Using a photodiode with low shunt resistance?

[FvM]

Referring to your original problem statement:

Summarily, how do you get high-sensitivity readings off an InAs photodiode?
It should be useful well into pA range, according to the descriptions.

I don't know the exact data of your diode, but I see a sensitivity in a 10 pA order of magnitude for 1 Hz bandwidth and room temperature according to NEP numbers of similar diodes. But I don't think that you'll be able to achieve it down to DC signals.

 

[Cpak]

Hmmm... sorry, nothing should change by adding a voltage amplifier after the initial photodiode amp because the problem with that initial stage is input referred voltage noise. That noise would then just get amplified by the second stage like it would have gotten amplified by a larger feedback resistor, so maybe it's a good thing that it didn't change.
I'm not sure I have any suggestions for you at this point. It's a good problem waiting for a solution. I read about building circuits for induced voltage pickup rejection that employed a differential architecture like in an instrumentation amp, using the CMRR of the last op amp stage to reduce common mode noise. I've been wondering if the input referred voltage noise would show up as common mode noise in this case if you have good resistor matching. It's figure 10 in TI app note SBOA35.

Sorry, I'm not sure there are any silver bullets in this case. It would be a good experiment to see what sort of changes you can observe by cooling the diode. You can use a little spray of upside down canned air on the diode package or cold spray if you have it.

 

[Artlav]

Well, apparently anything better than 1e7 range of V-to-A amplification is not going to work well for this diode, and a single stage transimpedance amp seems to have the least amount of extra noise.
Fortunately, i can change the optics to shave off the rest of the sensitivity requirement.
The drift problem is solved by obscuring the diode every second, thus getting a reference point - the drift is slow enough to assume it's constant over one second.

Thanks everyone for help.