0
\$\begingroup\$

In PID control an integrator action is represented by \$\dfrac{1}{s}\$

enter image description here

Does anyone know what would be the effect of using \$\dfrac{1}{s+k}\$ instead of \$\dfrac{1}{s}\$, where \$k\$ is a constant? What would this controller be called?

\$\endgroup\$
2
\$\begingroup\$

Expanding on Ben's answer, let's presume we construct a PI with the integral replaced with a single pole low-pass filter. The TF for our controller is $$C(s) = K_p + \frac{K_f}{s+k} = \frac{K_ps+kK_p+K_f}{s+k},$$ so you see you've essentially turned your PI into a Lead-Lag compensator with pole/zero combo $$p = -k,\ \ z = -\left(k+\frac{K_f}{K_p}\right).$$ The main difference between PID controllers versus Lead-lag compensators is whether or not you want to regulate the time (PID) or frequency (lead-lag) domain responses.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ interesting, do you know the general form of HPF and if they are similar? \$\endgroup\$ – Bajie Jun 7 '16 at 4:21
2
\$\begingroup\$

"Single-pole lowpass filter", typically.

The difference is that the lowpass filter has finite gain (1/k) at low frequencies, while the integrator theoretically has infinite gain at DC.

| improve this answer | |
\$\endgroup\$
2
\$\begingroup\$

One reason for using an integrator in the forward path of a closed-loop system (if integration is not present, inherently) is that it gives unity 'DC gain' or 'steady-state gain'. That is, the controller's I term forces the system steady-state error signal to zero for a step input. And, of equal importance, the I term also gives complete rejection of steady-state disturbances. A lead-lag controller or a simple first order lag, \$\large\frac{1}{s+k}\$, has neither of these attributes.

For example, a velocity control loop, where the controlled device is a DC motor, does not have an inherent integration (apart from the trivial zero-friction/pure inertial load case). So the controller would require integral action in order to ensure that there is no offset between required and actual velocities for a steady input command, and that any external constant disturbance will be reduced to zero.

However, the same motor in a position control loop would not require controller integral action in order to provide those desirable features, as there is an inherent pure integration built-in to the motor: displacement is the integral of velocity.

| improve this answer | |
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.