2
\$\begingroup\$

enter image description here

This is the step response of the power stage of a buck converter. I tried tuning it when i added a proportional term, the rise time and settling time improved.So kept on increasing Kp and the response never started to oscillate. I added ki and the steady state error vanished for ki=1 but the rise time and settling time worsened (from microsecond range it became in seconds) Please any suggestions

\$\endgroup\$
4
  • \$\begingroup\$ Perhaps Kd is too high? \$\endgroup\$
    – Void Star
    Jun 1, 2013 at 22:15
  • \$\begingroup\$ Is this the open loop step response or closed loop with a specific set of P, I, and D terms? You want to start with the former, but from the ringing is looks like you have the latter. \$\endgroup\$ Jun 1, 2013 at 23:15
  • \$\begingroup\$ It is a open loop step response. \$\endgroup\$
    – user22348
    Jun 3, 2013 at 5:58
  • \$\begingroup\$ flickr.com/photos/93878172@N03/8932833729 This is the response to using only Kp and Ki and the rise time and settling time keep on decreasing but the response never oscillates \$\endgroup\$
    – user22348
    Jun 3, 2013 at 6:05

1 Answer 1

2
\$\begingroup\$

Please note the following: As you know, the PI controller is responsible in modifying the steady state response. That is improving the steady state values and settling time. You are simply adding a pole at the origin.

On the other hand, the PD controller will help you in improving the transient response. That is overshoot, rise time and peak time.

Now it is important to remember that adding a PI controller will 'usually' increase the order of the system which will cause a change in the previous steady state value.

Similarly, adding PD controllers will change the transfer function of your original system due to exposing the system to a zero. This means that the root locus 'if plotted' will be different.

Therefore; there will always be a compromise -trade off-. The best you can do is add margins of errors or inequality ranges for accepted values for both steady state and transient parameters.

Long story short, adding any controller to the system to improve the steady state will effect the transient and vice verse.

\$\endgroup\$

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