In phase-shifting Fizeau interferometers, nonlinear motion of the phase shifter and multiple-beam interference are the most common sources of systematic errors affecting high-precision phase measurement. A new class of algorithms with extended compensating capability for these errors is proposed. Measurement errors for the new algorithms and two groups of conventional algorithms: discrete Fourier algorithms and the Schwider-Larkin-Hibino algorithms are estimated as a function of the number of sampled images when these systematic error sources are equally dominant. It is shown that the conventional phase-measuring algorithms produce significant errors when the reflectivity of the testing surface exceeds ten percent. Also, these algorithms have an optimum number of samples at around seven with which the residual errors become minimum. The new class of algorithms shows a substantial reduction of the residual errors when the number of samples exceeds ten. There is no optimum number of samples for the new algorithms. For fewer than six samples, discrete Fourier algorithms which have no error-compensating capability for the nonlinearity of phase modulation give a minimum error.

This article cannot obtain the latest cited-by information.