MOSFET Double-Pulse Testing: Measuring Switching Energy and Loss

What the Test Adds to a Design Estimate

The MOSFET Gate Drive Guide develops first-order transition and switching-loss estimates. Those equations are useful for choosing a candidate device and driver. A double-pulse test instead observes the actual voltage-current overlap, reverse-recovery interaction, ringing, and gate-drive behavior of a particular assembled commutation loop. It does not replace a full converter efficiency or thermal test.

The usual fixture contains a DC link, an inductive load, a switch under test, a freewheel path, and a gate driver. The first on-pulse raises inductor current; after a short off interval, the second on-pulse commutates the established current. The end of the first pulse provides a turn-off waveform. Pulse durations, current, bus voltage, and freewheel device are part of the result, not incidental setup details. This is a high-energy laboratory method: use rated probes, isolation, discharge procedures, current limits, and a reviewed test plan.

Integrate Simultaneous Voltage and Current

For an illustrative measured Eon of 80 µJ and Eoff of 40 µJ, a hard-switching stage at 100 kHz has approximately 12 W of switching-event loss if every cycle resembles the measured condition. That is not 12 W total device loss: hot-state conduction, gate-drive power, and other terms remain. Feed measured event energy into the MOSFET Switching Loss Calculator only when its input model matches the available measured quantities. Use the MOSFET Total Power Dissipation Calculator to assemble nonoverlapping terms, not to add recovery or capacitance energy a second time.

Probe Placement and Timing Can Dominate Error

Measure VDS across the device terminals or the defined Kelvin source reference, and measure the relevant drain or commutation current with a sensor whose bandwidth and placement suit the transition. Measure VGS at the actual gate-to-source terminals, not against distant board ground. Respect the common-mode and voltage ratings of differential probes. Minimize added loop area; a long ground lead can invent ringing or alter the circuit. A current probe on a different branch may include capacitor charging or diode current that does not match the intended switch current.

Deskew voltage and current channels using a known reference before multiplying waveforms. During a fast transition, even a small relative delay can create a false high-power interval. Capture enough pre- and post-transition time to establish steady baselines, then integrate identical, documented window rules across comparisons. Check oscilloscope bandwidth, sample rate, clipping, noise floor, probe calibration, and sign. Repeat captures and compare waveforms before reporting a single precise energy number.

Make Test Conditions Comparable

Turn-on energy may include the effect of the freewheel path recovering or charging. Whether a datasheet separates that term depends on its measurement definition. An Eon result from a diode-commutated fixture cannot be transferred unchanged to a different synchronous switch, dead time, or soft-switching topology. Compare at several operating points, then interpolate only within characterized bounds. Do not extrapolate a single 25°C trace to the maximum junction temperature.

Turn a Trace Into an Engineering Decision

First, verify VDS overshoot and VGS excursions against device limits and review the applicable safe-operating-area and transient-thermal context. Second, compare Eon and Eoff at the intended bus, current, driver setting, temperature, and freewheel path. Third, estimate switching power for the application event distribution. Fourth, add independent conduction and drive losses and assess thermal behavior using the MOSFET SOA and Thermal Design Guide. Finally, validate the complete converter with measured input/output power and temperature. An apparently low measured Eoff is not a pass if voltage overshoot violates limits or EMI is unacceptable.

• Using unaligned voltage and current channels.
• Integrating different time windows when comparing devices.
• Counting freewheel recovery twice in Eon and a separate loss term.
• Assuming a room-temperature point represents hot operation.
• Moving a probe or DC-link capacitor between comparison runs.
• Calling Eon plus Eoff the entire MOSFET or converter loss.
• Ignoring VDS overshoot, VGS ringing, or safe operating area.
• Running the fixture without a reviewed high-energy safety procedure.

Engineering Sources

Infineon device application note illustrates a double-pulse fixture and Eon/Eoff integration intervals. Infineon double-pulse guidance discusses matching bus voltage, current, and test conditions. Use the actual device datasheet and qualified laboratory procedure for limits and safety.