Any idea what the stator resistance is? I gave couple number in the post above this one.
Back-EMF is a non factor for this, I agree.
Interesting appnote! With distributed sensing, it seems like they can still accomplish the goal. If that is even needed. With the proper heat sinking, the die will not get to the max temperature point, even with imbalance. Basically, all that needs to happen is to derate the max current by the max temperature delta on the part and the transfer function.
4 kW with 3 phases = 1.33kW each = 350V * 4 Amps. Using the NTD12N10 curve for reference, 4 Amps at 25C is 5 VGS, at 100 C that is 6 Amps, so even with a 75 K temp gradient, there is only a 50% power difference. That would not happen since the junction to case/ heat sink thermal impedance works both ways.
Otherwise, you are tying to the get heat from this to the pack somehow, and it is build to shed heat to the environment. Unless that passage is for flow though rotor cooling and they are using it with inductive heating?