Persistent 2mm VFAs on X-axis

[Slarm]

Because I have been printing translucent objects whose cosmetic appearance I finally realized just how bad my VFAs are and I’ve been pulling my hair out trying to solve the problem. I have read numerous threads on it and tried pretty much every shred of advice I can find, but very little helps.

Bambu support has been rather unhelpful on it. Their typical response is “just print faster” which is not always an acceptable solution. In my case, I require the material to stay as transparent as possible and printing faster causes it to be matte. Likewise, printing the speeds required to reduce the VFA requires thinner layers which requires more layer changes for supports which massively increases the print time. They also refuse to do anything about it and have actually ghosted me in the support conversation because my printer is out of warranty, even though I can demonstrate looking at old prints that the issue has always been present.

In my case I see significant VFAs from 90-160mm/s with 120-130mm/s being the peak.

Things I’ve tried:
• Replace drive pulley with new, taller one. Later model Bambu XY motors come with tall pulleys, where the old ones are short. The short one caused the belt to grind against the flange. It did not improve the issue.
• Check axes for square using a set of tools provided on Printables. My printer is within 0.5mm of square which I think could just be the resolution of the test (I have designed some better clips to try to improve this.)
• Tune the belts to ensure equal tension. The official tensioning method produces a resonant frequency of around 150Hz across the ~250mm span. This did not improve anything.
• Adjust the skew of the pulleys to ensure the belt did not rub on the drive pulley or idler pulley flanges. No improvement.
• Test printer on wood workbench top, test on polypropylene felt pads, test on a concrete block on felt pads. No change in any case.
• Do carbon rod cleaning multiple times in case I hadn’t done it enough during my ownership of the printer.

The only thing that has had a noteworthy change was dropping the tension from the stock to around 100Hz and it was somewhat improved. Other CoreXY printers use 100Hz over a 150mm span as a standard which comes out to around 2lb of tension, but the default Bambu is around 7lb of tension. I am thinking I might try to tune it to something like 3lb, but it will require a solution to ensure a consistent belt length while tuning without hampering the vibration.

The belt still seems to be rubbing on various idler pulleys throughout the system and I can’t see any way to adjust for that.

I am in a bit of an uncommon position where I have access to several Kickstarter-era units as well as several later units and I’ve been able to run my VFA test on several of them. My unit is significantly worse than another Kickstarter unit, and both these are inferior to one that was bought more recently.

I am really disappointed that Bambu has not stepped up to help me resolve this issue which was present when shipped, and while I don’t expect free hardware support, I had hoped they would at least be willing to help troubleshoot the problem when it would benefit other users (another thread here with 400+ posts says it is a really significant problem.)

Other than trying to swap the entire carbon assembly for a new replacement, or further reduce belt tension, I’m out of ideas here. I would love to get some help from anyone with the same issue, or who has resolved these issues on a Bambu or another CoreXY printer.

Send help!

[Slarm]

Old print showing VFA:

Comparing standard tension with reduced tension:

Another example of the VFA with all standard configuration/settings:

Bambu support has seen these images and just insists the solution is to print faster even though it requires compromising strength, surface finish, time etc, and despite the fact that most printers are performing better than this.

[Olias]

It certainly seems like you’ve done your homework and made a herculean effort to minimize these artifacts. You didn’t mention it but did you try something low tech to rule out vibrational harmonic feedback from the printer surface? Maybe try placing the printer on a very rigid surface. This would eliminating any chance of harmonic vibrations being introduced from, let’s say the table surface.

At any rate, it will unlikely change anything but it would rule it out. I would do a simple test. Remove the AMS if it’s attached. Place the printer on the floor. Run the same test and see if it changes the result. Also, if you have the antivibration feet option, try it with and without and see if that also changes matters.

[zviratko]

It looks like he’s done his homework in this area.

What about lubricating the belts a bit? The top and bottom where those could be rubbing. It improved my prints a bit on and old Ultimaker (but my goal here was to make it quiter, also it didn’t run anywhere near the speed of our machines). Don’t get anything in the groves (though it’s bound to soak in somewhat), and be very gentle and use something sticky. I used some random satchet of O-ring grease I had for a coffee machine that made the surface just a little bit more slippery.

Since some of these VFAs come and go randomly (at least on my machine), I think at least part of the problem is some sort of software rounding/compounding error. I bet it’s bed leveling + frequency compensation + some sort of resonance from the motion system and ball bearings (I can literally feel those when moving the rods by hand. Hmm, but I just noticed you are talking about X - what about just changing the carbon rods? Somebody else here posted a picture of their faulty carbon rods (unfinished machining) and those had similiar pattern, so I wonder if the internal structure and the way they are made have something to do with it, maybe yours have this pattern hiding somewhere? Slight uneveness in the surface finish? Try a sharp light/laser from a side. I saw some imperfections on mine, but since the bushings are rather long I doubt something random could affect the print quality, but if there’s any sort of pattern to it then that’s another story…

[KanneKaffe]

This is obviously a problem with the printing system hardware.
I last retensioned the belts on my CR-20, installed a new extruder, and replaced a torn cable from the X-Stepper. With that, this printer ran “flawlessly.”
Since I have only been dealing with the X1C for the last few weeks, I took a look at the object last printed with the CR-20 and compared the surfaces. In doing so, I noticed exactly these artefacts. The X1C and CR-20 printed the outer walls at comparably low speeds, up to 50mm/s. The artefacts produced by both printers are nearly identical.

