Questions about IGBT Double-Pulse Test Method

Newbie Guide Basic Q&A
,
1

I’m conducting the IGBT double-pulse test, which simulates the phase-to-phase short-circuit time of IGBTs—namely Type II short circuit—under conditions of large load and high inductive reactance. I have several questions to ask:

  1. Why does the current need to rise to the IGBT’s rated current during the first pulse phase?

  2. After the first pulse is turned off, why should the second pulse make the current rise to twice the rated current?

  3. How to determine the turn-off time between the two pulses, and is it related to the dead time of the full bridge?

  4. When measuring turn-on loss, turn-off loss, and the stray inductance of the IGBT, why is the turn-off of the first pulse used to calculate the turn-off loss, and the turn-on of the second pulse used to calculate the turn-on loss—especially for the calculation of stray inductance?

  5. If the reverse voltage spike has reached the rated value when the second pulse is turned off, but the current has not reached the required value, how should we handle it?

Are there any theoretical bases for the above questions?

2

You haven’t clearly explained what you’re designing or why you need to test short-circuit conditions. Is this about implementing short-circuit protection in motor drives to prevent user errors? This is a common scenario. Which specific type of short circuit are you referring to? Phase-to-phase short circuits or ground faults? Do you need to consider both?

Double Pulse Testing (DPT, sometimes called two-pulse testing) is commonly used to evaluate various power electronic systems and their dynamic characteristics. If you’re new to this type of testing, searching for the keywords mentioned above will yield substantial resources. Infineon’s whitepaper on this topic is particularly well-written (I recall it was also published in Bodos Power magazine):
https://www.infineon.com/assets/row/public/documents/60/54/infineon-double-pulse-testing-bodos-power-systems-article-en.pdf?fileId=5546d46271bf4f920171ee81ad6c4a1f

When starting tests, select an inductance value matching your system expectations - for example, the output inductance of a UPS filter. For motor drives, this value will vary depending on the motor characteristics your product needs to support: asynchronous motors typically have higher inductance, while permanent magnet motors (especially high-speed types) have lower values. The selected inductance value, together with the DC bus voltage, determines the current change rate di/dt according to the formula:
V/L = di/dt (slope of test current)

The power stage must operate across the entire current range (from minimal current to overload conditions). I typically test near-zero turn-off scenarios to check issues like diode reverse recovery. Short-circuit conditions can also be tested - I start cautiously with extremely low voltages (the inductance L in short-circuit makes V/L very small).

One of the trickiest aspects of setting up a DPT system is finding a signal source capable of generating pulse trains. Some function generators can do this, but I usually build my own using 555 timers and logic circuits, debugging this logic section separately before connecting the power stage.

Exercise extreme caution! Improper operation could lead to catastrophic failures. I prefer using a low-power variac supply, followed by a rectifier bridge and capacitor bank to generate DC power. The capacitor bank should closely match actual design specifications. If safety concerns are significant, I add an isolation transformer after the variac to float the power stage while adding series impedance to the front-end supply (voltage will drop rapidly during IGBT faults).

Pay special attention to oscilloscope grounding! I prefer using differential probes for isolated measurements, preventing ground faults in the DPT system even when using multiple probes.

When starting tests, begin with low voltages and currents to verify gate drive functionality before gradually increasing levels. While IGBTs have reasonable fault tolerance, caution is still required. Early in my career, I once singed the hair on my finger joints - a painful lesson that emphasized the dangers of power electronics, though I still find this field an excellent career path full of excitement.

Good luck with your project.

3

That’s an excellent set of questions regarding the IGBT double-pulse test! This test is indeed a cornerstone for characterizing the dynamic switching performance and safely operating limits of power semiconductors.

Here are the theoretical bases and explanations for your questions:

:light_bulb: Rationale for Current Levels

The current levels are specifically chosen to characterize the IGBT’s performance at its rated conditions and its Safe Operating Area (SOA).

  • Why does the current need to rise to the IGBT’s rated current during the first pulse phase?

    • The primary purpose of the first pulse’s turn-off is to measure the turn-off loss (\boldsymbol{E_{off}}) and turn-off characteristics at the device’s nominal (rated) operating current (\boldsymbol{I_{C, rated}}).
    • Since switching losses are highly current-dependent, testing at I_{C, rated} ensures the measured loss is representative of the device’s performance under normal operating conditions.
    • The first pulse must be long enough to magnetize the load inductor to achieve this target current: I_{C, rated} = \frac{V_{DC}}{L} \cdot t_{on,1}.

  • After the first pulse is turned off, why should the second pulse make the current rise to twice the rated current?

