VC3.1 300 - Chaos on the Y Axis
Hi there,
Have been putting my VC3.1 back together after moving house but I am seeing chaos on the Y-Axis input shaper graphs.
It's running RatOS 2.1, the stock motors with 4 Beam Ruland Decouplers, BRS Belt Tensioning M2 in PC-CF, Toro-Tube & Toro-Rail, Honeybadger medium-preload Y-Rails, and new bearings on all points.
After putting all the parts back where they belonged I cleaned my X & Y Rails with IPA and re-greased with EP1. All running smooth with no weird issues felt.
Re-aligned the Y and X-Rails to make sure they are parallel according to the commisioning guide.
Then went through the belt tensioning process which seems to be good enough for now.
At input shaping I'm seeing a clean-ish X-Graph but absolute chaos on my Y-Graph and I have no idea what to look for. Tried to see if anything was rattling using the resonance tester but nothing I could find.
See attached the results of where I am at now.
Any ideas what might be causing this?
26 Replies
Have kept fudging around and loosening & tightening all the nuts & bolts that facilitate the bearings. And that has helped immensly. In this case - loosening the X-Y Joiners was a large part of the problem. But the y-rails were a bit tight in the middle, and the bearings near the motor mounts as well.
But thereis still some noise that I can't yet fully place. Any thoughts would be very welcome.
X-has also gotten a bit noisier but that might also be because the resolution of the graph has changed.
Interestingly - had to reboot the machine and my old, albeit slightly better, graph came back.
and it's back again after loosening the bearings up a bit more again.
Really confused and the graph is still a mess. Any idea's would be very welcome
The Y graph used to be ok? Are you certain that you have equal tension and that the gantry is square? For example, what are the frequencies of the belts starting from the front of the toolhead to the left and right, when the head is in the middle of x gantry? Are they within 2%? And there is no more than 0.5mm difference in distance from the y carriages to the front extrusions?
Are the labels on the graphs correct? In the y graph you seem to have quite a lot of z-movement at 80Hz. Does it seem like the head would be twisting around the x-gantry at that frequency or moving up/down?
The Y-Graph was OK before I replaced the Y-Axis rails with the new honeybadger rails. I didn't have the time to fully re-tune it before moving house so that might be an issue.
According to the belt graphs the tension is equal but plucking does show a difference in tone.
The Frame is square, just now re-measured with a laser distance meter. The largest difference measured is 0.001mm
The X-axis gantry was square, just noticed it shifted by 3mm. Will re-calibrate and re-run
Labels on the graphs are as the Macro generates them based on the accelerometer config generated by the 2.1 Configurator
[resonance_tester]
accel_chip_x: adxl345 toolboard_t0
accel_chip_y: adxl345 toolboard_t0
probe_points:
150,150,20
Tuned my belts to be the same pitch giving attached belt graph. I guess there is an issue with the number of teeth because getting those two lines to line up creates a noticeable p[itch change.
The X Input shaper graph is OK - similar to previously. Y is again a mess.
I believe some of the guides are a bit mis-guiding. Could you try 1) first align the gantry square, without the belts affecting at all. If need be, loosen the joiner screws, and re-tighten once square. Just make sure that in the free-moving state, where you or the belts apply no torque, the gantry is square. This is your neutral position for the gantry, and the squareness/neutral position will only stay that way if the the tensions are equal. Any unequal tension will turn the gantry away from neutral. So, with you neutral position, fix the gantry in place (push it against the front or smthng), and tighten the belts simultaneously, or like half a turn at a time, while making sure the frequensies (and thus tensions) are equal. The plucked paths need to equal in length. Once you get to the desires tension, your gantry will still be in the neutral position (square), and the tensions will be the same. There is no need for the belts to be same length to the tooth, and the belt tuning graphs are just a quality control check to see that frequencies there are identical/close to one another, and that the higher frequency peak (x) is higher than the lower fq peak. The graph must not be used to tune the belt tensions with this method.
My method is unorthodox and blasphemy, but it gives good results and should be correct from the theoretical PoV. But I fully understand if you do not want to try it, it is not a sanctioned one 🙂
And there is no telling whether you problem is in any way related to the squarenesss or tension. However, since it is quite easy to fix those, perhaps it would be good to try?
It's an interesting idea - if I understand it correctly the idea is that instead of 'forcing' the x-gantry to be square and then tightening that down, you square the joiners to the natural state of the x-gantry instead?
Yes. That is the only way it really is possible
..if you want equal tension, that is
In the meantime I double checked my belts and I found two small issues in the rear and front holders - both being 1 tooth shy. Double checked and the gantry is still square.
Again - pitch tuned belt tightness around 80Hz.
What is interesting is that the X axis stays the same albeit at a lower amplitude, whereas the Y-Axis has lost the 25Hz peak and gained a small 125Hz peak
It's nice that the X & Y lines are lining up a bit more - but the Z-axis movement still doesn't make any sense. Thinking of going back to the old rail, see if that helps, or maybe even the old 2020 extrusion.
Yes, I tried one of the lightweight x-tubes, but it had an absolutely horrible torsional resonance very close to the spring-mass linear resonance of the belt/head combo, so I could not use it. Perhaps just an issue with my extruder/gantry combo
Might try that - switching to the 2020 is something I can do without having to take the printhead apart (again).
New tests with the 2020 installed again tuned by sound.
It's a bit better - the Z-movement has been halved and the signal in general is cleaner.
Hmm.. there are still some serious-looking issues... The y looks bad, and the belt tuning graph does not show the two resonances. And the Y resonance peak would seem to be at the same frequency as the y (around 65Hz), which should not be the case. What size is this? I have first-hand experience only from a 300.
This is a 300 as well
Then I believe you should be able to get cleaner graphs
Yeah - I found some old graphs when the machine was stock and those were much better. Going to go back to the stock parts step by step and see what happens and work through it.
Mine is not the cleanest I've seen
This was the Y 😄
Not super but alot better than what it currently is.
Yes! And you must have a very lightweight extruder/gantry, with Y resonance of 61Hz
That was the stock 2020 & rail.
Right.. I am running extrusionless, so it might quite be that the rail alone is too wobbly to get to higher accelerations. But you have a nice target from past now 🙂
Aaaah - that is an exciting way to do it 😄 But thank you for all your time and help!
Going to close the ticket and just chug through the process or rebuilding.