RE: VFA Artifacts on X+Y Straight Edges
That's the post @yblaser linked to yesterday and which we briefly discussed here. Directly above in this thread.
RE: VFA Artifacts on X+Y Straight Edges
Sorry, only have my phone… Pain in the ass to read all this stuff. Don’t get old
RE: VFA Artifacts on X+Y Straight Edges
[q]Now, “improper belt meshing”. As the belt teeth mesh into the pulley, there is a small transition period when a new tooth engages from pushing to pulling, creating a minuscule variance in speed. This applies to every belt under the sun with straight parallel teeth. A good read on this phenomenon is a paper from Masanori Kagotani (Influence of Idler on Transmission Error in Synchronous Belt Drives (Under Transmission Force)). We identified this part as the most likely culprit on the severely affected printers. A side note: This happens only on motor pulleys. As there is no force transfer on idlers, it makes no difference whether the pulley is smooth or teethed[/q]
This would lead me to think out of tolerance pullies... Of course, I'm not a 3D printer engineer.
I am wondering if I can get a bunch of prusa pullies and send them out to my favorite machine shop and have them test runout.
RE: VFA Artifacts on X+Y Straight Edges
I am wondering if I can get a bunch of prusa pullies and send them out to my favorite machine shop and have them test runout.
The question of runout as a potential root cause came up here recently. It seems an unlikely candidate, since you would expect to see a variation in the VFA pattern on the length scale associated with the revolution of the pulleys, i.e. 32 mm.
But one actually sees only a pattern with a 2 mm period, i.e. the pitch of the belt teeth. So it seems more likely to me that we are looking at a mismatch between the belt's tooth profile (or pitch) and the profile of the pulleys.
RE: VFA Artifacts on X+Y Straight Edges
Which could be a problem with the tension of the belts, they evidently stretch as mine arrived with the belts at 107 Hz. I reduced this to 90 Hz then after a month and a bit of use found this had dropped to 77 Hz, now back at 86Hz so which is correct as each different tension will alter the pitch off the teeth ever so slightly.
Normal people believe that if it is not broke, do not fix it. Engineers believe that if it is not broke, it does not have enough features yet.
RE: VFA Artifacts on X+Y Straight Edges
The problem comes from the tooth of the belt not entering its groove on the pulley in a clean way. In an ideal world that meshing would simulate a flat pulley to the rest of the belt in the system and the printhead attached to it. But in reality the action of meshing causes a small bump in the otherwise smooth motion as the tooth shape and height does not perfectly match the corresponding groove perfectly. The end result in the small repeating disturbance is the 2mm ripple pattern on the print surface as the printhead oscillates slightly within the prescribed path.
[q]Now, “improper belt meshing”. As the belt teeth mesh into the pulley, there is a small transition period when a new tooth engages from pushing to pulling, creating a minuscule variance in speed. This applies to every belt under the sun with straight parallel teeth. A good read on this phenomenon is a paper from Masanori Kagotani (Influence of Idler on Transmission Error in Synchronous Belt Drives (Under Transmission Force)). We identified this part as the most likely culprit on the severely affected printers. A side note: This happens only on motor pulleys. As there is no force transfer on idlers, it makes no difference whether the pulley is smooth or teethed[/q]
This would lead me to think out of tolerance pullies... Of course, I'm not a 3D printer engineer.
I am wondering if I can get a bunch of prusa pullies and send them out to my favorite machine shop and have them test runout.
RE: VFA Artifacts on X+Y Straight Edges
I don't have a Core One but an MK4S which is displaying the exact same issue (the XL was mentioned here a few times too). Interestingly, the MK4 didn't. Could be that MK4S and Core one use the same motor pulley type/batch but that is just a wild theory.
I disregarded the statement that belt tension is the cause as a very vague guess at solving the issue but I want to share that I was able to significantly decrease the visible VFA by modifying the tension. The Prusa belt tuner app seems to be far off of any ideal tension, at least on the MK4S's Y axis. I tried 70, 80, 90 & 100 hz with a suited test model and saw almost no differences. In the last test, I used only my hearing and made it sound like a "bass guitar" and interestingly, this was the trick. I also used a spectrum analyzer (although just a phone app) and saw that it's very hard to determine which of the frequency peaks is the belt. I assume the Prusa app filters certain frequency ranges and assumes that the belt will be within a certain range. Anyway, I found using any kind of sound-related approach very hard to reproduce.
