RE: VFA Artifacts on X+Y Straight Edges
I had about decided I either needed to just live with the visible print quality issues or sell it and build something else.
https://www.reddit.com/r/prusa3d/comments/1m995lw/psa_vfa_update_3 from Josef mentions a few additional mitigation items. Still not sure what "non belt originated VFAs" means related to what some of us are seeing.
His summary:
To summarize:
What is here now
- New belt settings
- New tuner with waterfall graph
What is coming
- New belt tuning procedure without a smartphone or specialized hardware
- Slicer eliminating non belt originated VFAs
What will be done to get ahead of the CoreXY standard
- Better cooling logic to stay out of the resonant speeds.
RE: VFA Artifacts on X+Y Straight Edges
Coming back to the accelerometer brainstorming: why not connect the phone via USB-C to the motherboard to get a "real time" reading of the sensors?
Sure, not feasible for input shaking, but the phase stepping calibration excites basically the whole machine, so it should work.
Write a little toolkit into the Prusa App and ask for accessing the accelerometers.
RE: VFA Artifacts on X+Y Straight Edges
Joseph's reddit post seems like a deflection from the main issue, with the bulk of the text referring to an unrelated input shaping optimization. It sounds like they are only offering optimal belt tensioning as a belt ripple VFA mitigation measure.
RE:
Josef has posted a third VFA update post on Reddit:
https://www.reddit.com/r/prusa3d/s/JrXCfVlfdk
Seems like they have some promising things in the works. Very glad that those changes/fixes will apply to the XL also.
Well this all sounds rather promising, although I wont be getting too excited until we see some positive real-world results.
I guess now everyone and their dog will be cranking up the tension on those idlers I hope Prusa have plenty of those PCCF replacements printed and ready to ship out 🤣
RE: VFA Artifacts on X+Y Straight Edges
This was noted as addressing VFA in some form. I counted three different things coming in the (hopefully near) future which are directly related to reducing VFA.
Joseph's reddit post seems like a deflection from the main issue, with the bulk of the text referring to an unrelated input shaping optimization. It sounds like they are only offering optimal belt tensioning as a belt ripple VFA mitigation measure.
RE: VFA Artifacts on X+Y Straight Edges
Yes, to be fair they also seem to be offering some slicer based fix to avoid the worst belt ripple speeds.
They don't seem to be addressing the main issue though, that being the pulley/tooth geometry which is the root cause of the problem.
This was noted as addressing VFA in some form. I counted three different things coming in the (hopefully near) future which are directly related to reducing VFA.
Joseph's reddit post seems like a deflection from the main issue, with the bulk of the text referring to an unrelated input shaping optimization. It sounds like they are only offering optimal belt tensioning as a belt ripple VFA mitigation measure.
RE: VFA Artifacts on X+Y Straight Edges
Yes, to be fair they also seem to be offering some slicer based fix to avoid the worst belt ripple speeds.
They don't seem to be addressing the main issue though, that being the pulley/tooth geometry which is the root cause of the problem.
This was noted as addressing VFA in some form. I counted three different things coming in the (hopefully near) future which are directly related to reducing VFA.
Joseph's reddit post seems like a deflection from the main issue, with the bulk of the text referring to an unrelated input shaping optimization. It sounds like they are only offering optimal belt tensioning as a belt ripple VFA mitigation measure.
Your assuming that's the root cause I would say it's not as changing pulleys had seemed to help some, but not others. I'd also say that even 1.5 belts is not a complete fix. It showed improvement but did not eliminate them.
I think the things Joeseph are saying make sense. We'll wait and see, but I have followed this issue from the beginning, and there doesn't seem to be 1 fix, and different things have improved it for different people. Based on that I don't think anyone can definitely say what the issue is.
The XL does have them as well, but they are worse on the Core One from what I've seen.
