Page 617 - Mechanical Engineers' Handbook (Volume 2)
P. 617

608   Servoactuators for Closed-Loop Control

                                                        KKKA
                                                             1
                                                           a
                                                         p
                                                  K Lp            loop gain
                                                       KB   A 2
                                                         2
                             For the case where the servovalve dynamics are negligible compared to the servomotor-
                          load dynamics, Eq. (111) reduces to the following third-order differential equation:

                                        	Y    1                  1                           (112)
                                                              2
                                                           2
                                        	E c  K p  (1/K )(1/  )s   ((2  /  )s   1s   1)
                                                                    s
                                                                       ns
                                                     Lp
                                                           ns
                          In the steady state, Eqs. (111) and (112) both reduce to

                                                       	Y           1                        (113)
                                                      	E           K
                                                         c steady state  p
                          The integration in the forward loop results in a system with a zero steady-state error for a
                          constant input and a steady-state output that is dependent only on the system input and the
                          position feedback gain. (See Section 7 in Chapter 12 for a discussion of the following errors
                          for systems with nonconstant inputs.) That is, the accuracy with which 	Y follows the input
                          	E depends only on the accuracy of the feedback measurement device, and not on the
                            c
                          accuracy of the forward loop elements.
                             When the change in the control input is zero (i.e., 	E   0), the steady-state load
                                                                          c
                          sensitivity is

                                                    	Y             K 2
                                                    	F           K K K A                     (114)
                                                      L steady state  p  a  1
                          That is, an increase in the feedback gain results in a decrease in the load sensitivity or an
                          increase in the system stiffness. Likewise, an increase in the cross-port leakage in the ser-
                          vomotor or servovalve (i.e., increase in K ) results in a decrease in the system stiffness.
                                                          2
                             In most practical cases the roots of the quadratic in Eqs. (111) and (112) are conjugate
                          complex (i.e.,     1), or in physical terms, this portion of the system is ‘‘underdamped.’’
                                      s
                          Consequently, when position feedback is employed, limitations exist in the maximum value
                          of the position feedback gain that can be used while still ensuring system stability. Limita-
                          tions also exist in the input–output sensitivity and the system stiffness to load disturbances,
                          since these characteristics are dependent on the position feedback gain.
                             System relative stability can be viewed conveniently by employing the root-locus tech-
                          nique. Figure 48 illustrates root-locus plots for three important cases: (a) valve dynamics
                          modeled by a second-order differential equation, (b) valve dynamics modeled by a first-order
                          differential equation, and (c) negligible valve dynamics. These plots illustrate that the loop
                          gain must be set below some value in order to ensure stability.
                             When the dynamics of the servovalve are negligible compared to the dynamics of the
                          servomotor-load portion of the system, the system model is third order. Considerable study
                          has been made of third-order dynamic systems. The Routh absolute stability criterion can
                          be employed to determine the maximum value of the feedback gain that can be used and
                          still maintain stability. For the simplified model given by Eq. (112), the maximum value of
                          the feedback gain is
                                                                     2
                                                         2   (KB   A )
                                                     K     sns  2                            (115)
                                                      p
                                                             KKA
                                                               a
                                                                 1
                          and the corresponding maximum value of the loop gain is
   612   613   614   615   616   617   618   619   620   621   622