[Lf] [Fwd: LF: DDS Sources in receivers]

[Andre' Kesteloot]

Talbot Andrew wrote:

> A few EMails to this reflector have recently been knocking the use of
> DDS sources in receivers.  Well the facts are as follows :
>
> A DDS does indeed generate spurious products, but these are precisely
> characterisable and at a level defined by the chip manufacturer.  Apart
> from image and harmonics, they are a function of the DDS process and NOT
> SUBJECT to poor layout construction etc whereas a phase locked loop
> synthesizer is unbelievably sensitive to noise etc in the control
> circuitry caused by poor design.   Furthermore, the spurs are usually
> well away from the wanted frequency so properly applied filtering will
> eliminate any possible spurious responses.
>
> Quantisation noise, cited on this reflector as being a performance
> limiter is a function of A/D resolution and specificed in the data sheet
> and is the only jitter present as mentioned by a previous posting.
> This is simple DSP theory and defined precisely by maths plus any
> degradation caused by manufacturers implementation, and Analog Devices
> are a chip maker who know what they are doing in the DSP world.  There
> is no magical 'other source' of signal degradations caused by the
> digital process.
> In the case of the AD9850 spurious products and quatisation are at least
> 60dB down with the specified output filter.  Even without the proper
> filter, the main products present are then the image response at
> Fclock - Fout  plus  harmonics, usually well removed from the wanted
> freq when at LF.  In some cases these 'spurious' products can be treated
> as the wanted ones and extracted by bandpass filtering instead, giving
> access to DDS derived accuracy and phase noise at quite high VHF
> frequencies.
>
> Phase noise is excellent on a DDS, in fact, it is equal to that of the
> clock divided down by the synth ratio. So if the clock is generated by
> something like a 100 MHz butler crystal oscillator as used by many
> microwavers, that is probably the best phase noise that will EVER be
> achievable at LF to HF, several orders better than a good LC osc and
> making the spectrum of a PLL synth look like wading through a cornfield.
>
> Phase noise is a vital parameter for LF receivers as it directly sets
> the minimum theoretical bandwidth of a signal.  Conversely, for a PLL
> phase noise is never that good unless a very high Q narrow bandwidth VCO
> is used.  In this case it becomes a specific design useful only for its
> one design purpose.  A common DDS design will serve as optimum for the
> entire output range
>
> Any one who advocates using a PLL just to 'clean up' a DDS is asking for
> the worst of all worlds.  A valid reason for doing this however, is if
> the DDS is used as a reference for a microwave synth, which then
> multiplies the high resolution signal to the microwave range whilst
> keeping a high reference comparison frequency to minimise phase noise.
> For example, a 3 GHz synthesizer using a 50MHz DDS derived reference
> will have steps of less than 1 Hz and a phase noise at bandwidths within
> its loop bandwidth - ie. close in - probably acceptable for SSB use.
>
> Andy  G4JNT
>
> --
> The Information contained in this E-Mail and any subsequent correspondence
> is private and is intended solely for the intended recipient(s).
> For those other than the recipient any disclosure, copying, distribution,
> or any action taken or omitted to be taken in reliance on such information is
> prohibited and may be unlawful.