Klipper input shaper wierd graph

Hi all

Im trying to improve my print quality so im fighting with input shaper now. On x axis i have no issiues and perfect print after shaper. But on y axis i have second peak on z. I dont know what could cause that second peak on z when im measuring y. Take a look on graph.

Printer is carthesian tevo tarantula pro with tmc2208.
Im using portable input shaper on rp2040 wired to hw spi.
Its from here
I designed and printed adapter to screw it to holes near wires connect to bed. So probably i can exclude mounting isuess.

Graph looks OK, nothing to worry

Spring mounted beds on bed slingers tend to do this.

The bed will have a rocking motion with a component on the Z axis. Mounting the accelerometer at the center of the bed could reduce the pickup of this vibration mode. You can use one of these adxl mounts that are printed with a brim and left adhering to the hot bed for the duration of the test. However, it might be desirable to have input shaping compensate for this.

If you’re targeting quality over speed, I recommend to accept these results and use a wide band shaper (MZV or 2/3HUMP_EI) at lower accelerations, as recommended by the sharper calibration results. Especially if you intend to print tall and heavy objects that will offset the resonant frequency.

Another potential mode of vibration in Z can be produced by the bed inertia pushing the mounting screws out of square with the bed and carriage, acting like a unintentional four-bar parallelogram linkage.

The bed stiffness can be improved by tightening the springs, using stiffer springs, installing these silicone spacers, or even mounting the bed rigidly. Also it can be worth checking that the v-wheels are tight enough and that there is no play in the bed carriage, but don’t over-tighten them.

Yes, this could help. But I think this is a basic “problem” of bed slingers and my position is (and will ever be): If something is wrong, don’t play with the symptoms.

As you mentioned, this is a reaction in Z-Direction to an Y-Movement: the table is tilting somehow around the (some) X-Axis. Reasons could be:

  1. Mechanical Issue at the table: Bearings worn out, screws loose, Guides loose etc. Table bends during test.
  2. Setup problem: Springs for table too weak or loose (check, but springs should be no problem, since momentum against the tilting line [mid of the table in most cases] should be large enough). Silicone dampers are used and silicone is too weak, etc…)
  3. XY-connection to base is loose or printhead is loose or bearing of X-Axis or Y-Axis is worn out/loose/weak.

What you could do: put some weight (50g-200g) to the front edge of the table (tape or clamp) and redo the test. Front edge: Because few mass will increase the momentum against tilting remarkably but will not affect the acceleration in Y-Axis that much. If the issue is there in the same dimension, you should look for 3.

But: Some behavior are “as is” because of the configuration of the axis itself. If you can live with it, don’t spend too much time on it. You accelerate a lot of material with this configuration and this has consequences.(“Graph looks OK, nothing to worry”).

Finally: Print with lower speed and use a bigger nozzle. This compensates by far a faster movement of the axis (for a 0.4mm nozzle for example). Could you upload a picture of ringing tower for Y-axis? Just to be sure…

Not so sure where this discussion is heading:

  • Graph is to be expected for a bed slinger
  • Certain Z portions you will find in a lot of graphs (the reasons why they are there are basically endless, beginning from mounting the IMU or its cable over to some loose stuff, up to a bit bumpy linear bushings / wheels). Even more so considering the mechanical design of the y-axis of this printer
  • Recommended MZV shaper is a “harmless” shaper and the remaining accel of 4500 more than enough for such a printer

Why? The recommended MZV shaper already leaves no vibration so there is no good reason to do so IMO

Why why? I did recommend to follow the calibration results and use MZV. My point is that trying to fit ZV on a bed slinger Y axis is pointless because of the resonant frequency shifting during print ("nothing to worry”). Wider band shapers also make sense on heavier beds like CR10’s.

The implicit question was:

And that is why I described the most common vibrations modes of beds in Z.

I would not recommend wider shapers than the logic already recommends.

Not intending to confuse or divert this discussion any further, but…

It is possible to have sharp single peak on a well tuned bed slinger with some (a lot of) work. My CR-10S Pro is still on the original Y axis configuration except I significantly upgraded the stepper (Moons’ 2A 0.9°), replaced the belt with Gates and installed better quality pulleys (Mellow & Gates). The motion system is still using the mechanically over-constrained double extrusion wheel system that is almost impossible to adjust without some animal sacrifices. I straightened the Y extrusions and substituted Openbuilds POM wheels for the standard Creality issue. I also eliminated the spring suspension of the bed and replaced them with silicone spacers. I then spent literally hours fine tuning the wheel adjustment and extrusion positioning to arrive at reasonably uniform (low) wheel preload so that there is absolutely no vertical play yet they move freely. If you want to do this on the dual extrusion CR-10S Pro make sure you have a case of beer (or something else that’s more effective) on hand and a lot of patience. With this approach, I get the following resonance curve using a well mounted ADXL:

