[time-nuts] Fury - Rubidium - PIS

SAIDJACK at aol.com SAIDJACK at aol.com
Wed Jul 28 04:52:53 UTC 2010


Hi Scott,
 
yes, Rb's tend to have many orders of magnitude smaller control ranges than 
 OCXO's, so it takes longer to phase lock. What's even worse is hysterisis, 
eg.  we have seen units that jump back and forth between two control 
voltage values.  This tends to happen in TCXO's and Rbs from what I can tell, not 
so much in good  OCXO's.
 
It seems that these units do not change their output frequency until the  
EFC voltage change reaches some threshold, then the frequency jumps in a 
large  value. OCXO's seem to react instantly to even the slightest LSB changes 
on our  24 bit DAC. Many Rb's and TCXO's seem to have this spring-like 
effect, and that  causes some "chasing" of the phase as the oscillators "ignores" 
the EFC voltage  changes until they reach a certain level, at which time 
they over-compensate and  the effect reverses. 
 
I am wondering if adding some dithering noise would help prevent this  
effect, where the noise bandwidth is below the Rb's control bandwidth so as not  
to add ADEV instability...
 
Does anyone know how popular Rb's adjust their frequency? Is it through  
linear methods such as Varactor diodes? Or do they use an ADC to sample the 
EFC  voltage, thus creating quantization errors that could be the cause of the 
 hysterisis we sometimes see?
 
Brian, on a locked DOCXO unit (after a day or so typically) we  would like 
to see below 5ns SD on the GPS TI. Good DOCXO units regularly achieve  
<2.5ns. No reason a good Rb should not be able to achieve that as well. The  
DOCXO unit running at the University of Colima for example regularly achieves  
~1.8ns SD. For some reason the TI indicator has not been visible for some 
months  now though, but the plot still speaks for itself. Notice also the small 
control  voltage range of less than 50 microvolts typically:
 
   _http://resco.ucol.mx/Fury/gpsstat.htm_ 
(http://resco.ucol.mx/Fury/gpsstat.htm) 
 
bye,
Said
 
 
In a message dated 7/27/2010 20:11:00 Pacific Daylight Time, smace at intt.net 
 writes:

After  _MANY_ hours of watching and playing with different 
settings, I came to  the conclusion that you just need to be patient when 
using Fury+Rb, don't  expect it to converge like an OCXO, but it will 
over a few hours and give  you a good result as long as you have stable 
power and shield it from air  streams. 


More information about the time-nuts mailing list