Forrest,
Thanks very much for the extensive comments, they're greatly helpful.
I thought I might say a little more about my goals, and add some notes
on points that you made.
As mentioned, my main focus at this point is measurement of HF radio
signals, with a goal of 1 millihertz accuracy. At 10 MHz that would
involve stability of 1 part in 10^10. 1 part in 10^11 would be ideally
desirable for reference.
I have some oscillators (oven-controlled, double-oven-controlled, etc)
but have been told that these won't be able to achieve the stability I'm
looking for, and it's been suggested that I think in terms of GPSDOs
and/or devices using radioactive sources. I've been told that a GPSDO
will maintain 1 part in 10^12 if powered on, locked and left to
stabilize for a few months, whereas unlocked they vary with temperature
by about 10 PPB, drifting perhaps 5 PPB per day.
My first thought on a GPSDO to act as a backup to the Thunderbolt was
one of Leo Bodnar's units, either the Precision GPS Reference Clock or
the Mini Precision GPS Reference Clock. It appears that the primary
difference between these is that the first unit outputs two reference
clocks, whereas the mini outputs one. I don't need two clocks, so if
their specs are otherwise the same the mini should be fine for me. The
BG7TBL GPSDO sounds interesting; my main concern about going in that
direction is the variation you mentioned in different editions due to
parts availability.
I have very poor view of the sky immediately outside the room my gear is
located in, so for the Trimble I've been using a run of about 50 feet of
coax to an active GPS antenna. If I run multiple GPSDOs I hope I can use
an antenna splitter rather than putting up multiple antennas.
One topic I'd like to understand better is how GPSDO stability is
related to run time. Is it a geometric curve, with stability improvement
the greatest in the first hours and days after being powered on, with
later improvement less pronounced? If a GPSDO stays powered-on without
interruption but loses lock for some reason (for example, the antenna is
disconnected), does its stability hold reasonably steady or does it
start over from scratch?
I haven't yet seen or worked with a cesium or rubidium source, but am
very interested in them. I'm told that, like the replicants in Blade
Runner, they are high-performance but have limited lifespans -- i.e.
they achieve better than 1 PPB and don't require the extended run-times
of GPSDOs to achieve optimal stability, but due to the decay of their
sources they will not last forever. For educational purposes if nothing
else it would be very interesting to experiment with one of these, if I
can find one that isn't too expensive and is in decent shape.
I do incidentally have a nice Siglent two-channel oscilloscope, so
that's available for trimming purposes.
Thanks for the caveat about noise. My ultimate goal is accuracy of
frequency measurement, but precision may be a higher priority for
selection of a clock source. I need to spend time understanding better
the relationship between these in the measurement procedures I'm looking
at using.
One other note -- even with decent gear I realize there will be other
factors such as atmospheric Doppler that introduce larger errors than
the ones mentioned above. I'll certainly have to deal with those too.
If I have anything wrong, or if any of the above suggests any further
comments or suggestions (from any one on the list), I'll be very interested.
Thanks again,
Frank
On 7/8/19 6:32 PM, Forrest Christian wrote:
Long winded answer to your question:
I'm sort of an "Advanced beginner" time-nut or maybe a bit beyond that.
In relation to your question: It really depends on your goals. The
following is a high level overview of your options, and is intentionally
lacking and/or simplifying some details. There are lots of twists and
turns one can go down on each of these options. And there's a few spots
which I'm making generalizations which may or may not be exact.
In the way of "precision" 10Mhz clock sources, I have 2 thunderbolts, a
PRS-10 rubidium oscillator, and a BG7TBL GPSDO. Oh and a couple/few OCXO's
which are undisciplined. I want a cesium clock but haven't happened to
end up with one yet.
You're somewhat familiar with the thunderbolts, let me compare the others
first.
The OCXO's are undisciplined, but aren't that bad of a 10Mhz reference.
Especially if you can trim them to a good quality reference - such as your
thunderbolt. Depending on your needs, this might be good enough for you.
Generally, once the OCXO is warmed up you can adjust it to match the
reference, then as long as it isn't shut off it should stay very close to
the same frequency. Of course it will drift over time and perhaps as a
result of other external forces (temperature, humidity, voltage, vibration,
etc). Of course you can then re-trim it to the right frequency. How much
and how often you need to do this will depend on the oscillator. A
surplus OCXO can be had for well under $100. A good quality pre-aged
surplus one is likely better than a new out of the box one (even of the
same time). Other members of the list have much more experience with
which are the best, but I personally have found that the oscilloquartz ones
are typically acceptable for my needs.
