[time-nuts] Beginner's Atomic Clock

[John Ackermann N8UR]

On 9/17/19 3:33 AM, Dr. David Kirkby wrote:
> On Tue, 17 Sep 2019 at 04:00, Dana Whitlow <k8yumdoober at gmail.com> wrote:
> 
>> The nice thing about a Rb is that its short term stability (seconds to
>> minutes and perhaps
>> even longer) is much better than that of a GPS timing receiver.  The bad
>> news is that Rb
>> standards exhibit long term frequency drift in the neighborhood of a few
>> parts in 10^11
>> per month.  A pretty fair compromise is to use an Rb standard that is
>> disciplined by GPS
>> PPS pulses with a loop time constant on the order of a day or so.
>>
>> Dana   (K8YUM)
> 
> 
> Is  there any advantage in using a GPS Rb disciplined oscillator vs a GPS
> disciplined high quality OCXO like the HP 10811A? I can’t understand why
> there should be, as a Rb source would use an OCCO in its output stage
> Therefore in each case
> 
> * Short term stability depends upon the quality of the OCXO
> * Long term stability depends upon GPS.
> 
> Perhaps there’s is period over which the the overall stability can be
> improved by adding a rubidium oscillator. I would be interested to know if
> that is the case or not.

There are a bunch of interesting tradeoffs in choosing a frequency
reference.

Any Rb (except the HP 5065A which is in a different class as a lab
instrument vs. the small telecom units) will be worse at short tau than
a good OCXO, and is also likely to have much worse phase noise.  A
typical telecom Rb will be around 1e-11 at 1 second while a good OCXO
can be one or even two orders of magnitude better.

At medium tau (say a few thousand seconds) the Rb will likely be in the
mid to upper 13s, which is better than any but a very good OCXO.

At long tau, the Rb should show at least an order of magnitude less
drift than even a very good OCXO.

A Cesium at short tau will typically be worse than either an OCXO or an
Rb.  The Cs only wins (a) at long tau since there is zero drift; and (b)
for absolute accuracy.  But at anything shorter than around 10K seconds,
it's not the best choice.

A good GPSDO is really the overall performance winner -- short term
stability and phase noise limited only by the quality of the OCXO, and
very good long term stability and accuracy due to the GPS lock.  It's
only in the mid range of a few hundred to to a couple of thousand
seconds, where the OCXO drift kicks in before the GPS discipline takes
over, that a GPSDO will underperform a telecom Rb.

In short, the GPSDO takes much of the fun out of time-nuttery. :-\

John