The focus of this paper is on estimating system states in the presence of repetitive measurement noise. This estimation technique is applied to a novel self-spinning high speed rotary on/off valve that is being developed for pulse width modulating (PWM) fluid flow. The PWM duty ratio is determined by the valve spool's axial position. Unfortunately, the optical position measurement is corrupted by a repetitive noise, induced by the spool's rotary motion. Two models are developed to represent the periodic noise: one discrete time time invariant model based on internal model principle and a continuous time model that uses a set of periodic basis functions. Kalman filters are designed to estimate the spool position and the periodic noise. This estimates are used with a PI with feedforward controller for the spool position reference tracking. Simulation and experimental results indicate the usefulness of estimating the periodic noise.