Motor overshot after autotune

Hi
copter was fly good with default PIDs without any motor overshot but after autotune i can hear we we we we sound from motors and its visible in RCOUT values

current PIDs

what i did before autotune :
-finished everything in tuning process instruction page
-configured harmonic notch filtering

config :
AirGear 450 motor set (2216 880kv - 20A ESCs - 1045 prop)
daya carbon fiber frame 550 size
CUAV Pixhawk 1 v2.4.6 with latest stable Arducopter

log with default PIDs
Autotune log
Test flight after autotune log

@Leonardthall could you please look at logs
every help will be appreciated
thanks.

I need a log with these settings. It looks like your pitch D term is too large. I would back it off to your roll settings.

1 Like

OK
i backed pitch D term off to roll and did a test flight
motor overshot has disappeared but it seems that i have noise on my pid values

this the testflight log wit requested log bitmask
@Leonardthall

Yeh, there is a 63 Hz oscillation in roll. This is also primarily from the D term.

It looks like you have done an auto tune but then copied the D term between roll and pitch but not the P or I term. What was the Autotune values? Based on your picture above it looks like you copied roll to pitch. Maybe that is why we are seeing this problem only on roll. You have already decreased the D term on pitch. I confirmed from your autotune log that you have copied the Roll D to the Pitch D.

Your harmonic notch isn’t quiet setup correctly:
INS_HNTCH_BW 42
INS_HNTCH_ATT 40

This may be partially because your harmonic notch frequency or REF isn’t quite right.

I would suggest going back to your autotuned values.
Drop both the angle P for roll and pitch to 10.
Then backing off your PID values on roll by 25% steps until the problem goes away.
You should double check in the logs that you don’t have pitch oscillations too with the autotuned D term.

This is what they look like:

You can see that it is mainly in roll but you may see it in pitch too with higher Pitch D term. In that case you would need to reduce Pitch in steps of 25% too.

Good luck and let us know how you go.

1 Like

Hi @Leonardthall
i did your suggestion again and now everything is now much better
copter is completely stable and flies good
but still i have a little bit motor overshot and i was couldn’t resolve this little overshot by playing with PIDs


i can see that its on pitch

do you have any suggestion for me to resolve this little overshot ? (its affecting my camera)

log file

thanks.

The question is what is causing this?

You could back off the pitch parameters by 25% and 50% and see if that improves things.

If it is a resonance in your frame or payload mount you could use a fixed notch to attenuate it.

This is where things start getting tricky.

1 Like

ok i will try again , i know my weakness in manual tuning and i’m trying improve it by your helps

totally everything on frame is fixed, camera attached by a vibration damper to frame

and i have question, is it true to say that with a full square frame we must have same PID values on both roll and pitch ?

My much less informed opinion is to set INS_GYRO_FILTER,20 and do another test flight and lets see the .bin log from that.
AUTOTUNE_AGGR,0.1 if you are going to run autotune again

INS_GYRO_FILTER is already 20

i did this may times but no luck
manual tune.param (17.2 KB)
today i manually tuned copter with these values
then did an autotune but new values very high again and copter oscillating
it seems that autotune will not help on this build but i will try backing off all PIDs generated by autotune with 10% steps to get good PIDs

OK finally tuning has been finished on this build and now copter fly good as possible
first autotune generated this PIDs


and this is my manual tune

log file
@Leonardthall i dont know that these information could help or not but I would be happy if i could help to improve ardupilot

Well done mate. I am glad you got something that is working well enough!

1 Like