Consistent attitude offset in CF2.0

[droneguy]

Hey guys,
I tried installing the bootloader and new firmware on one of the Crazyflie 2.0s and now i see a persistent bias in the attitude.
The artificial horizon seems to reflect the manual tilting of the robot, but when kept on a level surface, the horizon shows the attitude bias again.
Note that this does not converge to zero when left as is. The bias is pretty stable which in my case is a 8 degree roll. (see image)

I have also tried resetting the eeprom using "python3 write-eeprom.py"

Also, here's the console output:

CF2_console.JPG (45.19 KiB) Viewed 2022 times

Any pointers?

[tobias]

I interpret it as the roll bias was not there before you updated the firmware? If you open the configuration dialog do you have zero for the trim values?

[droneguy]

Yes, and as I mentioned in my original post, I rewrote the default eeprom settings using the write-eeprom.py script.

In any case, here's the config window:

[tobias]

To me it then sounds like a software problem. Could you go back to the old FW and test? What new FW did you flash?