Line filtering in transimpedance amplifier

[RawCode]

Hello there.
Considering a transimpedance amplifier, which is sensing the current of a very resistive device.

(X3 is a current buffer)
Since is not possible to capacitive load the "- terminal" due to stability issues(C9 is the load's parassite capacitance), how is it possible to filter out the 50Hz line interference picked-up by the resistive load without altering the current measurement?
I tried to put a twin-t notch filter between X1 and X2, but the circuit becomes unstable.

[Vovk_Z]

50 Hz is quite a low frequency. I'm not sure we can filter it and make a transimpedance amp to do it's work simultaneously.
An additional filter after the amp will of cause work.

[RawCode]

That's true, but if it's filtered after the opamp, the latter can still saturate due to 50Hz noise.
it's necessary to filter it before it reaches the opamp input, but how?

[magic]

I tried to put a twin-t notch filter between X1 and X2, but the circuit becomes unstable.

Not sure why some many opamps in the feedback loop.
Adding anything between the opamps surely can't help because it only affects open loop response. Once you close the loop with the 10MΩ resistor, the first opamp will attempt to drive your notch filter in such a way to reproduce the 50Hz signal on the output regardless.
Add a 50Hz bandpass (lowapass/highpass if that's enough) in parallel with the feedback resistor.
Or add some filter on the input and isolate its capacitance from the opamp's input with a series resistor. But this will increase the input impedance of the TIA, as seen by the current source.

[RawCode]

If i understand well, if i put a bandpass filter at 50Hz, this interference will not be amplified by the amplifier, so it should not saturate the amplifier, right?
I put X2 to get a x100 gain, avoiding the use of a 1GOhm resistor (is it a good idea?) and X3 to have a current boost  to be able to read larger current when i change the feedback resistor to a smaller values.

[magic]

Yes. If the feedback network has low impedance at 50Hz (rather than 10MΩ like now) then lower output swings will provide the same input current and the output will be less sensitive to 50Hz noise.

Your second opamp only increases the open loop gain of the first one by 100x. You should connect the 10MΩ resistor to point X1 if you want 100x more closed loop gain.

[RawCode]

Thanks for the suggestions
I placed a bandpass filter in the feedback path, but in AC analysis there isn't a "hole" at 50Hz. It seems that it doesn't changes a lot in the response. Is it normal?

I removed X2 from inside the loop as you suggested, thank you
Do you think that with a 10MOhm resistor in the feedback path and a 100X amplification would be possible to sense nanoamperes?

[magic]

This filter is backwards (input swapped with output) and it's not the kind of thing needed here. You need something with low impedance at 50Hz and high impedance at signal frequencies.

If signal frequencies are <50Hz, that's easily realized by increasing C2.
If signal frequencies are >50Hz, well, I'm out of ideas because I have little clue about active filters and the obvious passive filter (inductor in parallel with 10MΩ) would be impractical, requiring a few H of inductance if my math is right. Sorry.