On the other hand my recently built Voron does pick up some quite significant movement in the Z axis when resonance testing the Y axis. This is despite it being adjusted to “perfection” within the limits of the design and the component tolerances. I am quite certain that the Z pickup is caused by the complete print head assembly responding to Y movement because its inertia is not evenly distributed above and below the X gantry extrusion. In other words either the gantry is twisting or the ABS components are slightly distorting under load in the Y direction:

At some point in the future I will be able to validate my hypothesis once I rebuild the gantry with much stiffer CF-ASA parts…

In the meantime, my recommendation would be to use the shaper that Klipper recommends, as long as you are reasonably certain that the ADXL was well mounted to the bed. If it was not well mounted and you suspect that it was picking up some stray resonance caused by the mounting, re-run the test with corrected / improved mounting.

I normally mount the ADXL without using any printed mounts. I use double-sided tape, insulating tape and aluminum duct tape on top to keep it firmly attached to the moving part(s). Keep in mind that accelerometers are frequently mounted using wax (or similar) in ad-hoc industrial, aerospace or military vibration measurement applications. Have a look here:

Avoiding Errors When Measuring Vibration | Brüel & Kjær (bksv.com)


These are candidates! Could also be the reaction of the frame of the original Voron! The frame is weak like cheese. I tuned my frame a little and this is my Y-reaction on a 350er standard Voron with improved frame stiffness.

see Rigid Voron Panel clamping by BergKaese - Thingiverse for a graphic showing the standard frame reaction (looks pretty much as yours), then blue line in the upper graph is with gusset plates (60x60x2) and the red one with the linked clamps and rubber 75 shore replacing the foam sealing. Add decent door hinges Rigid Door Hinge Voron or 2020 profile by BergKaese - Thingiverse (today: almost 400 Downloads, so should be acceptable) and you are done… Just as a proposal before you drink another couple of cases of beer!! You may check it, if you do the measurement in different Z-heights. Cheers!!

Finally i had some time to do something on printer.

After last test i printed new (different) tensioner for y axis. During replacing i found that y axis 2040 profile is very weak. In this printer its bolted to horizontal bar just by 4 bolts near center of y axis. But the end where tensioner is mounted have a play of about 2mm in z direction. Probably i will need to design and print some stiffener.

Anyway new graph with new tensioner looks even beter.

Its time to do some test prints.

I dont have any springs/silicone under bed. I mounted it directly to y carriage. In tarantula carrage have bended arms which also could acts as springs.

Hi @ReXT3D, have you mentioned those here?

Hi. I am not sure that I understand what you mean - you quoted me mentioning it here…

Very interesting, thanks for the links!

I previously fully characterized the resonance characteristics of my Voron by measuring at room temperature & then at stabilized chamber temperature, both at three deferent heights: just above the bed, mid height and just below the max Z travel. There naturally were differences but not significant enough to suspect the fame flexing. My Voron 2.4 is 300x300 spec in XY and 350 spec in Z:

Bed Height

Mid Height

Max Height

For kicks I also measured the resonances at room temperature on my print desk, on carpeted concrete floor and finally on concrete floor.

I do a lot of weird stuff with my printers to understand how they behave under different conditions while I set them up and fine tune them :slight_smile:


@ReXT3D Do you have a picture or a link from your silicone spacers? I wonder how they look like.

Kind regards, hcet14

That’s right, they look “ok”. What is completely different to mine Voron is that “visible” second peek at 67Hz. I have to review my original measurements to see, if the peek also exists on my system… But my feeling is, that my graph was quite “clean”…

#*# [input_shaper]
#*# shaper_type_x = mzv
#*# shaper_freq_x = 58.6
#*# shaper_type_y = mzv
#*# shaper_freq_y = 56.2

Funfact: For my old Anycubic Mega S, I put 1,5kg of lead bars to the top (the traverse connecting the two Z-Axis) and get rid of some ugly vibrations). I really though of doing that with my Voron,too… A nice 10kg steel plate on top? I start to like the idea… :thinking:

You can have them in black, green, yellow… :wink: But you will never know the shaw stiffness, so get them, love them or leave them…

1 Like

I tried a couple of them on two printers over the years (replacing the bed springs) and I found them to be highly overrated as they made it harder to level the bed and didn’t offer any kind of noticeable improvement in print quality. Of course, YMMV.


@Berggipfel @mykepredko

So stiffer springs are probably best for bed slingers?

Thanks a lot, hcet14

That was my experience.


Ordered a 25 KG granite test plate today for building further printers.
This could help as well when put on top. :smiley: