I just recently bought an Anycubic Kobra 2 Max. For those who might not know, it’s a large bed slinger that has a supposed 500mm/s max print speed. It performs its own input shaping, but can’t automatically tune linear/pressure advance. The reason I bring this printer up is because 1.) it’s cartesian and 2.) none of the belt teeth ride across smooth idlers anywhere on the printer. It uses slotted metal rollers over metal rods for the X and Y axis. The rollers are tensioned just like with POM wheels. I’ve printed several large prints now, and after performing some filament specific tuning, I get wonderfully smooth prints with no artifacts. This is printing at a print speed of 230mm/s, 0.4mm nozzle, 0.15mm layer height and 0.36mm layer width.
I feel like the latest firmware update might help with stepper motor noise showing up on the print, but again, ultimately I feel the basic design of the belt is to blame here. I haven’t installed the latest firmware update yet so I can’t speak definitively yet.
Thinking more about this in general, I feel it should be possible to eliminate this “belt resonance” using input shaping, but not with the way it’s currently executed. Vibrating the bed and print head gets you the resonances of the machine as a whole. However, this cannot take into account the vibration caused by the belt teeth (ie vibration caused by the movement system itself) since the print head does not move far enough during calibration to detect this imho. You can feel it yourself. Hold the print head and move it quickly from side to side. You will absolutely feel that vibration from the belt teeth. Moving it foward and back feels smoother, or at least it does on my machine. I’m sure this could be compensated for with input shaping if the calibration included moving the print head its full range of motion in each direction at different speeds in addition to everything else (diagonally maybe?) Or maybe I’m just way off in my line of thinking? Again I haven’t installed the latest firmware yet, so bear that in mind. At any rate, it would be interesting to put an accelerometer on the print head to see if it would pick up on the oscillation causing our VFAs. Perhaps then it could be compensated for to reduce it or even eliminate it?
I will say another thing to try if anyone does go into swapping those pulleys out or making toothed covers for them - make one idler a tooth bigger than the other if there’s clearance for it.
That way it’ll reduce the compounding effect of both belts getting sync and exciting each other. (Might be why some printers are better than others depending where the belts are cut/finished.)
The same number of teeth are stepped regardless of the size of the pulley and one being larger would not affect sync other than altering the length of one of the belts. The speculation is that for teeth over smooth the Bambu idlers are too small though so larger pulleys in general might be an improvement.
If you move it left to right slowly with barely any pressure you can see/feel the toolhead actually wobble back and forth between the pressure of the teeth on opposite sides of the gantry. While the toolhead doing the sweep in the center with such a short movement does not trigger this tooth-roll vibration. But I do not think input shaping would deal with this anyway since it is extremely variable depending on the movement itself, the vector, and the position within the envelope. Likewise, I am not especially knowledgeable about input shaping, but my understanding is that this is not the type of issue it accounts for.
I gave the new firmware a try and ran a test print before and after the update (the right print is after). The lower noise is impressive but it doesn’t do anything for the VFA.
Can you please verivy one thing for me between the different printer models, that you are mentioning here? i´m interessted, if there is a difference in the diameter of the pulleys… Toothed or not toothed may not be the exact answer constructionwise… I would also belief, that adding tape, could create an offset to the axis anglewise. so maybe the parallel construction of the belt to the carbon rods is not the best way to go. I know, sounds obstruse, but who knows…
Parallel is considered to be the most ideal setup, so in theory altering the diameter of the pulleys here is detrimental. The buffer to the teeth as mentioned in the previous comment is the main effect and greatly outweighs any drawback from pushing out of parallel (negligible I think anyway with less than 1mm divergence over a 30cm span.)
even though it appears on the x axis, the vfx sounds like an movement on the y axis, so an offset even under 1mm pulls the printhead against the rails and by that is minimizing the tolerances in the bearings. so yes i don´t give a ■■■■ aboudt considerations about most ideal setup… works in theory, not in the real world with hundreds of parts with each one having it´s own tolerances… other idea i had, just get rid of the teeth on the section that goes over the pulley… this section should not reach the motor at all and the belts are cheap.
