Transistor Module Overvoltage Protection Control: Methods That Actually Save Your Devices

Overvoltage kills transistor modules in ways that overcurrent does not. A current fault gives you time — microseconds, maybe tens of microseconds — to react. An overvoltage event can destroy a gate oxide in nanoseconds. There is no recovery from that. The device is dead before your protection circuit even wakes up.

This is why overvoltage protection for transistor modules is not a nice-to-have feature. It is a survival requirement. And the methods that work in practice are very different from what you read in textbooks.

Where Overvoltage Actually Comes From in Transistor Modules

Most engineers picture overvoltage as something that happens on the input side — a bus spike from the supply. That happens, but it is not the main killer. The real overvoltage threats come from inside the power stage itself.

When a transistor turns off, the current flowing through the circuit has nowhere to go instantly. The parasitic inductance in the loop — from PCB traces, bond wires, and module internal wiring — fights the current change. The result is a voltage spike on the collector or drain that can reach two to three times the bus voltage. For a 600-volt IGBT module, that spike can hit 1800 volts. For a 1200-volt SiC MOSFET module, it can exceed 3000 volts.

The second source is the Miller effect during turn-off. As the drain-source voltage rises, the Miller capacitance couples that voltage back to the gate. If the gate drive impedance is too high, the gate voltage rises with the drain voltage. The transistor can partially turn back on — a condition called false turn-on or parasitic turn-on. This creates a shoot-through current spike that destroys both devices in the bridge leg simultaneously.

The third source is inductive load kickback. When you switch off a motor winding or a transformer primary, the stored magnetic energy generates a voltage spike with no current path. That spike appears directly across the transistor and can easily exceed its breakdown rating.

Passive Snubber Circuits: The Old Reliable

Passive snubbers are the first line of defense against turn-off voltage spikes. They absorb the energy that would otherwise manifest as overvoltage.

RCD Snubber Across the Transistor

The most common passive snubber is an RCD network — a resistor, capacitor, and diode in series — connected from collector to emitter (or drain to source). When the transistor turns off and the voltage spikes, the diode conducts and charges the capacitor. The capacitor absorbs the spike energy. The resistor dissipates that energy on every cycle.

The capacitor value is typically 1 to 10 nanofarads for IGBT modules and 0.1 to 1 nanofarad for SiC MOSFETs. The resistor is 10 to 100 ohms. These values must be tuned to your specific operating point — bus voltage, switching frequency, and parasitic inductance.

The trade-off is real: the snubber clamps the voltage spike, but it dissipates power every switching cycle. At 20 kHz switching frequency, even a small snubber can waste several watts. For high-efficiency designs, this loss is unacceptable. But the alternative — no snubber — means the transistor sees the full spike every cycle, which shortens its life dramatically.

RC Snubber on the Gate

A small RC network across the gate resistor slows down the dV/dt seen by the gate during turn-off. This prevents the Miller capacitance from pulling the gate voltage up and causing false turn-on.

Typical values are 10 ohms in series with 100 picofarads. The resistor must be small enough not to slow down normal switching. The capacitor must be large enough to shunt the Miller current but small enough not to load the driver.

This is a cheap, effective fix for false turn-on problems. It does not protect against collector-emitter overvoltage — it only protects the gate. Use it in combination with an RCD snubber, not as a replacement.

Active Clamping: Faster and More Efficient

Passive snubbers waste energy. Active clamping recycles it.

How Active Clamping Works

An active clamp uses a secondary transistor and a clamp capacitor to absorb the turn-off spike and return the energy to the supply rail instead of dissipating it in a resistor. When the main transistor turns off and the voltage rises, the clamp transistor turns on and diverts the current into the clamp capacitor. The capacitor voltage rises, but the main transistor voltage is held to a safe level — typically the bus voltage plus the clamp capacitor voltage.

