Deskewing Voltage and Current Before Measuring Switching Energy

Switching energy is often calculated by multiplying measured device voltage and current, then integrating over a transition. If the measurement paths have different delays, the multiplication pairs values that occurred at different times. A small timing error can then produce a large apparent change in loss.

Define the measured boundary

Identify exactly which device voltage and current represent the intended power flow. Record probe polarity, current direction, and reference nodes. The Tektronix power measurement application note explains how unequal voltage and current probe propagation delays affect switching loss measurements and why deskew is necessary.

Also check whether a measured current includes displacement or another branch current outside the chosen device boundary. Timing alignment cannot correct a fundamentally incorrect measurement location.

Establish delay independently

Use a suitable deskew fixture or supported procedure that presents a known timing relationship to the measurement paths. Include the actual probes, cables, attenuation, and relevant acquisition settings. Record the correction rather than treating it as an invisible instrument adjustment.

Do not arbitrarily slide the traces until the switching energy looks smallest. The real circuit can contain a physical delay between voltage and current changes. A correction should remove measurement delay, not erase device behavior.

Check offsets and dynamic range

Zero the measurement channels using the appropriate procedure and verify polarity. Small offsets can contribute energy when integrated over a long interval. A voltage range suitable for the off-state voltage may have limited resolution for the low on-state drop.

Inspect bandwidth, sample rate, and probe response. Deskew aligns timing but does not make two unequal frequency responses identical. Avoid reporting more precision than the complete measurement chain supports.

Set and preserve integration limits

Define turn on and turn off windows consistently, including how ringing or commutation tails are handled. Save raw voltage and current as well as the calculated power and energy. Repeat representative events and report their spread rather than selecting only a favorable transition. When converting event energy into average power, use the actual event repetition rate and include any skipped cycles.

Test sensitivity to a small timing variation consistent with the deskew uncertainty. If the resulting energy range changes the design conclusion, improve the measurement before comparing device options. Record operating voltage, current, temperature, gate network, deskew method, and integration boundaries together. The result should make clear which loss is being measured and how strongly it depends on instrument timing, so a later layout or transistor comparison uses equivalent evidence.