Current control problem...

[tedlarson]

I am mostly a digital guy....but sometimes I run into analog problems I can't solve easily....so here is one bugging me.

I have this little circuit with takes photodiode input to drive a dummy load. It works, but not how I expected...and I can't figure out why. When I have constant input at the photodiode, I should be able to adjust the gain using R25. The problem is, the gain isn't linear. If I go from 10k to 20k, the current doesn't go down by 1/2 at R1....it seems exponential. I almost have to go to 2 meg in order to 1/2 the current at R1.

What am I missing here? Any ideas would be greatly appreciated.

Thanks,

-Ted

 

[mister_rf]

You may read some info here:
**broken link removed**
http://www.ti.com/lit/an/sboa061/sboa061.pdf
http://www.jensign.com/transimpedanceamp/
some examples:
**broken link removed**

 

[D.A.(Tony)Stewart]

YOu have a very non-linear design.
1. R25 determines the gain of the PD, ( current to voltage) the voltage gain of U8 , and the frequency response of the PD also .
2. the reference voltage is offset to V/2 which is not the zero light condition.
3. the output needs an offset for Vbe, which is not matched by the input Vref since you have open loop DC gain

What is the light source and level? is it infrared, optical or sunight?
Why DC response? For better sensitivity and rejection of stray light, pulsed IR with BPF is best.

I would use negative feedback from Q5 emitter for DC feedback.
What characteristic do you want high gain variable threshold switch with hysteresis, or none. or linear gain across dummy load over a range of light input?

 

[tedlarson]

Thanks for the replies. I liked that transimpedance amp document.....I hadn't seen that one. It was a good read.

SunnySkyGuy.......all your questions bring me to the "back story" on the circuit. Basically, it is a piece out of a laser diode driver circuit. The PD in this case is the feedback PD in a laser diode. The load is the laser. Another guy designed the circuit, I ended up with a fabbed PCB with it on there, it didn't behave right, so then I bread boarded it out to try to figure out how to fix it. My breadboard circuit behaves the same as the one on the PCB with the laser on it, so I figured I had a nice simple test case I could use for figuring out what I would need to tweak, without having the laser in the puzzle.

I certainly could easily just go and throw the whole circuit out the window...but then I wouldn't learn anything about why it doesn't work. I think what you have pointed out in your list are all the reasons it doesn't work for sure.

What I really want is a nice linear gain, so I can easily have a nice method to tweak the output power linearly. Right now what I have works....it is just totally nonlinear, exactly like you have pointed out.

Knowing the new info....would you still use negative feedback from the Q5 emitter for feedback?

Thanks,

-Ted