After the spike passes, the clamp transistor turns off and the clamp capacitor discharges back into the bus through a diode or a resonant circuit. The energy is recovered, not wasted.

This approach is common in high-frequency DC-DC converters and resonant topologies. It gives you hard clamping with soft recovery — the voltage spike is limited to a precise value, and the energy goes back into the system instead of heating up a resistor.

The downside is complexity. You need a secondary driver, a clamp capacitor sized for the worst-case fault energy, and a control circuit that times the clamp transistor turn-on precisely. If the clamp triggers too early, it steals current from the main transistor and reduces efficiency. If it triggers too late, the spike has already damaged the device.

Active Miller Clamp for False Turn-On

False turn-on during the Miller plateau is best handled by an active Miller clamp on the gate driver. During turn-off, when the gate voltage stalls at the Miller plateau, the clamp actively pulls the gate below the threshold with high current — typically 1 to 5 amperes.

This overrides any noise or Miller-coupled voltage that tries to push the gate back above the threshold. The clamp must be fast — under 10 nanoseconds response time for SiC MOSFETs — because the Miller plateau on a SiC device can be as short as 50 nanoseconds.

A passive Miller clamp (just a negative gate drive voltage) is not enough for SiC modules. The Miller current on a fast SiC device can be several amperes, and a passive -5 volt rail cannot sink that current fast enough. You need an active clamp with dedicated sink current.

TVS and Zener Clamping: The Last Resort

Transient voltage suppressor (TVS) diodes and zener diodes provide hard voltage clamping at a fixed threshold. They are fast — nanosecond response — and they absorb energy directly.

Placing the TVS Across the Transistor

A TVS diode connected from collector to emitter clamps the voltage spike to its breakdown voltage. When the spike exceeds the TVS breakdown voltage, the diode conducts and shunts the current. The voltage across the transistor never exceeds the TVS clamping voltage.

The problem is energy handling. A TVS diode can absorb a large pulse — maybe 100 to 500 joules for a high-power device — but it cannot handle continuous energy. At high switching frequencies, the snubber energy that a passive RCD would dissipate in the resistor now hits the TVS. The TVS heats up, its clamping voltage drifts, and eventually it fails short — taking the transistor with it.

TVS diodes work best as a secondary protection layer — they catch the rare, extreme spike that the primary snubber misses. They are not a replacement for a properly designed snubber or active clamp.

Zener Diode on the Gate

A zener diode from gate to emitter clamps the gate voltage during turn-off. When the Miller effect tries to pull the gate above the threshold, the zener conducts and holds the gate at a safe voltage — typically 15 to 18 volts for IGBTs and 20 to 25 volts for SiC MOSFETs.

This is a simple, cheap fix for false turn-on. But the zener must be fast — standard zener diodes have too much capacitance and slow down switching. Use a low-capacitance TVS or a dedicated gate-clamp zener with capacitance under 100 picofarads.

The zener also degrades over time. Each clamping event stresses the diode, and after thousands of cycles, the clamping voltage drifts. Monitor the gate waveform periodically — if the clamping voltage has shifted by more than 10 percent, replace the zener.

Gate Drive Techniques That Prevent Overvoltage Damage

The gate drive circuit is not just a signal path — it is an active participant in overvoltage protection.

Negative Gate Drive During Turn-Off

Applying a strong negative voltage (-8 to -15 volts for IGBTs, -5 to -10 volts for SiC MOSFETs) during turn-off does two things. First, it speeds up turn-off by actively pulling carriers out of the channel. Second, it raises the threshold for false turn-on. The gate has to climb from -10 volts to the threshold voltage before the transistor can turn on again. That takes time — time that the Miller spike may not have.

For SiC modules, the negative rail is critical. The Miller capacitance of a SiC MOSFET is small, but the dV/dt is enormous. A -5 volt gate drive during turn-off gives you a 10-volt margin before the gate reaches the threshold. Without it, even a small Miller spike can cause false turn-on.

