1

I am trying to amplify an optical signal which comes from a photodetector in the range of 30 µA with a transimpedance amplifier (OPA857) followed by a variable gain amplifier (THS7530). The problem is that the output is experiencing undershoot and I have figured out that this mostly comes from the first stage and maybe because of the input parasitic capacitance. Followed is the schematic of my design. Is there any way to compensate for this?

Schematic of Design

EDIT: Here is the simulation output when I increase the parallel capacitor to 2pF:

simulation output

And the circuit output: circuit output

PCB layout: (IC4 is the first op-amp and IC3 the second. PD is the photodetector)

enter image description here

Adding a parallel RC from the output of the first stage to its input (Feedback) compensates the undershoot to some extent but it still remains and is not negligible. It also broadens the pulse width which is not desired.

Farzin
  • 11
  • 2
  • 2
    Welcome -- please include waveforms of the circuit and simulation, and your PCB layout. What have you tried (if anything yet) to change the response? – Tim Williams Jul 21 '23 at 15:39
  • What's the photo detector? You've modeled it as 350fF in parallel with a current source, but that seems unrealistically small. – user1850479 Jul 21 '23 at 18:24
  • What photodiode? No bias? – evildemonic Jul 21 '23 at 19:26
  • @TimWilliams -- I included the outputs. Adding a feedback parallel RC compensates the undershoot to some extent while decreasing the total gain. – Farzin Jul 22 '23 at 05:50
  • @user1850479 -- It is an avalanche photodetector and the problem starts when the parallel input capacitor is in the range of 2 or 3pF which I think is because of the PCB or other parasitic capacitances. – Farzin Jul 22 '23 at 05:54
  • What is C4 doing? – user1850479 Jul 22 '23 at 14:06
  • @user1850479 It is normally not mounted but for test purposes. Do the pads affect the input capacitance? – Farzin Jul 24 '23 at 08:21
  • Pads have capacitance, as do traces and with a 2-3pF load you can't ignore it. I would have rotated the photodiode towards the detector to make that trace as short as possible and not put anything else on it. Not sure if it is your only problem here but it's a good idea. – user1850479 Jul 24 '23 at 13:37

1 Answers1

0

What you need to do is lower the capacitance. Reduce C2 to a lower value and check the layout, you should have a guard trace and minimal space between the photodiode and the U1 opamp.

Voltage Spike
  • 75,799
  • 36
  • 80
  • 208