Power Segmentation Optimum: a novel technique for continuous optimal efficiency tracking in Monolithic DC-DC converters

This paper presents a novel mixed-signal architecture for the extraction and online tracking of the maximum efficiency point in generic integrated DC-DC converters, called Power Segmentation Optimum. This optimal condition is achieved by dynamically adjusting the size of the power switches to minimize total losses. Proposed system continuously tracks the optimal efficiency point under all operating conditions, adapting to both external factors (such as input/output voltage, switching frequency, temperature and output current) and internal parameters (such as MOSFET ON-resistance and gate capacitance). The operating principle consists in a A/D mixed signal implementation which continuously extracts and compares a proportional value of each loss contribution, deriving the corresponding configuration which minimizes the total losses. This technique is particularly suitable for IoT applications, characterized by long idle periods and high current pulse transients. In order to validate the theoretical results, proposed method is implemented inside a 5A DC-DC fully-integrated buck converter realized in 180nm technology with 5V input voltage rating and switching frequency up to 4MHz.