Dampening issue on Y-axis (VC3.1 - 300 enclosed)
Hi I have a dampening issue on the y-axis (I think). I believe that the left linear rail starts "rubbing" once I tighten the gantry bolts and have tension on the belts. I have been figthing this ever since I have the printer 2 years ago. usually, I can "fuss about" intill it's gone but not today. Any tips? (X-axis seems smooth as babies bottom...)
Solution:Jump to solution
btw, made a first successfull print. not 100% content with the quality yet but good enough to mark this solved. Thanx @mazas !
44 Replies
I am still hoping to get some assistance on my issue. I HAD a printer that showed nice curves. one steep peak at x and 7, high accelerations and suggested MZV shaper on both X and Y. I used it for a year and last christmas I tried to install the x and y motor decouplers from toro-3d. I was not happy with the reduced speeds and accelerations (and the noise reduction was minimal) so I removed them again.
Since then, I have horrible X and Y peaks and bad shaper recommendations. I have in the mean time:
- Re-squared to whole printer (to learn that was not out-of-square anyhow)
- Taken apart the whole Eva3. Tightend all bolts using a bit of loctite,
- replaced the belts with new ones
- replaced the lineair rail and carriage
- removed all side panels
- Checked and rechecked all bolts,...
still my shapers are horrible:
@mazas apologies for tagging you in here. I am somewhat desperate (I have been trying to solve this issue for three weeks now) for assistance and saw you had some good suggestions in other posts...
No problem. Have you read https://mazas-3d.notion.site/CoreXY-Belt-and-Gantry-Tuning-with-VCore4-3-ed6eb94671d54c44a7228eb45eeb4cf9?pvs=4 ? I’d need to update it, but most of the stuff should be usable.
Yes, I have read that. Most of it is "common practice" for me.
I started here: https://www.youtube.com/watch?v=h7ZTq2dDTkM
than watched this for 15 times https://www.youtube.com/watch?v=M-yc_XM8sP4&t=1701s
3D Printers & a Whiteboard
YouTube
How to Read and Analyze Input Shaper Graphs Generated by Klipper
This video is supposed to be a brief overview on what the input shaper graphs can tell us. It is the first video of a series of videos which will give a more in depth view on the kinematic system and input shaping in Klipper.
As always: feel free to jump to those parts, that help you with your problems:
00:00 Intro
00:57 What we can learn from...
I Know something is off, but I just can't find it...
My X-carriage is super smooth without belts but with the belts, it seems like the carriage bearings cause some unevenness,... no idea if there is evidence in the graphs for that.
When tightening the belts, what method do you use? Is the belt tension the same for both belts? What is the tension, how many Newtons? If you have to sacrifice squareness of equal tension, then the equal tension is the more important, as it makes sure the y carriage does not get unnecessary load.
But overall, the graphs look so weird that there probably is something other than just the tightening. So you’d say that the gantry moves smoothly without the belts, or with the xy-joiner screws loose? Have you ”honed” the x gantry so that printed parts that attach to the Y-rail carriage plates are flat and parallel?
Tightening is done methodically by starting with 0 tension and then tightning each side one turn at the time. I usually loosen the xy joiners to monitor squarenes and equal tension. Once tight enough (and square), i check y-gantry smoothness (as there usually friction in bearings starts). Then I tighten the xy-joiners. No idea how I should measure the actual force. I try to tighten just tight enough that the belts are not vibrating in resonance against the frame.
I dont understand your last sentence regarding honing. Could you elaborate?
Also, how big is the chance that it is related to software? The IS macro now goes left to right while vibrating an front to back while vibrating. Id never did that before... so maybe the macro now is faulty? @miklschmidt might know?
starting with 0 tension and then tightning each side one turn at the timeSo how do you actually measure that the tension is the same?
maybe the macro now is faultyHighly unlikely 🙂 . It might reveal some issues in a different way.
Well, my theory is that, if the xy-joiners are loose, unequal tension throws the gantry out of square, so tightening with loose xy-joiners and keeping things square should result in a square and equally tight gantry.... Realy open to other suggestions here,...
Yes, grasping at straws here... I know... You have seen the above graphs,... they are VERY consistently screwed...
No this is intended, new resonance tester functionality in klipper 🙂
there's indeed something very odd going on in your system.
So two things i can gather from your tension graph.
1. Belt lengths (Between clamps) aren't equal, causing you to unevenly tension them (forcing the gantry out of square) to match the peaks in frequency.
