AD9142A Internal NCO Poor Phase Noise

Hi EHort ,

Thank you for using AD9142A.

Kindly allow us some time to review your concern, and we will provide an update accordingly.

Thanks and kind regards,

Alex

Thanks! Let me know if any additional information on our setup is needed... I am getting very repeatable results that look nearly identical to my original post (no NCO -> great phase noise ... with NCO --> much worse).

Hello, just pinging to see if there is any update on this inquiry? Thanks

Hi EHort ,

Apologies for the late response.

The behavior you’re seeing is normal and commonly observed when comparing a baseband tone to one generated by the internal NCO in high-speed DACs like the AD9142A.  Enabling the NCO introduces extra digital processing stages (as per Figure 47 of rev B datasheet) that create spurs and artifacts that can look like degraded phase noise.

To reduce phase noise using NCO, use high interpolation (4x or 8x) to move images farther away and smooths ripple.  Slightly adjust the output frequency, as certain ratios produce more truncation spurs.  Below are results for lower and higher interpolation settings.

Thanks and kind regards,

Alex

Alex,

Okay so the data you show maps to what we are seeing as well... 

I still don't understand why this is expected behavior nor why the interpolation factor would impact the phase noise introduced by the NCO...

My understanding is that the NCO operates at full FDAC rate and interpolation factor should only effect the input data up conversion and filtering prior to being mixed with the NCO...

So with a DC input... It seems like the NCO performance (theoretically) be independent of the interpolation factor...

Eric

Hi EHort ,

Okay... So we have in-house NCO solutions that do not introduce artifacts like this ... So this seems like a design issue within the AD9142A.

And this amount of spurs across the entire close-in phase noise spectrum is likely unacceptable for our application.

I would like to say we can just run without the internal NCO and run only at 2x interpolation for a maximum output of roughly 460MHz... But that is fairly limited versus the advertised "1600MSPS" (~800MHz 1st Nyquist).

2x interpolation -> FDAC max = 1150MSPS -> 2x HB filter -> 0.4x BW ~ 460MHz.

This does seem like a major limitation of the part though... Recommend that the ADI team considers adding this limitation into the datasheet to inform future customers.