Optimal Sequencing to Shape Non-Linearity in Reset Control System, Application for ASMPT’s Wire Bonder Motion Platform

Abstract

Linear controllers are widely preferred in the industry for controlling motion platforms used in semiconductor and electronics manufacturing machines. However, as production demands for semiconductor devices increase, the performance of linear controllers reaches a plateau due to their intrinsic limitations, making further improvements increasingly difficult. Thus, the reset control has been explored as a potential solution as it can overcome the limitations and it consists of the frequency domain analysis tools.

Past research has shown that the CgLp (Constant in gain, Lead in phase) is the most advanced reset based filter and has shown performance improvements for an industrial machine. The non-linearity of these filters, which gives an advantage over linear filters, also makes their placement within a control loop highly critical. While the inherent non-linearity of these filters provides certain advantages, it can also be detrimental to system performance.

This thesis aims at developing architectures based on sequencing strategies to improve the performance of reset based filters by optimally utilizing the non-linearity. Also, the disadvantages of existing frequency domain analysis methods have been identified. Finally, the method of filtered architecture which can shape the non-linearity of reset based filters have been explored.