In your initial model, you say that you use a motor with a max speed of 400 rpm. However, in your simulation, the simulation is failing closer to 6000 RPM, so you are driving the motor significantly faster than it was desigend to handle. In general you couldn't spin a PMSM this fast just by hooking it up to a constant voltage changing frequency. Specialized control is required for high speed motors. Also, this is a lumped parameter motor model, so it is effectifly acting linearly, and motors pushed to their extremes never act linearly, so extrapolating out this far is also likely to provide poor results. There isn't anything wrong with your model, you are expecting behavior that you should not see in this scenario.