I also switch the Motor Pulley. I ordered a no-name pulley and this also decreased the visible VFA artifacts. I noticed that the teeth of this pulley are not sharp but rather flat. Maybe the flat surface provides clearance during the belt movement which results in less torque peaks but that is also just a wild guess.
RE:
I followed the aliexpress link a few pages back to the 16T mellow pulleys. It was recommended previously to order a couple of pairs, but they are only allowing the purchase of one pair at a time. I ordered a single pair to have on hand once I find the time to build my conversion kit, and like others here I will swap them out if my printer exhibits the signs of strong belt ripple with the stock ones.
I'm really intrigued by the idea of going to a GT1.5 belt, but I'll only go down that path if a clean solution to the scaling issue is found or provided by Prusa.
As of now, I'm expecting to find satisfactory results with either the stock or mellow pulleys.
-J
RE: VFA Artifacts on X+Y Straight Edges
Firmware/stepper driver configs can really only affect low speed artifacts coming from the motors themselves independent of the pulley/tooth geometry, in the 0 to ~50mm/s range. The 2mm belt ripple being discussed here, which is proven to be coming from the pulley/belt teeth geometry occurs on all corexy printers in the ~70-160mm/s range.
Printing with SPEED profiles (200mm/s perimeters) is the way. Or even faster. Core One prints up to 350mm/s without tuning. There is a plenty of room to avoid VFAs with speed. Like other CORE XY printers does.
RE: VFA Artifacts on X+Y Straight Edges
We know that if you print at MACH4 there won't be any VFA, but because of geometries, limits or even to get strong parts, sometimes you have to print slowly, or even to get a nice consistent aspect on the printed part.
Printing at high speed should not be a solution to get a nice looking part.
RE: VFA Artifacts on X+Y Straight Edges
This is just it, Prusaslicer automatically slows down layers if they're below a certain duration, and if you disable that you're basically sacrificing dimensional accuracy.
I mean thats fine if you're just printing those daft articulated dragon things and other silly toys, and if that's all I'm printing I'd buy a Bambu...I bought a Prusa because dimensional accuracy is of high value to me and what I use my printer for.
RE:
The "just print fast" recommendation seems to imply that you can either have stable and dimensionally accurate parts, or parts with a non-rippled surface. Many applications will indeed require only one of the two, but it's certainly not a general "solution" for the VFA problem.
RE: VFA Artifacts on X+Y Straight Edges
Especially, when looking at the 45° walls, produced by only my left motor, which are almost VFA-free at all speeds, especially when printing slowly.
I've found that VFA (& noise) vary, depending on the direction the motor is moving and the position of the extruder. My left motor varies from "flawless" to very mild VFAs. My right motor varies from awful to "only a bit worse, than the worst walls of the left motor".
I guess CoreXY printers with 2.5 belts and stepper motors will always have some kind of ripples for some speeds and positions, but it's crazy how one motor + pulley combination can produce absolutely awesome walls, while the other is awful.
I'm gonna contact the support and will hopefully swap my current two motors this week to report back on motors vs pulleys vs belts vs position.
RE: VFA Artifacts on X+Y Straight Edges
Printing faster doesn't apply to a printer that ships with:
- official "best looking" profile using 45mm/s
- official filament profiles for silk filaments with severe flow limits
- official filament profiles for Prusa manufactured silk filaments (Prusament PLA Blend)
They sell a printer with official profiles printing slow, they manufacture silk filament in-house, they officially state these filaments are compatible with Core One.
Therefore it is their responsibility to make these combinations look better than a beat-up Ender 3.
We're literally printing exactly as Prusa wants us to, using their official profiles and their own official filament. Literally can't get more "official" than that. If these combinations are producing bad prints, then the printer is at fault and suggesting faster printing is downright insulting.
RE: VFA Artifacts on X+Y Straight Edges
I guess CoreXY printers with 2.5 belts and stepper motors will always have some kind of ripples for some speeds and positions, but it's crazy how one motor + pulley combination can produce absolutely awesome walls, while the other is awful.
Now here's the thing - take your best pulley and put it on the left motor. Print an angle that involves movement only from the left motor. Then move this same pulley to the right motor and print an angle that involves movement only from the right motor. You will see the right motor produces more VFA than the left even with the same pulley.
