Minimum-Time Transient Recovery for DC–DC Converters Using Raster Control Surfaces

Fast power converters are desired in microprocessors, audio, and servomotor power supplies. Fixed-topology dc-dc converters have a minimum time in which to respond to line, load, and control change. This response depends on converter control, topology, and circuit parameters. Minimum-time control (MTC) considers these three items when it predicts postdisturbance steady-state points and directs capacitor voltage and inductor current in the shortest time. In this paper a nonlinear model-predictive control steers energy using geometrical curved control surfaces derived from piecewise linear (PWL) and bilinear models. The MTC differs from previous work on the topic in that it considers lossy components (a cause for nontriangular ripple) and is an open-form solution (as originally proposed by LaSalle in 1959). The different approach revealed a quantitative link between the needed control surface memory and state/parameter resolution. Results are generalized for two-state dc-dc converters and all parameter/control disturbances. An unconventional digital method was devised to accommodate the open-form solutions. It involved a priori control surface calculations, A/D scaling, surface quantizing, and image compression. The final control executed no real-time arithmetic operations. MTC performance was compared with other pulsewidth modulation controls and time-optimal control in simulations, and tests were performed in hardware using synchronous buck and boost converters.