On-Chip Concurrent Device Aging Prognosis and Dielectric Failure Detection for GaN Power Devices

Lixiong Du,Yuanqing Huang,D. Brian Ma
DOI: https://doi.org/10.1109/tpel.2024.3413730
IF: 5.967
2024-07-19
IEEE Transactions on Power Electronics
Abstract:Despite promising figure of merits, as emerging wide bandgap devices, gallium nitride (GaN) power devices face significant reliability challenges. This article presents an on-chip condition monitoring (CM) approach addressing these challenges by seamlessly integrating device aging prognosis and dielectric failure detection into a unified circuit platform. Specifically, leveraging the device turn-on transient time (ton) as a reliable aging precursor, we propose a closed-loop ton sensing scheme with self-regulated sampling to enhance precursor readout accuracy while mitigating sensing delays. Furthermore, we improve CM accuracy through junction temperature (TJ) calibration, effectively eliminating temperature-induced effects on precursor measurements. Concurrently, we develop a gate leakage current (IGSS) based sensing scheme for dielectric failure detection, sharing the same circuitry of TJ calibration. Leveraging a proposed reconfigurable tri-mode gate driver, this CM approach minimizes impact on normal system operation. A power IC prototype was implemented on a 180-nm BCD process. Demonstrated on a GaN half-bridge power converter, the proposed CM exhibits the precise detection of both device aging and dielectric failure.
engineering, electrical & electronic
What problem does this paper attempt to address?