I haven’t done and ArduPilot tuning in a while - so maybe I’m just forgetting some things.
I did a test flight today on a 400” class quad-copter using a Cube Orange. I’m using bi-directional d-shot, so have rpm as the basis for my notch filter. I’m also using a 3 bladed prop.
For some reason, the Filter Review tool doesn’t allow selecting the Post filter data. And it has the “estimated post” filter selected by default - but it doesn’t show lines for that either.
I think I have all the parameters set properly - but maybe I’ve missed something. Below is a link to the log file on google drive.
I’d appreciate any suggestions. Thanks!
Note - my notch filter settings for this log file are still what I had set for a two bladed prop. Since this is for a three bladed prop, I need to change INS_HNTCH_FREQ. Hover rpm is 5100. I’d appreciate some guidance for setting INS_HNTCH_FREQ properly.
Logging is misconfigured. You have a case of “check all the boxes” with the Log Bitmask and are not taking proper advantage of raw logging with that Flight Controller.
Set the Log Bitmask back to default (disabling Raw IMU and Fast Harmonic Logging anyway).
Disable these:
INS_LOG_BAT_MASK, (set to zero)
INS_LOG_BAT_OPT, (set to zero)
And then enable this:
INS_RAW_LOG_OPT,9
AMC refreshes your memory with the correct sequence of configuration steps, and when only one correct value exists, like the ones above, automatically sets parameters to their correct value.
@jstroup1986 you think you are saving time by not using the ArduPilot methodic configurator.
There is more stuff that is not in the wiki, but is in AMC.
Would you please care to share your information about the errors in the AI report?
If my research is finding the wrong answer, it would be helpful to know the right answers. And everyone else reading this post would appreciate it too.
It’s hard to stay abreast of the rapid change in open source software such as ArduPilot. Some AI tools can really help finding information not otherwise easy to find.
I included the AI report to show I’d done due diligence on research. Undoubtedly there will be errors - as there are in some of these posts. In my opinion, it’s better to share what is found - and give others the chance to correct the record. After all, sometimes AI tools find published documentation that has errors - that might otherwise gone unnoticed.
People can easily do their own research using their favorite LLM. This is now a basic skill set. Filling up a thread with it is unnecessary and has always been frowned upon.
Quoting a relevant recent reply from a Developer “No one really wants to just talk to a robot with you as the proxy”.