Separate .xml thermal description of MOSFET+Diode into two independent

PLECS
, , ,
1

Dear Mr/Ms,

As I couldn’t find how to set diode electrical parameters in the MOSFET with Diode default block, I decided to try to separate the thermal description of the component into two independent files and build my own MOSFET with antiparallel Diode with their respective .xml.

The problem arises when I cannot match the turn-off losses between the original and the new setup and it seems that it comes from the tables. The procedure followed to fill the two independent files from the common one is:

MOSFET:

The conduction of the mosfet depends on the state of the gate terminal. Consequently:

  • Turn-on losses not modified.
  • Turn-off losses no longer dependent on the diode recovery. These losses will appear in the Diode file.
  • Turn-off losses no longer dependent on negative current values. If the switch already conducts through the diode, Zero-Voltage-Switching. Negative current values eliminated.
  • Conduction losses with gate off eliminated. These losses correspond to the diode when is forward-biased.
  • The thermal resistance left at the default value to not alter the losses and temperature calculations.

DIODE:

In order to obtain a ‘standard’ diode model, the negative values of currents and voltages are transposed in the corresponding tables.

Mosfet constants and gate tables/variables have been eliminated.

  • Turn-on losses are considered negligible.
  • Turn-off losses dependent on the recovery and deadtime.
  • Unique conduction losses, which correspond to the diode when is forward-biased.
  • The thermal resistance left at the default value to not alter the losses and temperature calculations.

Here the three attached files:

FF1MR12MM1HP_B11_mosfet.xml (26.0 KB)

FF1MR12MM1HP_B11_diode.xml (16.7 KB)

FF1MR12MM1HP_B11.xml (35.8 KB)

Thanks in advance!

2

Hi alefarri,

Could you provide a little more detail on the discrepancy you are seeing? Are you probing the switching losses of the three components in identical conditions and seeing the energy pulses to be significantly different? Are there any junction temperature differences for the three components? A PLECS model would also be helpful in debugging.

The process you are describing seems appropriate and with a quick glance the data seems to be be parsed correctly for the individual components.

A few things to consider:

  • For the high-voltage applications in which these devices are typically used, it is worth considering whether the diode forward voltage drop will significantly impact your electrical simulation.

    For example, with a 500 V bus, a diode forward voltage drop of 5 V represents only about 1% of the voltage in the KVL loop. Additionally, this effect is present only during the dead-time interval, which is typically less than 1% of the switching period.

    As a result, neglecting the diode forward voltage drop would likely introduce an error of less than 0.01% in the overall electrical simulation.

  • If you determine that the forward voltage drop is important to model, a potential simplification (as implemented in the Microchip thermal models) may be useful:

image
image322×242 4.07 KB

Hope this helps!

3

Hi Munadir,

Thank you for your time and responde.

It is a quite atypical converter topology that works in series and the magnitudes of currents are higher than in voltages. For this reason, voltage drops caused by the diode take more importance in this case and we wanted to consider them.

I coudn’t find the way of setting the electrical characteristics of the body diode, so I decided to buid my own mosfet with its own .xml thermal description.

Tank you for your proposal. I think it would be useful to have the diode characterised in both domains.

Best regards,

4

Glad I could help. I’ve also relayed the information to the development team as a feature request.