Why?
Take a look at the pulley placement on these motors. The left motor has its pulley relatively close to the motor, the belt runs roughly in the center of the shaft. Even force, very nice.
The right motor has its pulley so far away from the motor that it no longer sits fully around the shaft, it extends. What is even worse, it's rotated so the belt sits even further away. The belt runs outside of the shaft length, exerting a strong force over it - like a lever.
I suspect this is why the same pulley can produce flawless walls on the left motor but ugly walls on the right motor.
They wanted to save money by reusing the MK4 motors, which clearly have way too short shafts for the way the belts run on a Core One. Now WE are paying for their penny-pinching.
RE: VFA Artifacts on X+Y Straight Edges
The right motor has its pulley so far away from the motor that it no longer sits fully around the shaft, it extends. What is even worse, it's rotated so the belt sits even further away. The belt runs outside of the shaft length, exerting a strong force over it - like a lever.
I suspect this is why the same pulley can produce flawless walls on the left motor but ugly walls on the right motor.
Is there any way to modify the motor mount to make it stay closer to the belt?
RE: VFA Artifacts on X+Y Straight Edges
I guess CoreXY printers with 2.5 belts and stepper motors will always have some kind of ripples for some speeds and positions, but it's crazy how one motor + pulley combination can produce absolutely awesome walls, while the other is awful.
Now here's the thing - take your best pulley and put it on the left motor. Print an angle that involves movement only from the left motor. Then move this same pulley to the right motor and print an angle that involves movement only from the right motor. You will see the right motor produces more VFA than the left even with the same pulley.
Why?
Take a look at the pulley placement on these motors. The left motor has its pulley relatively close to the motor, the belt runs roughly in the center of the shaft. Even force, very nice.
The right motor has its pulley so far away from the motor that it no longer sits fully around the shaft, it extends. What is even worse, it's rotated so the belt sits even further away. The belt runs outside of the shaft length, exerting a strong force over it - like a lever.
I suspect this is why the same pulley can produce flawless walls on the left motor but ugly walls on the right motor.
They wanted to save money by reusing the MK4 motors, which clearly have way too short shafts for the way the belts run on a Core One. Now WE are paying for their penny-pinching.
You obviously didn't read Jo Prusas 'Deep Dive' last week. He's gonna teach us how to fix it all by tensioning the belts a special way 😂
RE: VFA Artifacts on X+Y Straight Edges
Is there any way to modify the motor mount to make it stay closer to the belt?
I'm not sure. The surrounding idlers are already barely fitting in that plastic part between the motor and the steel gantry, it can't really get any smaller.
And the motor cannot be mounted from the other side because of the steel gantry. It has no hole for the shaft to go through.
RE: VFA Artifacts on X+Y Straight Edges
Now here's the thing - take your best pulley[...]
I didn't think about the belt positioning on the motor, nice additional thoughts!
I really need to swap motors + pulleys and repeat my tests..
Have a read here, about what I already did (basically what you say):
https://forum.prusa3d.com/forum/prusa-core-one-general-discussion-announcements-and-releases/prusa-core-one-not-usable-inhouse-sorry-but-the-sound-was-to-intensiv/paged/7/#post-754897
Swapping the motors (and probably swap the pulleys for a second test suite), should give us a lot more insight on motors, pulleys, belt positioning.
I'll just need to find the time to do it...
RE: VFA Artifacts on X+Y Straight Edges
The right motor has its pulley so far away from the motor that it no longer sits fully around the shaft, it extends. What is even worse, it's rotated so the belt sits even further away. The belt runs outside of the shaft length, exerting a strong force over it - like a lever.
We have discussed it a couple of times -- in my opinion, run-out (whether from the pulleys or the motor axles) is an unlikely candidate for the root cause for belt ripple/VFAs.
If it were the cause, you would expect to see some pattern in the prints which repeats with the period of one motor revolution, right? Ripples going from strong to less pronounced to strong over a 32 mm period (the circumference of the pulleys) or half of that. I don't think such a pattern has ever been reported. It's just the pitch of the belt teeth which shows up in the prints.
To me this suggests a mismatch of the tooth profile on the belts vs. the pulleys. Possibly exacerbated by vibrations in the belt which really "exercise" the play between belt and pulley teeth. These vibrations might be excited by stepping irregularities of the motors, so the motors do potentially play a role in this.