Rubidium clock module unsuitable as a lab standard?

[tkamiya]

I hate this when it happens....  I stumbled upon an interesting article and I cannot find it again.

I was browsing yesterday and happened upon an article that said certain model of rubidium clock modules are unsuitable as lab standard on its own.  The article seem to indicate it requires additional something to make it work the way we'd want it.

That's it.  I didn't read the rest and I cannot find it again.  Does anyone know what module this piece of article is talking about?  I know one seller on eBay modifies (such as temp compensation) the unit before selling it.  His rational as described was it would be a "Mess" when GPS tries to steer the oscillator at the same time temperature compensation does the same. 

I know there isn't much to go on....  and I don't have any additional clue.  Darn!

[ArthurDent]

Where rubidium standards are designed to be used for frequency standards, it is hard to see what the objection might be. If it were set up for an odd output frequency, required a digital rather than analog tuning signal, or output a square rather than a sine wave, that might be a problem depending on the application. Most of the standard LPRO-101 class units require a EFC voltage and output 10Mhz sine wave so they are a good choice if you're buying a cheap used rubidium standard. A buffer/DA is always a good idea as well.

[jpb]

I hate this when it happens....  I stumbled upon an interesting article and I cannot find it again.

I was browsing yesterday and happened upon an article that said certain model of rubidium clock modules are unsuitable as lab standard on its own.  The article seem to indicate it requires additional something to make it work the way we'd want it.

That's it.  I didn't read the rest and I cannot find it again.  Does anyone know what module this piece of article is talking about?  I know one seller on eBay modifies (such as temp compensation) the unit before selling it.  His rational as described was it would be a "Mess" when GPS tries to steer the oscillator at the same time temperature compensation does the same. 

I know there isn't much to go on....  and I don't have any additional clue.  Darn!

I think it is probably referring to phase noise/spurs generated by a lot of rubidium clock modules. The additional thing may be a good OCXO to phase lock to the rubidium module. Quartzlock produces (or produced) a commercial system for this purpose.

[edpalmer42]

Related to jpb's comment, Rb by itself is probably not suitable for a reference if you're planning to use it as a source for a frequency multiplier or phase-locked oscillator due to its relatively high harmonic content.  Even if the output is a sine wave, it's probably not clean enough to multiply it up to the GHz range, for example.  You'd need either some good filtering or a clean-up oscillator to make it usable for that application.

Ed

[tkamiya]

Amazing!  I found it. 

http://www.ka7oei.com/10_MHz_Rubidium_FE-5680A.html

It's about use of FE-5680A as reference to microwave transverters.  Author  claims low level audio frequency phase modulation on output.  Additional comments are spread over the long article. 

So Ed was right. 

What I have is Efratom unit, but I'd have to check that out before implementing it.  I do plan to buffer but not add LC filters.  My understanding is doing so will invite temperature dependent phase shifts.

[edpalmer42]

No, he's describing something different.  I'm talking about simple harmonics of the 10 MHz output.  They're relatively easy to filter out without running into temperature-dependent phase shifts.  A simple 15 MHz low-pass filter might be all that's required.  The next step up would be a bandpass filter from 5MHz to 15 MHz.  Still easy.

I'm not familiar with the 'low level audio frequency phase modulation' he's talking about.  He didn't give any details about that so it's hard to make any suggestions regarding the cause.  It could be just power supply noise.  Or, every Rb that I'm aware of uses a low audio frequency in the range of a few hundred Hertz in the Rb feedback/locking loop.  Maybe some of this is leaking through.

We run into this sort of thing whenever we try to reuse commercial Rb standards or GPSDOs.  We don't know what the design criteria were so we don't know if a particular model will be suitable for our application until we buy one and test it.  Forums like this are great for sharing this info so we build up a stock of information on the different models.

Ed

[Wolfgang]

Hi,

I have an ELPRO 101 and I cannot complain at all. Phase noise it not all that bad, frequency stability is awesome when compared to a GPS locked source. When I have phase noise sensitive stuff I use an OCXO that I synchronize to the ELPRO and the GPS just before my measurements. That has worked all the time.

[tkamiya]

Oh....  I'll just buy a Hydrogen Maser and be done with it....  (ha)

[emece67]

.

