Sorry for the late contribution. I always want to add something and forget where to add it in the forum
The wavelength shifting that rdagger mentioned is mainly due to the fact that he is using the LLP method. As Seth asked (long ago ) why other ppl have not complained about this effect. The optical explanation would be that laser emission and led emission greatly differ. A laser output is monochromatic, the peak wavelength differs from the 0 emission wavlengths by only 0.5-1%. This means that a laser of 850 nm will emit mostly at this level.
A LED however has a greater wavelength spread that can be as large as 80 nm. When you combine the laser output to the (real) problem of the angle of incidence on the BP filter, you get the shifting effect. With LEDs you can still get 50-60% transmission efficiency even at the wavelength of the filter no matter how much it shifts. This is why FTIR users did not encounter this effect, or if they did, it is easily corrected through proper software filters.
In this case, FTIR users can opt for narrower band filters than LLP users.
I’m not sure that after so much time this explanation serves anyone but better late than never