Well you don’t design toothed belts to run against a smooth idler because as the teeth meet the surface, you do have a gap transition, thus vibration. Would perfectly explain the gap distance changing based upon how fast the printer (hz) is relaying the vibrations.
At least, as far as I’ve seen from my minimal exposure to the design, nothing else is “sqaure meets circle”?
I was curious, does anyone have a picture of the internals of the carriage?
You’re right. I was looking at the pulleys intently last night and realized only two have teeth riding over them. I skewed the tensioner pulleys to force the belts to track straighter over those pulleys and there was no improvement. I do not believe it is the motor because the cogging feeling comes from the X axis not the motors themselves.
When moving the X axis realllly slowly, the toolhead clearly wants to pull forward and backward alternately so I do believe the belt teeth are still imparting that force.
Would perfectly explain the gap distance changing based upon how fast the printer (hz) is relaying the vibrations.
Can you explain this a bit more? I’m not sure what exactly you mean but it sounds like it might describe how it feels staggering +/- Y when I move it slowly in the X directions.
I would like to try toothed pulleys but it appears that you can only remove them if the carbon axis is fully removed - the lefthand idler pin can potentially be forced out the bottom but the right hand has access blocked by a metal ledge.
I think I saw a picture of the internals somewhere, but they definitely had to access it destructively.
Tried new things:
• Tried tightening the bed screws and re-tramming as suggested in another thread - no effect.
• Tried aggressively decreasing tension on the belts to a point where the printer complained about low resonance on both axes. The amplitude of the artifacts was not diminished but the speed they appeared at was lowered.
• Added 4 quarters (~22g) to the toolhead. It SEEMED like it decreased the resonance slightly and maybe lowered the speed at which it peaked, but it was just as bad on the rear face of my test model as before.
Because the pattern always appears to be worst on the rear face of the test models, I am beginning to think it might be related to the direction of travel. It seems to always travel counterclockwise on single-walled rectangular shapes, so I plan to make a new test which will force it to travel in both directions to see if that has any bearing on the artifacts.
That’s actually coming from the motion system, it’s how CoreXY works. I only half-understood that after changing the belts and looking at how they are actually routed. If you want to excersise just one motor, you need to pull the toolhead diagonally 
Your printer might have some snags mine didn’t (surprisingly, after 4500 hours of printing), but most if not all of the resistance I felt in the belts was coming from the magnets in the motors, no pulleys, even though they had debris on them, nor the belts that looked like they snagged on the idlers, it was buttery smooth. 95% came from motors, about 5% came from the Y axis bearings. The bushings on the X axis ride surprisingly smooth when no belts are in place, but once you tension it there’s always some amount of force in one direction or another trying to rotate the whole gantry and there’s always more tension on one side of the belt compared to the other.
The fact that those artefacts (or rather the speed where they appear) change with weight/tension/speed indicates to me that those are not coming from the motion-ing part of the motion system but from somewhere else (resonance). I take this as a proof that the frequency compensation is not that effective at all speeds, but we might also be chasing ghosts (it’s ghosting after all :)) and just accept that it will never be perfect. I just choose filaments that hide the flaws and don’t worry about it…
I have had the A motor out and there’s certainly not that much cogging happening from the motor, especially since the same cogging cannot be felt at the drive pulley while under full even beyond full tension. I believe it is most likely the teeth rolling over the idler pulleys on the X axis and may not even be the source of the cogging, but that could be part of or unrelated to the overall issue producing the artifacts. I find it hard to imagine it could be the motors that make this unit so appalling compared to both its contemporaries and later production runs.
Your case may be very different from mine as there is no filament out there that can hide the artifacts my printer produces. On a matte print the front and sides could be nearly defect free, yet the rear shows 0.03mm variations you can feel with your eyes closed. Using a glossy, transparent, or silk filament makes it even worse, and I need to be able to use clear filament and there is no printable speed which eliminates the 2mm VFA.
I do not think you can rule out the vibration coming from part of the motion system. The fact that the resonant peaks occur at different speeds for different tensions speaks to the fundamental frequency of the belt. Anything in the machine could be causing the vibration and the belt is just resonating at its natural frequencies.
There’s no reason to think the motion system can’t create vibration, such as belt teeth rolling over an idler, the teeth snapping over a seized idler, or the linear bearing skipping on the rod like dragging a chair over a floor. Since the motion system is explicitly that which is moving, it must be what is creating the vibration which is resonating in the system. Outside of that there are fans, but those would not generate such irregular yet consistent patterns. I think I accidentally ruled out the toolhead fan because it had become unplugged and some of my tests had no part cooling fan because of that yet demonstrated the same artifacts. Regarding the repeatable irregularity in the patterns, the direction of motion is a candidate, and the reasons I’ve thought of why that could be a factor are that the linear bearings are asymmetrically defective or (a friend’s suggestion) that the forces imparted by the umbilical are causing it to manifest in certain places but not others.
