[time-nuts] Newbie with new questions: Rubidium

rlutwak at comcast.net rlutwak at comcast.net
Fri Aug 5 16:26:36 EDT 2005

10-20 years is pretty good reliability for any electronic system.  Unlike the cesium beam clocks which have a natural wearout mechanism (cesium depletion), the rubidium physics package can, in principle, last forever.

Usually, Rb clock failures are electronic.  Failed components are generally the same you'd expect on any electronics.  Check first the power supply capacitors.

In order to maintain lock, both the resonance cell and need to be temperature stabilized.  If the baseplate (+ internal temperature rise) exceeds the setpoint of either oven, the ovens lose control and the (physics) frequency may deviate by enough exceed the tuning range of the oscillator.


Robert Lutwak, Senior Scientist 
Symmetricom - Technology Realization Center 
34 Tozer Rd. 
Beverly, MA 01915 

RLutwak at Symmetricom.com (Business) 
Lutwak at Alum.MIT.edu (Personal) 

(978) 232-1461 (Desk) 
(339) 927-7896 (Mobile) 
(978) 927-4099 (FAX)

-------------- Original message -------------- 

> > Recently I acquired 10 compact Rubidium modules cheaply 
> > (http://www.dl-car.com/~2038/time/rubidium1.jpg etc.) 
> > made by Efratom between 1991 to 1997. They look identical 
> > internally and perform equally good by my preliminary tests. 
> > By comparing outside appearance with the Rb module FRS-C 
> > inside SR625 (http://www.dl-car.com/~2038/time/rubidium8.jpg), 
> > They look very similar. 
> There have been many Lucent quartz and Rb telecom 
> frequency standards on the surplus market. I think 
> the Rb are often FRC-C (some later ones are LPRO). 
> The FRS-C size and pinouts have become a sort of 
> de facto standard; that's why the PRS10 and others 
> are also the same size and have the same power pins. 
> > These modules are taken down from Lucent communication 
> > equipments which must be running 24hrs a day. 
> > 
> > My preliminary test results: 
> > Frequency deviation: 1E-9 max (1hrs warm-up) 
> > Starting supply current: 1.6 - 1.75A 
> > Lock time 145 - 160 second at 28C room temperature. 
> > Current consumption after warm-up: 0.36 - 0.41A 
> > However, I still got many questions. 
> Sounds about right. 
> > Q2A. Are these actually FRS-C? or others module? 
> > The module in my SR625 is the FRS-C made in 1992: 
> > http://www.dl-car.com/~2038/time/rubidium9.jpg 
> > Five of my modules labeled 'part # 100334-005', two are 
> > 'part # 100334-004', two are 'part # 100334-005'. 
> ? 
> > Q2B. The first test I did is to measure the frequency because 
> > the accuracy might be indications for their past drifts. The 
> > result is very good: max deviation from 10MHz is less than 1ppb 
> > against my HP58540A (locked to GPS for many days) by using 
> > HP53131A at a gate time=150 seconds. An example is, #28979 was 
> > manufacture in 1995, frequency deviation is 0.39ppb, factory 
> > set at 0.05ppb. Can I say that the long time drift for this 
> > particular module is <5E-10 in ten years? I know I cannot say 
> > its 5E-11 per year for sure. 
> These tests may not reflect past drifts since you 
> have no way to know what EFC voltage was used 
> prior to the oscillator being removed from the circuit 
> board it was mounted on. Also, I'm not sure if the 
> Rb drift rate when stored has any relation to the 
> drift rate when operated. 
> Still, powering up within 5e-10 after ten years is a 
> good sign. 
> > Q2C. Is the life expectancy for rubidium 10 to 20 years? 
> > if so, some of my modules are about to die even if they are still 
> > functioning well? I came across "A Note about Rubidium Oscillators" 
> > at: http://www.amtestpl.com/DATUM/GmbH/Rubidium-Note-e.pdf 
> > saying that life time is virtually unlimited. I'm confused. 
> Me too. 
> My guess it depends a lot on the design or on 
> the manufacturing. 
> > Q2D. Anyone have the electronic manual for these Rb? 
> > I'd like to know the pin out specs as to how to use other pins. 
> I'll see if I can scan it for you. 
> > Q2E. Power supply. Is a switched mode power supply will be Ok? 
> > What are the side effects if I supply the heater power and 
> > reg power from the same source? 
> The Datum, Austron, and Lucent boards seem to 
> use switched power supplies so it must be OK. 
> > Q2F. Is it possible to replace the OCXO in the HP58540A with 
> > the modules to Achieve better holdover result? 
> I believe there have been several attempts to replace 
> GPSDO OCXO with Rb. Yes, holdover performance 
> will be much better. Short-term performance may be 
> much worse so you have to ask yourself why you 
> want to do this. It may take some magic to match 
> the Rb EFC gain (dF/dV) to the OCXO EFC gain. 
> This may be more complicated than a resistive 
> divider, and if so it may introduce additional instability. 
> Please try it and let us know how well it works. 
> > Q2G. What is the maximum operating temperature as regard to 
> > ambient and base plate? Do aging Rb modules require cooler 
> > temperature? I power up some of the modules without any heat 
> > sink for about 10 hours and 3 of them stop working. Its seem 
> > that they are over heated at not more than 55C base temperature. 
> > When cooled down they begin to work again. 
> Rb get quite hot and, yes, you must keep the 
> base plate from temperature runaway. 
> > I a good gesture for the rich information here and for all the 
> > people who made the contribution, I'll give away two pieces of 
> > these to any list members who are first and second come to China 
> > from now on. Drop me a e-mail (or follow the thread) before you 
> > leave, and preferable give me the name and phone # of your 
> > Chinese friend. After I confirmed, I'll ship the module (at my 
> > own charge) before hand so that you can get the module once you 
> > 've landed in China. The information as to who and when have 
> > received the module will be posted here. 
> > 
> > Have fun. 
> > Lymex Zhang / bg2vo 
> Where in China are you located? 
> /tvb 
> _______________________________________________ 
> time-nuts mailing list 
> time-nuts at febo.com 
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts 

More information about the time-nuts mailing list