ADS124S06: System Calibration and self calibration

Hi Suzuki-san,

The system calibration can be either offset or gain to adjust for external component error. For example, if you have an amplifier in front of the ADS124S06 that has an offset that you wish to remove from the measurement. By setting the input to what would be considered a 0V input you can adjust for the sensor path and ADC offset. The offset value is stored in the OFCAL calibration registers to be subtracted from subsequent conversions. Similarly the system gain calibration can be accomplished by setting the input to full-scale and issuing the system gain calibration command. Once completed the scaling factor is saved to the FSCAL gain calibration registers. Always issue the offset calibration prior to the gain calibration.

In many cases there is only the sensor that is connected to the input. In this case you only need to compensate for the offset of the ADC. When issuing the self calibration for offset (which this is the only self calibration for the ADS124S06), an internal short is created at the device mux to short the AINP and AINN inputs together. Any offset difference from 0V is placed into the offset calibration registers which will then be subtracted from all subsequent ADC conversions before they are read out of the device. There is no need for a self gain calibration as the device is factory trimmed for the various PGA settings.

Best regards,
Bob B

Hi Suzuki-san,

As I mentioned in my previous response, there is no self-gain calibration because the ADS124S06 is factory trimmed for each of the PGA gain settings.  In other words, the gain is already calibrated for the device.  System gain calibration would take into account the gain error of devices outside of the ADC such as the sensor, amplifier, etc. and would correct for the total system gain error.

Gain error is the difference of the actual output slope relative to the ideal and is most pronounced at full-scale.

If the gain is corrected by calibration (either by system calibration of factory calibration) so that the endpoints match the ideal, then the INL error is the maximum deviation from the ideal slope.  If the gain error is uncorrected, then you need to do a RSS of all the error sources.  So the gain error is the shift of the linear slopes and the INL error is the deviation at the maximum point.

Another way to think of it is the gain error is the shift in the slope of two linear lines and the INL error is the deviation similar to a curve (or how closely the ADC fits the ideal response) from the ideal linear line.

Best regards,

Bob B