Help needed with off the shelf amplifier solution

[Cicada]

Hi

I know what I am asking for below can be easily designed and built. Depending on experience level this could be done in an iteration.  But I do not have the time to do so. And due to deadlines and procurement constrains at my company buying a cheap solution would be by far the best option for me.

With all the above in mind. Can someone point me to a solution that I can buy from a Digikey, Mouser, TME, TI or Analog devices that will just do the job below for me?

SPECS:
We have OEM equipment that has 4 x 10k resistors connected to metal plates in which current are generated when ionising radiation hits these plates. The currents from these plates will flow through the 10k resistors and will be measured by their equipment (high input impedance) as well by ours over the same resistors. We want to do this as accurate as possible after the system has been calibrated for the exact values of each 10k resistor. We thus need our equipment to have an input impedance that is as high as possible. I think our input leakage current at the highest input voltage should not exceed 1nA. 10's of pA would be better.

The currents from the plates and voltages over the resistance will look as follows:
1. 100nA  = 1mV
2. 1uA  = 10mV
3. 10uA  = 100mV
4. 30nA  = 300mV

The 1mV to 300mV voltages above will oscillate at about 8kHz and must be amplified by a gain of about 35 times for the voltage digitiser module that will be used to read it.
So the voltages fed from this module to the digitiser will be:
1. 1mV x 35 = 35mV
2. 10mV x 35 = 350mV 
3. 100mV x 35 = 3.5V
4. 300mV x 35 = 10.5V

It would be good if the gain of the equipment is variable or if I can change all 4 channels to say a gain of 30 by just changing resistors on the board.

All 4 channels does not have to be on the same board.

Even though the circuit will be used in air conditioned surroundings the temperature can still fluctuate so if good 10ppm and below resistors are not used you can get quite substantial changes in gain.

Are any one here aware of commercial boards that won't cost an arm and a leg that might fulfil the requirements above. I plan to built hem into a metal enclosure with BNC connectors interfacing to the outside world. This box can then also contain a AC to DC power supply for these boards. Or maybe a wall wart with DC to DC or linear regulator inside.

I have looked for solutions but can only find evaluation boards that are not suited.

Thanks
Cicada

[TimFox]

Your 8 kHz oscillation:  is this because the radiation source is pulsed at that repetition rate?
If so, what is the pulse width and duty factor (fraction of ON time)?

[Cicada]

Your 8 kHz oscillation:  is this because the radiation source is pulsed at that repetition rate?
If so, what is the pulse width and duty factor (fraction of ON time)?

I hope the picture attached give more clarity.

The deltaT periods in the signal graphs represent the time the beam move from one quadrant plate to the next. During this period the beam spot is on both plates.

The rising edge of the signals is the highest frequency components in the signal. Much higher than the 2kHz at which the beam sweeps.
I reckoned 8kHz might be a more correct representation (or safer bet) of a 2kHz signal with a 25% duty cycle. Didn't think this trough.

Cicada

[TimFox]

The application appears to be alignment of the radiation beam.
What response time do you need to changes in the beam position?
Do you have a spec on the stability of your signal amplitude measurement?

[Cicada]

The application appears to be alignment of the radiation beam.

Correct.

What response time do you need to changes in the beam position?

To be honest I don't know. Someone else will be doing that part but my guess would be no faster than 500us purely based on the 2kHz. But knowing what I know about the rest of the system then I would say much slower . Maybe on the order of 100ms to correct.

Do you have a spec on the stability of your signal amplitude measurement?

No.

[TimFox]

Would you be able to use four of these boards?  I know nothing about this company, but you could check their website from the amazon listing.

https://www.amazon.com/Teyleten-Robot-Millivolt-Microvolt-Instrumentation/dp/B08QM9SM1W/ref=asc_df_B08QM9SM1W/?tag=hyprod-20&linkCode=df0&hvadid=564730145269&hvpos=&hvnetw=g&hvrand=14611257232723781977&hvpone=&hvptwo=&hvqmt=&hvdev=c&hvdvcmdl=&hvlocint=&hvlocphy=9021749&hvtargid=pla-1629748634375&psc=1

[Cicada]

A simple circuit like this would have done the trick probably.
Just make it 4 channel and add the power supply circuitry.
But as usual there is no time so I have to buy.

[Cicada]

Would you be able to use four of these boards?  I know nothing about this company, but you could check their website from the amazon listing.

https://www.amazon.com/Teyleten-Robot-Millivolt-Microvolt-Instrumentation/dp/B08QM9SM1W/ref=asc_df_B08QM9SM1W/?tag=hyprod-20&linkCode=df0&hvadid=564730145269&hvpos=&hvnetw=g&hvrand=14611257232723781977&hvpone=&hvptwo=&hvqmt=&hvdev=c&hvdvcmdl=&hvlocint=&hvlocphy=9021749&hvtargid=pla-1629748634375&psc=1

This does look interesting. Looking in the AD620 datasheet I see Ios = 1nA. This I can live with but the Vos of 50uV is a bit high. This equates to an additional 5nA error. 50Uv/10K = 5nA. So the two errors together = 6nA. This is a 6% error when measuring 100nA. So yes 6% total error at 100nA is probably not that bad.

[TimFox]

That board has an offset trimpot for voltage offset.
Since your source is pulsed, you can consider AC coupling the output from the amplifier:  the peak-to-peak will go to zero when the signal vanishes.