Autotune is not giving a proper gains for CUAV x7+ and taking long time for tuning

May I know what exactly affecting the angle as I am facing this issue for first time. I need guidance to better tune my drone.

You didn’t set the fixed notch filter as @amilcarlucas suggested. And posting logs of the Auto Tune flight is not very useful. Fix the Notch filter and make another flight w/o Auto Tune.

1 Like

Yeah I will set and do another flight test. Will it solve the angle issues? May I know at what conditions or situations should harmonic notch 2 enabled? and why?

If there is a mode of vibration that doesn’t correlate with Motor RPM. This might be the case with the ~!38Hz mode but all we have been looking at is Auto Tune logs which is not a good reference to set the filter.
Personally I would be manually tuning by now but that’s what I always do.

How good is the manual tuning compared to Autotune sir. And is manual tuning done by using the method transmitter based tuning? And do filter need to applied for manual tuning?

That depends on you PID tuning skills. Have you done PID tuning often? I assume you have no skills, so manual tuning will not be good.

yes.

Of course, otherwise it will not get any performance out of it.

As a beginner your best bet is to stick to the automated stuff provided by the ArduPilot Methodic Configurator.

The problem so far is that you have not followed the instructions (for example no fixed notch activated), so your results have not been great. Start following the instructions correctly and your results will improve.

1 Like

Thank you …
I prefer most of the time doing manual tuning, as in hill station area we do manual tuning for most of the drones.

For copter 4.3.6 and below we never used filters and tuned manually and those drones are flying good in fields but recently we started using filters and did autotune well. But recently we are facing issues like autotune is not doing good in angles and giving low gains pids due to filters.

So I did followed your suggested filter and did flight test. Soon I will be uploading the log file.

Agree with all of @amilcarlucas replies.

@amilcarlucas As per your suggestion I did a flight test and maneuvered in althold and loiter mode. Here is the log. May i know what is the next step?

The next step is imputing ALL these parameters into the 19_Notch_filter_results.param file using the ArduPilot Methodic Configurator GUI:

Once that is done and the parameters have been uploaded proceed to step 20.

@amilcarlucas Could you please brief me how did you arrived at these filters values, as I understand the filters play a crucial rule in better tuning the UAVs. And these are not mentioned clearly in documentation. Moreover I want to know more on when second notch filters to be enabled. Why not first notch filter solved the problem in this case. As this is the first time I enabled 2nd notch. Thanks in advance.

I poured many years of experience as a control engineer into the ArduPilot Methodic Configurator to automate tasks like this. I did it so that I do not have to explain myself over and over again.
There are many good books on the topic:

Discrete-Time signal processing - Oppenheim
Feedback control of Dynamic systems - Franklin

1 Like

Thank you so much… I will refer to those books. And will update on the future tests.

@amilcarlucas Today I did one more flight test as i see at 49hz there is one more spike which affecting the tuning. please have a look.

Yes, I see it.

Post picture of your vehicle, from all angles.

It looks like a harmonic of the very small 23Hz peak. I’d avoid putting an additional filter down that low though.
It’s odd that there’s a 23Hz peak and 35Hz peak, plus their harmonics, so close together. Usually this happens when there’s a motor output or two significantly different to the rest (like twisted motor mounts or significant weight imbalance). There is a slight difference in motor outputs (motor 3), but not much. It could be the issue…

Until you can find and fix that physical problem (if possible) use these settings:

for this result:

I guess those landing legs/motor arms are vibrating at different natural frequencies. Hard to tell without photos.
It’s an hardware issue that you are trying to solve in software instead of tackling the problem at the origin.

@xfacta As per your suggestion I did a flight test in loiter and althold mode maneuvers. here is the log, May I know is there any hardware issues so that I can fix for better tuning?

May I know the issues with the hardware…as we directly mounted motors to the arm and there is no twist in the motor mounts. We manufactured the carbon frame and arms inhouse. Please let me know any issues in hardware.

I am not an expert, but the four bolts which retain the arms to the main body are too close together. I would try and put them as far as possible, i.e. you can spread them by a few milimeters closer to the edge of the arms and certainly separate by 1…2 cm in the longitundinal sense.

Obviously, one can not tell visually if the frame is rigid or not, it may well be that it is rigid enough, but just by increasing the separation of bolts you could obtain roughly the double rigidity…

Also perforating arms at midpoint for the legs is the worst place where the bending stresses are st maximum. I would put the legs much closer to the motors. But again, that issue would not be the cause of tuning problems.