Does anyone have any idea how long the sine lookup table is in the AD9951 and AD9834? Maybe I missed it in the datasheets, but I don't think it's in there. I have a noise problem that I can't get to the bottom of using the AD9834 @ 66.666MHz clock to generate spot frequencies from 30kHz - 1MHz.
For best sine waveform reconstuction, am I better filtering at Nyquist or just above 1MHz? (I realise I will not attenuate a lot of the spurs at Nyquist, but I was thinking of the interpolating gradient between 'steps').