This is pretty baffling indeed. Just for a sanity check. Maybe try another calibration model. Try out this quick-print wobble tower test linked below. Run it two ways, first using your current settings then change print to spiral vase mode. See if that has any impact. Make sure that you use zero infill, wall loop=1 and top layer shells=0. This will ensure a hollow structure like the one you’re using for calibration. I assume that’s coming out of Orca?
The sole difference between running that in standard vs spiral vase, is that spiral vase will move the head continuously whereas standard will pause when changing layer heights thus causing a seam.
If the tower below prints smoothly, then it raises the question of a corrupt calibration model. Not likely I agree but that’s probably another area to look at.
https://www.printables.com/model/393668-simple-z-bandingwobble-test-tower
Also, if you haven’t already seen this video, it might give some other ideas.
I’ve tried the VFA test that is built-in Orca Slicer, made some DIY ones, downloaded one, made one that expresses specifically 45 degree and 90 degree movements, and my newest one is a rectangle that guarantees the test will include movements both counterclockwise and clockwise. I’ve also used one of my single-walled models for a plant pot since it means I’m not wasting material one every test.
Some of these I ran on 3 separate machines and had my friend run on his machine, so 4 machines all ran one of my models to test and all produced somewhat different results. I ran my original model that had an unacceptable VFA-afflicted side on it as a spiral to check and was the sane. Unfortunately, the slicer recognizes height range modifiers are distinct objects, they are incompatible with spiralized so I switched it in favor of non-spiralized since at a given speed the results were the same.
Bi-directional VFA test:
I watched the video, and there’s definitely some meat here if it turns out to be the change in steppers from the Kickstarter Bambu X1C to the later models that is the culprit. Worth noting in the video that the prints of the parts he’s using to install the thrust bearing actually demonstrate the exact VFA I am fighting!
Check this out:
On both those faces you can see vertical lines all spaced exactly the same distance apart. That’s the 2mm VFA which seems to result from the 2mm spacing on the GT2 Belt.
What I was referring to about the hz is, depending upon how fast or slow the belt is moving, it would strike the idler and create the marks, and then they would expand or retract. Thus the moments of impact would resonate through the carriage, toolhead, nozzle, wobble the filament.
The other thought I had was depending upon how the bearing works in the carriage. If it’s a slick flat surface that slides on the rail, and the rail- another slick flat surface, it could entirely be possible it’s just flat surface against flat surface vibrations. Like when you have a tube with some similarly sized cylinder in it, it doesn’t just slide right out, it kinda bounces back and forth off the walls as it almost gets stuck. Of course depending upon how the belts are dragging the carriage back and forth, or against the back or front, of the rail, would dictate the oscillations waveform. (distance between lines on the print)
But I have no idea how they constructed the bearings in the carriage.
@Slarm On both those faces you can see vertical lines all spaced exactly the same distance apart. That’s the 2mm VFA which seems to result from the 2mm spacing on the GT2 Belt.
That was my thought exactly. It looks like the belt.
I think we’re on the same page here then!
They’re 12mm diameter aluminum, 7.5mm interior flange to flange and 9.5mm exterior with bearings in the bore riding on a 5mm shaft. The shaft is gently press-fit into the plastic housing. I am assuming all the idlers are the same based on the tensioner idlers, and the way they look to be assembled and size would support that.
I finally tried something that actually had a significant effect…
I wrapped electrical tape onto the idlers with the teeth riding the smooth face.
I couldn’t believe it had worked so I unwrapped it and tested again, then rewrapped it two more times. Consistently it improves the surface and seems to do so a little more when it is loosely wrapped.
There is some increased friction as it passes over the area where the tape overlaps itself an inconsistent number of times, but every test was superior despite that. On the first test I believe slight ridges where the tape wrinkled imparted a cyclical pattern to the print so there are a few lines that stand out where the standard 2mm have reduced.
Images show everything standard vs the first time I wrapped it with tape.
Images with the notches on the corner are the exterior shell printed counterclockwise.
No notches are the interior shell printed clockwise.
The faces with an R are the rear side and the sides facing the Y axis are not pictured but seem to be slightly better though it’s harder to say since they were not too bad to start with.
I think what this points to (despite people saying they tried it and saw no improvement) is that the teeth on the smooth pulley is indeed problematic. It is still possible that it’s the teeth rubbing the flange or just the teeth on the smooth surface in general, and this still doesn’t explain why some printers are really bad and others are fine unless some of the early bearings could be performing really poorly.
The three permanent solutions I have in mind:
• Add a soft rubber sleeve to the existing pulleys. Maybe a stretchy silicone tube.
• Make a firm but somewhat flexible toothed pulley (like polyurethane or one of the FormLabs flexible sorts of resin) since I can’t find any such thing on the market.
• Replace the existing pulley with a toothed version in plastic or standard aluminum.
