[/img]

[Polygonhell]

Implies it's related to the infill.

[lordbinky]

Yep, which leaves you with a couple fixes. One is you can decrease the bulge by increasing your jerk setting (which might also affect the perimeter to infill adhesion) or simply avoid all that add another perimeter that will be less affected by the infill.

[JFettig]

Implies it's related to the infill.

If someone who's having the issue real bad has S3D, they could try the random infill option to see if its banding still, infill is mostly junk with this setting but it will randomize it.

[dunginhawk]

so im having this issue on larger prints with infill... going up to 4 shells one suspects may resolve the issue? thats a sucky workaround though

[KAS]

And that's the confusing part for me to grasp. My infill is 10% triangle and it doesn't match the vertical bands at all going up the wall. Matter of fact, you can see the banding in the infill it's self and that's only one layer thick. crazy stuff...

[Khalid Khattak]

Actually a complex phenomena.. Infill also behave like a support material for the outer layer... You can see that the unsupported area between two infill point may behave differently due to shrinkage etc..

[KAS]

Actually a complex phenomena.. Infill also behave like a support material for the outer layer... You can see that the unsupported area between two infill point may behave differently due to shrinkage etc..

As much as I would like to just take for granted that these evenly spaced bands are caused by shrinkage or support layers/infill, that's to random to produce something so specific and uniform.
For example, in the large part above has been created twice with 10% and 25% infill both covers exactly 5 vertical bands within the same distance. Those 5 vertical bands line up exactly with the part on top that effectively has a solid three layer base with .07 overlap. Now the bottom part infill which produces rather large interior open gaps because of the 10% produces the same vertical banding as the face of perimeter wall at different angles. The small part perimeter produces no vertical banding after it passes the layered base with no infill at all.

The infill of the large part produces a wall the same thickness of the small parts perimeter.

This vertical banding appears to be extremely common throughout the Detla style printers. Obviously it's not an easy fix, and most just seem to loose interest and find a setting that minimizes the effect or deal with it. The only somewhat fact I can find is that it is more prevalent in smaller nozzles. That might be the reason that I did't notice it as much in the stock .5 nozzle vs the E3D v6 .4

Dont get me wrong, while the images I show looks horrible it's because I'm finding that perfect angle this light reflection to show the banding depth. In normal light, it's doesn't look anywhere near that bad.

Anyways, I'm going to keep tweaking my prints until I'm happy with the results then move on to the move difficult filaments. I'm really interested in the wood and metals.

[dunginhawk]

so this is a major issue that a lot of us seem to see.

what would be the 1 setting that people are finding the biggest change from?

I definately want to make any changes to try to resolve this.
thanks

[KAS]

Interesting read. Different printer/style same issue: http://www.printrbottalk.com/forum/view ... =15&t=5721" onclick="window.open(this.href);return false;

[Jimustanguitar]

As much as I would like to just take for granted that these evenly spaced bands are caused by shrinkage or support layers/infill, that's to random to produce something so specific and uniform.

I think you're exactly right. A vibration or some other motion/motor artifact wouldn't repeat so consistently for dozens and dozens of layers either...

I'm starting to believe the software/math/polygon theories.

[Jimustanguitar]

So I just had a thought... We've been talking about how these machines run one segment at a time instead of continuous lines and curves... This would also be true for the extruder.

Could this be die swell from the start and stop of each line of GCode? The speck of time in between each line of GCode with new extrusion instructions?

Perhaps a test with the flow multiplier would change the frequency of the ridges? How can we test this?

[Khalid Khattak]

Thanks for enlightment..but with my Delta Printer i am not getting any banding/texture---- I use ABS 0.4mm Nozzle, 0.2mm print layer 50mm/sec speed...

[Tonkabot]

So I just had a thought... We've been talking about how these machines run one segment at a time instead of continuous lines and curves... This would also be true for the extruder.

Could this be die swell from the start and stop of each line of GCode? The speck of time in between each line of GCode with new extrusion instructions?

Perhaps a test with the flow multiplier would change the frequency of the ridges? How can we test this?

Die Swell ??? But yes I think the rambo/smoothie is constantly interrupting to update the new step parameters. Like in the thought I had that put forth the idea that we are seeing the stair-stepping of the steppers (or an artifact of it), I explained that doing a move from (-100,0) to (100,0) that the 'z' stepper would not move, and 'x' would be constantly decreasing in speed while the 'y' would be increasing. I suspect that (unlike an cartesian which only has to move the 'x' motor at a constant speed for that move) the controller has to move at one speed for a short time, then constantly be updating the speeds/step rate. The 'x' step rate is doing a stair step to try and fit the ideal curve, and the 'y' step rate is also doing the same, I don't think it matters if they are being updated in sync. They are stair stepping to try and fit the curve, and they are only at one point per step-rate-change exactly on the curve, the rest of the time they are either high or low.

The problem: the delta axis (axii?) are always travelling on a 'curve' - even to make a straight line. and the current software can only approximate a curve by cutting it up into segments of straight moves. Even if the axis is making a straight move it still will have to match to a constantly changing speed CURVE.

