In order to characterize the ADC I'm working with, I try to create and 
calculate a worst-case-error-budget.
For that I use Gain Error, Offset Error, INL, DNL and Quantiziation 
Error (all in mV) and add them to my ADC-Conversion result. From the 
information about the ADC specified in the µCs datasheet, i use the 
maximum values for those parameters. For example the maximum gain error 
is 25 LSB, which corresponds to a voltage of about 8.5 mV.
I also learned, that because the Gain Error becomes greater, the higher 
the input voltage, I have to convert the absolute value to a relative 
gain error, so that I have a value for every step of the transfer 
function.
So my formula is: V_Out = (1 + Gain_Error_rel) * V_in + Offset_Error + 
INL + DNL + Quant_Error
Is that correct? If not, I'm looking for a correct formula as well as an 
explanation as to how which error is handled in calculations like that. 
Sources on that matter, especially on how to handle Gain Error in such 
calculations would be greatly appreciated.