OPA392: Transimpedance amplifier for photodiode using OSI Optoelectronics PIN-10DPI and OPA392

Hi Zisimos,

here are my recommendations:

1. Check with a fresh OPA392 and a fresh PIN-10DPI.

2. Do not connect the output of OPA392 directly to the ADC input but insert an isolation resistor to isolate any capacitive load from the output of OPA392. This can stabilize your circuit by restoring the phase margin.

3. 200pA across 2M4 would mean an voltage at the output of OPA392 of 480µV. Take another OPA392 and fabricate a buffer providing a gain of 100. Connect its input to the output of U1 (don't forget the isolation resistor). You should measure a voltage of 48mV at the output of this buffer (don't forget the isolation resistor). If this is the case, the error is at U1. If not, then it is an issue of the ADC. When doing this have U1 connected to the ADC and have U1 disconnected from the ADC. Check both scenarios.

4. The detector capacitance is with 9.8nF very very high. This can make trouble with the phase margin. 2M4 and 100p in the feedback loop would still be stable:

But there's a very low phase at 85Hz:

And with 24k and 10n or 240R and 1µ in the feedback loop the circuit will become instable:

zis_opa392.TSC

5. Connecting the photodetector to the input of OPA392 via a cable can make the circuit detect and demodulate EMI. This can result in an unexplainable offset voltage at the output of OPAmp. You can check this by replacing the photodiode by a 8.2...10n capacitor. Take care that you use exactly the same setup when doing this, as if you would connect the photodetector.

6. In a second step omit the cable and connect the photodetector directly to the input of OPA392. Keep in mind, that by connecting the cable on and off, all the time, you risk to damage the OPA392 by ESD.

Kai

Hey Zisimos, 

Just looking at the 2.4M condition, it should be stable, the big phase dip at lower F is normal for Zt - what matters is at LG crossover. 

Oddly, i was just in a discussion with OSI on C vs reverse bias curves on these parts - they don't have that, but they do warn you not to reverse bias them very much or permanent damage might result - I recall something like a 0.5V is ok, but something like 2.5V might cause damage. I wonder if maybe at power turn on you might have transiently damaged the diode - perhaps a BAV99 clamp diode on the op amp inverting input to gnd might be prudent. one in each polarity. 

Also, I think you need supply caps on each supply to ground at the op amp, I only see supply to supply cap here,