The solution: Implement actual curved moves. So that stepper axis are able to follow a speed curve without having to chop it up into segments.

[Jimustanguitar]

Die Swell ???

Differing or pulsing "pressures" on the filament would cause corresponding artifacts in the printed part... I may be using the wrong terminology, but I thought die swell was pressure's effect on extruded diameter... Sorry about that.

the delta axis (axii?)

Axes with a long E.

[KAS]

Thanks for enlightment..but with my Delta Printer i am not getting any banding/texture---- I use ABS 0.4mm Nozzle, 0.2mm print layer 50mm/sec speed...

You're welcome. Care to post an image so we can compare? It would be nice to have a goal or at least see whats possible.

[dunginhawk]

so i tried three tests tonight.

1. 2 outter shells, 50mm/s printing a rectangle. normal bad banding.
2. 5 outter shells, 50mm/s printing a rectangle . normal bad banding. In fact its like ballistics of a bullet/gun its in EXACTLY the same spots as the other print...
3. Circle with 2 shells - 50mm/s . banding, but not quite as bad. I assume if i do vase it would be gone.

Not that its super proof, but i really think it has to be the fact that the deltas cant do straight lines well. circles are better for it. Doesnt make me happy as my large print job right now is a rectangle part

[bvandiepenbos]

Maybe a odd number feed could have a positive effect? so maybe there are sweet spots, 41 or 53 etc.
Seems like everybody is typicaly going 40, 50, 60, 100 etc.
How about 33 since there are 3 motors?

[dunginhawk]

so for those running a delta printer without banding issues, what are you doing away from the stock parts and configs?

[JFettig]

I haven't really had much of any banding ever, never really did much of anything special. I never print PLA. I print from 20mm/s to 100mm/s and it doesn't really make a difference. I always print from the SD card(now trying octoprint).

Completely stock machine never had any issues, now dual extrusion, TL arms, etc. still no issues.

[DavidF]

I haven't really had much of any banding ever, never really did much of anything special. I never print PLA. I print from 20mm/s to 100mm/s and it doesn't really make a difference. I always print from the SD card(now trying octoprint).

Completely stock machine never had any issues, now dual extrusion, TL arms, etc. still no issues.

yes but what size nozzle are you using? A smaller nozzle dia increases the x-y resolution making the banding more apparent....

[KAS]

My printers apart right now getting a good cleaning and doing a better job on the cable management. Installing the v27 belt tensioners, and I think I've decided to remove the dampers while I'm at it.

At least that way I have a base line on how it prints both ways.



[img]http://i482.photobucket.com/albums/rr189/Onuxis/20150215_204933.jpg[/img]



These were linked on the forums, but not sure who made them. Thank you who ever you are!

rostock max v2 belt tensioner v27.stl

(159.26 KiB) Downloaded 239 times

[JFettig]

I haven't really had much of any banding ever, never really did much of anything special. I never print PLA. I print from 20mm/s to 100mm/s and it doesn't really make a difference. I always print from the SD card(now trying octoprint).

Completely stock machine never had any issues, now dual extrusion, TL arms, etc. still no issues.

yes but what size nozzle are you using? A smaller nozzle dia increases the x-y resolution making the banding more apparent....

.5mm from seemecnc and now a .4mm from E3D. I have a .25mm nozzle but I don't have a need to install it right now.

[Tonkabot]

Die Swell ???

Differing or pulsing "pressures" on the filament would cause corresponding artifacts in the printed part... I may be using the wrong terminology, but I thought die swell was pressure's effect on extruded diameter... Sorry about that.

the delta axis (axii?)

Axes with a long E.

If the patterns stem from the extruder, then if we do something to change the speed of the extruder (like print thinner or thicker layers) the patterns would change [size].
That should be pretty easy to test.

I think these patterns all boil down to the fact that we are using steppers which are almost always going to have a 'notchyness' due to being steppers [ and that micro-stepping does not eliminate the notchyness]
If we upgrade to higher [native] resolution steppers like .9 instead of 1.8, or even try the exotics of .36 or .18 I think it the notchyness will be less visible.

If we use a smaller number of teeth on the stepper pulley, that will help. Maybe if we go to a finer pitch belt we can go to an even smaller number of teeth on the pulley

If we upgrade to geared steppers that would also increase the step resolution - however there will be added backlash thrown into the mix.

If we make the software run the steppers even more smoothly, that should also help. I think that the accelerations must be smooth for better print, and I am questioning how the accelerations are implemented. Also I think if we implement smooth accelerations [and curves] then the moves should be unsegmented and made as long as possible - and prints should be more artifact free.

I know of a memory intensive way to implement simultaneous smooth stepping on lots of steppers at once. It would be interesting to try it, although I am pretty sure non of the current boards can handle this technique. Maybe I should add that capability to my board I am designing. It would mean moving to a BGA cpu package that can do DDR memory. How does 256MB Ram sound?

[dunginhawk]

anything that fixes this banding issue sounds good to me