This paper explores the combination of a Hybrid Reluctance Actuator (HRA) with a Hybrid Tunable Magnet Actuator (HTMA) to realize a high bandwidth actuator that can generate low-frequency forces with greater efficiently. The HTMA allows desired forces to be sustained without continuous coil heating by manipulating the remnant magnetisation of an AlNiCo magnet. This enables the actuator to exert force through two modes of operation: by magnetisation of the Tunable AlNiCo Magnet (TM) or by inducing a proportionally force-dependent field. The second mode may furthermore be used to compensate for unwanted variations in forces during magnetisation. Although FEM analyses provide an understanding of the actuator behaviour in steady states, it is inefficient to integrate transient FEM models with accurate hysteresis models. Hence, firstly, an analytical framework is presented to determine the transient behaviour and the comparative energy efficiency of the two actuator modes. Then, a control strategy is presented for the operation of the combined actuator to track a reluctance force step reference. An experimental setup is designed and tested to validate the concept and control method.

Recent studies have shown that tunable magnets (soft permanent magnets) can significantly reduce Joule heating in electromagnetic actuators. To achieve high motion accuracy and repeatability, this article proposes a novel actuator design with a linearized force-flux relation. In prior designs of variable reluctance tunable magnet actuators, the force and flux are related quadratically via a C-shaped actuator. Hybrid tunable magnet actuators based on biased fluxes are developed using lumped parameter models. Using finite element analysis, it is shown that the force-flux relation is symmetric linear around the mid position depending on the magnetic flux direction in the magnet. Within a position range of ±500 μm and a force range of ±20 N, the linear fit produces a negligible error of 0.08 N. Finally, this linear relationship is validated with a 0.03-N error in an experimental setup.