[hpsdr] PA3AKE Project update - DDS

[Grant Hodgson]

Interesting that on a discussion forum for High Performance (SD) Radio 
there are far more comments about web browsers than about the subject 
matter - which is one of those nasty RF-techy type things; sign of the 
times I suppose?

Martein should be congratulated on doing a superb job, not just on 
finding a design flaw in the AD9910 but also in the meticulous way that 
he has documented his design and development work.

As for the AD9910, and presumably the 9912 - it may not be all that bad. 
  In a nutshell, the AM noise of the DDS exceeds the phase noise by a 
considerable margin, which is very unusual and just about unprecedented. 
However, I think it is reasonable to assume that Analog Devices assumed 
that the 9910/12 would be used as some sort of LO, in which case AM 
noise should not normally be an issue as it should be suppressed by the 
action of a mixer, limiter or both.

It seems that Martein's use of a 74AC04 may have introduced some 
(considerable) AM-PM conversion, and therefore it may be prudent for 
anyone planning to use one of these otherwise superb DDS ICs to take 
care with the limiter, particularly with regard to AM-PM conversion.  It 
will be interesting to read Martein's future work on this.

On a related topic it appears that the spurs from the 9912 are 
predominantly AM spurs, NOT FM spurs - which is also very unusual.  The 
consequence of this is that if the DDS output is frequency multiplied, 
many of the spurs do NOT increase at the rate of 20 Log N as they do 
with other DDSes - they appear to stay at the same level - regardless of 
the multiplication factor.

One consequence is that there could be some novel ways of generating LO 
signals for use at VHF/UHF or even higher, by using a 9912 at a 
relatively low frequency (somewhere in the HF range) and multiplying to 
VHF.  This might generate less spurs than running the 9912 directly at 
VHF/UHF, with little degradation in phase noise (compared to running the 
9912 at VHF).

Note - this applies only to the 9912, not the 9910.  The 9912 has a 48 
bit wide phase accumulator which I believe is the reason for the very, 
very low levels of FM spurs.  The 9910 has only a 32 bit accumulator.

If AD could be persuaded to modify the design of the 9912 to improve the 
AM noise, then the AM spurs might get even better still, but don't raise 
your hopes.

regards

Grant G8UBN



> 
> A shame, too, because he's got some great content there.  The business about
> the missing DAC bandgap-reference bypass pin on the AD9910 is particularly
> interesting.  If that assessment is correct, Analog really slipped up!
> 
> -- john, KE5FX
>

 1231090526.0