Discovery of Loss Imbalance in SiC Half-Bridge Power Modules – Analysis and Validations

Benjamin Futtrup Kjærsgaard*, Gao Liu, Thore Stig Aunsborg, Dipen Narendra Dalal, Jannick Kjær Jørgensen, Bjørn Rannestad, Hongbo Zhao, Stig Munk-Nielsen

*Corresponding author for this work

Research output: Contribution to journalJournal articleResearchpeer-review

109 Downloads (Pure)


It is commonly assumed that power semiconductor switching losses are the same for high-side and low-side devices in a half-bridge power module. However, this article reveals that the high-side SiC mosfet in a medium-voltage power module exhibits over 40% higher switching energy compared with the low-side SiC mosfet. The loss imbalance is attributed to the parasitic high-side gate capacitance in the power module, which contributes to the equivalent high-side Miller capacitance. A physics-based switching energy dissipation model is, therefore, proposed, distinguishing between low-side and high-side switching energy dissipation. Double pulse testing demonstrates that high-side switching energy dissipation increases by 5 mJ per 2.5-pF increment in the equivalent Miller capacitance, aligning closely with the analytically calculated increase of 6 mJ per 2.5 pF. Continuous power module testing shows a 10 °C increase in high-side junction temperature. The findings from this article offer crucial insights into research, design, and manufacturing of half-bridge modules enabled by SiC MOSFETs.

Original languageEnglish
JournalI E E E Transactions on Power Electronics
Issue number5
Pages (from-to)5806-5819
Number of pages14
Publication statusPublished - 1 May 2024


  • Capacitance
  • Capacitive couplings
  • Couplings
  • Double pulse testing
  • Half-bridge switching dynamics
  • Medium voltage SiC MOSFETs
  • Miller capacitance
  • Multichip modules
  • Power module modelling
  • Silicon carbide
  • Switches
  • Switching loss
  • Switching losses


Dive into the research topics of 'Discovery of Loss Imbalance in SiC Half-Bridge Power Modules – Analysis and Validations'. Together they form a unique fingerprint.

Cite this