What I am not looking forward to is that while the left pulley appears to have a bearing shaft that drops out the bottom, the right side shaft needs to come up and there’s part of the printer frame in the way. Unless I’m missing something the entire X axis has to come out to change these bearings and I am not looking forward to that.
If anyone has done the X-axis swap or pulled those idlers and has pointers or any shortcuts, please share!
When I first got my x1c, i overtightened the belts because of a suggestion i read here. Caused the same marks. loosened belt tensioner screws and allowed the tensioner spring to do the work. Marks went away.
The pulleys cannot be removed without impacting the carbon rods. The carbon rods overlap the exit point of the press fit pin so you’ll need to drill them to open up enough space for the pins to come out. I’ve got a spare assembly and just checked. Also pushing them out from the bottom side will be a challenge as there is no through hole to insert a tool.
So if you want to remove them, maybe a drill bit to clear the carbon rods overlap from the top and a 90 degree Allen key can do it.
For me the 2mm VFAs are only visible on X axis movements. I presume that there is some pressure or “stiction” on the X axis because the belt forces try to twist it - they pull from the top and bottom side of the assembly. If you don’t apply any twisting force to the assembly it’s butter smooth. If you slightly twist it, it feels a lot “stiffer” hence I presume that increase in force causes these artefacts. This friction could be eliminated by using metal rods but then you’ll trade off speed.
I’ve also seen the same artefacts on the k1 and k1 max in several YouTube videos and Reddit posts , which uses a metal rod there. So maybe that is not even it, don’t know…
Personally I’ve quit chasing this. I try to orient the long way of a model parallel to the Y axis. For most models the VFAs don’t really show. If the occasional model is more susceptible to it (eg petg) I simply reduce external wall speed to 40mm/sec and they are gone and replaced with minor stepper noise.
This is really useful, thank you. I have seen the overlap of the rod and the pin from above but I did not even think about it being there on the other side as well. At this point I am not afraid to drill out the carbon rod if I have to - the issue is really bad on mine and affects X severely, and diagonal travels moderately.
I believe it affects the X-axis primarily because only travels along X require the tooth side of the belt to roll across the smooth idler. There is still some of this noise in Y travels but it could be the varying inflexibility of belt even smooth to smooth or the teeth of the belt lightly snagging the drive pulleys.
Unfortunately for me this isn’t an option. Two of the large models I need to be printing are squares and circle cross sections and the print that precipitated in this hunt looked 95% good on 3 sides and looked horrifying on the 4th. I’m talking tangible ripples all across the entire vertical planar face, deep enough to measure with a cheap caliper.
After what you’ve described in the X assembly tape might become a more permanent solution. Also, how did you happen to come by an extra X assembly? They aren’t exactly a cheap part so I imagine there was a reason for it.
I bought it because of the VFAs- in case my bearings were shot. But haven’t come around to installing it yet…
Any chance you can post a picture of the tape “mod”? I’ll give it a try too.
Maybe it has something to do with the belt tension that it occurs more on some printers and less on others. You can and should tighten the belts, if I understand correctly (Bambu Wiki).
Did you ever find a different means to reduce the VFAs (other than conscientious orientation of parts)? I ran a VFA test with the tape at a 45 degree angle which elongates the lines but they’re still present. In that configuration though they are quite a lot more subtle but they do appear on all sides of the print instead of on X axis.
Certainly though it does not look like much! Cut a length of tape and slit it lengthwise. With the gantry on the opposite side stick it to the pulley and keeping tension on the tape slide the toolhead sideways to roll the tape in. If it’s slack it will create ridges where the teeth settle, but seems prone to creating wrinkles. Tighter is more consistent but less damping. I haven’t been able to do a wrap without overlapping the tape unevenly since it is rather elastic.
Unfortunately the belt tension has little effect. As documented earlier in this thread all the belt tension does is change the speed band where the 2mm VFA appears but has no effect on amplitude. The automatic spring tensioners apply around 3lb force and I’ve manually set it to around 2lb and 3.5lb to see what effect it might have. It did not improve the situation alone, but might be useful in combination with other techniques.
[quote=“Slarm, post:39, topic:30883”] Unfortunately the belt tension has little effect. As documented earlier in this thread all the belt tension does is change the speed band where the 2mm VFA appears but has no effect on amplitude [/quote].
So the belt tension has an effect. That’s not what this was about. As already documented in this thread, these artefacts depend on the design of the machine / system.
And again:
My two machines, different filament, different slicer software, but same problem with the artefacts. The only thing that is the same on both my printers is the basic construction: stepper motors, toothed belts, deflection rollers (apparently not toothed but smooth on the X1 etc.). In addition, the functional principle of steppers does not allow for a uniform movement. They can only achieve a uniform movement through speed. This is different from a DC motor with a gearbox, which turns without jerking.
If you want to eliminate the artifacts, you have to rebuild the machine or choose settings where the artifacts occur less.
