Shaper graphs have different peaks for Y and Z
I just reworked my whole X-axis on my V-Core-3.1, 400mm. Based on the belt graphs, I think the tension is about right-- both peaks are close in magnitude on 55 Hz (can I tell if they are tight enough from the frequency? or only if they are equally tight?)
For input shaping, the Y-graph weirds me out a little bit. The Y and Z have different peaks, and Z is actually quite a bit taller than the Y vibration. As I type this out, I realize that must mean the whole tool-head has rotational vibration around X axis as it vibrates at 43 Hz? I presume that is not normal? If it's fine, which frequency should I use for input shaper?
If it matters, I'm using the toro3d tube and rail that I won in a Discord raffle forever ago. It is 340g lighter than using regular rail w/ 2020 aluminum.
14 Replies
My understanding is you should have two different spikes on your belt graph. Something is binding most likely. On the belt graph you should see a spike for the X motion and a separate one for the Y motion. This is a VC4 500 and you see a smaller spike at 35hz and the larger spike at 70hz.
I have some noise out at 125hz so dont look at that 🙂
with the motors idle, how is the gantry feel if you move it back and forth. Is it smooth feeling or binding. It should be the same tension from front to back. Also does the toolhead stay in the same place or does it try pulling left or right, etc.
I just squared it up and tried to equally tighten the belts, which got me to this point. Nothing feels abnormal to me.
I think my belt chart does have both peaks, they're just pretty close so I think they merged into (almost) one, due to the extra-light X-axis, which raises the frequency of the Y-axis test peak to be closer to the X-axis test peak. At least that's my intepretation of it.
I realize now I should've marked specifically what I'm looking at on my chart: "Y-axis test graph" that shows different peaks for measured Z-vibration (blue) and measured Y-vibration (green). The peaks are distinct, and the measured Z-vibrations (blue) is much bigger than the Y (green).
I think that means the toolhead is moving up and down, or rotation around the X (I posted here looking for feedback, but I think I'm figuring this out as I type). I did just confirm that the print head has quite a bit of rotational play about the X-axis. I think that explains it. This linear rail has a serious problem!
I realize now I should've marked specifically what I'm looking at on my chart: "Y-axis test graph" that shows different peaks for measured Z-vibration (blue) and measured Y-vibration (green). The peaks are distinct, and the measured Z-vibrations (blue) is much bigger than the Y (green).
I think that means the toolhead is moving up and down, or rotation around the X (I posted here looking for feedback, but I think I'm figuring this out as I type). I did just confirm that the print head has quite a bit of rotational play about the X-axis. I think that explains it. This linear rail has a serious problem!
you might be right on the lightweight gantry. I just am not sure if thats causing things to merge together or not.
What are you using to measure the resonance? I suspect its mounted in a way that is not correctly aligned on the right axis. I dont see how you would have that much Z movement and not have some major issues
that is my Y on my VC4 500 and you see little to nothing on the Z. Maybe you have some with a messed up rail but I think you need to check on how the ADXL is mounted. Do you have a separate ADXL chip or part of a toolboard. How is it mounted?
The accelerometer is mounted on the standard mounting holes on the front-top of the EVA3 toolhead (upper left in the video).
In the clip you can see how insanely bad this problem is. I'm hardly applying much force. The glide block is clearly rotating with the toolhead. The bearing balls must be the wrong size or something.
So my hypothesis is: the center of mass is not centered on the glide block, then the Y-oscillations will induce rotations about the X axis. Given the position of the ADXL, that would probably look mostly like Z movement.
I won this assembly in a discord raffle from Toro3D. This one must be defective. This is the first time I've tried assembling the gantry with it. Haven't tried printing anything yet. I suspect this will cause major problems as you suggested.
In the clip you can see how insanely bad this problem is. I'm hardly applying much force. The glide block is clearly rotating with the toolhead. The bearing balls must be the wrong size or something.
So my hypothesis is: the center of mass is not centered on the glide block, then the Y-oscillations will induce rotations about the X axis. Given the position of the ADXL, that would probably look mostly like Z movement.
I won this assembly in a discord raffle from Toro3D. This one must be defective. This is the first time I've tried assembling the gantry with it. Haven't tried printing anything yet. I suspect this will cause major problems as you suggested.
oh wow yah that is not good. time for a new rail 😦
Gonna see if the Toro3d guy will send me a new one. I weighed the whole X-axis before putting it back in:
Old X-axis: 610g
New X-axis: 270g
About 40% of that weight savings is the linear rail. I'll probably just have to suck it up and go back to the old rail for now.
Thanks for listening to me figure this out, haha
can you use your old linear rail on the titanium extrusion?
Yeah, that's my next step. I'll rerun the ADXL stuff after I do
good luck!
SOOO, you were right that the Y- and Z-axes of the accelerometer were switched. I forgot about that. So (actual) Y was dominant frequency in the initial Y-resonances chart. But (actual) Z was still relatively huge, and my conclusion is valid (just not as extreme).
I swapped the rail and re-ran everything, and now it looks a heck of a lot better. There's still a tiny Z-spike (there's still a tiny bit of rotational play in the toolhead), but overall everything looks peachy now. Thanks for the help!
your welcome! happy printing!