Answers are grounded in this paper and cite their sources.
A wavelength calibration method for low-resolution diode array spectrometers is described. The method was developed for routine calibration of 0.1-meter-focal-distance spectrometers having 35- or 38-element silicon diode arrays, normally operated in the 340–700 nm spectral range. Each diode of the array is approximately 1 mm wide, giving an instrumental bandwidth of ∼10 nm per diode. The calibration method requires two well-separated monochromatic spectral lines, their central image locations on the array, and the grating groove frequency. This method is compared with nonlinear regression (least-squares) methods, with multiple calibration lines fitted to quadratic or cubic polynomials. The predictive accuracy of the wavelength-pair method compares favorably with the regression methods. A calibration accuracy of ∼±1 nm is expected for the instruments considered here. The method described could, in principle, be applied to instruments with higher resolution, such as those having self-scanned photodiode arrays with 25-μm or 50-μm-wide pixels. For such instruments, however, a large number of calibration lines should be resolved. In this case, the regression method, which averages diode position measurement errors, is probably more accurate. The wavelength-pair method is most useful for low-resolution instruments, for which regression methods may not be practical.
No takes yet. Share an insight, caveat, or question.
J. Thomas Brownrigg (1993) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: