I’m confused.
Your response states:
"Typical signs of underdamping are slow convergence to the setpoint and no overshoot."
I thought that it was the exact opposite, that overdampened prevents overshoot (desirable), but slows convergence (as an unfortunate side effect).
RR
Duh… silly of me. You are correct in that overdamped systems don’t overshoot. I got confused there.
But I don’t agree that an overdamped system is a good thing. It can make response unnecessarily slow and it may even fail to track the input.
A better thing is to have a critically damped system, but that can only happen with up to 2nd order systems (if memory serves me right), which in our case is a a false assumption.
No problem. I think we’re saying the same thing. I plan to make some modest adjustments before the next flight:
I will reduce pitch damping (PTCH2SRV_D) by a small amount. (currently at .063, will set at .05)
I will increase PTCH2SRV_TCONST by a small amount (currently at .45 sec, will set at .6)
I’ll respond back here with the results.
I’ll also keep looking for better log graphing tools. MATLAB looks pretty neat and I see that they have an analysis plug in for control systems, including PID loops. But my experience with programs of that complexity is that the learning curve is high.
RR
@RogerR I think you should also consider increasing the throw on the tail servos. So increase SERVO2_MAX to 2000, SERVO4_MAX to 2000, SERVO2_MIN to 1000, SERVO4_MIN to 1000.
You have some odd throws at the moment btw. SERVO2_MIN is 1187 and SERVO2_MAX is 1852, whereas the other tail servo is 1100/1900. Is there some reason for that? You want it to be symmetric.
Tridge,
In the Full Parameter List page on Mission Planner, I see the min for both SERVO_2 and SERVO_4 shown as 1000 and the max as 2000.
RC2 and RC4 also have the same values. I also see logged values of 1005 to 2000 for RCOU.C2 and 1071 to 2000 for RCOU.C4 in the log file we’ve been using (I assume that the logging of SERVO values is using the old RCOU names).
I think you’re looking at Servo 12 and servo 11. One of those has no function, the other controls the servo that drops my landing gear 
RR
If you want to make a control system faster, you must lower its time constant. The time constant is a characteristic time interval of the system’s response. Short time constant leads to fast response.
But this would be the last thing I would change, because usually this parameter affects every other aspect of the PID controller tuning and might cause unexpected changes.
Makes sense, I’ll put the TCONST back at .45.
Thanks!
RR