IntroductionTilted-wave interferometry (TWI) is a novel measurement technique for the highly accurate optical measurement of aspheres and freeform surfaces [1][2][3][4][5][6][7]. It combines interferometric measurements with ray tracing and mathematical reconstruction procedures. The first results of a sensitivity analysis were given in [8]. The results support the feasibility of the basic TWI principle, and good reconstruction results were obtained for surfaces whose deviations from the design topography are described by Zernike polynomials. The TWI reconstruction procedure is divided into two parts. The first part reconstructs the long-wave deviation of the specimen from its design form. The second part aims at the reconstruction of high spatial frequencies [7]. This paper investigates the robustness of the long-wave TWI reconstruction method. Therefore, we extend the sensitivity analysis to surfaces whose deviation from their design topography no longer corresponds to a Zernike polynomial. We utilize a simulation environment that was developed at PTB for assessing optical measuring systems [8,9]. Reconstruction accuracies are then investigated in dependence on the variation of the specimen's topography and the measurement noise. Different variations of the specimen's topography with different spatial frequencies are studied. Furthermore, the influence of the deviation of the vertex radius of the design topography is investigated. Throughout all simulations no errors in the characterization of the TWI were assumed to be present which corresponds to a perfect calibration.
Mathematical Reconstruction ProcedureThe TWI reconstruction principle for the basic form of the specimen utilizes measurements of the optical path length (OPL) in order to calculate the deviation of the surface under test from its design topography. The idea is that small changes of the surface topography lead to small changes in the OPLs. The long-wave