Hi Prathik,

That's an extremely reasonable concern: it simply comes down to how the losses are calculated in PLECS. As you are likely aware, PLECS completely disconnects the thermal domain from the electrical domain for simulation speed. For turn-off switching losses specifically, it looks at the values of your input state variables (e.g. current, voltage, temperature) at the time step where switching occurs (i.e. your gate goes from logical 1 to zero), and finds the energy in the lookup table for those conditions. It then injects the correct losses into your thermal system. Since Q1 does not have a gate transition, the model does not consider there to be any switching event to pull a loss value from the lookup table. If I instead switch Q1 on when Q2 is off, you can see we now get switching losses in Q1:

As you have realized, this behavior means that switching losses are normally uncounted in asynchronous converters. In the example, conduction losses have dropped by 70% in Q1 since the device is gated on when conducting (which is far higher performance than using the body diode). While I would not recommend asynchronous switching due to the high losses of the body diode, if you do need to investigate asynchronous losses, you can do so by making this modification to the simulation:

In this case, Q1's gate turns on briefly (for the length of T_d) right before Q2 turns on. You can see the conduction losses have returned to their previous level while we still have switching losses. In some sense we are tricking Plecs into counting the switching losses here, but it should be fairly accurate.

One note, while testing your simulation, I realized I had made a mistake in our "Eoss" table which is tucked into the model. I've made a new version that fixes this issue by clamping the Eoss losses to zero for negative current.

Let me know if you have any other questions!

Thanks,
Blake