0
\$\begingroup\$

I am trying to make a bare minimum transimpedance amplifier for a very large Si photodiode (Thorlabs FDS1010). My circuit is just like in the picture except my digipot is an AD5272-100kohm and my amplifier is half of an LM358LV. My LM358 is powered from my microcontrollers 3.3v regulator. On a solderless breadboard the signal measured at my ADC is stable. This is also true for my two layer PCB. On my four layer PCB, at low gain setting (low digipot register and high illumination of the PD) the output is also stable. However, on the four layer board specifically, and with very high gain (50kohms feedback) and low illumination intensity of the PD the output begins to continuously oscillate from 0 to 1023 (max for 12bit ADC). Changing the analogRead sampling frequency changes this oscillation from anywhere between 1khz to 1hz.

This seems to me that the breadboard has provided enough capacitance to smooth things out but I am unsure how to rectify the PCB. I tried a 10uF capacitor parallel to the PD, a 100nF decoupling the LM358 VCC pin, and I tried using the second opamp on the LM358 as a voltage follower (unity gain amplifier?) between the first opamp and the ADC.

It would be super helpful if you explain to me like I'm 5 as I am a biologist not an electrical engineer.

enter image description here

\$\endgroup\$
2
  • \$\begingroup\$ Have a read through this TI App Note SBOA268A ti.com/lit/an/sboa268a/sboa268a.pdf \$\endgroup\$
    – vicatcu
    Commented Apr 11 at 20:43
  • \$\begingroup\$ Jim Todson and Bonnie C. Baker both used to work at Burr Brown and have written up some nice white papers on photodiode transimpedance amplifier front end stages -- both for noise and for oscillation considerations. But perhaps the first thing Bob Pease writes about, in "What's All This Transimpedance Amplifier Stuff, Anyhow?" is oscillation. \$\endgroup\$ Commented Apr 11 at 20:56

2 Answers 2

2
\$\begingroup\$

Big fat diode, high capacitance, makes a phase shift around the loop. Put a capacitor in parallel with the feedback, between output and (-) input. I'd suggest 1 nF.

\$\endgroup\$
3
  • \$\begingroup\$ This is it exactly. Capacitance on the inverting node is causing oscillations and is fairly typical. Follow John's advice on the cap across the feedback resistance, but also keep your decoupling cap on the Vcc pin of your opamp. No capacitor across the PD, that will only hurt you. \$\endgroup\$ Commented Apr 11 at 20:48
  • 1
    \$\begingroup\$ It's even worse if using a digipot for the feedback as it has its own phase shift. \$\endgroup\$ Commented Apr 11 at 21:06
  • \$\begingroup\$ wow that was exactly the problem! I went with a 1.2nF and all is stable for all my gain settings and illumination levels. This saved my project and sanity so thanks again! \$\endgroup\$ Commented Apr 13 at 19:53
1
\$\begingroup\$

I am going to guess that when installed in a breadboard, the PD is relatively far away from the ground trace (with the emphsis on "trace"). The PCB version has the PD right snug up against the ground plane, separated only by the thickness of the photoresist. Both conditions increase the capacitance between the Vin- and ground. Since you aren't big on circuit theory, I'll just say that, at some feedback gain this will cause the circuit to oscillate. As has been suggested by others, the cure for this is to increase the feedback capacitance. This is done simply by connecting a capacitor between the output and Vin-. Frankly, I'm not willing to do the work to give you a recommendation as to the cap value, although I'd recommend 100 pf ceramic as a starting point. Start at low gain, and increase until you get oscillation. Add some capacitance and increase the gain some more. When you get to a value that just works, add another 50% on the cap. This will provide insurance against op amp variations, temperature-induced component variations, and other bizarre effects you haven't run across yet. Increasing the feedback capacitance (actually, the feedback resistance-capacitance product) will slow down the overall response, but if you are using this configuration you should not expect great speed.If you do need more speed than you can get, you should look into the photoconductive circuit version. You are using the photovoltaic.

\$\endgroup\$
1
  • \$\begingroup\$ I went with 1.2nF and now all is working great thank you! \$\endgroup\$ Commented Apr 13 at 19:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.