Autotune...tuning for dummies?

Since I don’t have a lot of experience tuning the Pixhawk and I’ve got 4 units to tune, I thought autotune would come to the rescue. My first unit is flying pretty strange. In stabilize it’s way more aggressive, which is great for me. But in Loiter, it’s corrections are extremely abrupt. When it corrects it literally snaps, look pretty unnatural lol! also, takeoffs are a little hairy. The motors are all spinning at different rpm’s and it’s pretty twitchy. But if you fly it like a DJI by punching it off the ground it’s fine. So should I just lower the Stb roll and pitch from 10.5 back toward it’s default 4.5? Or is there something else to consider?

Tarot 680 Hex
Sunny Sky 4108s motors
Afro 30A esc
Pixhawk, firmware 3.2
Props carbon 13-4

Hi ,

This might seem like a stupid question but have you calibrated the ESCs? Occasionally, with the Afro 30 Amp ESC, I have had the odd motor which seems to respond differently. What I usually do now is plug the motor ESC pair into the throttle output of the receiver I am going to use and perform an individual calibration. This also gives me the opportunity to change motor direction if required.

I also tend to re flash all the ESC to the same firmware before I install them. This lets me know that they are working and identical.

Using the auto-tune feature requires that the ship can already fly to some extent in the Stabilise Mode. This can normally be achieved by completing the basic setting up in the Mission Planner software.

Auto tune is very good at tweaking the PIDs but is not a substitute for the basic setup procedures.

Beyond that I have no idea :slight_smile:


I calibrated all the esc’s as you do from channel 3 on the multi’s rx. And with Afro’s, I flashed all of them with the latest firmware as the firmware supplied has sync issues with larger pancake motors. The new firmware fixes it and Afro’s are so easy to flash : ) Before autotune it flies pretty good so I have no idea either.

If you want a softer response you can reduce the STB_*_P gains back to 4.5.