tombrazier
commented
1 year ago For anyone who does not want to delve into the Python script to find out how to use these calibration patterns, here is the blurb in the documentation inside the script:
This script generates gcode for one of three patterns to tune input shaping. Each pattern
performs a number of frequency scans, starting from 0Hz and going up to some maximum (default
60Hz). A frequency scan is a zig-zag line which is drawn at steadily increasing speed.
Horizontal lines are oscillating in the Y direction. e.g.
^^^^^^^^^^^^^^^^^^^^
And vertical lines are oscillating in the X direction. e.g.
>
>
>
>
>
>
>
>
Don't get them confused! You read X frequencies off the vertical lines and Y frequencies off
the horizontal line.
At the resonant frequency, the oscillations become noticeably larger and there may even
be layer shifts. If this is a problem, reduce the value of the amplitude variable.
The "freq" pattern draws one frequency scan for Y and then one for X. Measure from the start of
the line to where the oscillations are worst. Divide distance in mm by wavelength (default 2)
to get the frequency.
The "zeta_x" and "zeta_y" patterns draw a series of frequency scans all in one direction. The
zeta value increases by 0.05 for every frequency scan drawn, starting with a value of 0.05.
Find the line which has the most uniform amplitude of oscillation across the whole line. This
gives you the X or Y zeta value.
revetuzo
KimmoHop
github-actions[bot]
Is your feature request related to a problem? Please describe.
No response
Are you looking for hardware support?
No response
Describe the feature you want
When I get time this is what I want to add to
M593but if someone else gets there before me, that would be fantastic...The Input Shaping calibration pattern we
stoleborrowed from Klipper does not do a good job. It uses a lot of filament and takes a long time to print. But, worst, longitudinal oscillations are often a lot more visible on the pattern than transverse oscillations meaning the X and Y resonant frequencies are overlaid on top of each other the value measured for X is often actually the value for Y and vice versa.A much better alternative is to draw a zig-zag line at increasing speed and, thereby, performs a frequency scan on just one axis. This is a single layer print and separates X and Y very well.
I have written a Python script to generate gcode for this. This gives much more accurate results and also makes zeta calibration a lot easier.
Sample generated gcode files are:
Frequency scans in Y and X. Zeta calibration pattern for X. Zeta calibration pattern for Y.
It would be fairly trivial to incorporate these calibration patterns into
M593but I do not have the capacity for that at present. In the meantime, this FR records what needs to be done.Additional context
No response