I am doing an Active Rectifier simulation. The closed loop control is Discrete and i am using a discrete solver with fixed step size (1e-7). The model is working fine, the problem is when i started doing loss analysis. I created a loss and thermal model of the IGBT from the datasheet. While the conduction losses settles to something reasonable, the switching losses keeps increase (>1e65). When i imported the same loss and thermal model to other model which has the variable-step solver, it gives the correct losses which means that the loss and thermal model is fine. How do i use that same loss and thermal model in a discrete solver system and get the correct losses?
For further assistance please post a representative model that one can use to recreate your observations.
Regards,
Bryan
Your controller has an underlying issue. If you look at the switching frequency with a fixed step solver, you’ll actually note the switching frequency of your PWM signals you’ll see it is in the GHz range (!) when you intend the switching frequency to be 500 Hz. This in turn leads to the thermal issues.
The fixed step solver masks this underlying issue that occurs when the Vref is close to zero. In the variable step solver you disabled all the error detections and warning messages that would point this out.
Also your PR block doesn’t work as you expect since all the delay blocks have a sample time of 1 second.
The duty cycle shouldn’t change faster than the carrier. You can put a ZOH block in front of the Vref signal to guarantee this. You can refine your control accordingly.