RE: VFA Artifacts on X+Y Straight Edges
It's been a problem for as long as 3D printers have existed to varying degrees, and is known without doubt to be caused by the belt tooth/pulley mesh and toothed belt on flat/toothed idler. Many variables are at play as to how bad it is on individual machines.
Yes, to be fair they also seem to be offering some slicer based fix to avoid the worst belt ripple speeds.
They don't seem to be addressing the main issue though, that being the pulley/tooth geometry which is the root cause of the problem.
This was noted as addressing VFA in some form. I counted three different things coming in the (hopefully near) future which are directly related to reducing VFA.
Joseph's reddit post seems like a deflection from the main issue, with the bulk of the text referring to an unrelated input shaping optimization. It sounds like they are only offering optimal belt tensioning as a belt ripple VFA mitigation measure.
Your assuming that's the root cause I would say it's not as changing pulleys had seemed to help some, but not others. I'd also say that even 1.5 belts is not a complete fix. It showed improvement but did not eliminate them.
I think the things Joeseph are saying make sense. We'll wait and see, but I have followed this issue from the beginning, and there doesn't seem to be 1 fix, and different things have improved it for different people. Based on that I don't think anyone can definitely say what the issue is.
The XL does have them as well, but they are worse on the Core One from what I've seen.
RE: VFA Artifacts on X+Y Straight Edges
I put the question to ChatGPT to see if it had any ideas. Some are interesting!...
🔄 1. Non-Circular (Cam Profiled) Drive Pulley
Concept: Use a deliberately non-circular pulley profile that counteracts the sinusoidal belt-induced positional error by introducing a complementary corrective motion as it rotates.
How it works:
Model the positional error caused by the GT2 belt (typically a low-amplitude sinusoid at 2mm period).
Design a lobed cam pulley that introduces a tiny corrective motion at the same frequency but opposite phase.
Use this custom pulley as the drive pulley on the motor shaft.
Feasibility:
Would require very precise design and possibly CNC or SLA printing.
Theoretical basis comes from control systems and mechanical compensators.
Could be combined with software compensation (see #4).
🔊 2. Ultrasonic Belt Dither
Concept: Use a high-frequency piezoelectric dither mechanism to shake the carriage in a direction orthogonal to ripple (Z or lateral), decohering the ripple pattern without affecting average position.
How it works:
Belt ripple arises due to regular, small amplitude positional oscillation.
If a controlled high-frequency dither (e.g. 20–40 kHz, sub-micron amplitude) is applied during motion, the artifact gets "blurred" out.
Similar to how dithering is used in ADCs or inkjet printing to improve effective resolution.
Pros:
Doesn’t interfere with normal motion.
Could be retrofitted onto the carriage using a piezo actuator and a spring preload.
Challenge:
Needs precise control to avoid degrading print quality.
Needs mechanical isolation from other axes.
⚙️ 3. Friction Roller Preload on Belt Teeth
Concept: Mount a small roller on a spring that rides the teeth of the belt just before the drive pulley, applying constant pressure to "seat" each tooth uniformly, reducing micro-position errors due to tooth engagement inconsistencies.
Why it could help:
Ripple may partly result from micro-jumps or inconsistencies as the belt teeth engage the pulley under tension.
Preloading the belt like a record player stylus damps this engagement.
Could be especially useful on the tight side of the belt near the motor.
🧠 4. Software Phase Compensation (Advanced Linearization)
Concept: Use calibrated software compensation that adjusts stepper commands in real-time using a stored correction curve based on belt ripple characteristics.
How it works:
Print a ripple test pattern.
Analyze ripple amplitude and phase using a high-res scan or camera.
Generate a lookup table or correction waveform (e.g. sinusoidal position tweak every 2mm).
Inject it into the motion planner (similar to pressure advance).
RE: VFA Artifacts on X+Y Straight Edges
Would there be value in getting a high-speed video (i.e. to slow down the motion) of the print head when VFA is occurring? Would uneven motion of the nozzle be visible?
