A double pulse test is a way of checking the time domain behavior of a half bridge power module (transistors and associated circuitry).
In this test the transistor is loaded with current for a very short period. This allows losses, energy and other parameters to be measured without building heat. The temperature can be considered as a control variable.
If desired, heat can be added with a hot plate or climate/environmental chamber to simulate different operating conditions.
It is called the low-side Double Pulse Test (LS-DPT) because the device under test is the low-side switch of the half bridge.
Why this is often used:
The low side transistor is turned on, this allows zero current parameters such as stored energy to be measured. The transistor is used to charge a large inductor with current. The inductor is usually air-cored to avoid saturation.
Fig. 1A: Equivalent current path for the first pulse in a standard low-side Double Pulse Test.
Fig. 1B: Idealized waveform response during the first pulse of a standard low-side Double Pulse Test.
The transistor is turned off this allows Turn-off parameters such as Eoff to be measured; the current circulates in the inductor.
Fig. 2A: Equivalent current path during low-side turn-off, showing inductor current commutation into the upper freewheeling path.
Fig. 2B: Idealized waveform response during step 2 of the standard low-side Double Pulse Test.
The transistor is turned on again. This allows turn on parameters such as Eon to be measured. The inductor acts like a current source (similar to many real loads).
Steps 2 and 3 can be repeated to measure different currents (Multiple pulses). Be careful as heat may start to build. Typically, <5 pulses are used.
Fig. 3A: Equivalent current path during step 3 of the low-side DPT, with the low-side device turning on against the inductor current source.
Fig. 3B: Idealized waveform response during step 3 of the standard low-side Double Pulse Test.
The transistor is turned off to complete the test and stop the current from rising. The current in the inductor will slowly decay due to resistive and body diode losses.
Again, Turn-Off parameters such as Eoff can be measured.
Fig. 4A: Equivalent current path during step 4 of the low-side DPT, with the inductor current commutated into the upper freewheeling path and beginning to decay.
Fig. 4B: Idealized waveform response during step 4 of the standard low-side Double Pulse Test.
Below is a real waveform from a “Triple Pulse Test” of GaN. Measured On an IWATSU DS-8000 Series Oscilloscope with PMK Probes.
Fig. 5: Measured multiple-pulse low-side DPT waveforms for ID, VGS, and VDS
Choosing the right high-voltage probe depends on the required voltage range, bandwidth, common-mode rejection, and measurement environment. Optically-isolated probes are ideal for the most demanding switching measurements, offering outstanding common-mode immunity and galvanic isolation. High-voltage differential probes provide a practical and versatile solution for many general-purpose floating measurements. High-voltage passive probes are a straightforward option for high-voltage measurements where simplicity, bandwidth, and probe handling are key priorities.
Selecting the right low-voltage probe for VGSmeasurement depends mainly on bandwidth, probe loading, reference conditions, and the required voltage range. Optically isolated probes provide outstanding common-mode immunity for fast switching environments. Single-ended and differential probes offer very high bandwidth for accurate gate-drive characterization close to the device. Fast low-voltage passive probes are a practical choice where low input capacitance, simple handling, and clean probing technique are essential.
Current measurement in double pulse testing requires the right balance of bandwidth, invasiveness, dynamic range, and physical accessibility. Ultra-fast current shunts provide the highest signal fidelity for capturing very fast switching currents with minimal insertion inductance. Rogowski coils offer flexible, non-intrusive current measurement for larger conductors and high di/dt events. Optical electric current sensors provide galvanically isolated current measurement with excellent immunity to high common-mode transients.
Fig. 6: Effect of oscilloscope bandwidth on the measured response to a very fast step input, showing how lower bandwidth increasingly smooths the edge and alters overshoot and ringing.
Fig. 7: various types of test point connections
In a high-side gate measurement, the switch node can move over a very large voltage range relative to ground, while the actual gate-to-source voltage to be measured remains small. For example, the high-side source may sit at a reference potential of 1000 V, while the desired gate drive is only 20 V. In that case, the probe must accurately resolve a small differential signal on top of a very large common-mode voltage swing.
This makes CMRR (common-mode rejection ratio) one of the most important probe parameters for high-side VGS measurements. If CMRR is insufficient, common-mode voltage is converted into measurement error, which can severely distort the displayed gate waveform.
In short: high-side gate measurements are not defined only by voltage range or bandwidth. They are defined by the ability to measure a small gate signal in the presence of a large, fast-moving common-mode voltage. The higher the CMRR and the better the isolation, the more trustworthy the measurement.
Fig. 8A: High side gate driver and reference, and low side gate driver and reference
Fig. 8B: High side reference voltage
Below example shows an overview of the test setup and probes used:
Fig. 9: Example test setup (40V GaN)
Manual analysis of the switching waveform parameters can be performed using oscilloscope math functions. However, the IWATSU DS-8000 Oscilloscope with DS-821 software application has a large range of configurable, automatic, switching measurement and reporting functions.
Pro tip: 𝐸_𝑜𝑛 and 𝐸_𝑜𝑓𝑓 are Energies and not losses - these parameters also include stored charges. At any given current 𝐸_𝑜𝑛+𝐸_𝑜𝑓𝑓=𝑠𝑤𝑖𝑡𝑐ℎ𝑖𝑛𝑔 𝑙𝑜𝑠𝑠 : the stored charge terms cancel.
Fig. 10: results showing Eon = 46 μJ at 168A (left side), and Eoff = 31 μJ at 181A (right side)
By repeating the measurement a different test currents, we can create data-sheet information for the power device and learn more about its operation.
By repeating the measurement a different test currents, we can create data-sheet information for the power device and learn more about its operation.
Fig. 11A: Example Switching Energy Comparison – 400 V High-Speed SiC Module
Fig. 11B: Turn-Off Voltage Rise Time – 400 VHigh-Speed Sic Module
Below we can see a typical zero current (initial) turn-on waveform for a GaN double pulse test.
Fig. 12: typicalzero current (initial) turn-on waveform for a GaN double pulse test
Below we can see a typical turn-on waveform for a GaN double pulse test.
Fig. 13: typical turn-on waveform for a GaN double pulse test
Below we can see a typical turn-off waveform for a GaN double pulse test.
Fig. 14: a typical turn-off waveform for a GaN double pulse test
At high current we can see a tail current on this device. GaN does not have a tail current? Is this a measurement error?
Take a closer look at VGS ......
Fig. 15: unusual turn off waveform showing tail current on GaN device
This can be explained as follows:
If the probes had overshoot or ringing this effect would be hidden. With our accurate probes and attention to test-point placement we can see this issue and investigate it further.
Fig. 16: graphical explanation of phenomena causing tail current
Send us an email: sales@pmkamerica.com
