| Abstract: | Valuation of photovoltaic devices depends strongly on the measured power output of the device. This quantity is usually determined under artificial sunlight in production line measurement systems or industrial or research test labs. A practical calibration chain is realized essentially with measurements at solar simulators. The measurement conditions are defined in the IEC 60904 series of standards. An important part of the standard testing conditions is the definition of a specific spectral distribution of the sunlight (AM1.5 global). The inevitable deviations of the spectrum of artificial light sources from the standard spectrum have to be taken into account by a spectral mismatch factor. The uncertainty of this crucial correction is spectrally dependent, in most cases unknown and complex and inconvenient to evaluate. In this article a randomizing method is proposed which allows one to calculate the uncertainty of the mismatch factor from the uncertainties of the input parameters determined with high spectral resolution. Based on a range of different spectral responses of solar cells on the one hand and variations of the solar simulator spectral distribution on the other, we are able to generalize the results to a broad set of measurement configurations. A sensitivity analysis reveals the crucial wavelength regions and thus allows the systematic optimization of simulator spectra and selection of reference cells. Copyright © 2010 John Wiley & Sons, Ltd. |
