Data-based feed-forward control in MIMO motion systems

For MIMO motion systems, a data-based feedforward control is derived and implemented on a wafer stage of a wafer scanner. On the basis of a quadratic objective function related to a performance-relevant time-frame of the servo error signals, the coefficients of a set of finite impulse response (FIR) filters are optimized using a Gauss-Newton method. Applied to the wafer stage both the optimization algorithm and the FIR filter structure provide the means to significantly improve upon scanning performance. This includes improved settling behavior, the reduced effect of cross talk by MIMO feedforward forces, and the generalizing properties of these forces at set-points other than for which is optimized.