The igbt devices aren't a mystical mystery device at all. If you know the characteristics of the old school bipolar power transistors, that's how igbt transistors handle the high power switching side. High maximum voltage and current handling, they have a minimum "on-state" impedance, and have a negative temperature coefficient. They can experience thermal runaway and self destruct. They're pretty impervious to voltage spikes on the power lines as well.
On the control side, they're pretty much exactly the same as a n-channel mosfet. Super high impedance, defined and narrow threshold voltage for switching, really stable over a wide temperature and power range. That's because an igbt is basically the same as a mosfet on the control side. In fact, an igbt is basically a power mosfet with an extra slice of material bonded to the substrate.
The only real significant difference kicks in at high switching frequency. The extra slice of material is p-type material, which creates a drain-collector pn junction. At high frequencies, the junction adds some turn-off lag to the switching action that a mosfet does not. This "collector current tail off" affects switching losses. The higher the switching frequency, the higher the losses.
Recent improvements in manufacturing have blurred the selection process as far as which device is better for doing the job at hand. Below about 25 khz, either one is fine. Above that, it becomes an exercise in choosing between conduction losses versus switching losses. Above about 100 khz, the switch time lag of the igbt is too slow, and the switching losses pile up against it.
|