Was yours a kickstarter unit or just really early?
Yeah i’ve read every thread I could about this to see if this was the normal. I believe this is the best thread on the subject though. I seen your pics and it looked the tape improved it quite a bit hey. I couldn’t see any pics of yours showing a 45 deg angle they all looked straight on. Sometimes i cant have just straight edges and the 45 degrees look really bad. I submitted a ticket so Ill wait a few days and see what Bambu have to say I guess.
What is the mod exactly?
Is this rolling the pulley with tape? I did not see it but I imagine it isnelectrical tape rolled around it so the belt rides on it?
Thinking out loud here. Would this add a cushion of sorts that dont transfer to the head?
If so, how about printing a pulley out of TPA to experiment, maybe someone can make custom pulleys with a plastic shroud of sorts and the core being metal.
There is something wrong with this design, it just blows my mind bambu won’t admit it
if you are thinking of returning it, do it very fast.
I am currently being ignored by support because my printer is 1 year old, even though is has very little use.
I am investigating what is the return policy for an unopened A1 I am about to receive, as I am thinking of leaving this ecosystem.
So I read this entire thread and am very impressed by what I’ve seen. I have an early model that I bought on pre-order just after the kickstarter ended. I received my X1C last November and have been printing with it nearly 24/7. My prints absolutely show the same 2mm VFAs on the X axis. It has never really bothered me, but I have always been curious about it.
I am going to suggest that what we are experiencing here is a CoreXY problem, not a Bambu problem. My Ender 3 did NOT exhibit this behavior, even when I was operating it with Klipper and pushing the platform beyond its limits. I suspect that this is because the printer used POM wheels in extruded aluminum for its movement across the X and Y axis. All CoreXY machines have the same basic design with these same pulleys and belts, and from what I’ve read up to this point, they all suffer in similar fashion.
From the beginning, I always suspected the design of the belt. It’s too uncanny to have the artifact match almost exactly the tooth spacing/pattern of the belt itself. It’s going to act exactly like driving over washboard on a dirt road. There are certain speeds that you can drive that helps reduce the vibration caused by washboard. If you drive super slow, you don’t notice the bumps very much, and if you drive crazy fast you sort of glide over the crests of each bump. However, everything in between rattles the teeth out of your head. Well, same concept here. You have a belt with evenly spaced, tiny ridges riding over smooth idler pulleys here and there throughout the machine. That is ALWAYS going to induce a vibration, and also explains why it’s better at super slow and super fast speeds, but never really, truly improves to an exceptible point.
You can absolutely bandaid the problem by cushioning the striking of the teeth. It is recommended to air down your tires when driving over washboard to help reduce the vibration. You’re essentially doing the same thing by adding teflon tape to the idlers, and it’s why you’ve seen the most improvement with that idea.
Unfortunately, the ultimate solution in my mind is to ditch the belt entirely. The toothed belt in the CoreXY assembly will always, always, always create VFA to some degree in the print. The belt has to be a different design. Look at timing belts for a vehicle. They are toothed belts, but the teeth only ride in cogged pulleys in order to reduce vibration (but doesn’t totally eliminate it). All of the idlers and tensioners touching the timing belt use the backside (the smooth side) of the belt. I feel like the reason some Bambu printers do better than others could be the quality or material of the belts being used. A softer, more pliable belt will exhibit less ringing than a hard, stiff belt (stay with me yall. Keep your minds out of the gutter!)
I believe a serpentine style belt is the only real solution here as it can still provide the most grip with the least belt slip as opposed to a totally smooth belt or a V-belt. The “teeth” will then run parallel with the pulleys, completely eliminating all vibration that otherwise would be produced simply by moving. I never really liked seeing the toothed belts used in CoreXY because I assumed the belts would totally resonate. It appears I was correct at least in part.
I’m sure there are other factors, but ultimately the style of belt being used is the largest problem. And yes, it would require an extensive redesign. It’s easier and cheaper to tell you to keep everything clean and print faster to reduce the effect than reinvent the wheel, so to speak.
