Short Circuit Withstand Time Power Device

# Short-Circuit Withstand Time

A power device that's supposed to be fully on, carrying normal load current, can suddenly find itself conducting the full supply voltage into what's effectively a short circuit — a load fault, a shoot-through event, any number of real-world failures that momentarily put the device under both high voltage and high current simultaneously. Short-circuit withstand time is how long the device can survive that specific combination before it self-destructs from internal heating, and it's a hard physical limit, not a design margin that can be engineered away — it's the window a protection circuit has to actually detect the fault and shut the device off before the device destroys itself.

Power dissipation during a short-circuit event is enormous precisely because voltage and current are both at their maximum simultaneously, which never happens during normal switching. In ordinary operation a power device is either blocking high voltage with near-zero current (off state) or carrying full current at a small on-state voltage drop (on state) — the two extremes of voltage and current essentially never coincide for any meaningful duration. A short-circuit fault forces both to their maximum at once: full supply voltage across the device while it's also carrying far more current than its rated on-state current, and that product — voltage times current — is instantaneous power dissipation orders of magnitude above anything the device sees in normal operation, concentrated in a tiny silicon or wide-bandgap volume with nowhere for that heat to escape fast enough.

Wide-bandgap devices generally have shorter short-circuit withstand times than the legacy silicon IGBTs they're replacing, which is a direct, non-negotiable consequence of why they're better devices in the first place. The same low on-resistance and thin active-region construction that gives GaN and SiC devices their efficiency advantage also means less silicon volume and thermal mass available to absorb a short-circuit fault's heat before junction temperature reaches a destructive level — a silicon IGBT historically offered several microseconds to tens of microseconds of withstand time, while many wide-bandgap devices offer meaningfully less. That's not a design oversight to be fixed; it's the direct physical tradeoff of the same thin, low-resistance structure that makes the device efficient, which means protection circuitry has to get correspondingly faster to keep pace.

Gate-driver-integrated desaturation detection is the standard fast-protection mechanism, because it senses the fault through the device's own on-state voltage rather than waiting for a current measurement to settle. A desaturation detector monitors the device's drain-to-source (or collector-to-emitter) voltage while the device is supposed to be fully on; during normal conduction that voltage stays low, but during a short-circuit fault the device is forced into desaturation and that voltage rises sharply — a detector tripping on that voltage rise can respond within the fault's first microsecond, fast enough to shut the gate off and clear the fault well inside the device's withstand-time window. Gate drivers for GaN and SiC devices specifically build this detection in with correspondingly tighter response-time specifications than legacy silicon gate drivers ever needed, directly matched to the shorter withstand times the devices themselves provide.

Avalanche ruggedness describes a power device's tolerance to a transient voltage overshoot event. Short-circuit withstand time describes something related but distinct — survival time under sustained simultaneous high voltage and high current — and together the two specs define the real-world fault envelope a power device and its protection circuitry have to be designed around jointly, not independently.