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

Charles Steinmetz csteinmetz at yandex.com
Wed Oct 29 10:02:44 EDT 2014

Karen wrote:

>I raised this question as a ADEV measurement result of my 
>GPSDO/NEO-7M is about 1E-9 for tau = 1sec and it isn't meet my 
>expectation.  In my understanding (before I published my question 
>here) the reason for this is not a good enough short range stability 
>of my frequency counter's reference OCXO that's why I had an idea to 
>use external rubidium source.

The Pendulum is a well-respected instrument for this sort of 
measurement, as you have now discovered by reading the 
datasheet.  Much better than the ADEV you are getting.  (In fact, it 
is probably better than the datasheet specification.)

>Now I see that even my internal reference OCXO provides a good ADEV 
>value 1E-10 for tau =1 sec (according Pendulum CNT-91 with opt.19 

>If so what can be a reason of not a good ADEV result of GPSDO/NEO-7M?

(1) poor performance of the particular OCXO you have, or (2) poor 
GPSDO design or construction.

To check (1), remove the OCXO from the GPSDO and measure its ADEV by 
itself.  A good OCXO should be better that 1e-11 at 1 second, 
certainly better than 1e-10.  I believe you said you are using a 
Morion MV-89.  Many of the surplus MV-89s available in the US are 
broken and have poor ADEV or lots of spurs.  The situation may be 
better where you are.

If the ADEV of the oscillator by itself is only 1e-9, then you need 
[at least] a different oscillator.  If the oscillator by itself is 
OK, then the GPSDO is not doing a good job of disciplining it.

The purpose of a GPSDO is to let the oscillator run without 
interference at tau where the oscillator stability is better than the 
GPS timing signal, then to cross over at longer tau and let the GPS 
control the oscillator at averaging times where the GPS stability is 
better than the oscillator.  Usually, this will be somewhere between 
100 and 1000 seconds.  So the control loop must be VERY slow, with a 
time constant in the hundreds or thousands of seconds -- it should be 
doing essentially nothing at tau = 1 second.  The ADEV of the GPSDO 
at 1 second should be the same as the ADEV of the OCXO at 1 second, 
or very close to it.

Best regards,


More information about the time-nuts mailing list