11.30 – 11.55
This presentation for Electrometric by Eindhoven University of Technology showcases an intelligent, self-learning gate driver for SiC mosfets. It explores how real-time feedback and lightweight learning algorithms can be used to automatically shape switching waveforms in SiC power converters. This allows the system to continuously improve EMC performance without requiring device models or manual tuning. The talk highlights the integration of multi-level digital gate-drive hardware, embedded optimization running on FPGA/MPSoC platforms, and reinforcement learning-inspired exploration strategies. It will be shown that such an adaptive gate drive is able to adapt to temperature changes, load variations, and device aging by improving the switching behavior of the mosfets. This allows the driver to adapt to temperature changes, load variations, and device aging by improving the switching behavior of the mosfets.The presentation will show measured results, which illustrate how the system actively minimizes overshoot and ringing when operating conditions change.
Speaker: Nart Gashi (Eindhoven University) on behalf of Electromatic