With faster PLA prints these artifacts do not occur or are not detectable or are very very difficult to detect. But these are then speeds 150 to 250mm/s about. With PETG I see them now especially because I print that at low speeds to get a stable print.

As I said, I think it’s system related.
I watched a video on YT where someone was supposedly able to minimize problem somewhat (cut visibility in about half) with higher quality steppers at other printer.

[Slarm]

I haven’t tried it and it is worth a shot. The printer is right now on a concrete slab floor of my garage, recalibrated and running my VFA benchmark. It’s around 5mm into the print and I can already see that the VFA are still majorly present. To further try to reduce resonance since the right panel is so sensitive to vibrating, I leaned a gallon of water against it to dampen the vibration as well. No luck with this one.

I never bought the anti-vibration feet, but the polypropylene felt pads I think fill this role adequately. We use them for the expensive laser cutters at work too. With the pads underneath it greatly reduced the hum produced when doing the frequency sweep directly on top of the workbench.

[Slarm]

Lubricating the belts is an interesting idea since rubbing on flanges is one thing I’ve read as a possibility but not found any way to eliminate. I think it’s unlikely since two of the newer post-kickstarter batch of Bambu X1C are also rubbing the idler pulleys. My concern is getting oil on the belt - I’m not sure if it will facilitate the degradation of the rubber. I would try this, but only I think once I’m ready to potentially replace the belt or the carbon X assembly.

I did take the risk of lubing the carbon rods since I noticed both the post-kickstarter units I was hassling have more slop in the linear bearings. I am putting some stock in that being a possibility even though the lube had no effect. (I figured it will be rather easy to remove the lube with IPA and I used a non-gumming turbine oil.)

Do you happen to have the link for the post about the unfinished carbon rods post? I would like to take a look. There are definitely some imperfections on my rods - I see a spiral of small dimples on one of them for example. I read somewhere about friction on the X axis potentially causing these issues hence the try with the oil.

[Slarm]

Yeah, this issue is supposed to peak starting from around 50ish mm/s up to around 160mm/s, though on mine it goes all the way to 200mm/s and is extreme around 120-130mm/s. The issue with the 1.8 vs 0.9 motors is a different VFA issue and produces smaller artifacts. This issue is 2mm spacing which matches the teeth, though I noticed on the diagonals it stretches to what I’m assuming is 1.4x the length because the belt pitch is moving across the diagonal of the square or something along those lines.

Something on mine and certain other printers is either receiving extra vibration from the teeth, resonating it more strongly, or failing to account for it in the frequency sweep.

[Slarm]

New efforts:
• Tried on solid concrete slab - no improvement.
• Lubricated carbon rods with turbine oil - no improvement.
• Verified with a new more precise system that the axes are square.

I previously improved it by arbitrarily decreasing the belt tension. This seems like a treatment of a symptom not the cause, so while not ideal I will take what I can get. I am currently trying to standardize the reduction in belt tension. Using the rear spans between the tensioner and the outer pulley, I have calibrated both belts to approximately 111Hz. On Spectroid this looks like it is the second order resonance, so assuming a fundamental frequency of 55.5Hz that calculates to approximately 2.2lb of belt tension. A test print is running now, but it does not look like it is much better though I will wait til its finished to verify.

If the lowered tension is not effective, I will try deliberately mismatching the belt tensions in case the belt resonance is constructive and differing tensions will be able to reduce some of the amplification. Will share results from that when I have them.

[KanneKaffe]

Just a idea. You can try reduce the motor current in GCode. If the artifacts are produced over the steppers, then decimating the current should have an effect.

[Slarm]

I dug a little into that and it sounds like people tried it on their Vorons while trying to diagnose similar issues and it was a bit of a red herring.

Also, and unfortunately, my new pulley slid off the shaft last night while I was testing an increase in tension so I can’t do any more tests in any case til I get that fixed. I do have a permanent metal pulley coming I can epoxy into place like the OEM Bambu one.

I also ordered some dry silicone lube to try on the pulleys as zviratko suggested as the silicone lube should not spread and should not hurt the belts at all. Worth a shot!

[Slarm]

Tried a few more things:
• Tensioned 31cm side span belts to 85Hz instead of default of 75Hz - no improvement.
• Relaxed tension on A motor to tensioner default - no improvement.
• Followed prescribed belt tension adjustment methodology but skipped frequency sweep - no change for better or for worse.

Have also observed that the belts are not rubbing on many pulley flanges. It looks like the tensioner pulley flanges are the only ones really getting rubbed.

Any other mechanical ideas to test on the existing hardware (options to replace carbon axis remain open assuming other ideas end up exhausted.

[KanneKaffe]

I would test speeds. And if at one speed, these artefacts are gone, I would use that speed for the outer wall. Stability comes over the inner walls, if there are enough. The melting temperature will probably also play a role.

