Different recommendations for X Input Shaping

Ran this command in console: GENERATE_SHAPER_GRAPHS AXIS=X The result was this:

Shaper calibration data written to /tmp/calibration_data_x_20230228_124812.csv file
**Recommended shaper_type_x = zv, shaper_freq_x = 45.6 Hz**
To avoid too much smoothing with '3hump_ei', suggested max_accel <= 7500 mm/sec^2
Fitted shaper '3hump_ei' frequency = 101.4 Hz (vibrations = 23.5%, smoothing ~= 0.080)
To avoid too much smoothing with '2hump_ei', suggested max_accel <= 7500 mm/sec^2
Fitted shaper '2hump_ei' frequency = 82.2 Hz (vibrations = 23.8%, smoothing ~= 0.080)
To avoid too much smoothing with 'ei', suggested max_accel <= 7500 mm/sec^2
Fitted shaper 'ei' frequency = 63.6 Hz (vibrations = 23.6%, smoothing ~= 0.080)
To avoid too much smoothing with 'mzv', suggested max_accel <= 7500 mm/sec^2
Fitted shaper 'mzv' frequency = 50.6 Hz (vibrations = 22.5%, smoothing ~= 0.080)
To avoid too much smoothing with 'zv', suggested max_accel <= 8100 mm/sec^2
Fitted shaper 'zv' frequency = 45.6 Hz (vibrations = 21.2%, smoothing ~= 0.080)
Calculating the best input shaper parameters for x axis

Shaper calibration data written to /tmp/calibration_data_x_20230228_124812.csv file
**Recommended shaper_type_x = zv, shaper_freq_x = 45.6 Hz**
To avoid too much smoothing with '3hump_ei', suggested max_accel <= 7500 mm/sec^2
Fitted shaper '3hump_ei' frequency = 101.4 Hz (vibrations = 23.5%, smoothing ~= 0.080)
To avoid too much smoothing with '2hump_ei', suggested max_accel <= 7500 mm/sec^2
Fitted shaper '2hump_ei' frequency = 82.2 Hz (vibrations = 23.8%, smoothing ~= 0.080)
To avoid too much smoothing with 'ei', suggested max_accel <= 7500 mm/sec^2
Fitted shaper 'ei' frequency = 63.6 Hz (vibrations = 23.6%, smoothing ~= 0.080)
To avoid too much smoothing with 'mzv', suggested max_accel <= 7500 mm/sec^2
Fitted shaper 'mzv' frequency = 50.6 Hz (vibrations = 22.5%, smoothing ~= 0.080)
To avoid too much smoothing with 'zv', suggested max_accel <= 8100 mm/sec^2
Fitted shaper 'zv' frequency = 45.6 Hz (vibrations = 21.2%, smoothing ~= 0.080)
Calculating the best input shaper parameters for x axis

Then, I used RatOS interface and ran calibration for X again. The result was this:

Fitted shaper 'zv' frequency = 36.6 Hz (vibrations = 3.1%, smoothing ~= 0.118)
To avoid too much smoothing with 'zv', suggested max_accel <= 5100 mm/sec^2
Fitted shaper 'mzv' frequency = 36.2 Hz (vibrations = 0.4%, smoothing ~= 0.155)
To avoid too much smoothing with 'mzv', suggested max_accel <= 3900 mm/sec^2
Fitted shaper 'ei' frequency = 39.4 Hz (vibrations = 0.0%, smoothing ~= 0.208)
To avoid too much smoothing with 'ei', suggested max_accel <= 2900 mm/sec^2
Fitted shaper '2hump_ei' frequency = 51.6 Hz (vibrations = 0.0%, smoothing ~= 0.203)
To avoid too much smoothing with '2hump_ei', suggested max_accel <= 3000 mm/sec^2
Fitted shaper '3hump_ei' frequency = 64.2 Hz (vibrations = 0.0%, smoothing ~= 0.199)
To avoid too much smoothing with '3hump_ei', suggested max_accel <= 3000 mm/sec^2
Recommended shaper is mzv @ 36.2 Hz