2. Waaaaaaaaay over tensioned going by that 115-120hz peak. I think the 60hz one is harmonics.
unequal tension thows the gantry out of squareThat is an incorrect assumption. I might suggest a re-read of the article. What would be correct is that no matter what the angle is, you will have equal tension automatically if you have completely freely rotating xy joiners. Unfortunately just loosening the screws might not give you a fully free movement. I would suggest that you get a frequency tuner app for the mobile phone, move the head to exactly in the middle of the gantry, and pluck the belts to make sure the frequency is equal to +-0.5%
Yes, Did that(frequency tuning) yesterday. Did not solve the issue... I am going to take apart the whole eva3 again, because I hear a light rattle coming from it. I suspect the bearing balls in the carriage of the lineair rails, however I replaced them 2 times in the last week with spares and it did not have this issue before I started to tinker with it...
What is the fq that you have for the front belts on the gantry?
I am going to check the belt lengths, however I think they are withing 5 mm of each other. I will Check!!!
And I am also STOMPED by the 120 Hz Resonance peak. It never did that and I always had good solid peak around the 60 / 64 Hz....
A 5mm difference by itself should not cause something like this
Right now,... 0Hz. Took the thing apart yesterday night (again), got stressed, stopped, took a beer,....(or 2). I tuned the belts on the front of the gantry with the printhead in the middle at (i think) 135Hz. However, I positioned the head at 150mm after homing but realised later that this does not automatically puts it in the middle of the gantry as Origin is defined relative to the homing point etc... need to redu this.
I Also do not understand the Y-gantry lower peak of 42 Hz,...
I (think I) need a new toolhead but cant print it because my printer is screwed. Also the Ratrig toolhead (for VC3.1) does not support Beacon...
however I think they are withing 5 mm of each other.That number needs to be 0, you don't actually need to take them out and measure (it doesn't really help you), you can release a tooth on the tighter one (the blue one - the one with the higher amplitude). Remember it's not the length of the entire belt, it's the length between the points where they attach to the toolhead. Repeat until amplitudes match. Start with MUCH lower tension though.
Ok, sounds like a plan. I'm running my 300 with about 90Hz on the gantry belts.
It's definitely a multitude of things, excessive tension + random belt lengths can make the funkiest of graphs.
especially on the 3.1 where you're literally bending the motor shafts.
Un-tentioned this is marvelous...
90 Hz on the front two belts, left and right of the print head, right? on the front as per the video above...
Yes
So you have the dual-rail system. I would check that the gantry is not twisted, and sand the bottoms of the white blocks to make sure they are parallel. (That is, if they are not parallel.) You do this by removing the gantry, taping a piece of sandpaper on a flat surface, holding one of the white xy joiners on the flat surface and then rubbing the other end/joiner against the sandpaper. To get more accurate, you actually should have a piece of cardboard below the not-on-the-sandpaper joiner, the thickness of which is the same as the sandpaper. Does that explanation make sense?
But not sure whether that will make a difference. Where is the accelerometer located? How do the prints look like? I'm just trying to make sure this not an issue with the measurement setup.
Yes, The explanation makes sence. I did this about a year ago and used the double gantry with reasonable results. Prints were decent but not perfect. As part of the trouble shooting, I removed the bottom carriage as it can be a pain in the butt to align the 2 carriages. It is now only riding on the top rail. As for accelerometer, I usually use the beacon one but I have very similar results if I use the EBB42 accel.. (as said, I am troubleshooting this for 3 weeks already,,....). I did not do a lot of printing with the current setup as I need to calibrate my gantry again, and all my bed meshes and I am not realy looking forward to al that work... Particularly while I know something is very wrong...
Althoug it looks like I am countering all your suggestions, I Realy appreciate your suggestions and am convinced that if I keep on listening to all the stuff you and others bring forward, I should be able to solve this (at least,... that is what I am telling my wife and myself for a while now.... 🙂 ?
We have something non-obvious going on here, so it is quite likely that the suggestions will not result in fixing the issue right away. Can you identify what resonates with the y-test? The 64Hz is high for y, especially since you have the heavier-than-stock dual rail system. My suspicion would be that it is different mode of resonance (twisting instead or back-and-forth) for the gantry, but the accelerometer readings do not seem to support that. As a troubleshooting step, can you vibrate the head in y at 64Hz and video the head in slow motion, at 240fps or more if possibe? I would like to understand how the head/accelerometer moves.
I will try to work on that when I am home,... my payed day-job is getting in the way of "hobby stuff"...
