Performance degradation of a Michelson interferometer due to random sampling errors

The performance of a standard Michelson interferometer is degraded by disturbances that cause the interferogram signal to be sampled at nonconstant time intervals. A formula that shows how the power spectrum of the random disturbances interacts with the signal to contaminate different regions of the measured spectrum is derived for the spectral noise. The sampling noise does not look conventionally noiselike because it is correlated over large regions of the measured spectrum, and adjustment of the unbalanced background interferogram to match the size of the balanced background interferogram minimizes the sampling-noise amplitude.