Hello,
I'm trying to use raw ADC count data to get spectral plots corresponding to those in the data sheet. Seems simple but my noise floor is about 40 dB off from that of the data sheet plots on a dBFS scale. So, I terminate the ADC input with a resistive load to ground, sample at 65.536 MSPS ((for a final spectral line resolution of about 3 Hz (took a lot of data)), scale the data by the maximum ADC count value (2^14 - 1 = 16383), do an FFT in MATLAB, take the absolute value of that, convert to dB (20*Log10(fft data) ), offset the plot by (20*Log(1/fs) = -156 dB) to get it to correspond to a Fourier Transform, So, my spectral noise floor is 40 dB too low. I'm not seeing a problem here except for the wrong answer. What am I missing? Thanks.