[hpsdr] High Performance DDS Oscillator source

[Grant Hodgson]

Gerd et. al.

As Chris GW4DGU mentioned, the best option for a UHF clock source would 
be a good crystal oscillator running at VHF and then multiplied up. 
Thanks to help from Chris, I'm working on exactly this - a very low 
noise VHF crystal oscillator and a series of low-noise, no-tune 
doublers.  Using a very high-spec. 125MHz crystal and 3 cascaded 
doublers should give a 1GHz source with just about the lowest possible 
phase noise.  I'm having difficulty simulating the phase noise with LT 
SPICE, although I think it's possible; I'll have a go at measuring it 
but that is not easy either.  Manual calculations indicate -152dBc/Hz @ 
1kHz offset which is about the same as the Wenzel ULN.

Just today (20th Aug) AD have published Rev. C of the 9912 datasheet but 
there's not really much data on the residual phase noise.  However the 
full datasheet for the AD9910 is available, and they seem to have 
characterised the residual phase noise quite well.  Assuming the 9912 
has a similar performance to the 9910, the residual phase noise will be 
about -151dBc/Hz @ 1kHz offset at 28MHz output, and will not be degraded 
significantly by a good clock source.  This could then form the basis of 
a very low noise Local Oscillator source for the HF bands, and would be 
in the order of 15dB or so better than the Orion - which currently has 
the lowest close-in phase noise of any HF transceiver available, 
although I haven't seen any measurements for the Hilberling yet.

I don't know what's used inside the Flex-5000, but given that the 9910 
only became available a few months ago, I doubt that it is used in the 
5000.

Therefore, the AD9910/9912 if clocked by a very low noise crystal 
oscillator/multiplier has the potential to give 'best in class' phase 
noise performance.  Spurs are another issue, but the preliminary plots 
published by Giancarlo indicate the potential for absolutely superb spur 
performance.

If Horton ever gets off the ground then this would seem a logical choice 
for the LO; lets hope that AD get the 9912 released soon - there could 
be a lot of interest in it.

regards

Grant Hodgson  G8UBN
>--------------------------------
> 
> Subject:
> Re: [hpsdr] High Performance DDS Oscillator source
> From:
> "Gerd Loch" <g.loch at nt-electronics.de>
> Date:
> Tue, 14 Aug 2007 09:49:32 +0200
> To:
> <hpsdr at lists.hpsdr.org>
> 
> To:
> <hpsdr at lists.hpsdr.org>
> CC:
> 'Tayloe Dan-P26412' <Dan.Tayloe at motorola.com>
> 
> 
> I am working on a design with the AD9912 and I asked AD which combination of
> clock generator frequency and clock multiplier is expected to give the
> lowest jitter or phase noise.
> The answer from AD was the best choice would be to use a 1 GHz sine
> oscillator.
> 
> Any suggetions how to get an affordable 1 Ghz stable and "clean" sine
> oscillator?
> 
> 73 Gerd, DJ8AY

 1187620092.0