I would like to calculate the expected bit accuracy of the AD974 if I use the internal reference. I see two tolerances that seem to dominate the accuracy error and that is the full scale error and the internal reference tolerance. Does the worst case full scale error of .5% tell me I could be off by as much as .0125V (.5% of 2.5 nominal), equivalent bits and the ref tolerance tell me that I could be off by +/-20mV (2.48-2.52) or equivalent bits. Also there is a DC code uncertainty described in the spec. Should I also include this error in addition to the linearity error, no missing codes and power supply sensitivity?
Thanks for the info.