An automated setup has been developed for spectral radiometric characterization of LED components with precise control of the settings of forward current and operating temperature. The automated setup has been used to characterize commercial LED components with respect to multiple settings. It is shown that the droop in quantum efficiency can be approximated by a simple parabolic function. The investigated models of the spectral power distributions (SPD) from LEDs are the strictly empirical single and double Gaussian functions, and a semi empirical model using quasi Fermi levels and other basic solid state principles. The models are fitted to measured SPDs, using the free parameters. The result show a high correlation between the measured LED SPD and the fitted models. When comparing the chromaticity of the measured SPD with fitted models, the deviation is found to be larger than the lower limit of human color perception. A method has been developed to optimize multicolored cluster LED systems with respect to light quality, using multi objective optimization. The results are simulated SPDs similar to traditional light sources, and with high light quality. As part of this work the techniques have been applied in practical illumination applications. The presented examples are historical artifacts and illumination of plants to increase photosynthesis.
Faculty of Science, University of Copenhagen, 2012