Why should we design a controller manually using Bode plots and control theory instead of using PI or PID controllers and tune them? Or more specifically when should I design a controller and when should I go for a PI or PID controller and tune them? I put forth this question especially for the case of designing a controller for Grid Connected Inverters, where I have seen many literature designing a controller for inner current loop and outer voltage loop and some have just used PI controllers.

  • \$\begingroup\$ Well, do you want to make design decisions up front OR, do you want to wing-it and fiddle with PID values until it's working? Both have their place. \$\endgroup\$ – Andy aka Jan 19 at 18:12
  • \$\begingroup\$ Stability of any system higher than 1st order has risk for overshoot and oscillation or instability. Trial and error PID is not safe for a GTI. you must analyze the sensitivity to what happens if gain drops to 0 from a saturated linear output and hysteresis. It necessary to understand how it responds to disturbances under different loads, which may affect the results greatly. Try this experiment. try driving backwards with a trailer using a delayed video on your phone. The instructor says, just adjust your PID gain in traffic. What is the sensitivity to anticipating or delaying error? \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Jan 19 at 18:31
  • \$\begingroup\$ I can understand the consequence of tuning PI or PID vs Manual controller design now. Then why are literatures still recommending designing PI or PR controllers for a GTI instead of manually designing one? What practice is used in industries for designing a Controller for GTI? \$\endgroup\$ – Aravind Khumar Jan 19 at 18:56
  • \$\begingroup\$ Some applications are amenable to PID + ZN rules, others are not. \$\endgroup\$ – Chu Jan 19 at 23:47
  • \$\begingroup\$ Calling your controller a PI or PID, and tuning it using control theory, are two different things. I've made a lot of controllers of the form \$H = k_p + \frac{k_i z}{z - 1} + k_d \frac{(1-d)(z-1)}{z - d}\$ that were tuned using measured or calculated plant responses and Body and Nyquist plots. \$\endgroup\$ – TimWescott Jan 20 at 22:01

Your Answer

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

Browse other questions tagged or ask your own question.