This thesis focuses on the design, analysis, and implementation of a predictive deadbeat current controller for controlling a Permanent Magnet Synchronous Machine (PMSM) drive system. In order to evaluate and compare the performance of the proposed deadbeat controller, a classical Field Oriented Control (FOC) PI current controller is also designed.

The basic theory of the drive system is presented, and mathematical models of the inverter and PMSM are developed. Two simulation models are developed; one based on the classical PI current controller, and another based on the deadbeat controller. Given inverter dead-time and DSP delay are influential for the performance of the deadbeat controller, compensation is also developed and implemented in the deadbeat controller model. Based on the results obtained from simulations, the controllers are implemented in a dSPACE 1103 laboratory setup. Finally, the controllers are tested for parameter sensitivity both in simulations and in the experimental setup.

Additionally, this thesis discusses how the inductance of the machine may be determined using the inverter zero voltage vector, and discusses how the core losses of the machine may be decreased, by weakening the magnetic field flux density.

As a result of this thesis, the predictive deadbeat controller is found to be a simple and intuitive controller, which offers outstanding dynamic performance that can compete with the performance of the classical FOC PI controller. The deadbeat controller is found to suffer from small steady state errors, possibly caused by its dependency on model parameters and inverter non-linearities which must be properly compensated for under different system operation points.