In the pictures below you can observe the same signal at two different NCO frequency.
Signal is a DVB-S2 QPSK at 200 MSymbol/s, Roll-off=0.20
- first picture: NCO @ 3.4 GHz, EVM is 8.5 %
- second picture: NCO @ 2.8 GHz, EVM is 2.3 %
So we have a ~6% increase in EVM!
Moreover, in the first the first picture we can observe a very strange artifact in the constellation diagram!
The only difference between the two captures is the main nco settings.
We tested this behavior on different AD9081s and we observed that some DACs tend to exhibit the same pattern, some more than others, so we are working under the hypothesis that this is a part-to-part different behavior.
The DAC frequency is 12 GSAPS, the JESD TX is configured with 8 lanes carrying 2 GSAPS (hence total interpolation factor is x6)
In the last image we can observe EVM by changing NCO values besides the two shown here (2.8 GHz and 3.4 GHz)
Is this an issue with DAC's self calibration? Is it possible to explicitly enable background self calibration?
