Experimental characterization and quadratic programming-based control of brushless-motors

A new torque control strategy for brushless motors is presented, which results in minimum torque ripple and copper losses. The motor model assumes linear magnetics, but contains a current limit which can delimit the onset of magnetic saturation, or be the motor amplifier current limit, whichever is reached first. The control problem is formulated and solved as a quadratic programming problem with equality and inequality constraints to find the nonlinear mapping from desired torque and position to the motorʹs phase currents. The optimal solution is found in closed form using the Kuhn-Tucker theorem. The solution shows that, unlike the conventional commutation with a fixed current-position waveform, the waveforms of the proposed controller vary in order to respect the current limitation in one phase by boosting the current in the other phases. This increases the maximum torque capability of the motor - in our particular system by 20% - compared to fixed waveform commutation. Experimental data from our brushless direct-drive motor demonstrates that the controller produces virtually ripple-free torque and enhances remarkably the tracking accuracy of the motion controller.