Can someone help me to understand?

[winter roost]

attach the log

[humble ruin]

https://www.mediafire.com/file/km6hnit7z35ktqp/btfl_002.bbl/file

Here is the log file

[humble wolf]

Woow DTerm BIQUAD LPF1 + DTerm PT3 LPF2 --> Added a ton of delay on DTerm filtering and can cause PIDloop feedback oscillation.

Change DTerm Filtering strategy:

• Enable back your DTerm Filter slider
• Set your DTerm (dynamic) LPF1 and (static) LPF2 to PT1

[winter roost]

the signals to be below -30db
This isn't a strict rule; each setup has its own noise threshold below which filtering isn't necessary. Generally, it all depends on how the drone behaves: if the motors aren't overheating and there are no strange noises or vibrations, you can keep the noise levels as they are. Personally, I'd recommend not using BiQAD and PT3 filters unless absolutely necessary. It is also better to use PSD vs RPM plots - they are more informative

[humble ruin]

Thanks, that's very informative.

[humble ruin]

Thanks, I've got one more question, When filtering, do we aim to reduce the noise after filtering the gyro until it is invisible? I can't all the time to understand when to stop filter more, for example when I see more noise that are amplified by dterm I want to push dterm filters even more

For example

O can still see some noise, but I was flying pretty aggressive this time

When I hover or fly more smooth, there are is no very much dterm noise

I understand that a lot of other factors also have an influence, such as propeller oscillations, propwash turbulence, etc., I want to understand if this graph is supposed to be all black or what is it supposed to look like?

[humble wolf]

can you post your log (zip it) here... even if you added a shared link... (I am not a big fan of this).

[winter roost]

one can just rename it to .txt

[winter roost]

ah, even so, txt is for github)

[humble ruin]

Yep I see hah, I'm already ziping it

[winter roost]

no, just choose db value, -30 for gyro and -20 for D term is like perfect clean. but you can get a good tune with slightly higher noise values

[humble ruin]

Can you tell me is there electrical noise in my build based on this log? How can I see that?

[winter roost]

It is impossible to say with certainty whether the noise in the log is electrical or of another nature. The only things that can be said with certainty are the RPM noise and, in some cases, physical resonances from the frame or other components. Electrical noise is usually referred to as broadband background noise, especially in the low-frequency range of 0–100 Hz, but no one knows the exact source of such noise.

[humble ruin]

So when we have noise below - 30db on 50hz it's pretty electrically clear, I mean my exact build

[humble wolf]

I would suggest here:

• RPM MinHz: back to 100 + CLI faderange back to 50
• RPM W CLI: 100,50,80
• DNF: N=2, Q500, MinHz 100, MaxHz 600
• Gyro Slider set to 1.3
• DTerm Slider set to 1.2 (Dynamic LPF1 PT1 + Statuc LPF2 PT1).

-30dB on post-filtered Gyro signal is a "baseline", not a must have. -10dB for post-filtered DTerm signal same ("baseline").

Here you are pretty good. Indeed, your filtering latency is good for a 5" (as you did not mention anything about your current rig specification).
Do not chase the rabbit hole.

[humble ruin]

Well thanks a lot, I would also like to ask how to see if drone is clean build on the logs. Maybe you can send me some log example of 5 inch quad that is clean build?

And one last thing, when to stop filtering? What the ideal chart should look like

[humble wolf]

Your drone is pretty clean yes.