[time-nuts] Behavior of a disciplined rubidium oscillator?

[John Miles]

> I have been observing the relative time stability of a Jackson Labs Ln Rb
GPS
> disciplined oscillator and a 5071A cesium reference (undisciplined) , as
> measured via the 1 PPS outputs and a TIC. See the attached plot for about
8
> days of history. Horizontal axis is seconds and vertical axis is
nanoseconds.
> 
> The Ln Rb has very good specs, qualfied by Jackson Labs with "no
vibration",
> "no air current", etc. I have been surprised to find that the unit
responds
> rather violently to external temperature changes.

Coincidentally, I just measured one of the LN Rb units for 24 hours in
undisciplined mode against a couple of 5065As.   It performed well relative
to its +/- 600 ns holdover spec, staying within about +/- 60 ns over 24
hours:

http://www.ke5fx.com/lnrb_undisc.png

I didn't attempt to maintain a constant temperature, just normal HVAC
activity.  About 80 ns of drift occurred between 75K-80K seconds after
opening a window a few meters away; opening the same window again near 85K
seconds started another downturn.  It's about -2 degrees C outside right
now, so opening the window dropped the room temperature substantially.  

The event near 31K seconds didn't correspond to anything in particular, but
it only amounted to about 15 ns. 

A couple of points to consider: 

- How long was the unit running before you began the measurement?
Traditional rubidium standards sometimes benefit from several days' worth of
aging before being put to the test.  I don't know if this is still the case
with a CPT-based device like this one.

- Small rubidiums are designed for low power consumption, so it's not too
surprising that tempco suffers to some extent.  If the relevant holdover
specification doesn't specify a temperature spread -- and AFAIK it doesn't
-- then it may not be treated as a key target for optimization.

- It might be interesting to measure your LN Rb in undisciplined mode in
order to rule out any peculiarities in GPS signal quality.  Ideally, if you
want to make a disciplined measurement, you could use a splitter to drive a
separate GPSDO whose performance is being measured at the same time.  That's
next on my list of things to try with this one.

-- john, KE5FX
Miles Design LLC