Identifying the impeller type is essential for enabling torque sensing in conventional agitators. Previous studies have demonstrated that using arrays of permanent magnets with like poles facing each other allows for cost-effective, non-contact sensors. However, these configurations create strong repulsive forces, complicating assembly during sensor fabrication. This study addresses the issue of poor assemblability by introducing a high-permeability ferromagnetic ball between the magnets. This ball not only reduces repulsive forces but can also induce attractive forces, making assembly easier. We analyzed the effects of ball diameter, magnet thickness, and the number of magnets on the inter-magnetic force using ANSYS Maxwell. To validate the finite element method (FEM) results, we conducted experiments, which showed that the measured values closely matched the simulation results. This confirmed that the ferromagnetic ball significantly mitigates the repulsion between magnets, and in some cases, reverses the force to attraction. These findings are important for enhancing assemblability in automated mass production. Additionally, the study identified an optimal steel ball size that minimizes repulsion while facilitating sensor miniaturization, providing a practical solution for compact sensor design.

The rapid growth of semiconductor and display manufacturing highlights the demand for fast, precise motion stages. Advanced systems such as lithography and bio-stages require accuracy at the μm and nm levels, but linear motor stages face challenges from disturbances, model uncertainties, and measurement noise. Disturbances and uncertainties cause deviations from models, while noise limits control gains and performance. Disturbance Observers (DOBs) enhance performance by compensating for these effects using input–output data and a nominal inverse model. However, widening the disturbance estimation bandwidth increases noise sensitivity. Conversely, the Kalman Filter (KF) estimates system states from noisy measurements, reducing noise in position feedback, but it does not treat disturbances as states, limiting compensation. To address this, we propose an Augmented Kalman Filter (AKF)–based position control for linear motor stages. The system was modeled and identified through frequency response analysis, and DOB and AKF were implemented with a PIV servo filter. Experimental validation showed reduced following error, jitter, and control effort, demonstrating the improved control performance of the AKF approach over conventional methods.