### Garage PID tuning

Awww yes finally time to do some PID tuning.  After my last field test it was clear I needed to replace my motors with better quality ones and do some more PID tuning.  After making a few mods to the airframe for added support and installing the new Jdrones Purple 2830 /12, 850kv motors with 10x4.7 props it was time to start testing again.  In the first part of the video the quadcopter wobble front to back ( I was making it do this with the elevator stick) then get stuck at 45 degree, this is when I chopped the trottle and have a rough landing.   This was from setting the D to low .09 is way to low.  The last hover is with the PID settings below.

So far my best luck has been with this setting. But, I still think there is room for improvement.

The Yaw is stable, but not very responsive to what tell it to do.

Roll & Pitch   P =( .3)        I = (.024)        D = (.13)   Yaw  P=(.505) I=(.007) D=(.085)

• it would be absolutely amazing if this could be integrated into APM (automatic PID tuning) ;)
• Matlab has a system identification toolbox (typing "ident" in the command window will bring it up). There you can import input/output data, select specific segments of data to work with, and compute a transfer function. it can then display a simulation of the transfer function using another segment of data comparing the simulated and measured results.

Once you have a proper transfer function, tuning the PID with classical methods is easy. You could also use Matlab's sisotool (type 'sisotool' in command window) and then have it compute a PID automatically (which works adequately most of the time).

• Alex, can you explain how to tune PID gains using input / output data. do you calculate the transfer function with matlab ?
• Would you be able to log the the roll and pitch angles, as well the the throttle controls during a flight?

With the input / output data, i could try to calculate some PID gains using Matlab.

- Alex