This contribution considers a control scheme consisting of Model Predictive Control (MPC) together with a feedforward control and a Maximal Torque per Ampere (MTPA) structure. The MTPA structure, which is in the outer loop of the control system, calculates the optimal direct and quadrature current from a given desired torque. This current represents the reference currents for the MPC which states the control loop. The MPC is adapted using a Dual Kalman Filter which estimates parameter of the electrical system in d-q coordinates. Simulations show the effectiveness of the proposed control scheme with respect to a standard PI controller in the presence of saturated inputs. In particular, simulated results in the presence of input saturations show that MPC is working without overshoot in the actual currents which leads to less needed power in the input. The reason of that is that the proposed MPC uses just an optimal proportional control and thus avoids windup effects. Nevertheless, using just a proportional action, a bias between the desired and obtained signal is present in the results.