Abstract

A given uncertainty of the output current vector of the four-detector photopolarimeter (FDP) leads to a corresponding uncertainty in determination of the Stokes vector of incident light. Using an FDP instrument matrix measured free from imperfections in the calibration optics, and for a preset level of error for each current, we calculate the maximum possible azimuth error, ellipticity error, and apparent depolarization as a function of the state of polarization of an assumed totally polarized incident beam. Experimental results are also presented that show the effect of reducing the power level of a light beam (from <1 mW to <1 μW, an attenuation range of 3.5 decades) on the precision with which the normalized Stokes parameters (NSPs) are measured. The rms error of each NSP is virtually independent of attenuation over an initial range of 2 decades and rises subsequently. However, it is the digitization (or quantization) error of the analog-to-digital converter that sets the limit on precision in our FDP. The accuracy with which the NSPs are measured is also essentially independent of light level, provided that the adjusted operational–amplifier gains are correctly accounted for.

© 1988 Optical Society of America