Analysis of a GaN based PWM AC-AC converter with an improved switch loss model

Enis Baris Bulut, Mehmet Onur Gulbahce*, Derya Ahmet Kocabas

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

6 Citations (Scopus)

Abstract

AC-AC converters have a wide range of use in industrial applications mostly with installed silicon switches. In last decade, GaN's are served to market with their high efficiency and fast on-off times, hence implementation of GaN's to any conventional power electronic circuit is a focus of interest. In this paper, effects of both GaN and MOS switches on efficiency of an AC-AC converter were analysed both numerically and practically. A new extended mathematical loss model proposed by the authors was used to calculate the switching-conduction losses. Laboratory tests were performed with same physical circuit for both switches having the same package. Calculated and measured efficiencies and loss results which are in great harmony proved that the proposed mathematical loss model is extraordinarily realistic. All-in-all, it was observed that GaN and MOSFET switches have both high efficiencies (greater than 96.5%) at low switching frequencies, even though they operate at low power levels. With increasing switching frequency, GaN's efficiency is slightly reduced (less than 0.5%) while that of MOSFET decreased dramatically almost by 3%. Besides, it was seen that at low frequencies, reverse diode has a significant effect on efficiency of a fast switch, although switching-conduction losses are extremely low.

Original languageEnglish
Article number153578
JournalAEU - International Journal of Electronics and Communications
Volume131
DOIs
Publication statusPublished - Mar 2021

Bibliographical note

Publisher Copyright:
© 2020 Elsevier GmbH

Keywords

  • AC chopper
  • AC-AC converter
  • GaN
  • PWM
  • Switch loss

Fingerprint

Dive into the research topics of 'Analysis of a GaN based PWM AC-AC converter with an improved switch loss model'. Together they form a unique fingerprint.

Cite this