I increased the belt tension on the CR-20. That did not bring any improvement there. Either there was no effect at all or a slight deterioration (dimensional accuracy or something else). That’s why I went back on the CR-20 with the belt tension and only tightened it so much that the belt was just tight. So I got the idea there to install tension springs to bring the belts to an equal tension where it works best. At that time I didn’t have an X1. When I got the X1, I saw that I was not alone with this idea. Increased belt tension increases the load on the motors and the pressure in the bearings, but it also stretches the belt. Therefore, the engine could run even more jerky and make the problem worse. And the dimensional accuracy of the pressure could be worse.

I would not handle oil. At high accelerations, I would be concerned that the belt might skip a tooth. But it could also promote an unclean pressure pattern if the timing belt suddenly becomes slippery. Oil always spreads, it never just stays where to apply it.

[Slarm]

Tested at all speeds - including Bambu support’s recommendation to “just print above 200mm/s.” The issue presents from 80mm/s-220mm/s with differing severity. While it diminishes above 160mm/s, it is still fairly bad and comes with side effects such as matte/opaque surfaces, weaker layer fusion, and of course high speeds cannot be used on partial overhangs so the issue will always show up. Printing faster is a crutch and a bad one because it compromises too many other things one might need in a print, including thicker layers for transparency.

Regarding tension, all the values I used are within spec for typical stepper motors and GT2 belts. Anything between 2-6lb should be fine, and I’ve only been in the 2-4lb range at the extremes.

Yeah, I am not keen to oil the belt or pulleys especially as I cannot see any significant rubbing. If I do, it will be silicone film lube and even that I am not going to do til I’m ready to buy a new X-axis. I am getting closer to that though as I believe I have eliminated just about everything else as being a factor.

Comparing kickstarter-era builds and later ones, there is noticeably more play in the carbon rods on the later ones. I think they may have relaxed the clearance and that had a big impact on the quality of finished prints. I have not checked the Y-axis but I do have some VFA there as well - just not nearly as bad as on the X-axis.

[zviratko]

@KanneKaffe

This is obviously a problem with the printing system hardware.

How do you know? It might be a messy frequency compensation, or even pressure advance (or combination of both). I saw weird artifacts change between firmwares and before/after calibration, without touching the hardware at all.

@Slarm

Do you happen to have the link for the post about the unfinished carbon rods post?

https://www.reddit.com/r/BambuLab/comments/16duhli/bambulab_says_this_is_due_to_lack_of_maintenance/
I remembered those grooves bigger, so probably not the same issue.

I wouldn’t advise lubing the belts themselves, just the idlers. So yes, it would still get on the belts, but you’re not going to soak them in oil. I changed the belts on my machine recently, it wasn’t simple but it wasn’t undoable, especially if you can glue the new ones in and avoid disassembling the chassis… (But I would disassembly it anyway).

Btw why would you need a new X-axis after that? At most it would impact the belts slightly, you can change those, but otherwise you can just wipe it off. (It’s true that when I was changing the belts I almost wished I bought the new carbon rods, as it would be just one simple step to change those after disassembling everything). Also on my machine those idlers were pretty dirty and had to be cleaned (IPA+cotton swabs was enough), I wonder if having deposits could produce some sort of pattern, but probably not a regular one.

[kungpaoshizi]

Sorry I didn’t read too much of what people said, but I think it’s fairly straight foward. I’ve not done it to mine yet, but change out the smooth idlers, with toothed idlers.
It’s the vibrations of the toothed belt on the flat surface of the idler pulleys.
Be sure to credit me when you do it

[Slarm]

Btw why would you need a new X-axis after that?

Because at that point assuming it had not worked I would have a greasy belt and I think the X-axis would be the last thing left to try. If I would be replacing the belt/pulleys I might as well just replace the X-axis since at this point the time I am spending is worth more than the cost of just pulling the trigger on the X-axis. The only thing is I have no guarantee that will fix it either - just a combination of the process of elimination and wishful thinking. I am less afraid to try silicone lube though since it won’t hurt rubber and should not spread.

[kungpaoshizi]

Maybe my suggestion above was stupid? Did anyone try that yet @Slarm

[Slarm]

Supposedly people did try changing to toothed idlers and it did not work. That said, I am not convinced that they installed ALL toothed idlers, and they used FDM-printed ones which I do not believe can hold tolerances enough to actually match the GT2 profile. The drive pulley I replaced stock with is 12K resin printed based off a model from McMaster Carr which matches the drawings for the GT2 belt spec.

That said, the majority of Bambu owners do not have such significant issues with the 2mm lines. I would like to believe that it can’t be from the pulleys since nobody else is running toothed pulleys. There would still need to be a separate element in the affected printers causing the pulleys to be an issue.

[zviratko]

You don’t need to replace all idlers, just the ones where teeth are going inward - so 2 in the X carriage and that’s it. But I doubt it does anything. Most of the uneveness I feel when moving X/Y by hand is coming from the motors. I still think it’s a frequency compensation issue, or at least something that it can improve on some printers and not others.