Filtering Derivative

[Arpad Toth]

Hi,

What are your thoughts on Filtering Derivative?

book Practical PID Control (Antonio Visioli) says:

"80% of the employed PID controllers have the derivative part switched-off (Ang et al., 2005))  Because it's hard to tune and the inherent amplification of the measurement noise represents a significant technological problem, because, if not properly filtered, it might cause a damage to the actuator.
 ..
 All these results confirm that the presence of the derivative filter in a PID
controller in ideal form cannot be neglected in general in the controller de-
sign phase (Leva and Colombo, 2001)."

One way to achieve it is to filter the feedback signal, but that can add lag, 

another is to filter the feedback signal only for D, either before or after D calculation.

Phil’s Lab did this:
https://youtu.be/zOByx3Izf5U?t=371
https://github.com/pms67/PID/blob/master/PID.c

Cheers

[relaxc Tuner]

Hi

it is a good question

science has made progress since 2005.
when controlling a process, there must be consistency between the expected position and its speed, otherwise there are overshoots.
100% of pid have  overshoots or too sluggish   (no active disturbance rejection). To have a well-balanced PID, there must be a D term.
The best is to filter the setpoint with 1 or 2 first order, which will avoid shocks on derivative and to calculate the term D with Relaxctuner.

But by dint of modifying the structure of a PID it is no longer called a PID

To have filters without delay, relaxc can do that too, see
https://www.researchgate.net/publication/347510415_Relaxc_vs_Kalman

best regards