IGBT simulation: SPICE or schematic issue ? (not urgent)

Schematic Simulation (Ngspice)
1

Hi!

Attached is a simple IGBT SCH that gives results inconsistent with the datasheet.
While the datasheet seems very ambitious on commuting times the results are too slow (few kHz) and also gives a strange shape.
I don’t know if it’s a SPICE model issue or something else.
Gate signal is +/-15 V though a 1k resistor
IGBT.zip (695.6 KB)

image
image1368×770 27.9 KB

image
image1420×1274 116 KB

Side questions I find IGBT models hard to get, are power engineers working full time on that making their own models ?

2

Indeed IGBT switching is relatively slow, a few 10kHz maximum.

I have checked some models, unfortunately did not find any which is compatible to ngspice.

IXYS uses a PSPICE NIGBT model, not available in ngspice (P-Spice Models).

Infineon models (https://www.infineon.com/cms/en/product/power/igbt/igbt-discretes/discrete-igbt-with-anti-parallel-diode/600v-1200v-trenchstop/?intc=202009_igbtselectiongraphic_en_ipc_600vtrechnstopperformance) are not compatible.

The Fuji Electric model (Power Semiconductors - Discrete IGBT P-Spice Models | Fuji Electric) does show an overly large gate current in ngspice, probably due to not understanding the RDS and N parameters (see ngspice warnings). This leads to a large voltage drop across the gate resistor R1, so VGate does not reach 15V. I have modified this model a bit to get rid of this current, see attached. It seems to be o.k. for now, but some model adaption is still due.
FGW50XS65.lib (979 Bytes)

3

I don’t know if it’s a SPICE model issue or something else. Gate signal is +/-15 V though a 1k resistor

I use IGBT (and power mosfet) in applications, but I don’t use simulation, so I can’t say something about the simulation aspect.

But:

  • switching of IGBT is slow compared to Mosfet, so some microseconds could be ok (depends on actual device - datasheet)
  • nobody uses a gate-resistor of 1000Ohm - this slows down the switching enormously compared to original datasheet-numbers.

I have only a preliminary FGW50XS65 datasheet and it states <1us fall/rise time (seems very fast for IGBT), but measured with a 10R gate resistor.

EDIT: could it be that you have exchanged gate-resistor and load-resistor? Using a 50A-IGBT with 50A load current (500V on collector-side with 10R load resistor) is not a really valid usecase.

4

Hello both and thank you for your support.
Indeed, I haven’t read properly the datasheet and wasn’t thinking a 1k gate resistor would completely waste the swith ON time.
10R gives results that are consistent with the datasheet.