Effect of Sensor Location, Web and Idle Roller Structure on Measured Tension Accuracy in Roll-to-Roll Manufacturing Systems

This paper investigates the effects of idle rollers, web span, and sensor location on web tension control performance in roll-to-roll (R2R) continuous manufacturing systems. The web dynamics are modeled as a mass-spring-damper system using a Kelvin-Voigt viscoelastic formulation and are validated through frequency response comparison with experimental data. The analysis shows that increasing the number of idle rollers lowers the first minimum resonance frequency, thereby reducing the achievable control bandwidth; specifically, the addition of a single idle roller shifts the first resonance from 11.5 Hz to 10.6 Hz. Sensor placement is also found to significantly affect system dynamics, including both resonance and anti-resonance behavior. A cascade control strategy with a notch filter is implemented to attenuate the dominant resonance and improve tension control performance. Experimental results reveal that when the loadcell is placed to cover all idle rollers in the tension zone, the system achieves a steady-state tension error within ±0.35 N and a settling time of 1.00 s at 40 mm/s web speed. In contrast, when the loadcell is located near the actuator, the step response is faster (e.g., 0.78 s) but tension regulation is limited to a partial zone, resulting in larger steady-state errors (up to ±2.0 N). These findings provide practical insights for optimizing sensor placement and mechanical configuration to enhance dynamic stability and tension regulation in high-speed R2R systems.