    • The second pulse’s turn-on is primarily used to measure the turn-on loss (\boldsymbol{E_{on}}) and reverse recovery characteristics of the freewheeling diode (FWD) when the current commutes back to the IGBT at the rated current I_{C, rated}.
    • Your assumption that the current needs to rise to twice the rated current during the second pulse is typical for a short-circuit test (Type II), which is a different (though related) test.
    • For the standard double-pulse test, the second pulse is kept short so that the current does not rise significantly above the I_{C, rated} level established during the first pulse. This ensures that the turn-on and subsequent turn-off of the second pulse occur at approximately I_{C, rated}. The second pulse must be short enough to avoid excessive self-heating and exceeding the Reverse Bias Safe Operating Area (RBSOA).
    • However, for your specific objective of simulating the Type II short-circuit phase-to-phase short-circuit, the goal is different: you are intentionally testing the IGBT’s ability to withstand and safely turn off an overcurrent condition. In this context, the pulse is typically long enough to let the current rise to a high multiple (e.g., 1.5 \times to 2 \times) of I_{C, rated} or the device’s maximum pulsed current (\boldsymbol{I_{CRM}}) to characterize its ability to survive a short-circuit fault and the corresponding turn-off surge.

:stopwatch: Turn-Off Time Between Pulses

  • How to determine the turn-off time between the two pulses, and is it related to the dead time of the full bridge?
    • The turn-off time (or commutation time / off-time) between the two pulses is determined by the need for complete diode recovery and stable measurements.
    • Primary Goal: The freewheeling diode (FWD) must fully recover its blocking capability and allow the current to decay only slightly through the large load inductor before the second pulse begins. If the FWD is still in reverse recovery when the second pulse starts, the turn-on measurement will be distorted. A typical duration might be in the range of a few microseconds (\mu s) to tens of microseconds, but it primarily depends on the FWD’s reverse recovery time (\boldsymbol{t_{rr}}) and the inductor value.
    • Dead Time: This off-time is conceptually similar to, but NOT directly the same as, the dead time of a full bridge/half bridge. The dead time (or blanking time) in a converter circuit is a short protective delay applied between the turn-off of one switch and the turn-on of the complementary switch to prevent a “shoot-through” short circuit. The off-time in the double-pulse test is typically much longer than the application dead time, as its function is to manage the magnetic energy and ensure stable current for the next measurement.

:chart_decreasing: Loss and Stray Inductance Measurement

  • Why is the turn-off of the first pulse used to calculate the turn-off loss, and the turn-on of the second pulse used to calculate the turn-on loss?

    • Turn-off Loss (\boldsymbol{E_{off}}): The turn-off of the first pulse is the first switching event that occurs at the desired test current (I_{C, rated}) and the full DC-link voltage (V_{DC}). Since the circuit starts with I_C = 0 at t=0, the device is fresh, and the current has built up cleanly.
    • Turn-on Loss (\boldsymbol{E_{on}}): The turn-on of the second pulse is the event that involves the reverse recovery of the freewheeling diode (FWD). The FWD is carrying the load current (magnetized by the first pulse) when the IGBT turns on. This diode reverse recovery current (\boldsymbol{I_{rr}}) causes a significant current spike and is an integral part of the IGBT’s total turn-on loss. Therefore, this is the only correct point to measure E_{on} under realistic operating conditions.

  • Especially for the calculation of stray inductance?

    • The stray inductance (\boldsymbol{L_{s}}) of the commutation loop is calculated using the voltage overshoot (\Delta V) during the turn-off of the first pulse.
    • During turn-off, the rapid change in current (\frac{di}{dt}) through the stray inductance induces a voltage spike (\Delta V):
      \Delta V = L_{s} \cdot \left|\frac{di}{dt}\right|
    • This current commutation is clean at the end of the first pulse and provides the most pronounced \Delta V for an accurate L_s calculation. The voltage overshoot is superimposed on the DC-link voltage, V_{CE, peak} = V_{DC} + \Delta V. The turn-on of the second pulse also has L_s effects, but the turn-off event provides a more direct measurement for the commutation loop’s stray inductance.

:stop_sign: Handling Voltage and Current Limits

  • If the reverse voltage spike has reached the rated value when the second pulse is turned off, but the current has not reached the required value, how should we handle it?
    • The reverse voltage spike (V_{CE, peak}) at turn-off must never exceed the device’s rated blocking voltage (\boldsymbol{V_{CES}}), as this will lead to destructive breakdown. The maximum allowable current the IGBT can safely turn off is limited by the Reverse Bias Safe Operating Area (RBSOA).
    • Action: If V_{CE, peak} reaches V_{CES} (or 80\% of V_{CES} for a safety margin) before the current reaches your target (e.g., 2 \times I_{C, rated} for a short-circuit test), you must stop increasing the current.
    • Remedies to safely increase current:
      1. Increase Gate Resistance (\boldsymbol{R_G}): A larger R_G slows down the turn-off \frac{di}{dt}, which reduces the \Delta V spike (since \Delta V \propto \frac{di}{dt}), allowing the IGBT to turn off a higher current before hitting the voltage limit.
      2. Reduce Stray Inductance (\boldsymbol{L_s}): Optimize the test circuit layout (shorter, wider busbars) to minimize L_s. Since \Delta V is directly proportional to L_s, reducing L_s directly lowers the voltage spike.
      3. Adjust DC Link Voltage (\boldsymbol{V_{DC}}): Lowering the V_{DC} slightly also buys headroom for the \Delta V spike before the V_{CES} limit is reached.

This video demonstrates the practical setup and process for the IGBT double-pulse test, which visually illustrates the two pulses and the resulting waveforms.

How to Measure Switching Loss - Double Pulse Testing