[hpsdr] High performance DDS Oscillator

[Grant Hodgson]

Gerd

The comments from AD about the noise of the VCO dominating the DDS phase 
noise performance only relate to using the on-chip clock multiplier. 
The best performance will be obtained with an external 
crystal-controlled clock.  Then the true potential of the 9912 can be 
achieved.

We need to wait for AD to provide a full datasheet with the residual 
phase noise fully characterised, as per the AD9910.

Phase noise can cause a number of problems; the most significant for 
HF/VHF is when you are trying to listen to a very weak signal which is 
very close in frequency to a very strong signal.  Assuming the 
transmitted signals are perfect, and have no phase noise of their own, 
then the phase noise of the receiver will mix with the (unwanted, 
strong, off-channel) signal and will raise the noise floor within the IF 
passband, which may mask the weak signal and make it unreadable.

For SSB this can be a problem with phase noise at offsets of a couple of 
kHz up to several MHz, for CW with narrow filters the phase noise at 
offsets as close as a couple of hundred Hz can be a problem.

You can sometimes hear this quite easily on some radios; just listen 
close to a very strong signal and you may notice the noise floor go down 
when the strong signal goes away.

If there is no strong signal, phase noise is not usually a particular 
problem.

Hope that helps.

regards

Grant

-----------------

Gerd wrote -

Grant,

meanwhile I received the information from AD that the phase noise of the DDS
itself would limit the total phase noise to a level which would make it not
be effective to minimize the phase noise of the clock oscillator to a very
low level:

<The exact oscillator doesn't really matter, as the noise pedalstal will be
dominated by the VCO on-board the <AD9912.

which does not mean that the noise level is not pretty low.

Do I understand the influence of noise of the DDS-oscillator in a
directconversion system correctly when roughly presuming that the noise in
the range of 100 Hz to 10KHz can result in noise in the audio range what you
hear (and see) and from 10kHz up to i.e. 196 kHz what you see on the
panadapter and noise and spurs above 200kHz can produce additional "mirror"
signals with real signals and spurs which can be received within the
receivers bandwidth?


73 Gerd, DJ8AY