I am hoping somebody can offer me some advice on this. I have built a new Quad X frame (similar to the F450) in JSBSim as well as created a new SITL interface class. I have tested the motors and checked motor assignments and spin directions. All seem to be correctly setup. Frame type and class are set as follows:
FRAME_TYPE 1
FRAME_CLASS 1
I also went through the initial tuning, setting the following to spin up the motors after arming, but not enough to take off.
MOT_SPIN_ARM 0.01
MOT_SPIN_MIN 0.012
Again, everything appears to be working. I have set the arming checks to require GPS and when arming the motors, this failsafe appears to work correctly.
Next, I set the mode to guided, arm the throttles, and then issue a takeoff command to 1m. The engine RPMs increase the drone lifts off, but then quickly starts to oscillate and crash.
Following the tuning guide, I have iteratively reduced the following values by 50% and repeated the process. Each time, the drone does the same thing… starts to oscillate and then crashes.
I started with the following values:
ATC_RAT_PIT_P. 0.135000
ATC_RAT_PIT_I 0.135000
ATC_RAT_PIT_D 0.003600
ATC_RAT_RLL_P 0.135000
ATC_RAT_RLL_I 0.135000
ATC_RAT_RLL_D 0.003600
And then reduced them by 50% until I reached the minimum values as follows:
ATC_RAT_PIT_P 0.01
ATC_RAT_PIT_I 0.01
ATC_RAT_PIT_D 0.0
ATC_RAT_RLL_P 0.01
ATC_RAT_RLL_I 0.01
ATC_RAT_RLL_D 0
But the response is the same, rapid oscillations and then a crash. The airframe is almost identical to that of the F450 except that I have added 3.8 hp electric motors and a matched 18in prop.
I am not sure what else to look at to fix the oscillations or if this may be a problem with the airframe in JSBSIM. Any help would be most appreciated.
One thing I did notice is that the F450 model has a simulation of the GPS, barometer, and IMU. These sensor are incorporated into the JSBSim configuration. I am not sure if when running Arducopter in SITL mode these external sensors are necessary as I think arducopter code is already providing these components. Is it possible that these simulated sensors are affecting the controllability of the vehicle?
Here is a video of the last flight with the pitch and pro PID values set to their minimums.