Switching losses - MOSFET, negative current

Question

Hi all,

I am trying to simulate a bidirectional half bridge with MOSFETs. I have a thermal model for the losses, however depending on the duty cycle the current across one of the switches is always negative (using PLECS convention), therefore the switching losses are not calculated and I am getting zero. I tried to modify the look-up table by mirroring the original table across the current axis (for example if 50A=2 mJ, -50A= 2mJ and so on) to account for negative currents but I am still getting zero switching losses. I did a similar trick for the conduction losses and they work ok.  Thank you in advance for your help.

Comments

Have you also looked at the voltage across the device when the current is negative and taken it into account when mirroring the loss table?

Kind regards,

Oliver Schwartz

---

Dear Oliver,

Thank you for your reply.  Vds is always positive as the DC link voltage does not change, therefore there is no need to mirror the voltage in the losses maps. However, depending on the duty cycle the current across the MOSFETs can be positive or negative (using PLECS convention).  For example if the half bridge is working as a buck converter, the current in the top MOSFET goes from drain to source (positive) during ON time, but the current in the bottom MOSFET goes from source to drain (negative) during OFF time. I mirrored my maps to account for negative currents but I keep getting zero switching losses whenever the current is negative.

---

Danilo,

if you want to dig into this yourself you could change the setting "Negative switch loss" in the Diagnostics tab of the Simulations Parameters dialog to "warn and crop", then look at the warnings generated during a simulation. The warning are displayed when clicking the yellow triangle in the lower right corner of a schematic during or after a simulation. This should give you a hint if and why any losses are cropped.

You can also send your model and loss table to support@plexim.com and we'll have a look at it.

Kind regards,

Oliver Schwartz