So, I have this differential ADC (AD7641) and plan to use AD8139 as his driver. The goal is to digitize a signal (a sine wave) of about 3.5Vpp and 36KHz. The signal is actually single-ended and centered at 0V, the differential ADC is for pick-up noise reduction. AD7641's input range is up to 2.048V its with internal reference.
AD8139 will be used for both scaling and biasing (I'd be glad if someone could confirm it works for less-than-unity gains) and the scaled-down output from a precision reference will provide the bias.
I'd like to reduce complexity by skipping input scaling and biasing and just use Schottky diodes (or maybe an active rectifier, but I guess that would defeat the complexity purpose) to clip the negative part.
The ultimate goal is to build a lock-in amplifier, so I will need magnitude and phase information from the input signal. Is it OK to just clip the negative part and sort of "mirror" half of the wave in software? What is the best practice?