Split-rail or single supply for ADC/PGA with differential input.

[bluespider]

I am redesigning a circuit that has an ADC IC - The existing version uses a +/-2.5V supply for the ADC and its integrated PGA. The input is isolated differential, and the circuit is battery powered. The have always been more parts on the circuit board that required a negative voltage supply (+/-5V). Those parts are no longer needed so the only negative voltage required is for the ADC.
The other option is to bias the mid-point of the differential input to Vref/2 instead of GND and have a 5v-GND supply on the analog side. This would simplify the power supply of the overall circuit. Are there any other considerations to going this route that a beginner might not have considered?

[Andy Chee]

What is the analog signal going into the ADC?  Microphone audio? Air flow/pressure sensor?  Light sensor? Magnetic field?

[bluespider]

Currently a geophone (magnet moving in a coil), but will possibly add a hydrophone (piezoelectric) later - will require a high-impedance input.

[Andy Chee]

In the interests of preserving low frequency fidelity, I would be inclined to retain any existing DC coupling, and keep the bipolar power supplies.

However, I'm not familiar with geophones nor the typical data it generates, so I'm making some assumptions about your frequencies of interest.

If you do decide to go with Vref/2, it is important that the input AC coupling capacitor has enough low frequency response, otherwise your data may be distorted.

[bluespider]

For our purposes >2Hz would be fine. I've attached the current input circuit (minus the clamping diodes) and an example given in a datasheet for a different (but similar) ADC - which shows the alternative.