The recommended calibration procedure is as follows:

1. Preset the offset and gain calibration registers to 000000h and 400000h, respectively.
2. Perform offset calibration in bipolar input mode by shorting the ADC inputs, or short the inputs at the system level to include the offset error of the external amplifier stages. In unipolar mode, perform offset calibration with inputs greater than 0 V to avoid clipped readings that occur with inputs less than 0 V. Acquire conversion data and write the average value of the data to the offset calibration registers. Averaging the data reduces conversion noise to improve calibration accuracy.
3. Perform gain calibration by applying a calibration signal to the ADC input or at the system level to include the gain error of the external buffer stages. For the standard input range mode, choose the calibration voltage to be less than the full-scale input range to avoid clipping the output code. Clipped output codes result in inaccurate calibration. For example, use a 3.9-V calibration signal with V_REF = 4.096 V. When operating in the extended range mode, the calibration signal can be equal to V_REF without causing clipped output codes. Acquire conversion data and average the results. Use Equation 16 to calculate the gain calibration value.
   Equation 16. Gain Calibration Value = (expected output code / actual output code) · 4000h

   For example, the expected output code of a 3.9-V calibration voltage using a 4.096-V reference voltage is: (3.9 V / 4.096 V) · 7FFFh = 79E0h.