Hi,
This phenomenon might be related to the LED’s semiconductor structure heating. A white LED is just a blue LED (450 nm) engulfed with phosphor rich medium. The phosphor converts part of the blue light into green, yellow and red. The lower the color temperature of the white LED, the higher the conversion efficiency.
When the LED heats up, the blue light shifts towards long wavelengths a bit. The shift is about 1 nm. This isn’t much for the spectral peak that has about 20 nm of FWHM.
Our MFTs do active light intensity stabilization. But that stabilization cannot prevent the mentioned spectral shift.
There are three major ways to mitigate the effect:
- Wait for the LED to warm up before you measure white reference standard and start an aging experiment. The scientific grade MFTs require more time (half an hour) than the regular ones to achieve thermal equilibrium.
- Use lower electric current driving the LED. This way the thermal equilibrium will be achieved sooner.
- Use LEDs with lower color temperature for the color determination.
Anyway, in most cases this spectral shift has negligible effect on the color coordinates $L^{*}$, $ a^{*}$, $b^{*}$ computed from the spectra of an evolving white LED and even smaller on the color change $\Delta E$.
You can run a simple test , as Anna suggested, by aging a white reference standard and observing the resulting $L^{*}$, $ a^{*}$, $b^{*}$. The standard does not age and all apparent aging can be attributed to the light source instability. This way you can determine how long the warm up time should be to keep your results to a satisfactory error level.
If you have already collected your results, it might be easy to estimate how much of the color change can be attributed to the spectral shift of the illuminant.
