I just ran a very interesting experiment with my Vero29 5000K COBs, my lux meter, and some colored gels used for stage lighting. I set the lux meter on a flat surface about 5 feet below the lights and turned it on. Then I put the red, yellow, green, and blue gels on top of the light sensor.
none…13960 lux…energy in specturm
When you compare these to Bridgstone’s published spectra data for their different color temps, it actually looks pretty good. All the colors add up to 12534 lux which is reasonable (about 10% loss) for going through the filter sheet without dye. (They did not give me a clear one.)
If you divide the yellow value by 6.5 you get a reasonable approximation to what you would expect from the spectrum. The other three colors match what you would expect pretty well! I suspect this is because the lux meter is especially sensitive to yellow, but no problem: We don’t care about yellow much!
So for a $10 lux meter and $5 worth of color filters we do have a pretty good tool for measuring the red and blue content, because you can calibrate it against a published spectrum. Of course it does not give you a real spectrum with peaks and valleys to match against the Chlorophyll absorbtion peeks, but neither does a PAR meter. In fact a PAR meter would tell you you are fine with just a brilliant green and yellow source, when you are not. This way you can get some numbers in the blue and some numbers in the red.