Hi, we have a prototype using the 18 bits ADC ADS8598H that we need to calibrate.
We are using the +-10V input range and we add 3 op amp differential to single ended analog front end (AFE).
We need this to isolate with a higher impedance the 4 bipolar independant signals that we need to measure for our application.
We want to calibrate the gain and the offset of the 4 channels of the ADC along with the analog front end together (channel 1, 2, 3 and 5 are used... and Channels 4, 6, 7 and 8 pulled down to the analog ground).
We use the classic linear calibration with Y = mX +b.
The gain "m" we find with the 4 channels is OK and stable all the time.
To find the AFE+ADC offset of each channels, we shoot a 0V that reads 0.000000133 V with the precision voltmeter.
With the circuit (AFE + ADC) we read offset values in the range of plus and minus 500 micro volts.
The problem is with the offset. The 4 channels will have a different offset depending on the day and the environment.
For example, one day we get say [-0.000687 V] offset on a channel 1 and the next day in a similar environment we get [+0.0001526V].
This is completely in the opposite polarity with the exact same channel.
Question:
Should the ADC offset be the same (or almost the same) all the time?
If yes, what could explain the offset variation that we observe on the prototype?
The variations are in the hundreds of microvolts.
Please find attached an overview of the design (note that the schematic is based on the ADS8586 but we use the ADS8598H that is pin and footprint compatible).
