[time-nuts] Low noise frequency multiplication

[SAIDJACK at aol.com]

In a message dated 3/2/2007 03:11:25 Pacific Standard Time,  
stephan at rrsg.ee.uct.ac.za writes:

Phase  noise of -115dBc/Hz @ 10Hz for a 100MHz carrier sounds a bit
steep when  compared to one of Wenzel's ultra low noise ULN series
which achieves  -125dBc/Hz @ 100Hz. (these oscillators are probably the
best you can buy).  I guess it will probably climb to about -112dBc/Hz
@ 10Hz. Your prediction  postulate that the close-in phase noise of two
devices, the digitally down  divided 100MHz and the state-of-the-art
100MHz low-noise oscillator, will  be comparable. I don't mean to
contradict you, since I am really not an  expert, but this spec sounds
a little suspect. Maybe there is something I  am missing? If this spec
is correct, I am strongly considering  it.


Hi Stephan,
 
you are probably right, I did make some aggressive assumptions about  the 
phase noise of the Fox oscillator and DDS, that's why I said  'theoretically'. 
But as Bruce described, when dividing a frequency by 10, the  phase noise offset 
also moves closer to the carrier by 10x. Also, a DDS is not  strictly 'a 
divider'. It's a bit more complex, especially since Analog Devices  didn't publish 
all of their specs yet :)
 
Will let you know what kind of phase noise Fox will claim for their Xpresso  
parts.
 
BTW: since you are using the signal as a 'digital' clock reference for a  
DDS, I think you really need to also look at the jitter spec (RMS, Total,  
Deterministic versus Random, etc) rather than just close-in Phase  Noise.
 
Wenzel has some  Excel spread sheets that calculate ADEV from Phase  noise, 
and I am not convinced that it is that easy to go between Phase Noise,  ADEV, 
and Jitter by such simple calculations.
 
For example, Wenzel's calculations don't take into account any spurs, which  
will affect your ADEV, and show up as deterministic noise on the Jitter  
measurement spreading out the Bell curve jitter histogramm, but you will never  see 
them on a close-in Phase Noise plot if they are outside of the Sideband  
noise spectrum being shown (for example if they are inside the other sideband  
bandwidth not being measured).
 
Also, phase noise measurements show all noise, including amplitude, phase,  
and frequency deviations of the analog signal within the measurement bandwidth. 
 Jitter measurements only show the noise present on the zero-crossing of the  
signal - but that with a huge measurement bandwidth, so these will likely  
measure slightly differently. Of course the zero crossings are what affect the  
digital DDS you are driving.
 
Take a look at the Wavecrest instruments (SIA-4000, DTS-2070 etc) these  
measure the signal essentially with a huge measurement bandwidth (2GHz -  >10GHz 
BW) so they capture all relevant noise that will cause jitter. This  compares 
to close-in phase noise which only has a measurement BW of 0.1 to 1MHz  
typically and disregards any noise sources present outside this BW.
 
Not sure what will affect your system more: jitter due to e.g.  deterministic 
noise, or close-in phase noise. You may have to do some  experiments to see 
how well theory matches reality.
 
bye,
Said
 
 
 
 
 
 
<BR><BR><BR>**************************************<BR> AOL now offers free 
email to everyone.  Find out more about what's free from AOL at 
http://www.aol.com.