we already used AD7710 with a very simple schematic as attached to read a pressure transducer output. Our board was a simple two layer PCB,without isolating digital and analog ground, no polygon, no low noise regulator for reference and no isolating from our micro controller to motor driver. the system resolution in this scenario was 16bit free of noise with 5 SPS.
In our new design we use AD7176-2 with ALL considerations which we didn't care on the previous one (the schematic is attached) + EMI film capacitors for excitations and also chebyshev digital filter on our microcontroller side. The accuracy of the system in this scenario is around 19bit free of noise with 1000 SPS BUT we face a big issue which is voltage drifting !
In the old system when we calibrate the system after even a long period of time we don't have even 0.5% voltage drift. However in the new system with the same mechanic and the same sensor specifications we have 3-5% voltage drift few hours after calibration and it is day to day varry. The interesting thing is when we reset our microcontroller (ARM SAM4E) the value drift 1-3%.
Anyone face this issue before? or can guess what could be the issue?