0

Preamble: I have a LED driven by an opamp-based constant current source (CCS). It works fine, I can set up current required and it is reasonably stable. The problem is, that LED forward voltage drops over first 5-10 minutes due to internal heating, optical output power drops with it. To compensate for this, I would like to use a photodiode (PD) as an optical feedback, which adjusts current through the LED to keep optical output stable (increase drive current as Vfwd drops).

Request: A circuit idea, which would allow to set initial output level as a function of CCS setpoint, which is measured by PD and then used for driving the CCS (in theory setpoint input shouldn't matter after initial setup). The sequence I envision could be:

1. power up
2. set current limit (turn the LED on)
3. photodiode gets connected in feedback (perhaps with a small delay)
4. photodiode feedback controls output power, initial setpoint can be ignored, since no changes in levels are planned

EDIT: Per Andy's request CCS circuit, borrowed from another question: enter image description here

EDIT2: Essentially I'm looking for a way to override one analog signal with another.

stiebrs
  • 809
  • 5
  • 9
  • 1
    Sounds like you are swallowing a spider to catch a fly. If the voltage drop is the reason for the initial loss in brightness then use that as an indication and do compensation based on that. – Andy aka Jul 17 '20 at 10:46
  • Hmm, that's an actual idea worth trying out. But my main problem is I don't know how to go about overriding CCS setpoint input with whatever feedback source I use – stiebrs Jul 17 '20 at 12:09
  • Well, I would help but I see no circuit. – Andy aka Jul 17 '20 at 12:18
  • That's my problem. I have no idea how to implement external feedback for CCS in the form "use a if b is not present, otherwise use b". I could post generic opamp+transistor with feedback from low side sense resistor circuit, but I don't think that's necessary. Question is about how to plug that feedback into opamp based CCS in such a way, that it overrides the setpoint – stiebrs Jul 17 '20 at 12:27
  • You're not helping yourself. – Andy aka Jul 17 '20 at 12:28
  • Added CCS :P Actually linked question is essentially on the same topic, especially valuable is @jonk reply. I've done similar tests for long term (hundreds of hours) drift as well as short-term and would like to flatten initial warm-up curve.

    Just now noticed that you were active there as well and I've reached the same conclusion you suggest - optically coupling. I would expect PD to be much more stable than LED simply because it's not actively driven. Needless to say, that my setup has 2 separate thermally stabilized sections - LED mount and control circuitry.

     – stiebrs Jul 17 '20 at 12:46
  • OK but a PD output can drift with ambient temperature so you need to check on this. – Andy aka Jul 17 '20 at 13:07
  • Or, don't explicitly control the current. Use your photodiode feedback all the time and just set the brightness. Your user control then isn't the current but rather a value to compare with the measured brightness. The difference between photodiode voltage and set voltage adjusts the constant current to maintain a set brightness level. – JRE Jul 17 '20 at 13:11
  • Andyaka: PD sits on the thermally controlled surface, so I wouldn't expect too much drift. At any rate, it should repeatable (predictable), because it shouldn't degrade when used in passive mode. This can be characterized and compensated for mathematically in measurement post-processing stage. JRE: I need repeatability of initial conditions - CCS setpoint has to drift with Vfwd of the diode, to maintain constant optical power. I've been thinking about simply analog multiplexing of initial value and PD to CCS, but I'm not sure about moment of transition between inputs. – stiebrs Jul 17 '20 at 13:26
  • Actually PD amp charging a capacitor as a memory cell could help with surviving the switching disconnect of multiplexer. A slight drop in power would be acceptable. Thanks for the ideas guys! If no one offers any better ideas and this one works, I'll write it up and post the circuit – stiebrs Jul 17 '20 at 13:41

0 Answers0