학술논문

An Accurate Predictive Method of Crosstalk Peaks Considering Dynamic Transfer Characteristics and Miller Ramp for SiC MOSFETs
Document Type
Periodical
Source
IEEE Transactions on Power Electronics IEEE Trans. Power Electron. Power Electronics, IEEE Transactions on. 39(5):5602-5613 May, 2024
Subject
Power, Energy and Industry Applications
Aerospace
Communication, Networking and Broadcast Technologies
Components, Circuits, Devices and Systems
Computing and Processing
Engineered Materials, Dielectrics and Plasmas
Fields, Waves and Electromagnetics
General Topics for Engineers
Nuclear Engineering
Signal Processing and Analysis
Transportation
Crosstalk
Silicon carbide
MOSFET
Logic gates
Voltage
Switches
Capacitance
dynamic transfer characteristics
miller ramp
mosfet<%2Fsc>s%29%22">silicon carbide (SiC) metal-oxide-semiconductor field effect transistors (mosfets)
Language
ISSN
0885-8993
1941-0107
Abstract
Silicon carbide-based metal-oxide-semiconductor field effect transistors (SiC mosfets) have been widely applied due to their lower on-resistance and parasitic parameters compared with silicon devices. However, the unexpected crosstalk phenomenon occurs and could lead to false turn-on or gate-oxide damage of the devices. In this article, an accurate predictive method of crosstalk peaks is proposed to provide theoretical guidance for crosstalk evaluation and suppression, which can improve the reliability of the systems. First, the generation of crosstalk and conventional predictive method are illustrated. Second, an improved crosstalk model considering dynamic transfer characteristics and Miller ramp is proposed. Furthermore, the dynamic transfer characteristics and Miller charge of SiC mosfets are obtained experimentally to realize the necessary supplement of the data provided by manufacturers. Finally, the crosstalk peaks predicted by the proposed method are compared experimentally with the actual measured values. The results have verified the proposed method can achieve accurate prediction of crosstalk peaks, and the relative errors of the predicted crosstalk peaks are below 10%.