Answers are grounded in this paper and cite their sources.
A current-limiting device (CLD) on a 400 Hz shipboard bus protects its load by turning an IGBT path off and diverting the current into a reactor. An IGBT that has failed stuck-on keeps supplying the load and appears healthy, so the loss of turn-off capability can be found only by testing the path in service, without creating a fault current and with the single Hall sensor already installed. This paper designs and validates such a self-test. Short high-impedance pulses at 45° and 225° force a current-path transfer in each of the two anti-series IGBTs, whose transient current drop ratio is the decision metric; a three-of-four criterion and a current-dependent deferral rule confine verdicts to currents the measurement chain can resolve. On production units, no healthy resistive-load capture failed at or above 7.5 A (0 of 25), and all three captures with a jumper emulating a both-device stuck-on fault failed. Fault cases above 10 A, where hardware injection was impossible, are evaluated in a chain model calibrated on healthy captures. A closed-form rule relating the lowest diagnosable current to the sampling period, sensor sensitivity and noise reproduces the measured 7.5 A limit and indicates how the test transfers to other hardware.
No takes yet. Share an insight, caveat, or question.
Park et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: