A Comparison of Control Architectures for Atomic Force Microscopes

We evaluate the performance of two control architectures applied to atomic force microscopes (AFM). Feedback-only control is a natural solution and has been applied widely. Expanding on that, combining feedback controllers with plant-injection feedforward filters has been shown to greatly improve tracking performance in AFMs. Alternatively, performance can also be improved by the use of a closed-loop-injection feedforward filter applied to the reference input before it enters the feedback loop. In this paper, we compare the plant-injection architecture with the closed-loop-injection architecture when used in controlling AFMs. In particular, we find that even in the presence of plant uncertainty, the closed-loop-injection architecture yields better tracking performance of a raster scan.