Vertical Banding Revisited

[Nylocke]

I search thingiverse for a file and then I forgot to link it.. oops

http://www.thingiverse.com/thing:168022/#files

[KAS]

Try slicing and printing this with your regular settings and then with spiral vase enabled. Then post the G-code and I'll try printing them and we can compare the results. KAS could try their hand in printing it as well, then we can all compare results. Make sure you post your diameter and flow rate settings from Cura so we can get our extrusion multipliers correct. From this we should be able to see the effect of x32 drivers and what, if any, effect the infill has on the surface finish.

Sounds good, link a file and I'll print and post results.

[Jimustanguitar]

I"m a Cura guy as well.

I'll try this with the shorter prism.

[KAS]

I'll start on trigonal_prism_h10 tonight.

[DavidF]

Should we pick one layer height and all stick with it, or does it not really matter?

[KAS]

Should we pick one layer height and all stick with it, or does it not really matter?

I'd say the same just for comparative reasons. It looked like a wider part, but maybe it was just the height that changed.

[Nylocke]

Well, the plan was for one person to slice it and then everyone just adjusts temps and extrusion, so we all use the same G-code.

[Jimustanguitar]

Interesting... I was already running a piece that I decided to play with the speed multiplier on and it did change the banding.

Here is 40, 60, 80, and 100 mm/s, respectively.
[img]http://i.imgur.com/wl8DdZq.png[/img]


It's not completely gone at the fastest speed, it's just a more matte finish because of the temp difference. My PID must not be tuned well because I'm set to 220deg and it's only extruding at 116 when I get into the higher speeds.

[KAS]

Well, the plan was for one person to slice it and then everyone just adjusts temps and extrusion, so we all use the same G-code.

I understand. I'll let one of the more experienced people start it off then just in case I have something completely screwy with mine

[Jimustanguitar]

Here you go. Sliced in Cura both ways.

[KAS]

bottom shows vertical bands in large print, next shows minor bands printed with same settings in matterslice, next shows about the same with spiral vase with no fill, top is the gcode from Jimustanguitar (standard Cura)

Cura showed more vertical banding but the print in person is much better. I'm going to match the layers in matterslice and try it again.



[img]http://i482.photobucket.com/albums/rr189/Onuxis/20150208_204559.jpg[/img]

[jdurand]

The thing I just printed has the pattern. Narrow spacing compared to the infill. You can see it in the attached picture.

Infill 10%
Slicer S3D
Speed default (50mm/s infill, don't remember what perimeters are).

[Nylocke]

I'll get to printing mine tomorrow. Got Pilot's EZstruder mount going right now. Excited to try it out.

[Jimustanguitar]

I was able to tease out some of the banding with a change in speed and temp, but it's not all better. (switched from 40mm/s @220 deg to 70mm/s @ 215 deg)

Before:
[img]http://i.imgur.com/qVihWzt.png[/img]


After:
[img]http://i.imgur.com/096bIrK.png[/img]

[KAS]

I was able to tease out some of the banding with a change in speed and temp, but it's not all better. (switched from 40mm/s @220 deg to 70mm/s @ 215 deg)

For shits and giggles can you weigh those and see if its a significance difference at all? I'm noticing that even the infill has a pattern, specifically larger parts with lower infill ratio.

[DavidF]

ok here is the one with infill just as sliced. hard to see the banding on such a small piece but its there.
[img]http://i1102.photobucket.com/albums/g446/davidflowers1/Mobile%20Uploads/20150209_094609.jpg[/img]

[Jimustanguitar]

So what's the conclusion of your test, and what should we test next?

I wasn't sure if the idea was to visualize the GCode and try to find ridges that way or to print the same file on multiple machines to see if it's identical...


BTW, I didn't mention above that I sliced those files for a .4mm E3D hot-end. If you try it with a stock setup, it no workey.

[Nylocke]

I didn't realize how small this stupid thing was. I was expecting it to be a good 5-10cm per side.

I used your exact g-code. I had to turn off the fan and bump up the temps, but it worked out fine. My results are more of "what good cooling fans can do for a small and fast ABS print"

When I said "I don't get much banding" I was only thinking of the non square parts I had printed. I don't print much square stuff (or things with large flat faces). Looking back on the few things that I have, I realize I actually do. Prime example of this I don't think is a good example, because its suffering from a few other issues (mainly over extrusion) so I think the banding is more pronounced.

