The manufacturer of this instrument, which uses visible and near-infrared radiation, had been receiving field reports of erroneous readings, but were unable to reproduce them in house. We performed a theoretical study of the signal processing methods embedded in the device, enabling us to predict the types of interference that might generate measurement errors. The predictions were then proven by experiment. Seventeen error sources were considered and quantified, including several types of light sources, external sources of electric and magnetic fields, and self-generated errors. Design changes were recommended along with the expected effect of each change on each error source.