So the photodiode mostly corrects for temperature drift, correct? If this trend is largely an exponential decay with reasonable lifetime I can just wait until its stable within 1% or something, but a linear drift will affect the measurement I'm trying to make.
I'll start with the basic LM317 CC...