The goal was to print on multiple machines to see what effects different hardware has on the same g-code. Since I have 32x drivers and TL arms I was expecting my banding to be less pronounced that someone else's without. Would be interesting to see someone with a smoothie.

The main question I have is how accurate your pictures are to the actual scale of the issue, Jim. Is that just the camera, or does that print you did 4 posts ago really look that banded?

KAS showed the same object sliced in different slicers. I'm wondering if the differences were due to each slicer calculating the paths/speeds slightly differently (assuming you used the same speed settings for each) or if some of them segmented each line into smaller segments for whatever reason? That would have to be checked in a g-code viewer.

After all that, heres my results. Its hard to get enough light to see everything properly.

Really blurry, but the flashlight I shined on it helped show the banding a bit better

Again not the best of pictures. It does show the benefits of ABS cooling on the left part though. Sped up to 300% speed (Cura slowed it down) and so I turned on the layer fan to 50%

(I didn't enable the layer fan until it got around 3mm up so it started mushrooming)

[Nylocke]

Taking a look at the Cura slice in RH, it does indeed not segment each side. Each perimeter is one G1 command. So that rules out that. Interesting how Cura provided the most banding in KAS's tests, KAS, would you mind posting your settings for each of those that you did? Speeds, filament diameter, flow rate.

Jim, you're using Cura, right? Maybe check how the paths of another slicer affect the banding? What I'm mainly intrigued by is how pronounced the banding appears to be in your parts. Most of the banding artifacts I see come from when I print my extruder design, and they aren't very heavily noticeable. I doubt your average joe would even notice them.

[KAS]

Filament Diameter 1.73 PLA @ 220c/60c
First layer height .2
Layer Height .25
speed inside 20 mm/s
speed outside 20 mm/s
speed infill 20 mm/s
flow ratio 1.0


I literally printed his Cura gcode as it was. I'm going to change the speeds to 20 and print it now.

[Jimustanguitar]

Yeah, my prints really are that banded. You wouldn't notice it half as much with another color, but this black PLA really has a glossy sheen to it.

I'll have to gather all of my settings and try translating them into a different slicer this evening. I've been using Cura exclusively for about a year, so I've made a lot of tweaks since the last time I opened one of the other programs.
It sounds like others in the thread have Kiss, Slic3r, and MatterControl covered, so I'll do my tests with CraftWare. I've been meaning to start poking around in it anyway.

I've got a paying print job that I need to keep the machine running for instead of making test parts with everybody else (I actually need multiple printers, it's a good problem to have!), but I'll be playing with the settings and tweaks that we find in this thread on those parts as I go.

[KAS]

I printed your gcode 50% slower and the bands are there but I'm telling you it looks better than mine. They are small uniform, almost reminds me of those 3d cards that you have to move back and forth to see the effect. I cant feel if with my fingernail.

What about simplify3d? Thinking about picking that up.

[img]http://i482.photobucket.com/albums/rr189/Onuxis/20150209_154534.jpg[/img]

[Tonkabot]

I am thinking that eventually someone should really implement the curve G-codes. So instead of always chopping everything up into straight line segments, actually running a spline curve, or bezier or whatever. Done right, it should improve the output quality and reduce the gcode file size.

Seems like a smoothie board would have more horsepower to do it.

I once wrote a screen saver that would 'scribble' a line on the screen. Instead of using segments some higher level function, it took the random bezier curve segments and then used fast forward differencing to plot the points. [FFD can plot all the points on a bezier curve without having to do much math per point, just adds and multiplies]
It was amazingly fast even on an ancient 386 without floating point. As in 1,000's of curves per second. I had found out about it because Silicon Graphics computers could do the FFD in machine language - these were way before graphics accelerator cards.

The FFD could be done on a smoothie (probably even an atmel) to plot curves [quick and efficiently]

Here is someone explaining the math. Which I knew how to do a long time ago - it gets fuzzier now http://www.scratchapixel.com/old/lesson ... ferencing/

[Mac The Knife]

If we could just get away from using stl files, that would be a bonus.

[Jimustanguitar]

If we could just get away from using stl files, that would be a bonus.

Most of the current slicers will already support AMF files. They're still a mesh, though (it is a better mesh, at least).