[time-nuts] Soundcard sampling

[Lux, James P]

> > The newer DDSes have lower power, faster clocks, lower
> spurs, more bits in the DAC, etc.  Although, I'm not sure how
> important all these things would be in this application.
> >
> > There are also some very low phase noise, low jitter XO
> available that
> > can drive these things (not temperature compensated,
> though, so as the
> > XO temp changes, so does the frequency)
> >
> >
>
> Using a low phase noise XO with a suitable frequency would
> probably be preferable as there would be fewer potential
> problems with spurs.
> With appropriate algorithms the drift in beat frequency with
> temperature shouldn't be too much of a problem.
>

There's low phase noise oscillators intended for the SONET market from Valpey Fisher, etc. that have fairly good performance. A VF161 is a <$50 sort of part and is speced at 1ps jitter (1 sigma, fj>1kHz).  VF has "jitter attenuators" too, which are a low jitter pll with <0.18ps jitter (RMS 12kHz-20MHz)..

Pick a XO at 200-300 MHz or so (whatever frequency is readily available), then set your DDS for a "spur free" frequency (where it divides nicely into the SIN table size)  (on the AD9584 running at 200MHz which has a 48 bit phase accumulator, 17 bit phase table, and a 12 bit DAC, these are about every 3 kHz or so)