EMRFD Message Archive 1658

Message Date From Subject
1658 2008-05-23 14:54:02 Weddig, Henning-C... AW: [emrfd] AD9912 DDS clock source - just a thought?

I partly disagree on Your statement!
It is true that by multiplying a low frequency to a high frequency signal the phaase noise (theoreticly) rises by 20 * log N where N is the multiplication factor.
This means that 1 GHz is 100 times of 10 MHz resulting in a growth of 40 dB.
For a PLL this is only true within the loop bandwidth.
Outside this bandwidth the phase noise of the "free running VCO" is dominating.
If the two phase noise curves are known, the loop bandwidth can be determinded at that offset frequency where the two curves (reference multiplied by 20 log N) and free running VCO cross.
A multiplier will rise the phase noise broadband!



Von: emrfd@yahoogroups.com im Auftrag von Art
Gesendet: Fr 23.05.2008 17:57
An: emrfd@yahoogroups.com
Betreff: Re: [emrfd] AD9912 DDS clock source - just a thought?

Hi Kevin,

Phase locked 10 MHz sources are not the best
source if a clean DDS output is needed. The 10
MHz output from any VCO is going to have
relatively high phase noise, and the
multiplication to 1 GHz is going to multiply the
phase noise as well as the original 10 MHz
signals frequency. Use the phase locked DDS if
you need accuracy of the frequency, but not for a
DDS clock if you need a clean output.

If you need a very clean 1 GHz source to clock a
DDS, you can buy a ready made partial kit from
I0CG. He makes a small PCB with a helical
resonator and a MMIC amplifier to produce an
output at 500 MHz (for the 9951 through 9954 DDS
chips). I do not know if he makes a similar product for the 9912 DDS.

I have Valpey Fisher oscillators for sale, which
are known to be clean and are much cheaper than
other high performance oscillators you can get.
Digikey has helical resonators in stock and I
have MMIC's for sale also. A small pcb to hold
the MMIC, oscillator and helical resonator should
not be difficult to fabricate since the layout is fairly simple.

I have been looking for the time to lay out a
board myself, since I have most of the
components, but the time has been illusive and I
haven't gotten it done yet. I just started doing
kits for the N2PK VNA, which is a major time
sink, so the outlook for finding teh time for the
DDS clock generator looks poor (in the immediate future).



>With the release of the AD9912 DDS there has been a surge of
>interest for a good 1Ghz clock source.
>Not fully understanding the issues, would the proposal below be
>worth experimenting with?
>Crystek have recently introduced a new VCO - CVCO55CX-1000-1000, the
>performance specs look
>pretty good to me. If this was phase lock looped to a 10Mhz GPS,
>would this be sufficient
>as a clock for the AD9912.
>10Mhz --- Phase det --- Diff Amp --- VCO -- Buffer Amp --- 1Ghz out