LED's are radiators with a spectrally narrow emission band generally less than 50 nm. Compared to broadband light sources, like lamps, their spectral power distribution can be classified as quasi-monochromatic. Photometer heads integrate the spectral emission of a light source for the whole visible spectral range, this generally averages out a great part of the error caused by the local spectral mismatch of the V((lambda) ) detector. On the contrary, if LED's are measured this local mismatch error appears entirely in the measurement results. Measuring red LED's with a photometer head of f₁' <3% the absolute error can be even greater than 5%. The same head produces for commercial broadband light sources absolute errors less than 0.8% (in many cases less than 0.3%). Concerning quality industrial measurements this level of high error is not acceptable. A simple method will be shown how to increase the measurement accuracy of existing photometer heads. A computer simulation using generated and real LED spectral power distributions as well as spectral responsivity curves of a large number of real photometer heads took place. The results show that this method provides much higher absolute accuracy than the classical one.

Date Published: 20 October 1998
PDF: 9 pages
Proc. SPIE 3428, Illumination and Source Engineering, (20 October 1998); doi: 10.1117/12.327955

Published in SPIE Proceedings Vol. 3428:
Illumination and Source Engineering
Angelo V. Arecchi, Editor(s)