Other Parts Discussed in Thread: LMC6482, LM324, LM358, LMP2011
I am trying to drive a high brightness monochromatic LED with precision. I ran across app note snoa837 which I suppose it started life as a National Semiconductor note. This note describes using a photodiode in an optical feedback loop to stabilize the light output from a laser diode.
There is a critical error in the final schematic shown in the note. In snoa837.pdf, figure 2 shows a simplified schematic to illustrate the concept. In the drawing, the photodiode feedback signal is connected to the inverting input of the op amp and the bias voltage (in this case ground) is connected to the non-inverting op amp input. This is fine. However taking a look at figure 3, it shows a schematic of the "practical implementation" of the circuit. In figure 3, the photodiode feedback signal is connected to the non-inverting input of the op amp and the bias voltage is connected to the inverting input. This is wrong and this circuit will not work as shown. The op amp inputs must be reversed.
Unfortunately, my boss told me to go straight to a PCB instead of breadboarding this. I spent way too long trying to debug it before I noticed this discrepancy. As they say in Massachusets, "light dawns on Marblehead!". I should have gone with my gut and breadboarded it first. Caveat Emptor. The app note should be corrected to save the next careless person who copies without breadboarding some debugging time...
By the way, with the op amp inputs corrected, it does work quite nicely. I found I needed a capacitor between the inverting input and the op amp output but that could be just my configuration. I used a PMOSFET instead of the transistor shown to get higher power without requiring the op amp to drive out a lot of current. And I had to change Rg to match my photodiode characteristics. It will work with almost any op amp with inputs that can go to the negative rail, even LM324/LM358's. I use an LMC6482 for it's full RRIO capability.
-Randy
