I’m trying to determine the non linearity of the ADC AD9655 with the “histogram” method. Although the achieved results are in the range of the data sheet descriptions, something looks wrong. Instead of a straight DNL trend, I got a "sine wave" trend. My first worry was that this behavior is caused by a not eligible input signal quality (I’m using (as input signal) a 20kHz square wave signal that is riding on a ramp signal (0.02Hz)) – but in the meantime l also have to consider the possibility that the error is caused by an (internal) calibration issue. Can somebody give a hint, how to exclude the (internal) calibration as issue source? Can somebody recommend a strategy to achieve the best internal calibrations?