Propagation of measurement uncertainties in mismatch factor correction for photovoltaic device calibration.

The mismatch factor uncertainty is one of the major uncertainty contributors in calibration measurements, especially for thin film devices, and the most complex to calculate analytically due to underlying correlations in uncertainty. Typically, empirical estimates or Monte Carlo simulations based on wavelength dependent uncertainty analysis are used and sampling is conducted by random walks. A different sampling approach is proposed based on fitting a sum of Gaussian distributions to measurement data and generating spectral response curves within the uncertainty envelope by altering the fitting parameters. The generated curves are smooth and approximate real measurement data. Presently, the sampling method is limited to SRs.