A Practical Switch Condition Monitoring Solution for SiC Traction Inverters

Abstract

As automotive manufacturers move toward silicon carbide (SiC) MOSFET-based traction inverters, practical online switch condition monitoring solutions are crucial to address potential reliability concerns. In this article, an end-to-end practical online condition monitoring (OCM) solution is proposed. An online sensing circuit is proposed, which enables online ON-state resistance ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$R_{\text {ds-}\mathrm{\scriptscriptstyle ON}}$ </tex-math></inline-formula> ) measurement for all six switches of the inverter. To address the challenge of periodic data acquisition alongside higher priority motor control tasks, a fast, code-efficient out-of-order equivalent time sampling (ETS) technique is also proposed. The obtained periodic, high-resolution <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$R_{\text {ds-}\mathrm{\scriptscriptstyle ON}}$ </tex-math></inline-formula> data are filtered by a Kalman filter stage. With the proposed measurement solution, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$R_{\text {ds-}\mathrm{\scriptscriptstyle ON}}$ </tex-math></inline-formula> obtained at the motor current peak has an error of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${&lt; }1.5\%$ </tex-math></inline-formula> . Furthermore, the symmetrical nature of the inverter’s operation is exploited to propose a Bayesian inference solution for independent online state-of-health (SoH) estimation for all six switches. This technique isolates aging-related <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$R_{\text {ds-}\mathrm{\scriptscriptstyle ON}}$ </tex-math></inline-formula> change from operating conditions-related changes. In particular, by automatically accounting for device- and system-level variations in the model, the proposed Bayesian SoH estimation solution eliminates the need for extensive system/device specific calibration. The efficacy and robustness of the proposed solution are tested by inducing bond-wire failure in several decapsulated discrete SiC MOSFETs.