Fitted shaper 'zv' frequency = 36.6 Hz (vibrations = 3.1%, smoothing ~= 0.118)
To avoid too much smoothing with 'zv', suggested max_accel <= 5100 mm/sec^2
Fitted shaper 'mzv' frequency = 36.2 Hz (vibrations = 0.4%, smoothing ~= 0.155)
To avoid too much smoothing with 'mzv', suggested max_accel <= 3900 mm/sec^2
Fitted shaper 'ei' frequency = 39.4 Hz (vibrations = 0.0%, smoothing ~= 0.208)
To avoid too much smoothing with 'ei', suggested max_accel <= 2900 mm/sec^2
Fitted shaper '2hump_ei' frequency = 51.6 Hz (vibrations = 0.0%, smoothing ~= 0.203)
To avoid too much smoothing with '2hump_ei', suggested max_accel <= 3000 mm/sec^2
Fitted shaper '3hump_ei' frequency = 64.2 Hz (vibrations = 0.0%, smoothing ~= 0.199)
To avoid too much smoothing with '3hump_ei', suggested max_accel <= 3000 mm/sec^2
Recommended shaper is mzv @ 36.2 Hz

Same command but different recommendations?

6 Replies

miklschmidt•16mo ago

You say you ran GENERATE_SHAPER_GRAPHS AXIS=X twice, but that can't be right. You must have run SHAPER_CALIBRATE AXIS=X and then GENERATE_SHAPER_GRAPHS AXIS=X. That's why you're getting different results. SHAPER_CALIBRATE is using the preconfigured v-minion smoothing target of 0.08, which will yield slightly different results given the same resonance. Furthermore i seem to remember SHAPER_CALIBRATE uses a slightly different algorithm. Assuming it's the v-minion here. Also if it is, that is a super low frequency for X. Well, it's a super low frequency for x on all types of printers. Either your belts are super loose, your axes are swapped or your strapped a brick to your toolhead 😄

war4peace•16mo ago

Nah, the belts were loose, let me pull the new graph...

war4peace•16mo ago

I applied this for as follows:

shaper_freq_x: 74.8
shaper_type: mzv

shaper_freq_x: 74.8
shaper_type: mzv

Now, when printing a test model (simple empty cube with two walls), Input Shaper seems to work, but there are still two huge bumps right at the beginning. I'll post a picture tomorrow morning. And it's still unclear to me how to measure Y with the same ADXL, I know I have to place it on the bed, but do I have to design my own mount? And which orientation should the ADXL be placed in?

miklschmidt•16mo ago

Ooh much better! 🙂 It should fit on the belt tensioner Orientation doesn't really matter, as long as Y is Parallel to X,Y or Z and X is parallel to X, Y or Z, etc etc. Ie. don't mount it at an angle. If you want you can remap the axes:

[adxl345]
#axes_map: x, y, z
#   The accelerometer axis for each of the printer's X, Y, and Z axes.
#   This may be useful if the accelerometer is mounted in an
#   orientation that does not match the printer orientation. For
#   example, one could set this to "y, x, z" to swap the X and Y axes.
#   It is also possible to negate an axis if the accelerometer
#   direction is reversed (eg, "x, z, -y"). The default is "x, y, z".

[adxl345]
#axes_map: x, y, z
#   The accelerometer axis for each of the printer's X, Y, and Z axes.
#   This may be useful if the accelerometer is mounted in an
#   orientation that does not match the printer orientation. For
#   example, one could set this to "y, x, z" to swap the X and Y axes.
#   It is also possible to negate an axis if the accelerometer
#   direction is reversed (eg, "x, z, -y"). The default is "x, y, z".

war4peace•16mo ago

Thank you, I looked around but somehow missed the belt tensioner holes. I'll find some time tomorrow to do Y as well, then see what can be done with the bumps on both axes, if they are still there. I will also print the Klipper model.

Gaming

Programming

Different recommendations for X Input Shaping