[time-nuts] Antique Rubidium Standard Questions

[Ed Palmer]

Magnus,

On 4/24/2012 10:49 AM, Magnus Danielson wrote:
> On 04/24/2012 07:55 AM, Ed Palmer wrote:
>> Hi Joe,
>>
>> On 4/23/2012 9:45 PM, J. L. Trantham wrote:
>>> Ed,
>>>
>>> I am not familiar with the Tracor units, only the 5061A and B as well
>>> as the
>>> 5065A. These units use the 2nd Harmonic as an integral part of the
>>> feedback
>>> loop.
>>>
>>> Without the 2nd Harmonic, is there another way to 'unambiguously
>>> determine
>>> that it is locked', other than comparing it to a 'known', 'locked'
>>> signal?
>>
>> Strictly speaking, the answer is probably 'No'. After all, why would
>> they include the 2nd harmonic circuitry if they didn't need it? There
>> should be 2nd harmonic and I hope to find some somewhere. Remember that
>> this unit is being brought back from the dead as a learning exercise so
>> a few 'minor' issues aren't a show-stopper. The unit has been running
>> for most of the day. I flipped the switch to open the loop. The
>> frequency went from 5 MHz to 5MHz +0.045 Hz while the error meter went
>> from 0 to -25 on a scale of 50. Close the loop and the frequency
>> returned to 5.000 000 000 MHz and the error meter went back to zero.
>> That certainly sounds like locking behaviour to me.
>
> What I was speculating was either of:
>
> 1) Locking onto the wrong hyperfine line. You would be off in 
> frequency, but the base response may be sufficient to integrate into a 
> locked state and maintain lock. Since the wrong hyperfine line will 
> have smaller amplitude, the response on both base and second harmonics 
> will be much lower, so having second harmonic response of close to 
> zero but still locking behaviour isn't all to hard to believe.
>
> 2) Similar to the above reasoning on the locking while lacking 
> (strong) second harmonic, you can have other deficiencies which give 
> similar behaviour on the main harmonic. For instance, if you have 
> incorrect alignment of lamp temperature and isotopic filtering cell 
> temperature, then the filtering will be miss-aligned so both D1 and D2 
> lines of the Rb-87 lamp will make it to the Rb-87 resonance cell, and 
> then the state-inbalance which the single line optical pumping is 
> intended to achieve is much less effective, so then the absorption 
> effect which is detected will have much less response, and hence there 
> will be a weak base harmonic and seemingly no second harmonic.

I knew I'd forget something.  I checked the temperature of the lamp and 
it's around 100C.  But I forgot to check the cell / cavity.  It's at 
55C.  Neither temp is specified in the manual but the cell temperature 
seemed low so I raised it to ~70C.  Bad idea.  The higher it went, the 
lower the maximum error amplitude went.  From a reading of 40 (50 is 
full scale), it dropped to 20.

> One way to separate between these two errors is vary the C-field. If 
> you are locked to get base feature, then you move a little, but at the 
> square of the of the C-field current. If you are locked to the side 
> features, then you move dominantly linearly with the C-field current. 
> One needs to measure at least three points to make any real conclusions.

I checked this and the frequency vs. dial setting for the C-field is 
definitely not linear.  Half of the change is concentrated in the last 
10% of the dial.

> Another way is to speculate on the temperature settings may have aged, 
> so both measuring them and compare to the values they should have. You 
> can also try to trim them one at a time and see if you can see if you 
> start to receive second harmonics and then trim to stronger signal.

I'll play with the lamp temperature.  The lamp frequency is within 
spec.  It runs at the unusually low frequency of 20 - 25 MHz.

Another thing has me puzzled.  When you take off the end cap of the 
physics package there's a hole that goes through two shields, two 
circuit boards, and the wall of the cavity.  The hole through the cavity 
wall isn't threaded so there wasn't a screw of some kind.  You can't see 
the lamp so it wasn't for lamp inspection.  The only other thing I can 
think of is a probe and since it goes through those other parts, it must 
be for use during manufacturing.  I built a probe by shorting the shield 
to center conductor on a piece of tiny coax and made some measurements.  
The SRD is driven at ~69.742 MHz and the 98th harmonic is the Rb 
frequency of 6.834 GHz.  The 92nd harmonic (6.416 GHz) was the strongest 
signal I measured.  The 93rd (6.485 GHz) and 95th (6.625 GHz) harmonics 
were also strong.  There was no trace of the 98th harmonic.  Shouldn't 
the strongest signal be the Rb frequency?  I tried moving the probe 
around, but although it changed the levels, it didn't show any different 
frequencies.

I just realized that I haven't mentioned that there's a partial manual 
for this thing online.  It's missing a few schematics, but is otherwise 
complete.  The URL is http://sundry.i2phd.com/ServiceManual_304b.pdf .

Ed

>>> I guess another way to ask the question is do you think you happen to
>>> have a
>>> particularly good OCXO?
>>
>> It's a 40 year old AT-crystal that hasn't had nearly enough recent run
>> time to work the kinks out. I would be astonished to find that it's that
>> good. But I realized that I've never looked at the oscillator by itself
>> so I did a quick test. I measured an aging rate in the range of 0.2 ppm
>> / day. If I cancel out all the aging, the results start to look like the
>> earlier attachment. But not when it's unlocked.
>
> Maybe it needs to settle a little.
>
> Cheers,
> Magnus
>