There is currently a pid_compute and a pid_compute_1. The original pid_compute uses a subtle anti-windup strategy, while the *_1 variant uses simple saturation for antiwindup, with the limit equal to a hardcoded fraction of the output bounds. Neither version saturates the output.
There should only be one pid_compute, and it should either use a single antiwindup strategy or offer the caller a choice of strategies. Further, the windup limit shouldn't be hardcoded.
There is currently a pid_compute and a pid_compute_1. The original pid_compute uses a subtle anti-windup strategy, while the *_1 variant uses simple saturation for antiwindup, with the limit equal to a hardcoded fraction of the output bounds. Neither version saturates the output.
There should only be one pid_compute, and it should either use a single antiwindup strategy or offer the caller a choice of strategies. Further, the windup limit shouldn't be hardcoded.