[time-nuts] Antique Rubidium Standard Questions

[Magnus Danielson]

Hi Ed,

On 04/24/2012 01:48 AM, Ed Palmer wrote:
> Hi Magnus,
>
> Please note, I goofed when I said that the drift was 30 ppt. It's 30
> ppb, i.e. 3e-8.

I assumed that.

> On 4/23/2012 2:08 PM, Magnus Danielson wrote:
>> Ed,
>>
>> On 04/23/2012 07:54 AM, Ed Palmer wrote:
>>> I'm playing with a Tracor 304-B Rubidium Standard from 1969. I'm using
>>> it as a learning exercise to find out more about the guts of a Rubidium
>>> standard and how it works.
>>>
>>> This thing is a beast! Rack-mount, 3U high, 39 pounds (~18 kilos), 9
>>> plug in circuit boards. The OCXO is an oddball frequency that is
>>> multiplied directly to 6.8 GHz. There's no synthesizer in that chain. A
>>> synthesizer is used to convert the oddball frequency to a 5 MHz output.
>>
>> OK. A bit different to some of the rubidiums, but looking around it is
>> not as uncommon as one might think. This technique is used in some of
>> the FEI 5680 and also the later GPS rubidiums.
>>
>>> It's sort of working. The error signal isn't up to spec, but it's strong
>>> enough to give a stable lock although there's no trace of a second
>>> harmonic signal. Allan Deviation is in the Xe-12 range from 1K to 10K
>>> seconds. The OCXO has a not-yet-resolved issue that is probably
>>> degrading the results.
>>
>> What is the OCXO issue?
>
> Due to drift, the AT crystal is so far off frequency that the EFC can't
> correct it. There is no frequency adjustment on the oscillator, but
> there is an adjustment for oven temperature. In order to bring the
> frequency back into the EFC range, it's running at a temperature below
> it's turning point.

OK. Not ideal but you should still have a useful OCXO. Possibly you 
could modify it to have a trim-cap.

>>> The lock frequency suggests that the Rubidium cell has drifted down by
>>> ~30ppt over the 40+ years since it was built. Is that reasonable? That's
>>> much more drift than the specification states, but I doubt if the spec
>>> was intended to be valid for 40 years!
>>>
>>> Could the drift be at least partially responsible for the lack of second
>>> harmonic? A message on the list (
>>> http://www.febo.com/pipermail/time-nuts/2006-April/020562.html ) said
>>> that you could peak the second harmonic by adjusting the cavity tuning.
>>> If the cell and the cavity are out of sync would that kill the second
>>> harmonic? How close to they have to be? If this thing has a cavity
>>> tuning adjustment I haven't found it.
>>
>> You should see second harmonics regardless, it's an effect of the dip
>> itself. I would check if you can observe the second harmonics on the
>> signal from the detector by some other means. If you seems to lock up
>> but does not see second harmonics, it just could be something in the
>> second harmonics detection which needs some LTC.
>
> I had the same thought so I replaced the Rb lamp with an LED that was
> modulated at the 2nd harmonic frequency. The signal showed up at the 2nd
> harmonic test point. I changed the modulation to the fundamental
> frequency and saw approximately the same amplitude at the fundamental
> test point. Since the signal paths are almost identical (I have the full
> manual with schematics), I concluded that the signal chain was working
> and would display the second harmonic if it was present.

Good strategy!

OK. Signal paths work.

>> There are many sources for shifting the frequency, including the power
>> supply (check voltage and clean-ness), the trimming of the C-field,
>> the leakage of helium, resonator tuning, temperature trimming of lamp
>> and resonator may shift amplitude and hence frequency through light
>> pulling.
>>
>> It would be good if you could hook up the modulation sine on X and
>> return signal on Y on a scope, that would give you a clear display of
>> the resonance dip.
>
> I just tried that, but had no success. I thought the actual dip was so
> small that you couldn't see it and that's one of the reasons why the
> modulation is used?

Yes and no. The modulation is used as it is a convenient way to lock 
into that feature, but true, it is a weak signal compared to the DC 
light detection, so AC signals forms a convenient way to allow a cap to 
block the DC light away, and after some amplification you should have a 
detectable signal.

>> Oh, and check if you have leakage problems around the integrator cap,
>> that would also shift the frequency.
>
> I'll check for that, but wouldn't that type of leakage just be corrected
> by the loop?

Partly, only partly. I would at least have a look there and see if there 
is a need for cleaning.

I think it is a bit strange you don't have your 2nd harmonic. If you 
managed to have it locked to the wrong hyperfine line, then this would 
be expected I guess. How does it react to the magnetic field? Linearly 
or quadratically?

Cheers,
Magnus