[Wolfgang]

Oh....  I'll just buy a Hydrogen Maser and be done with it....  (ha)

Why not. Make sure you join the TEA group of this forum 

[Wolfgang]

I learned too that the FExxx units are having jitter problems. LPROs dont seem to be afffected, at least mine (I have two) are not.
I got both from eBay, but they worked.

[tkamiya]

This is amazing! 

I just connected it all up.  My counter that references GPSDO reads 10.000.000.001,9MHz without doing any compensation.  That's only after 2 minutes or so warm up and establishing physics lock.  I'm impressed!  Calibrating this will be a challenge.  I'll probably have to do 24 hour count to ride out bumps on GPSDO.

This is Efratom 1000318 from Ebay.

[tkamiya]

I'm sure there is better, but I think anymore zeros will be just academic....

Oh, wait...  I'm being REASONABLE!  I can't have that!  More Zeros!  Zeros!  Zeros!  I'm feeling sick....

[EEVblog]

This is a common requirement and it's called a Rubidium disciplined oscillator. i.e. the Rubidium standard keeps the accurate reference but it's not the oscillator that provides the low phase noise output. The 1Hz output is often used for this. GPS 1Hz references are often used for this too.

[tkamiya]

But aren't these Rubidium modules like Efratom already a Rubidium steering oven crystal oscillator??  Or am I mistaken?  Should their output have noticeable jitters as indicated?

[Wolfgang]

Yes, all Rubidium units you can buy are in fact governed crystal oscillators. The control loops works by a crystal oscillator folowed by frequency multiplier that creates a microwave signal at around 6GHz. This signal is radiated into the Rubidium cell. Some electronics sweeps the crystal frequency up and down until a "dip" (Atomic absorption) is detected, and then it tries to lock at this frequency. This control loop involves high multiplication factors and the "dip" is quite weak so a lot of noise and jitter can occur if you dont design very carefully. The output frequency of an LPRO or other brand unit is always coming from a crystal oscillator.

[edpalmer42]

But aren't these Rubidium modules like Efratom already a Rubidium steering oven crystal oscillator??  Or am I mistaken?  Should their output have noticeable jitters as indicated?

That comes right back to the question of what the original design specs were.  Many of these units were used to keep cell sites on frequency during outages of the main sync source.  As long as the jitter was within their jitter mask, they passed.  There was no reason to install a standard that exceeded their specs.  That would probably cost more money.  Some Rb standards might have been used just to maintain an accurate clock for time-of-day before GPS or cell service was available for that.  They wouldn't care about jitter at all, only frequency accuracy.

Ed

[tkamiya]

That makes sense.  That explains the reason why Efratom had two models.  One for instrumentation and one for telecom.  (FRS-A and FRS-C)

[David Hess]

Emece67 summed up the problem.  It is the conversion between the native rubidium disciplined oscillator frequency and desired output frequency which is the problem.  If not done properly, excessive phase noise and spurs make the output unsuitable for low noise applications like local oscillators in a superheterodyne receiver or transmitter.  The solution is to use the rubidium reference to discipline a suitable oscillator.

Lab equipment which operates using an external 10MHz reference is unlikely to be affected because if it matters, internally it will be disciplining its own low noise oscillator to the external 10MHz reference anyway.

[BigMark]

Its not uncommon for the higher level rubidium standards to have a 1PPS input. The 1PPS can be sourced from a GPSDO or a off air frequency standard service. So a mix of long term and short term stability.

[FriedLogic]

Amazing!  I found it. 

http://www.ka7oei.com/10_MHz_Rubidium_FE-5680A.html

It's about use of FE-5680A as reference to microwave transverters.  Author  claims low level audio frequency phase modulation on output.  Additional comments are spread over the long article. 

So Ed was right. 

What I have is Efratom unit, but I'd have to check that out before implementing it.  I do plan to buffer but not add LC filters.  My understanding is doing so will invite temperature dependent phase shifts.

There's a phase noise comparison between one particular variant of the FE-5680 and some other rubidium oscillators on https://febo.com/pages/oscillators/rubes/

An error of 1E-12 in the output frequency of a rubidium shifts the phase by 3.6ns/hour, and depending on your rubidium, the frequency could change by several times that for 1ºC temperature change.