No smoking gun (i think) but I pulled Eva apart and found 2 cracks around heated inserts. One that is used for Beacon (left) and the other to connect the fan shroud middle with bolt in it.... I need to repair this temporarily and try to print a new part...
Phoe,.... I can not tell you how much I hate this printer.... the volitivity of the design is horrible... tighten one thing, another thing goes wrong, make a 1/1000 of mm misalignment on one side, something starts rubbing.... Man what a waste of money.
Anyway,...assembled the whole lot. Put new belts on and now the belt on the right-front XY joiner creeps up when the head goes to the right, causing the belt teeth to interfere with the flanged bearing. Bought other belts: no change. Used Kevlar reinforced: no change, Steel Cable reinforced.... also crap.... about a 1200EUR paperweight if you ask me. Happy that it is a hobby and not my business...
And it didn't do it before... So another thing to solve....
That movement is to be expected. The geometry on the pulleys does not self-center (as per their design), and the small variations shaft angle can cause that. Also, the spine fibers in the belt are not absolutely straight. The belt is cut spirally from a large-radius tube, so there is ever so slight tilt in it.
So, i would not be overly worried about that. Ofc if the rubbing is strong, you should check that the shafts are reasonably straight.
Besides the belts, do you somethig else rubbing?
Hmm.. is it possible that the dual-rail xy joiner is somehow curled/bent? I do not know what is the orientation it was printed in, but can you check that there is no warping?
One rather tedious debugging thing could be to switch to the the stock gantry. Or you could do what I do and leave gantry extrusion away altogether, and use just a single rail. That might not be the final solution, but least both of those should produce nice IS graphs. And I realize that you had food graphs earlier, but somehow we’d need to now figure out what changed
So, what i did was remove some shim plates between the 1mm precision shim and printed part. That now has 0.5mm play and now the rubbing is no longer obvious (it showed on the resonance graph before. now it doesnt)
As per your recommendation, I tuned the belts at 90Hz but that resulted in resonating belts. I added some tension and now they are at 120-ish Hz. Also, I find it hard to see if the 120 hz is the first order or second order (i.e. resonant frequenty.). If it is resonance, than my tuning now is at 60-ish Hz, obviously
What do you mean by resonating belts?
they start "flapping" against the frame at 40 to 60 Hz
Yesterday I repaired the bottom part of eva (the lower duct) with some ASA filament and soldering iron. I re-melted the heatpress inserts and carefully assembled the lot.... I believe I am getting somewhere and I start to think that I might have solved the issue,... (not all issues, but the funky graphs now only have one little issue,... One Moment Please,... Graphs are rendering,.....)
Y-axis still has an issue....
Ok, that belt flapping is also to be expected, and not a major concern.. i believe mine does that as well. That flapping does not seem to be a real problem in printing, and neither can I say that I could see it in the graphs. But 120Hz is probably ok, as long as the motor shafts can handle it. If my memory serves, that will cause the force to go over the spec for the motors though.
Ah,.... I never actually tried just printing with the "flappy belts". Might try that later still.... Can I ask you opinion on the Y axis graph and particularly the second bump?
The graphs are looking better, but I am concerned about the second bump. Can you run the graphs with another accelerometer? The beacon is fixed with two screws, and that provides only minimal support for movement in y. It just might be that the second is a phantom.
Well, actually i tried to say that with y movement there is a possibility for twisting/rocking, but that should be visible as z-acceleration. Well, nevertheless, if you can try another location for the accelerometer that would be good.
Good Idea! I'll try that tomorrow (I now am in whisper-mode as my kids are in bed....)
again, @mazas Many thanks for your support! as you can read, I actually am ready to throw the stupid thing out of the window,... but still,... here I am, here we are,... strugling onwards... It is not an easy printer, this VC31
I can relate to that feeling, although my biggest issue with the printer was probably building it square :). Also, at this point, since graphs are starting to look reasonable, I might try checking the max acceleration/ringing by printing that klipper l-shaped test print. Doing the IS with the accelerometer is only an approximation, as it does not measure the nozzle tip, comes with its own set of errors and there is no filament or part involved. It might quite well be that you have a well-functioning printer but we just do not know it yet 🙂
Ha, yes I might have been focussing too much on the IS... Just wondering if the absence of filmant in the reverse bowden could cause the Y-peak... (it should be vissable than on the x-graph as well, I assume...)
Solution
btw, made a first successfull print. not 100% content with the quality yet but good enough to mark this solved. Thanx @mazas !