In the end, this is the big drawback to fast printing. There are absolutely going to be physical limitations that will act as speed limits to how fast a 3D printer or any 3D printer could possibly print. The tradeoff will always be the quality of the print with CoreXY as it stands. A bed slinger like a Prusa will always outperform the Bambu in print quality because the basic design of the printer is completely different with the design inherently favoring the final quality of the print while sacrificing print speed. If you crammed Bambu’s awesome tech into something like a Prusa machine in an effort to promote quality over speed, you’d have yourself an incredible printer, albeit slow. I see the A1 as an unintentional step in that very direction through its use of linear rails instead of belts to move each axis. But that’s just my 0.02 I’m no scientist for sure. Just a dummy huffing plastic fumes constantly lol
2 Likes
It was a kickstarter unit.
On the final turning pulleys where the teeth ride the smooth face you can wrap it with a buffering material. My early tests were with electrical tape worked well, but it’s thick enough that the overlap can cause a hump/thump artifact which is why I switched to teflon. It is thin enough to do overlap with negligible impact, but you do have to wrap it more times to have as much effect.
Replacing the pulleys is a nightmare because the X-axis was never meant to be disassembled. The pins securing the idlers cannot be removed without drilling/removing the carbon rods, and that itself requires a complete teardown of the machine to remove the X-axis.
As far as I understand this is correct, but on other machines it is MUCH easier to calibrate and adjust things. For example the Bambu has no skew adjust or ability to actually tune the belt tension - just dummy mode on the tensioners and hope for the best. There are ways to force it to do what you want, but not ideal. I have talked extensively about this issue with a friend who builds his own Vorons and he said the issue had appeared but he was able to resolve it much more easily.
All but one pair! On left and right side the teeth of the toolhead the teeth ride the smooth face on 2 pulleys. This is where the issue almost entirely comes from because these are the only ones that need alteration to manifest a huge improvement.
I do not think the Bambu belt is a real Gates and suspect the wandering could be the result of an unstraight cut. If some belts are cut straighter and do not rub the flange, this could be part of the explanation. I ordered Gates neoprene belt to try, and eventually I will replace mine and see if there is any improvement.
I love that idea, and I imagine with the appropriate method for wrapping the drive pulley it would have enough friction to work. Alternatively, if it is actually the flanges rubbing as my tests seem to indicate, then a belt with dulled tooth corners or teeth only on center of the belt would also help.
What were some of the things your buddy said/did with his Voron? I’m not overly familiar with them as I never built one, but always had a great appreciation for them. I might still do it just for the fun of the hobby. Anyway, what adjustments and such was he able to perform?
This will be very interesting to see if there’s any improvement when using a belt of known good quality.
[quote]Alternatively, if it is actually the flanges rubbing as my tests seem to indicate, then a belt with dulled tooth corners or teeth only on center of the belt would also help.
[/quote]
I wonder if you could feather the corners with some emery cloth to see if that would further improve anything. I also love the idea of a “toothed” V-belt hehe!
It was basically a matter of having control over all factors such as:
-Belt choice
-Ability to square X and Y axes
-Belt tracking by pulley adjustments
-Control of belt tension
-Choice of pulleys
1 Like
I ended up trying the teflon tape wrap, a few different belt tensions, a couple jigs to try to “square” the XY gantry, played with twisting the belt tensioners slightly to get the belt tracking better, and I’m not sure if I made any difference with anything I tried. I really wish we had some better diagnostic utilities on this system, like if we could do a movement in X and have it display the accelerometer data in Y then we would at least have some numbers to go off of.
I can confirm that you never run belted teeth side over smooth pulleys, never. This is one of the problems with the SV06 too. You should always use a toothed gear for the teeth side of the pulley. Then you only need to ensure the belts do not hit the flanges and ride as smooth as possible.
It will always result in VFAs if not a strait setup. What is worse however while that is a no no as far as I am concerned in my X1C I only get it in the X axis and not y, which makes no sense.
Also while no one wants to hear BL tell you to run faster, there is some truth to it that running it faster will help but it in only widens the gaps of the VFAs and distorts the patterns. So its not a real fix. You can test it by dropping your accell to below 10k and maybe outside perims, and you will see the VFAs stack nicely. In any case for a company that kickstartered over $8mil you think they could have through about how belts go the other direction over pullies. It would been as simple as to to just turn the belts around but I bet the routing setup they did will not allow it. On Vorons the only part of the teeth that touch a gear are on the stepper motors only the rest of the routing is smooth on the back of the belt.