I would very much like to see this, I think I calculated the resonances to be around 40-50Hz. I would really like to see high speed macro video of the motor shafts to see if they are oscillating as well.
RE: VFA Artifacts on X+Y Straight Edges
I would very much like to see this, I think I calculated the resonances to be around 40-50Hz. I would really like to see high speed macro video of the motor shafts to see if they are oscillating as well.
Would it not be a variable frequency over a wider range, given the fairly wide range of speeds at which it occurs?
RE: VFA Artifacts on X+Y Straight Edges
With the new belt tuning I was able to eliminate most of the VFA issues. Make sure that you verify proper torque of all your assembled components, verify proper belt alignment and then retune. It was a pretty impressive difference. I have only done it on smaller prints with PETG but its noticeable.
RE: VFA Artifacts on X+Y Straight Edges
With the new belt tuning I was able to eliminate most of the VFA issues. Make sure that you verify proper torque of all your assembled components, verify proper belt alignment and then retune. It was a pretty impressive difference. I have only done it on smaller prints with PETG but its noticeable.
Would be very interested in seeing the results of a proper VFA test print from your machine.
RE: VFA Artifacts on X+Y Straight Edges
I have finally started my conversion. Just did the Nextruder. I'm going to try and finish the Core One without any of the panels, so that I still easily reach around the XY system. Squaring the gantry worked flawlessly, I needed just a few mm of correction. Belt tuning worked very well via the new method, gantry is still perfect at the correct belt frequencies.
Right now I am using all stock parts, to see how my unit behaves. I do have Mellow 3D idlers and pulleys here, that I can install to check the results.
RE: VFA Artifacts on X+Y Straight Edges
I have finally started my conversion. Just did the Nextruder. I'm going to try and finish the Core One without any of the panels, so that I still easily reach around the XY system. Squaring the gantry worked flawlessly, I needed just a few mm of correction. Belt tuning worked very well via the new method, gantry is still perfect at the correct belt frequencies.
Right now I am using all stock parts, to see how my unit behaves. I do have Mellow 3D idlers and pulleys here, that I can install to check the results.
I guess it's too late for the initial installation, but I hope you remembered to lubricate the tension screws. If not, it's worth doing the next time you need to adjust the belts for whatever reason.
RE: VFA Artifacts on X+Y Straight Edges
Actually I did. I'm not sure it's enough, I just used some regular machine oil. I did notice there is quite a bit offset between the screw and the pulley, causing a significant moment on a rather small thread length. So far haven't had problems but I'm not too sure it will hold up long term.
RE: VFA Artifacts on X+Y Straight Edges
Actually I did. I'm not sure it's enough, I just used some regular machine oil. I did notice there is quite a bit offset between the screw and the pulley, causing a significant moment on a rather small thread length. So far haven't had problems but I'm not too sure it will hold up long term.
What do you mean by offset? I'd have to verify specifically, but by memory mine seemed to line up well.
-J
RE: VFA Artifacts on X+Y Straight Edges
The screw for tensioning the idler is not in line with the centre of the idler pulley. There is 11/12 mm offset, resulting in a moment. Although I will say, the geometry and direction of the moment is such that part of it will be taken up by the wall that guides the tensioner itself.
RE: VFA Artifacts on X+Y Straight Edges
The actual rotation is limited by the hole that the tension screw passes through in the tensioner mounting block. But nevertheless, there will still be a lateral force between the tensioner square nut and the screw, presumably adding to the friction between the threads on one side (and reducing the friction on the other?). Everything seems to stack up against that poor little nut - small number of threads, stainless steel on stainless steel, higher tension after the VFA update, lateral force from the offset...
But here I go again - taking the thread off topic!
RE: VFA Artifacts on X+Y Straight Edges
Oil may result in loosening over time and may not prevent galling... Until someone sees it this is just an experienced guess though.
Got my pccf, waiting on inserts... Paranoid about touching things till I have replacements now.