| Summary: | This paper examines the practical challenges of simplified setups aimed at achieving high-power IGBTs’ <i>I<sub>C</sub></i>–<i>V<sub>CE</sub></i> curve. The slope of this I–V curve (which is defined as on-resistance <i>R<sub>CE</sub></i>) and the point where the <i>V<sub>CE</sub></i>–<i>V<sub>GE</sub></i> curve visibly bends (threshold gate voltage) can be suitable failure precursor parameters to determine an IGBT’s health condition. A simplified/affordable design for these specific measurements can be used for in-situ condition monitoring or field testing of switching devices. First, the possible I–V curve measurement methods are discussed in detail in order to prevent self-heating. The selected design includes two IGBTs in which the high-side IGBT was the device under test (DUT) with a constant gate voltage (<i>V<sub>GE</sub></i>) of 15 V. Then, the low-side IGBT was switched by a short pulse (50 μs) to impose a high-current pulse on the DUT. The <i>V<sub>CE</sub></i>–<i>V<sub>GE</sub></i> curve was also extracted as an important failure-precursor indicator. In the next stage, a power-cycling test was performed, and the impact of degradation on the IGBT was analyzed by these measurement methods. The results show that after 18,000 thermal cycles, a visible shift in I–V curve can be seen. The internal resistance increased by 13%, while the initial collector-emitter voltage and voltage at the knee point in the <i>V<sub>CE</sub></i>–<i>V<sub>GE</sub></i> curve slightly changed. It is likely that in our case, during the performed power-cycling test and aging process, the bond wires were most affected, but this hypothesis needs further investigation.
|
|---|