Adaptive Gate Drive Strength

Some modern gate drivers adjust their output current based on the operating condition. During normal turn-off, they use moderate drive strength to balance switching loss and dV/dt. During a detected overvoltage event, they switch to maximum sink current — pulling the gate low as fast as physically possible.

This adaptive behavior gives you the best of both worlds: low loss during normal operation, and maximum protection during fault conditions. The driver needs a fault input — typically from a desaturation detection circuit — to know when to switch to aggressive mode.

Detecting Overvoltage Fast Enough to Matter

Detection speed is everything. If your overvoltage detection takes 5 microseconds to trigger, and the spike lasts 200 nanoseconds, you have already lost. The device is dead before your circuit even notices.

Direct Voltage Sensing on the Collector

The fastest way to detect overvoltage is to sense the collector-emitter voltage directly. A resistor divider from collector to emitter feeds a comparator that triggers when VCE exceeds a threshold. The comparator output drives the gate driver into shutdown mode.

The divider must be fast — use low-value resistors (10 to 100 kilohms total) to minimize RC delay. The comparator must have propagation delay under 100 nanoseconds. The total detection time from spike onset to gate shutdown should be under 500 nanoseconds for IGBT modules and under 200 nanoseconds for SiC MOSFETs.

Using Desat Detection as an Overvoltage Proxy

Desaturation detection, which is normally used for overcurrent protection, also responds to overvoltage. When the collector-emitter voltage rises above the desat threshold (typically 7 to 9 volts for IGBTs), the desat circuit triggers. This happens during both overcurrent and overvoltage events.

The problem is that desat detection is slow — it has a blanking time of 2 to 5 microseconds to avoid false trips during normal turn-on. This blanking time is too long for overvoltage protection. You need a dedicated overvoltage detector with no blanking time, or a desat circuit with a much shorter blanking time and a lower threshold.

PCB Layout for Overvoltage Protection

The protection circuit is only as fast as the signal path between the transistor and the driver.

Keep the voltage sense trace from collector to the comparator input under 10 millimeters. Long traces add inductance that slows down the voltage rise seen by the comparator, delaying detection. Every nanosecond of trace inductance is a nanosecond of delayed protection.

Route the sense trace away from high-current power traces. Capacitive coupling from a switching node into the sense line can inject false voltage spikes that trigger false shutdowns. Use a grounded guard trace alongside the sense line if the distance exceeds 5 millimeters.

The snubber components must be placed as close to the transistor pins as possible. The RCD snubber capacitor should sit within 5 millimeters of the collector and emitter pins. Long traces between the snubber and the transistor add inductance that defeats the purpose of the snubber — the spike happens at the transistor pins, not at the snubber.

Common Mistakes That Kill Overvoltage Protection

The first mistake is relying on the transistor's built-in avalanche rating. Datasheets specify an avalanche energy rating — maybe 10 to 50 millijoules for IGBTs. This is the energy the device can survive in a single avalanche event. But avalanche degrades the device. Each avalanche event creates microscopic damage in the silicon. After a few dozen events, the breakdown voltage drops, and the device fails at normal operating voltage.

Do not design your protection to allow avalanche. Clamp the voltage before it reaches the breakdown rating. The avalanche rating is a last resort — a margin for when everything else fails.

The second mistake is using a snubber that is too small. An undersized snubber clamps the voltage but does not absorb enough energy. The capacitor voltage rises cycle after cycle until it reaches the bus voltage, at which point the snubber stops working entirely. Size the snubber capacitor for the worst-case fault energy, not the normal switching energy.

The third mistake is neglecting the negative rail during turn-off. Without a strong negative gate drive, the Miller effect will cause false turn-on on almost every switching cycle at high dV/dt. The transistor will not fail immediately — but it will overheat from the shoot-through current, and it will fail within hours or days. The negative rail is not optional. It is the cheapest, most effective overvoltage protection you can add.