A rubidium oscillator is like an OCXO on steroids. But if you think of it
as a very very very low drift OCXO then you've probably got the right
idea. For comparison a good OCXO will drift in a day about the same as a
rubidium will drift in a year. You can often get a Stanford PRS10 in good
condition for around $250 or so. The challenge is that any atomic clock
has a certain limited life due to the physics package "wearing out" in some
form or another, so it's a bit higher risk than just buying a good quality
OCXO.
Before I get to the GPSDO's let me mention about the relative ease of
trimming the above two types of references. If you have a two channel
oscilloscope, it is rather easy to trim these. You plug the reference
clock into one channel, set the scope to trigger off of that channel (so
the waveform is stable). Then you plug the clock you are trimming into
the second channel. You then adjust or 'trim' the oscillator until the
waveforms do not move (or beat) in relation to each other.
If you don't have a two channel oscilloscope, there are other ways to do
this as well.
Now back to the GPSDO's.
The thunderbolt you have is a OCXO 'disciplined' by the GPS's 1PPS output.
It's the same as trimming the OCXO on a continual basis such that the OCXO
output always has 10 million cycles per 1 second. There are other GPSDO's
that use other types of oscillators as well (rubidium, vcxo, etc). The
method of trimming the oscillator varies from GPSDO to GPSDO, but the
effect is typically the same: that is to adjust the 10Mhz output to
something close to 10Mhz.
As I mentioned I have a trimble thunderbolt and a BG7TBL GPSDO. The
BG7TBL GPSDO is a GPSDO designed by a Chinese ham, and each edition seems
to be a little different, typically using a u-blox GPS receiver and
seemingly whatever surplus OCXO they can get their hands on. Mine happens
to have a russian OCXO in it. The quality of these seem to be rather good,
although there is a bug which affects mine and other earlier ones which
cause the long-term frequency to be very slightly off. The very slightly
off means it runs at ~9.999999999 Mhz instead of 10Mhz. I beleive that
There are other designs out there as well from other vendors. A search on
a certain popular auction site for "GPSDO" reveals a lot of options. I've
had discussions with a couple of people who have compared several of these
and apparently the consensus is that they're typically all decent for
general use, and some are better than others, but none that they've tried
are especially bad. Certainly they should be good enough for a reference
clock for most people. If I didn't have a collection of oscillators
already I'd probably pick up 1-2 of these of varying types. Maybe a
thunderbolt and one of these.
One final caveat, which may or may not be applicable to you: In some
applications, certain types of noise can be a problem. Depending on how
you're using the clock you may actually find that a disciplined oscillator
is not the right solution. Generally I'm timestamping events and I'm more
interested in stability instead of precision. So, even if the frequency of
my clock source is off by 1-2% it isn't a big deal, as long as it stays off
by exactly the same amount (within some level of precision of course).
But if my clock frequency is always changing, then I'm introducing noise
into my measurements that I can't adjust out. Depending on the quality of
the disciplining method, the very act of disciplining a clock can introduce
more noise than if you just let it freerun.
Hope my rambling helps.
On Mon, Jul 8, 2019 at 2:12 PM Frank O'Donnell <[email protected]> wrote:
Hi, I'm extremely new to the topics covered on this list, but am finding
reading the message traffic to be highly educational. I have some
questions that I'd like to put out, and apologize in advance for them
being at a greatly simpler level than most of the discussion here.
I came to time-nuts by way of an interest in frequency measurement as a
radio ham. To supply a 10 MHz signal to my gear I bought a used Trimble
Thunderbolt. I always like to have a backup, so am thinking of obtaining
another device (either a GPSDO or something else that can provide a 10
MHz signal). For a beginner, is there something else that offers
accuracy/precision at least as good as a Thunderbolt and doesn't cost
more than a few hundred dollars at most that you might recommend?
I also have some questions about using the Trimble, but will put those
in a separate message.
Thanks,
Frank O'Donnell
_______________________________________________
time-nuts mailing list -- [email protected]
To unsubscribe, go to
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
