[time-nuts] Disceplining a Rubidium with a Thunderbolt

[EWKehren at aol.com]

Bill
 I agree with what you say. I am a fan of Brook Shera's  loop and  have 
used it with a Rb and between a Rb and a D 1000. I did modify the  input to 
"fool" it to get smaller steps and the only problem is the AD 1861.  
Availability is not the problem, they are readily available at a price much  less than 
the original catalog price.
 
The problem is its performance. It was never intended for what we use it,  
but it was perfect for us at that time and it has no DNL specification. 
Richard Mc Corkle has run some tests and DNL is 8 bits which is not good  
for any digital loop let alone a loop with very long time constant. A 16 bit  
LTC 1655 would be perfect for Rb's with resolution, tuning range and aging  
over 10 years. I like to use my D 1000 as clean up for particular tests and 
 since those tests are time limited again the 1655 would be perfect. I 
whish  Brooke would consider a software change or authorize some one to do so.
I am now a TB fan and use it as a frequency source for all my equipment and 
 even as a 1  PPS GPS receiver. It will take some time but I plan to 
compare  TB with the best affordable GPS receiver controlling a Rb when I find a 
way to  get around the AD 1861 or some one else's loop.
In the mean time I am looking forward to some one publishing test results  
for a Rb TB combination. But without a clean up OCXO I for one do not 
consider  it an improvement as you pointed out.
Bert Kehren
 
Hi David,

Just to clear the AIR, all Rubidium frequency standards  have a crystal 
oscillator as
the primary signal source within the Rubidium  device.  The Rubidium 
portion of the
standard is just a very high Q  filter whose properties can be controlled 
such that
it's filter's center  frequency has extremely small drift.  That small 
drift factor  is,
typically, way less then the resulting factors that control drift in a  
Quartz
resonator.

In order to gain the properties of the Rubidium's  longer term stability 
and the short
term noise properties of a very good  Quartz oscillator you would need both 
items.  You
select a very good  Quartz device and phase lock it to a really good 
Rubidium (with its
own  Quartz oscillator).  You would adjust the loop constants to correct at 
a  very slow
pace consistent with the quality of the very good Quartz  oscillator.

To get to the next level (connection to the Nation's  reference), you would 
discipline
the Rubidium against a GPS device with an  even slower loop.  So, in the 
end you have
two separate loops with three  separate devices.  This is not your "Nickel 
& Dime
store"  plug-and-play set up.  It would have to be set up with care and  
some
experimentation to get it right.

For a reference on the basic  process, you should read the QST article on 
Brooke
Shera's GPS disciplined  oscillator system.  Contained within it is a 
description of
the loop  process I referred to above.  To utilize his method would require 
 upgrading
his circuit design (some parts not available any longer) and some  software 
upgrading
as well to account for those changes.

To obtain the  QST article go to Shera's web site at 
http://www.rt66.com/~shera/
Also click  on the "more information line" for further  reading.

Bill....WB6BNQ