[time-nuts] HP-5065A frequency adjustment advise and Common (all in) view GPS comparisons proposal.
magnus at rubidium.dyndns.org
Sat Oct 20 16:08:03 EDT 2012
On 10/19/2012 11:06 PM, Edgardo Molina wrote:
> Dear Group,
> Good afternoon. I just realized that my two HP 5065As have been
> running for about a month and their operation has been closely
> monitored. Now I assume both clocks are stabilized. I want to
> start doing some phase and time comparisons between them and
> with respect of a TrueTime 5 MHz GPS receiver signal. I only
> have an old Tracor 895A linear phase/time comparator for this
> purpose. Luckily it has 5 MHz as a frequency input. At least
> it will allow me to get close to a decent synchronization as I
> continue searching for better measurement instruments. Next week
> I will be receiving my first HP 53132A and Prologix GPIB-USB
> interface to start characterising my clocks and doing
> Allan Deviation work.
You should have a good starting-point then. Pull in TimeLab!
> My question here is how to finely adjust the 5065A frequency
> to bring it in phase to the 5 MHz GPS derived signal. I read
> somewhere that it can be fine tuned using the C Field control,
> but I am afraid of moving it as both controls have been set
> to each instrument's factory values. My oscillator fine
> frequency adjustment pots are set to 250 and locked there as
> per manual recommendation.
This is a fine recommendation when you use them as stand-alone
instruments. However, it will limit their ability to achieve SI second
(UTC second back in the days when you intentionally frequency offset it)
to be the current instruments properties and long term drift.
If you use a UTC source of sufficient quality (i.e. GPS clock such as
Thunderbolt) you can remove systematic frequency offsets to within a
much tighter specification.
> May I get some advise as which is the best procedure to adjust
> these clocks to start matching them regularly to GPS reference?
> I just don't want to mess with the clocks during the process.
The manual gives a good description of how to coarse tune the HP5065A
into a state where the OCXO is close to correct frequency, locks, has
good signal properties to maintain lock over time and not add to much
noise. You have a stable frequency source, but not "on mark". The
C-field correction is to get the final touch, to correct the frequency
to be very near what the SI second should give you.
For manual trimming, straight frequency measurement is a good start for
things like oven oscillators, but then you better start to use Time
Interval (TI) mode. Recall that you want to measure over such lengths
that the systematic error dominates over the noise. TimeLab is a good
tool to do this, as it's phase and frequency plots will do linear or
quadratic estimates, draw the trend curves and you can get a good feel
for it. You can also get the Allan Deviation and Time Deviation to see
how your phase and frequency drift compares to the noises.
> Also related to my previous question, is it common to use phase
> shifters with the 5065A not to touch the frequency standard adjustments?
> I saw NMIs using them to adjust Cs clock time scales without touching
> the Cs adjustments.
You are comparing apples and oranges.
Good Caesium atomic beam clocks does auto-trimming with a separate
control loop to steer the C-field. Old caesiums, such as HP5060A and
HP5061A is in this sense "open loop" (the term has been used in this
context before) and need C-field corrections. With such auto-trimming,
the systematic error due to C-field error can be significantly reduced,
if not effectively eliminated. Other systematic effects is still there,
but quality clocks take care to handle them one way or another.
Even good Rubidium gas cell clocks does not have this C-field locking,
as the C-field correction is used to overcome the systematic effects of
wall-collisions, buffert gas pulling (which is used to coarse-correct
the wall-collision effect), cavity pulling and ligth amplitude pulling.
I'm not talking about other standards using caesium and rubidium, but
most of the effects is dues to the standard type rather than atom type.
So, old Caesiums and Rubidiums require the C-field corrections to align
up, taking out their systematic effects, and also to compensate drift in
them, which comes natural from aging.
Good quality modern Caesium clocks on the other hand doesn't need this
correction to the same degree, but it is also trouble some to measure
the clock if you measure it after corrections, so by monitoring the
non-corrected clock and then create a corrected variant for the
time-scale realization is the preferred variant.
> I would like to explore some interesting possibilities as a
> byproduct of this enterprise. Is anybody interested in common
> (all in) view GPS clock comparisons with me? I just learned the pro
> way of doing it at CENAM over the last couple of days during the
> symposia of time and frequency that I had the luck to attend. With
> some coordination there has to be a way to reproduce those
> experiences here, not only as isolated comparisons, but continuous
> periodic measurements. Probably automated in the future. I am
> afraid we will not be able to achieve legal traceability to UTC,
> as we are not doing this directly with an NMI or following a legal
> traceability chain to it. Or am I wrong? Does anybody has direct
> legal traceability using common view GPS with an NMI? Despite this,
> we could still build a network of inter comparisons using GPS as a
> way to continuously monitor our labs performance, just as the SIM
> network headed by NIST, NRC and CENAM do across the american
> continent NMIs.
What for instance NIST do in their system as a service is nothing very
secret, it is just taking off the shelf components (VP Oncore), some
measurement infrastructure and some software and apply many of the
principles covered in a whole range of articles. GPS common view as such
isn't all that hard to do. Some of the trick is to do coordinated
logging and to steer which satellites is being used. Some others is
trivial stuff as using a good quality antenna (they now use the Novatel
700 pin-wheel antenna if I recall correctly).
It would be nice to set up such a thing, cook up the software to run it
and operate it continuously.
> If the idea is welcome, please answer to the thread. I am more than
> enthusiastic. If there are currently efforts to achieve this, please
> accept my apologies for the lack of knowledge in the group's work.
> Still I would like to cooperate if possible. Other comments regarding
> feasibility and technical implications are surely welcome.
I haven't seen anyone do it with such systematic rigour, but it would be
a nice little lab to do. :)
> By the way, while attending the symposia, I noticed with pride that
> the Time-Nuts community is highly regarded among the crème de la crème
> Time and Frequency circles. Thank you all for allowing me to
> participate with the group.
We all contribute with information, questions, answers, experience,
tinkering and experiments. Nice to hear that they have taken notice.
I always take the opportunity to talk warmly about the group. I think I
may have recruited yet another guy while being in Atlanta this week.
> Thank you!
Magnus - just home from a loooong travel
More information about the time-nuts