I am currently attempting to characterize the AD7689's SNR.
The problem I am encountering is that I do not seem to be able to reach datasheet values.
This is my setup :
Instrumentation generator delivering ~19977Hz sine wave (110dB THD, 100dB SINAD), it's output buffer is AD797BRZ
...sampled coherently on AIN0 on the AD7689, 10cm of PCB trace after, @ 200kHz. VDD = 5V, Vref External = 5V, both coming from different very low noise PSUs
REFIN has 1 100nF X7R capacitor a few millimeters from the component.
REF has 100nF + 10uF X7R capacitors a few millimeters from the component, same for VDD and VIO.
Since the sampling is coherent, no windowing is applied to the resulting FFT.
Register configuration is the following :
1 -> CFG=1 overwrite contents of register
1 - Unipolar, INx referenced to COM = GND ± 0.1 V
0 - selecting ch0
1 - full BW
1 - external ref, internal buffer on, everything else disabled
0 - sequencer disabled, reading only ch0
1 - do not read back
This is the FFT I manage to get with 64k samples and a signal that is "almost full scale".
Here are my 3 questions :
1) Do you see anything wrong in the used setup here?
2) How many points do you use for your FFT, in characterization and in testing?
3) Is there an "optimum configuration" that you used for your datasheets that we could use to reach those specs?
Thanks for your help.