I'm designing an amplifier for a sensor that has extremely high impedance and outputs low frequency (not above 20Hz) currents in the picoamp range which I must measure. It's an ion cell and the cathode (nominally at ground potential) collects positive ions.

I am considering two approaches:
(1) Using a very high impedance op amp (LM7721) with the cathode connected to the inverting input and a single high value feedback resistor (~1G) with the non inverting input grounded. This is effectively a trans-impendance amp.
(2) Using the same op amp with the cathode to the non-inverting input and a high value resistor to ground so as to avoid charge build-up and the inverting input connected to the output to form a unity gain buffer stage.

My ion cell produces a stable background current which I need to offset with great care.

I know this sensor is viable because, many years back, I got acceptable performance with older devices and with the cathode connected to the inverting input via a 10M resistor and a 10G (high quality, glass) feedback resistor - non-inverting input grounded.

Now however, I want to take advantage of developments since and minimize resistor noise.

I attach two candidate circuits and would value comments.

 

LMP7721 looks ok.

I would be inclined to use the LMP7721 in a unity-gain non-inverting configuration.
Also I would mount the chip so that pins 1 and 8 are free-floating off the PCB.

 

I’ve not seen the offset technique you have used before – more common is to feed the offset op-amp output voltage via a resistor to the input of the final output stage, with the output of the input stage also fed via a resistor. This will lead to a more stable output offset voltage which is not dependent on the final output impedance.

Since you are only interested in a frequency response up to 20Hz, it might be worth adding a low pass filter at the output to reduce any higher frequency noise.

 

MrChips said:
LMP7721 looks ok.

I would be inclined to use the LMP7721 in a unity-gain non-inverting configuration.
Also I would mount the chip so that pins 1 and 8 are free-floating off the PCB.

Thanks.
What do you see as the downside to the trans-impedance arrangement?
I thought to put a couple of Teflon standoffs and the board milled away beneath the two pads.

I’ve not seen the offset technique you have used before – more common is to feed the offset op-amp output voltage via a resistor to the input of the final output stage, with the output of the input stage also fed via a resistor. This will lead to a more stable output offset voltage which is not dependent on the final output impedance.

Your input is appreciated.
My reasoning on this was that, in that case, I'd be using and amplified offset signal (where the final stage has gain) and the current draw from the input to the next stage is extremely small. Swings and roundabouts? Thanks anyhow, I'll give it some thought again - and, in fact, it is better to do what you say with the unity gain final stage option.

 

Luckily these two arrangements can be executed in a single PCB - with the addition of a couple of optional jumpers. The key difficulty (in my option 2), with an op amp of such huge impedance, is that my input line is effectively a capacitance to ground which will simply keep charging unless I provide a leakage path; otherwise putting the chip into saturation. Doing that, though, is counter-intuitive because the device was picked partly because of its input impedance and the measured current will be a function of the input capacitance.

To this point I can see no downside to using the trans-impedance strategy (my option 1) which has the advantage of directly telling me the actual value of my ion current at any moment - i.e. how many ions are being collected.