Wonder if we could figure out routing with the belts flipped.
1 Like
I did more or less all the same experiments as you listed. Verified gantries square with jigs, varied belt tension between lower and higher than default, locked the tensioners in crooked to try to track the belt straighter, added weight to the toolhead, and tried some stuff that people had suggested in other threads that did not seem related (and weren’t.) In the end the only thing that changed anything was varying belt tension, and all it did was change the speed range where the artifacts showed up.
The only thing that made a difference for me was pulley wrapping and while it is not perfect it did reduce the intensity from being measured on a basic caliper at 0.03mm to not being measurable. I have been using around 7-8 wraps of varying length on each of the two pulleys for maybe 120 hours of printing and it seems to still be holding the quality static.
Fully agree it would be nice and helpful to see more technical info on the operation side. But the problem existing and not being given any attention as well as obfuscating how things work and what is actually happening speaks to the apple-ification of the platform. Ease of use at the cost of refinement and functionality for power users.
On Bambu the only place it is tooth on smooth is 2/4 of the final turning pulleys for the X-axis. Movements in Y have a hint of it I assume from the varying flexibility for teeth through the smooth side, but any travel on X has it rolling the teeth over smooth. That includes combination X-Y moves which distorts the length of the striations by the factor of the vector of motion. If it’s moving 1:1 in X and Y the striations are scaled by sqrt(2), so 2mm becomes closer to 3mm.
I think they basically got to a ‘good enough’ point where the number of people who will have a problem and be bothered by it is so small the financial impact is less than the cost to fix it.
I think it would be possible to flip the belt on those two individual pulleys, but since the travels don’t leave much space on the Y belt spans, it might cause problems toward Y +/- limits. It needs to be looked at!
I keep seeing this mention of using micro-v or serpentine style belts as a solution - a serpentine belt is absolutely not usable for a printers motion system without having closed loop encoder feedback on the absolute position of the printhead. You’re not going to be able to do that here.
Serpentine belts drive by slipping - any time they’re under load they slipping by a percent or two, so your accuracy and precision would be gone in a few minutes of running alone.
There’s a reason everyone uses cogged belts for 3d printers and not flat or micro-v belts - it’s because even without the massive slip that jerk motion would cause, they slip, constantly, it’s inherent in how they work, the same as tyres on your car are always slipping slightly under load.
The best option would be herringbone drive belts and matching pulleys, which I use on some bigger drive systems, but I’ve never seen them small enough for consumer size printers.
3 Likes
Just to add to that, I think the reason the tape seems to help even though the toothed pulley swap doesn’t, is more to do with the pulley being undersized for the belt/belt path. Softening the impact with the teflon tape and making the pulley slightly bigger helps more than the toothed outer does.
It’s a packaging compromise.
I do not trust the conclusions from toothed idler tests that have been performed so far. They have all been sleeves placed over the existing pulleys and printed with FDM. The design for GT2 pulleys calls for tolerances finer than consumer FDM printers can hold, plus there will be seams and a split for the way it is installed.
It would need to be proven by installing proper GT2 pulleys in place of the existing ones which requires extensive alteration of the X-axis assembly.
3 Likes
That’s fair enough, a resin print might be close enough to get meaningful results, or the old trick of a bit of belt glued on inside out (the profile is wrong but it usually softens things enough to solve belt noise in other applications)
I do think resin prints are quite good enough actually. I was running a pulley I printed on one of my drive motors for a month or so and as long as I glued it to the shaft it was reliable. When my work receives the Formlabs silicone resin I might ask the lab to print some silicone pulley sleeves!
An alternative might be to print them in TPU just so there’s some give in the design. It’ll reduce the motion accuracy a bit but for a quick test it might help.
So Bambu Lab announced motor noise cancelling option for the X1 series in the upcoming firmware update and they stated that “it slightly improves Vertical Fine Artifacts (VFA), resulting in smoother printed walls.” I’m curious how much improvement this feature will bring.