[time-nuts] Low noise frequency multiplication

SAIDJACK at aol.com SAIDJACK at aol.com
Thu Mar 1 13:13:14 EST 2007

In a message dated 3/1/2007 05:50:30 Pacific Standard Time,  
stephan at rrsg.ee.uct.ac.za writes:

Hi  Said,

It seems there are indeed many ways to kill a cat. What happens  to the
close-in phase noise using this method?  



Hi Stephan,
you are asking the right questions :)
The near-carrier phase noise depends on the 1GHz reference  source:
   1) If it is a crystal such as the Xpresso 1GHz series from  Fox, then the 
phase noise of the 1GHz reference will actually be reduced by 20dB  for every 
1:10 reduction in output frequency!
   2) If you are using a standard (noisy) VCO, then the phase  noise of your 
10MHz reference source will be increased (multiplied) by the  standard 20dB 
for every 10x increase in output frequency.
This is due to the fact that when using a VCO, the PLL will use  the 
reference crystal to actually reduce the phase noise of the VCO. You will  likely use 
a loop filter with 1 - 15KHz BW. So you get the 10MHz  reference noise 
multiplication inside this loop filter BW. Outside the BW,  it's the DACs and VCO's 
noise floor.
But for a crystal 1GHz reference, you would use a simple PLL with an  
extremely small loop filter BW (say <1Hz) so that your total phase noise is  only 
dependent on the 1GHz crystal reference, and not at all on your 10MHz  reference 
anymore. The PLL could be as simple as an Exor gate in this  case.
I have been trying to get phase noise/jitter specs from Fox for their  new 
1GHz Xpresso crystals, but it is hard to come by. I am waiting for  feedback 
from them. They only specify "UI"!?
But let's say these are as good as advertised, and for me that would mean  
say better than -95dBc/Hz at 10Hz offset from 1GHz carrier, then by reduction  
through the DDS, we would theoretically get -20dB below this, or -115dBc/Hz at  
100MHz output.
I would assume that at that point the noise floor of the DDS DAC chip is  the 
limiting factor at <-150dBc/Hz as it is for the well-known AD9858 chip  (it 
actually measures about -155dBc/Hz).
AD has new chips coming this month, see for example the AD9957 with a 14  bit 
DAC (9858 has 10 bits), I assume these will have an even lower phase noise  
floor and very important lower spurs than the AD9858, the datasheet  says TBD:
<BR><BR><BR>**************************************<BR> AOL now offers free 
email to everyone.  Find out more about what's free from AOL at 

More information about the time-nuts mailing list