Fully digital hysteresis modulation with switching time prediction

This paper proposes a digital hysteresis modulation technique based on switching time prediction. Sampling controlled variables several times within a switching period, it ensures a dynamic performance comparable to that obtainable with an analog hysteresis modulation. Compared to conventional digital hysteresis modulation, it avoids frequency jitter since it predicts switching transitions. Compared to hysteresis modulation based on the detection of the zero-crossing of current errors, it avoids external analog circuits. Compared to PWM techniques, it ensures faster dynamic response. These advantages are obtained at the expense of an increased signal processing requirements and of control complexity. Switching frequency stabilization and synchronization with an external clock can be achieved extending the techniques proposed in the past for analog hysteresis modulations. The proposed predictive algorithm does not require knowledge of load parameters and only a rough estimation of the inductor filter, which can be easily self-adjusted. The proposed solution is suited for high-performance current (or sliding-mode) control where the digital hardware has enough computational power to allow multiple samples within a switching period. The proposed modulation technique has been applied to a sliding-mode control of a single-phase uninterruptible power supply (UPS). Experimental results confirm the effectiveness of the proposed approach.