I'm trying to compare the SVPWM code with what we have today, and as a precaution I also looked at the inverse park routine. I noticed that MCM_Rev_Park uses a different form than most others use, and the MCM_Park transform seems to match this. Comments showing a comparison between CMSIS and MCSDK code is shown below.
Valfa= Vq*Cos(theta)+ Vd*Sin(theta)
Id = Ialpha * cosVal + Ibeta * sinVal;
Iq = -Ialpha * sinVal + Ibeta * cosVal;
I'm looking at MCSDK V5.0.