Stability of a class of hybrid computer models of dynamical systems

Hybrid computer programming automation research is giving rise to a renewed interest in standard hybrid programming configurations. The stability of implementations whereby only time integrations and first-order prediction compensation are performed in the analog section is examined as a function of: (1) the damping ratio ζ of any one of the conceptual model's second-order eigenvalues, and (2) the normalized sampling rate R, expressed in samples per cycle of its natural undamped frequency. Stability boundaries in the (ζc,R) plane indicate a great dependence of stability on C and the existence of an optimum value of compensation factor, also a function of ζ.