Yeah those are both with the tape mod, and I didn’t see much of a change when I added the tape. The teflon tape at least seems to be holding up fine and has nice little dents where the teeth go (I still can’t believe it doesn’t just unravel and get tangled after 5min). The lighting for these is also about as harsh as I can make it, I’ve got a very bright flashlight basically parallel to the surface. These are also at 45 deg angles so that only one stepper is turning at a time, from the bottom, 60mm/sec, 120, 120 reverse direction, 240, 240 reverse direction. Here’s the file (not mine but it ran fine on my X1C).
How many wraps did you do? 1 was subtle but noticeable and the more I added the better it got. I think I left it at around 7 wraps of unequal lengths (to prevent the overlap from stacking up in one place.) It really is wild how well it clings though. I probably have 100+ hours on the tape on mine at this point and no sign of deterioration.
I did a test where I tried to wrap the pulley barrel and tried to wrap the flanges separately. Of course there was some overlap but I saw a much greater improvement trying to wrap the flanges which leads me to believe it is the corner of the teeth striking the flanges. Rather than stripping the whole teeth away (because I think that would lead to a harsh line when the belt suddenly offsets by a millimeter) I am curious what would happen if you remove JUST the corners of the teeth - give them a little chamfer?
exactly. I own a p1s and i see this vfx also appearing on my PETG prints. More like holograms, but still there… hence why i´m in this post
Yes i think, it would be enough, to get rid of the roundnes of the teeth.
I think the reason for the problem is plain simple, yes also because of the flat pulley, but replacing them on your own could add new problems, so as long bambu lab is not releasing a gantry with updated pulleys, our hands are tied, when it comes to that problem…
So my simple explanation would be, that the teeth deviate the belt just a slight bit, so the belt is not parallel to the rods every tooth. This in combination with the tolerance in the bearings of the carbon rod, moves the Toolhead on the y axis. More Speed means more inertia in the main direction and so less deviation on the y axis.
So getting rid of the round end of the teeth on the Belt could help with the problem in two ways. First more contact surface on the pulley and less deviation from tooth to tooth. And second, a little tension in one direction on the toolhead what minimizes fibrations created by tolerances.
I’m assuming that you see it more on PETG than on PLA prints is because the flow cut-off is lower on PETG. This means that PETG prints are printing slower. And from what people have been saying, that slower speeds seem to show the artifacts more clearly.
last polymaker polilite 260 degrees print was with 150mm/s. would fit in the 130-160mm/2 bad range. Its there but more like holograms…
I think its a combination of different factors and as i read more into the carbon shrinking topic, i learnd, that you can minimize and also negate the shrinking in different ways, so its a question of what carbon and resin treatment( alignment, different resin mixtures (resin expansion coefficient)) Bambu lab is using.
So higher temperature can cause higher tolerances could be a true statement, or completely fals. Idon´t know.
I just interpretet the statement about filaments, that are more vfx friendly and ohters and looked for a correlation.
But you´re right, speed is an realy important factor.
And my personal favorite soloution for this vfx problem would be.
Bambu lab figures out, how to make a belt with the two end sections in the same thicknes as the teethet part and just sell it, as an belt uprade.
It´s the cheapest solution. Single part, every one has to replace it based on printhours anyway and its easy to replace, cheap, and easy to distribute in big quantitys.
So if the production is easy that would be a great way.
Or maybe one of the third party faction could see a possibility.
As long as thats not happening, maybe shortening the teeth would help…
If someone want´s to try. best way would be using double sided tape on a board or plank about 2 inch wide and glueing both belt ends along the corners of the board. So it´s more difficult to ■■■■ up the alignment of the belts.
Then drag over fine sandpaper as you feel.
