Hi all, I’m trying to get my QAV250 frame with a PixHawk 4 Mini flight controller to go through the AutoTune sequence. I get a couple recurring messages:
“AUTOTUNE: ROLL RATE D UP 0%”
repeated every couple seconds, followed by
“AUTOTUNE: FAILING TO LEVEL, PLEASE TUNE MANUALLY”
I don’t observer any obvious twitching going on. The current tuning seems fairly solid in Stabilized and AltHold, but I’d like to get AutoTune to dial things in tighter. I have it set to only attempt to tune the roll axis at the moment, since I don’t think the battery will last through all three axes.
Any ideas how to get AutoTune to run? I’d prefer that over manual tuning.
A 5", or whatever you have on that frame will not be Stable on the PID’s you have set and the result is oscillation on the Motor outputs and very poor initial tuning. I would reset to default and start over as this will never work:
ATC_RAT_PIT_I,0.135
ATC_RAT_PIT_P,0.04
ATC_RAT_RLL_I,0.135
ATC_RAT_RLL_P,0.04
Or:
Set those to .07
INS_ACCEL_FILTER to 10Hz
These to the tuning guide based on MOT_THST_HOVER:
PSC_ACCZ_I
PSC_ACCZ_P
Setup to configure the Dynamic Notch Filter:
INS_LOG_BAT_MASK,1
INS_LOG_BAT_OPT,4
Then do a AltHold hover flight for a minute or so and configure the Notch Filter as per the Wiki.
Auto Tune
Yea with P at MOT_THST_HOVER. I think it’s best to suggest a look at the Tuning Wiki as there is a lot of other valuable information there in the Initial Tuning Flight section before you get to those parameters.
Thanks for the suggestions, I was able to get the drone to enter Autotune. However, it visually appears somewhat unstable in flight still. I’m not sure if this is just to be expected or if it’s possible to tune out. Here’s a video of it in flight post-autotune, I’d be very interested to hear your thoughts!
Ah, an Auto Tune flight log. My bad for not being clear. A flight log of a flight after Auto Tune was completed. The actual log of the Auto Tune session isn’t that valuable unless you are trying to pull something out of of it.
Ah, I missed setting those parameters (PSC_ACCZ_I and PSC_ACCZ_P). I’m not able to set PSC_ACCZ_P equal to MOT_THST_HOVER though, since MOT_THST_HOVER = 0.13 and the minimum allowable P value is 0.2. So I set that P value to 0.2 and I set the I value to 2x MOT_THST_HOVER. This alone seems to have made the flight characteristics a lot more stable, thank you!
With the two harmonic notch filters, I was just trying to follow the process on the wiki and on Andy’s videos. The FFT showed two peaks on the first main peak, so I set each filter to be centered on each peak, and then followed the process through for each individual filter as described in the wiki. Would it be better to just set the first harmonic notch filter to be roughly centered on the overall peak?
That’s interesting to hear Andy, I didn’t realize ESC telemetry driven notches would be critical for a small drone. I do have the BLHeli32 ESCs. I’ll look into the ESC telemetry based tuning process and report back.
I’ll include a bin file of the new tuning flight, hopefully today.
Got it on the P value. I updated it to 0.13. Here’s a log of the IMU sampling flight from today.
I’ll move on to figuring out the ESC Telemetry based HNOTCH filter this evening, thanks everyone for your help!
The HNOTCH is working OK in it’s current form, but might struggle when you get into more dynamic flight, so do persist with changing to bi-directional DSHOT firmware and related changes.
