Improved dynamics in DC-DC converters for IoT applications with repetitive load profiles using self-calibrated preemptive current control

摘要

This paper presents a novel approach to improve the dynamic response of inductive dc-dc converters in applications having repetitive load profiles. In many Internet-of-Things (IoT) applications, such as wireless sensor networks (WSN), the load current profile has a periodic nature, and is therefore predictable by the power management circuits. This unique nature is exploited by the proposed Preemptive Concurrent Controller (PCC) to achieve a dynamic response superior to the theoretical limits of time-optimal control. The preemptive controller ramps up the inductor current prior to the occurrence of a load step and reduces the required output capacitance. The non-inverting buck-boost converter is used in this work and operates with a freewheeling mode that avoids output voltage overshoot during the preemptive inductor current ramp. Two hysteric control loops operate concurrently to define the freewheeling interval. A simple digital calibration scheme is demonstrated to extract timing and amplitude features from a load current profile in order to optimize the timing of the preemptive current reference in the next cycle. Freewheeling is thus minimized to increase system efficiency. The PCC and associated load profile learning algorithm is experimentally verified and uses 10× less capacitance compared to the time-optimal control benchmark.