Has anyone faced the same issue of not being able to damp down the Y-axis vibration measured by the MicoAir flight controller, such as MicoAir743v2-AIO-45A? The annoying part is that the x-axis and z-axis are well below the suggested peak values, 15 m/s/s. We have redesigned the frame three times and improved the fastening, still only the Y-axis captures high peak vibration, 40 m/s/s.
If you have, have you managed to solve it? How did you solve it?
Have you tried the PID notch filters, that are specifically there for the purpose of removing single axis frame resonance?
I suggest you delete your LLM generated post and refrain from it’s use here.
Thanks for taking time to read my issue.
I believe it is another definition that you are referring to: the Dynamic Harmonic Notch Filters. Yes, I did, unless I had set it up inefficiently.
We tried referencing the Vibration Damping page to see if Y-axis vibration can be further reduced, but were not quite lucky.
So, one of the examples of our Y-axis vibration issue. Arducopter 4.6.3.
No, PID notch filters (AMC step 27 and 28) are not the Dynamic Harmonic Notch filters (AMC step 25)
AMC explains that and explains how to use them. PID notch filters are not documented on the ArduPilot wiki yet.
Discussed here PID Notch
There is Webtools PR from me to add the DFF and the follow up that will make some of this more explicit. Would be great if @andyp1per and @iampete would review and merge them.
Thanks for taking the time to understand my issue and provide the guidance.
Do you know why @andyp1per in the video identifies the Roll vibration at around 307 Hz but he sets up the filter 1 at 180 Hz?
Thanks. I did not turn on the FAST IMU log, or else I would already have the flight data to identify the frequency for the PID notch filter.
Fly a short test flight again and inspect the logs in the Filter Review Tool before making a change if necessary. Filter Review Tool: is it referring to ArduPilot PID Review Tool or mavfft_pid.py? Are they the same?
Another question is, I am always confused: for Y-axis vibration, I should apply the filter (e.g., FILT1_) index to ATC_RAT_RLL_NEF, right?
Y is pitch. I would first try adding a notch to pitch before doing roll - often they are coupled and if you fix the most dominant the other goes away. Before you start adding PID notches make sure that your noise profile is as clean is you can make it - its easy to chase peaks that are actually the remants of poorly attenuated noise.
This is really confusing.
Yes, and thanks for the reminder.
So, I take it as set FILT1_NOTCH_FREQ = 80 Hz, and a second PID filter to suppress the vibration at 163 Hz and apply it to ATC_RAT_PIT_NEF first. Wrong, ATC_RAT_XXX_NEF can only points to one filter index.
Yes, that is a current limitation of the code.
You can only filter one peak out.
Interesting. Now, I have one question. So, in terms of vibrations, do the sensors pick up left-right movement along the Y-axis or rotation movement about the Y-axis?
So, based on the discussion, applying the filter to ATC_RAT_PIT_NEF to attenuate Y-axis vibration seems to suggest vibration about the Y-axis.
Accelerometers do the former and Gyro’s the later. Notch filters are applied to the Gyro signal.
