[time-nuts] FE-5650A option 58 tuning word for 10 MHz output

wb6bnq wb6bnq at cox.net
Wed Jan 4 04:16:07 EST 2017

Hello Mathias,

I did mis-speak.  You're correct that only 32 bits (8 hex digits) are 
used to program the DDS chip.  As Mark points out they store a 64 bit 
number and only use 56 bits but he does not elaborate on what is done 
with the bits beyond 32.  Also, after reviewing my comment on the “R” 
number, I see that I did not really convey what my mind was trying to 

Let me try again and see if I can be clearer.  The Rb physics package, 
itself, is an extremely narrow filter that, under certain conditions, is 
stable to some finite degree with an extremely small drift.  One 
function that can modify the filter is a magnetic field.  By shielding 
all external magnetic influences except for what is purposefully 
introduced, the finite frequency of this filter can be varied by a small 
amount.  This is what the C-field adjustment is doing.

Yes, I do think the outer can covering is a MU-metal shield.  The bottom 
plate where the connector is located is not.

A Rubidium frequency standard is a controllable crystal oscillator fed 
through the Rb filter with a feedback signal from the filter to keep the 
crystal oscillator centered on the Rb filter.  By making very small 
changes to the Rb filter, the crystal oscillator frequency can be 
steered.  Over the long term of many months the Rb filter is more stable 
than a crystal oscillator and thus becomes a higher level standard 
relative to a good crystal oscillator for the long term (months).

In the scheme of things, the Rb filter is narrower than the crystal is 
and these adjustments are very small compared to the Rb frequency (6 
GHz).  So setting the C-field to minimum and adjusting the final output 
frequency with the DDS to the closest step just below the desired 
frequency and then adjusting the C-field, by external comparison, to the 
correct wanted frequency achieves the desired result.

I know the calculator that comes with Windows XP will produce the 
correct mathematical results.  I think the Windows version 7 does the 
same.  I do not have Windows 10 and therefore cannot address that one, 
if there is one.  Even EXCEL spreadsheet does not do the job properly.  
So use caution with your calculations.

However, with all that said, it means nothing if you cannot properly 
measure the final value against an external standard of greater 
accuracy.  Acquiring the equipment to do the external measurements is 
where the real cost comes in.

At the minimum you will need a good “timing” GPSDO and I would highly 
recommend the Trimble Thunderbolt (properly set-up of course).  The next 
item is a good time interval counter like a HP 5335 with GPIB 
capability.  Add a GPIB controller to acquire the needed data, 
specifically one of these two 
https://www.sparkfun.com/search/results?term=gpib would be an excellent 
choice.  With these three items you can make a computer version of a 
strip chart measuring the delta difference in frequence between the 
Thunderbolt and the Rb.

Hopefully the above helps to clear up your query ?


More information about the time-nuts mailing list