Interfacing bipolar amp with an adc - voltage shift?

Hello.

I'm diving into analog biporal opamp circuit design for the first time. Basically I am trying to measure a faint voltage on my microcontroller, so I figured I needed to amplify the signal and then read it by using an ADC. I plan to use the INA116 instrumental amplifier to amplify the signal from a pH probe. The ADC will be AD7791. Originally I planned to use a single supply, but it has been suggested to me that it would be safer to use a dual/bipolar power supply. Now assuming I am powering the INA116 with a +/-6VDC, and I have to amplify a signal of -2000 to +2000 mV, I can set the gain to 3 (Rg = 24 kOhm) to get nearly the full range at the output for maximum resolution. The thing is I now have to lead this signal, which can stretch almost from -6V to +6V to an ADC that can only measure 0 to +5 VDC. How can this voltage range be converted to fit into the 0 to 5 range? The ADC cannot accept negative voltages on the analog input pins. How can I clamp the voltages on the ADC so that they don't exceed the max/min ratings (0 to 5V)?

Best regards!

Datasheets:
INA116: https://www.ti.com/product/INA116/description
AD7791: https://www.analog.com/en/analog-to-digital-converters/ad-converters/ad7791/products/product.html



Update: I kind of came up with this circuit while taking into account some suggestions. Basically I decided to power the INA116 (U1) with a dual supply. The negative input pin, and one of the sensor wires is purpously grounded (coax cable). The gain resistor is set so that the gain = 3. This will amplify the signals from +/- 2000mV up to +/- 6V.

Next I used another amplifier that acts as a signal conditioner. It should have a gain of 5/12 = cca 0.4, and a reference voltage of 2.5V. R1 and R3 are 100kOhm, while R2 and R4 are 43kOhm. So technically this should reduce the input signal range from +/- 6V to +/- 2.5V, and shift it up by another 2.5V, so that we get nearly the full 0 to 5 V range. I purpously used a single-supply amp here because it is impossible for its output voltage to exceed its power supply voltage (0 to 5V). This way the voltage is clamped to the proper range. Finally, an ADC (U3 = AD7791) picks up this signal and converts it to digital data, which is then forwarded over SPI to a microcontroller (not shown). Both opamps, and the ADC use 100n ceramic decoupling caps.

Since this is my first analog circuit of this "magnitude" I'd like to consult some other people about it. I would still like to place a RC filter somewhere. Should I put it in front of the signal conditioner (U2) or after it (just before the ADC)? Also, can I connect both +2.5 voltage references (A nad B) onto a single IC or is it better to use two separate IC's? My concern is U2 where the voltage might rise above 2.5V, and the current will be forced to flow backwards (into the VRef IC).

Finally, what is the recommended way to do a dual/pipolar power supply? I've seen circuits that use two diode bridges and two linear regulators, but I've also seen some IC's that can produce bipolar power out of single DC supply. I think they're called charge pumps?

Thanks in advance for any suggestions and/or comments.

In theory, a pH probe produces about 59 millivolts (mV) per pH unit, and at pH 7 (neutral pH) the probe produces 0 volts. A typical pH probe conducts 1~2nA thus represents an impedance ~ 600mV/1nA=600M approx.
Naturally a FET input Op Amp is needed such as a TL071 but the INA116 gives better CMRR for stray Efield noise.

You must choose a better refV for pH7 to match your ADC midpoint of 2.5/2 and use that for terminating R1 instead of 0V and remove R4.

U2 can then be a 2nd or LPF or 50 or 60Hz notch filter with LPF and a series R output to limit current into built in clamp diodes of ADC.

Can you please elaborate why should I change the circuit the way you suggested?
I was trying to implement this signal conditioner (level converter): https://www.cnblogs.com/shangdawei/p/3190899.html

EDIT: also see circuit #2 in this document: https://www.ti.com/litv/pdf/sboa097

*NUDGE*
I did some more research, and it would seem I misunderstood the meaning of "differential input". Apparently my circuits can only measure 1/2 of the ADC's range since one of its inputs is clamped to a static voltage reference. Looking at the components again, the in-amp produces a bipolar single-ended signal that has to be converted to a unipolar differential signal for my ADC, but is it really worth the effort (even for a 16-bit ADC like ADS1100)? I've been unable to find an IC that does it all in one go, and I do not want to introduce too many components to keep the errors/noise to a minimum.

I did find the LT6350 ADC driver though.