Alleviating oscillations in GaN power circuits through active gate drive control and evolutionary multi-objective optimization

Abstract

With GaN power devices being able to switch at rates in excess of 100 V/ns, this rapid change inevitably drives high 𝑑𝑖⁄𝑑𝑡, which induces voltages over the various sections of the parasitic loop inductance in a power loop, triggering ringing, which is a source of unwanted electromagnetic interference (EMI)—making adaptation and integration of GaN switching devices challenging. This thesis explores the possibility of deliberately increasing loop inductance to cause a partial reduction of switching loss and the use of an active gate driver to alleviate parasitic oscillations induced by the rapid changing turn-on current. In this work, a simulation model of an active gate driver is developed for the dynamic transient shaping of a hard-switched GaN bridge leg circuit. This work provides a strategy to use an Evolutionary Multi-objective Optimization (EMO) algorithm to discover gate drive patterns across Pareto-optimal solution space using a multi-objective Nondominated Sorting Genetic Algorithm (NSGA-II). Three key measurable indicators are used to give each gate pattern a performance metric. These include: GaN device drain current overshoot, energy loss, and current ringing spectral analysis with Fast Fourier Transform (FFT). Both ideal and additional parasitic inductance circuit configurations are compared with a discreet gate pull-up resistor and active gate drive controller. Subsequent use of inequality constraints allows to refine search space by
circumscribing the optimization region, resulting in a faster convergence towards higher impact switching optimization strategies. The achieved results reveal that the active gate driver controlled by an NSGA-II is capable of finding unique gate drive patterns, improving current overshoot, switching losses, and compensating for additional parasitic loop inductance by reducing drain current ringing.

Date of Award	29 Sept 2022
Original language	English
Awarding Institution	University of Bristol
Supervisor	Bernard H Stark (Supervisor) & Saeed Jahdi (Supervisor)