[time-nuts] 10MHz Rubidium reference source for frequency counter

Charles Steinmetz csteinmetz at yandex.com
Fri Oct 31 19:39:35 EDT 2014

Karen wrote:

>All OCXOs have been tested in free-running mode after 40-50 min of
>All OCXOs have been installed in isolated close box to exclude external
>Cases of all OCXOs had not any contact with external metallic parts of the

Then it appears that something is wrong with your method of measuring 
the ADEV.  I would suggest trying a few of the oscillators again (no 
need to try them all), using the Isotemp OCXO as the time base for 
the Pendulum by connecting it to the Pendulum's REF input and 
selecting External Reference from the front panel menus.  This will 
tell you if there is a problem with the internal time base in the 
counter.  [NOTE that the REF input needs a sinewave between 0.1 Vrms 
and 5 Vrms.  If the Isotemp has a square-wave output, choose another 
OCXO that fulfills the Pendulum specification.]

If those tests produce ADEV measurements of 1e-11 or better at 1 
second, the internal time base is too noisy for what you are trying 
to do.  I expect that the internal time base will prove to be OK, but 
you need to verify that before proceeding.

However, I note that in the manual, the internal time base of the OPT 
19/90 counter is rated for a Root Allan Variance of 1e-10 at tau = 1 
second and 10 seconds (see p. 8-15 of the 2014 User's Manual, 
Rev.18).  OPT 30/90 is rated at 1e-11, and OPT 40/90 is rated at 
5e-12.  So, you may be getting about all of the performance the OPT 
19/90 can deliver, if its OCXO just meets specification.  (HP5370Bs 
are also specified at 1e-10, but they often deliver ADEV = 1e-11 or 
1e-12 at 1 second.  I do not have enough experience with Pendulum 
counters to know what kind of stability is typical.)

Assuming that the test above (with the external REF) gives similar 
results to what you posted today, there is something wrong with your 
measurement setup.  Are the input signals triggering the counter very 
stably near the middle of their peak-to-peak voltages?  Is the 
counter adjusted to give maximum resolution for the sampling you are 
doing?  Does your sample interval give the counter time to process 
each new sample?  Have you read Chapters 3, 4, and 5 of the User's 
Manual carefully, and set up the counter properly for your 
measurements?  Are you sure your TimeLab setup is correct?  (I'll 
leave it to John to comment on that.)

Best regards,


More information about the time-nuts mailing list