Page 77 - Modern Control of DC-Based Power Systems
P. 77

42                                  Modern Control of DC-Based Power Systems


             However, this PID controller has a transfer function with an increas-
          ing asymptote at high frequencies due to the derivative action [22].
          This in practice has a disruptive operation of the PWM due to the
          propagation of the switching ripple causing failure of the closed-loop
          converter even though the voltage controller has been correctly
          designed to have the required closed-loop performance. To avoid this
          malfunction, high frequency poles ω P1 and ω P2 must be present in the
          real voltage controller to reduce the gain of the controller transfer func-
          tion at high frequencies. This can be obtained with the following trans-
          fer function.

                                                   !
                                        K i   K d s     1
                          G c sðÞ 5 K p 1  1                          (2.74)
                                        s    1 1  s  1 1  s
                                                 ω P1    ω P2
             The first pole ω P1 flattens the gain of the derivative component, while
          the second one ω P1 allows the gain to drop. As a good practical design
          rule, ω sw   ω P1 {ω P2   10   ω P1 , where ω sw is the switching frequency
          in rad/s. Solving (2.70) and (2.71) does not lead to the solution because
          three coefficients need to be found, i.e., K p , K i , and K d , from two equa-
          tions. Therefore, to solve (2.70) and (2.71), the designer needs to set one
          of the coefficients and solve for the others. Typically, it is convenient to
          set the value of the integral coefficient K i . This is chosen relatively low,
          let us say 1{K i {10, to prevent antiwindup [23]. Finally, taking into
          account the voltage control transfer function (2.74), its control coeffi-
          cients are calculated as follows.

                                                                      (2.75)
                        K p 5 cos arg G c ðjω c VM Þ = G vd ðjω c VM Þ

                                                          K i
                      K d 5                           1               (2.76)
                              sin arg G c I ðjω c VM Þ

                                                              2
                                                        ω c VM

                           ω c VM   G i L d OL ðjω c PICM Þ
          2.5.1.3 Simulation Example
          A buck converter is now considered as an example. The buck con-
          verter switching frequency is 20 kHz, its input voltage is V g 5 400V,
          output voltage is V 5 200V, and circuit parameters are L 5 3.5 mH,
          C 5 50 μF, and R 5 30 Ω. A MATLAB script is provided in the
          Appendix that is able to perform the design of the controllers in VMC
          and PICM_FB.
   72   73   74   75   76   77   78   79   80   81   82