Skip to main navigation Skip to search Skip to main content

Self-Adjusting Feed-Forward Ripple Cancellation in a Hybrid GaN/Si Interleaved Inverter

  • Mowei Lu
  • , Jurgis Reinotas
  • , Xiaoyang Tian
  • , Yejing Fan
  • , Nima Tashakor
  • , Florin Udrea
  • , Stefan M. Goetz*
  • *Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

In hybrid interleaved wide-bandgap (WBG) / silicon (Si) structures, the Si part could shoulder most of the load, while WBG contributes only fractionally and primarily cancels out the current ripple of the Si bridge. However, such a mode required perfectly known inductances of each bridge. It also needed accurate ripple measurement and fast closed-loop control with extreme demands on the sensing, filtering, and control bandwidth. Additionally, previous use was limited to dc conversion, where the quasi-stationary output voltage simplifies operation. This paper proposes a novel control strategy for a hybrid GaN / Si interleaved inverter, which uses low-bandwidth feed-back control for fundamental current sharing in combination with a feed-forward term for ripple cancellation. This approach decouples high-frequency current ripple from low-frequency fundamental currents and achieves fast response and efficient ripple cancellation without the excessive demands of full-bandwidth feedback control. Furthermore, we introduce a self-adjusting or auto-tuning algorithm. It dynamically updates the control coefficient through real-time ripple indicators to adapt to inductance variations and ensure optimal ripple cancellation.

Original languageEnglish
JournalIEEE Transactions on Power Electronics
DOIs
StateAccepted/In press - 2026
Externally publishedYes

Keywords

  • Interleaved switching
  • low-bandwidth control
  • pulse-width modulation
  • ripple compensation
  • wide-bandgap transistor

Fingerprint

Dive into the research topics of 'Self-Adjusting Feed-Forward Ripple Cancellation in a Hybrid GaN/Si Interleaved Inverter'. Together they form a unique fingerprint.

Cite this