AntennaTracker 1.0 Tuning - CR Servo Deadband?


I am currently using a highly geared-down pan/tilt unit (7:1) in conjunction with high torque, relatively low-speed continuous rotation servos (HSR-264CRH) in both yaw and pitch. A Pixhawk 1 is driving the assembly, running AntennaTracker 1.0.

My antenna tracker configuration is “functional”, but I am having difficulty tuning the PID for yaw, to obtain sufficient response rate. Part of this is an apparent issue in which the tracker points within 20 degrees of target, such that the PWM output to the CR servos is very close to neutral (~1550-1450), for which the servos sit there and buzz without effecting any movement. And so, while there is no true electrical deadband, there is an effective deadband from 1450-1550us (as the servos do no provide significant torque at such a low input). This primarily influences the Yaw axis, as Pitch appears to track sufficiently well.

I have attempted to correct for this by tuning the yaw PID, but as of yet have not found the right combination of sufficient response rate, damping of over/undershoot, and system stability (lack of oscillation).

Aside from the standard approach to iterative PID tuning, has anyone had similar difficulties with slow tracking and steady-state error, and overcome them? I am willing to modify the source code if necessary, but the current Github source is quite ahead of the latest AT release (1.0).

I have included my parameter file.

TrackerParamPost.param (4.0 KB)

Thank you.

As an aside, compiling the current state of Antenna Tracker (GitHub source) results in an apparently broken build, that will not arm. Tracker is stuck in (SAFE) mode.

Does anyone know where I can find the source code for Antenna Tracker per the 1.0 release?