I would like to bridge the gap between the two grayed out zones (I’m assuming the first and second harmonics) as I see some noise there as well. I was under the impression that the INS_HNTCH_BW parameter defines the notch bandwidth and thus would widen the gray zones if I increased it. But apparently this isn’t the case as I see no impact of this parameter on the notch visualization when I change it. Could someone please explain how can I bridge this gap, do I need to set up 2 notch filters (I have an F4 FC)? And what exactly is it that INS_HNTCH_BW controls.
Here is my last log for reference. I haven’t auto-tuned my vehicle yet.
P.S: I created a topic with the same name and tried to delete it due to errors in the post but I’m not sure if the deletion fully succeeded as I can still see the topic on the forum. Apologies if I’m breaking any forum rules!
You dont necessarily need to aim for bridging that gap. The trick is looking for the noise that got past the filters, INS_GYRO_FILTER being the primary low-pass filter. See the estimated or post-filter lines on the graph for what actually needs additional filtering.
From the log, change INS_HNTCH_FREQ to 90
and INS_HNTCH_HMNCS to 1 since you dont need the second harmonic at all.
Thanks Shawn, I will try with a single harmonic. Just out of curiosity could you please explain what INS_HNTCH_BW does? The ardupilot documentation just mentions that it is the notch filter bandwidth.
It was quite gusty today when I flew, could it possibly be due to the imperfect tuning combined with these gusty conditions? Otherwise, in your experience, do you have any ideas on what could possibly be causing noise around such relatively low frequencies?
Usually that low frequency would be from very large props (and suitable motors) or something like a cooling fan on a piece of equipment.
It’s not there or quite low at the start of the flight, so it got worse. Maybe it is related to wind, but that would be in conjunction with something vibrating because of the wind.
The notch filter attenuation profile is a V, and not a |_|. Either way, the bandwidth determines how wide about the center does the filter affect. This is a consideration when using a filter that tracks a state that changes over time, like the throttle. Not only will the filter centerline track the peak, but also 1/2BW on either side of the peak. At low throttle, your filter will be applied as low as centerline-1/2BW. It’s important to watch this and make sure you understand how low the filtering will be applied, the lower the filtering the more directly you are muting real craft motion as opposed to noise.
I haven’t really been able to determine what the gray zones represent when you enable tracking. The black line doesn’t show the filter’s centerline, but perhaps the average position in the filter over the timespan you have selected, with the min and max (of the centerline) on either side of the gray box? But I don’t think it includes the ranges of the BW