kb8tq@n1k.org said:

Why is that?

I would have guessed that a GPSDO would average over many GPS pulses thus
reducing the noise.

Is it something like GPSDOs are normally designed for good frequency rather
than good time, so when they find the time is off, they use a small frequency
offset for a long time to correct rather than a big frequency offset for a
short time?

Are there any GPSDOs designed for good time?  Or any with parameters that can
be tweaked to provide good time?

--
These are my opinions.  I hate spam.

Hi

The controller gets in the way. If you want good frequency stability you have long
time constants. The local reference is free to (and does) wander inside that time
constant.

It does and it does. However that does not help the accuracy of your 1 second
time tick. A good GPS module with sawtooth. can get you down to a fraction
of a nanosecond (ADEV) on your 1 PPS. There is no need to average that noise
down any further.

There are a few. They steer the 1 PPS so it follows the PPS output of the module.
If all you want is time, just use the module PPS and save on the electric bill.

Bob

Hi,

On 2021-03-10 14:23, Bob kb8tq wrote:

Indeed. It is different to optimize for ADEV or TDEV performance.

However, even when phase/time is interesting, a frequency optimization
has benefits when it comes to hold-over performance, i.e. how stable it
will be after loosing tracking. You very quickly come to a situation
where frequency error dominates the phase error, and only later does
drift and environment start to kick in.

Actually, you go below that if you average long enough, but the sawtooth
for sure helps. There is enough of noise and systematics to make things
move around. It's still not very good thought, so you end up with a
fuzzy border about there. As always, it depends. :)

Indeed, unless you need hold-over capability, that's when an OCXO or
Telecom Rubidium may kick in and be relevant. As time-nuts, we do it
regardless, because it is fun, but that is a different account. :)

Cheers,
Magnus

Hal,

The older (and probably the newer models, too) Trimble Thunderbolt GPSDOs
have
a user-adjustable time constant accessible via the serial port using a
program like
"Tboltmon.exe" (from Trimble).  I suspect that  "Lady Heather"  may also do
this.  I
am fortunate in owning a still functioning PC with an actual hardware RS232
port
and a usable O/S (Win XP), so running Tboltmon is a trivial exercise for me.

In addition to time constant, tboltmon also lets one examine (and set,
where appropriate)
a large number of other items, and it's reasonably intuitive to use.  I've
never had to resort
to a manual. (if there even is one) to do what I needed.

My own experience is that a time constant around 50 sec works best in my
environment,
providing the best compromise between filtering PPS jitter from out of the
unit's GPS
receiver and tamping down OCXO wanderings due to my home HVAC system's
cycling.
Setting a far longer time constant value (say, much longer than 500 sec)
tends to lead
to "funny business" so I just don't go there.

I have convinced myself that the PPS output from my Tbolt is derived from
the produced
10 MHz output, because if I trigger an o'scope from the PPS output, the 10
MHz sinewave
shows very little time jitter, perhaps 1 or 2 nsec.  So, I'm pretty happy
with the T'bolt, with
two (minor) exceptions:

receivers (my unit
was apparently made in the early 2000s).

called 'AMU',
for which I've been unable to find a definition.

Dana

On Wed, Mar 10, 2021 at 6:53 AM Hal Murray hmurray@megapathdsl.net wrote:

Bob-

As a rank amateur e astronomer, I am a  lurker. I am amazed at what I have
learned here. I know that there are differences between the meaning of
precision and accuracy, but please correct my understanding if I am
imprecise.

I have a need for precise time, as all sorts of calculations are dependent
on precise geocentric position, and of course time to convert to other times
e.g. sidereal, utc, etc., as related to the motion control of a large
telescope.

I have an old hp z3805a; seems to be really precise, agreeing with my
location (surveyed). Other gps's that I have seem to wander more.

My question is thus: It seems that procuring a more precise PPS/time output
unit is quite a bit more costly than what I have; even more costly is a unit
that has both more precise PPS/time output, and a really stable 10 Mhz
output ( I might add that I am a Ham, where 1 uhz  error is detrimental).

Assuming I can afford an upgrade, would  getting a more precise PPS/time
unit then and feed that data into separate OCXO? Getting both seems out of
my league.

Seems that  could the best of both worlds.

All advise welcome; it's how we learn....

Thank you,

Charlie
N6CFH

-----Original Message-----
From: time-nuts [mailto:time-nuts-bounces@lists.febo.com] On Behalf Of Bob
kb8tq
Sent: Wednesday, March 10, 2021 5:24 AM
To: Discussion of precise time and frequency measurement
time-nuts@lists.febo.com
Cc: hmurray@megapathdsl.net
Subject: Re: [time-nuts] World's most precise.... wall clock

Hi

using

GPSDO.

The controller gets in the way. If you want good frequency stability you
have long
time constants. The local reference is free to (and does) wander inside that
time
constant.

It does and it does. However that does not help the accuracy of your 1
second
time tick. A good GPS module with sawtooth. can get you down to a fraction
of a nanosecond (ADEV) on your 1 PPS. There is no need to average that noise
down any further.

rather

frequency

can

There are a few. They steer the 1 PPS so it follows the PPS output of the
module.
If all you want is time, just use the module PPS and save on the electric
bill.

Bob

http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com

time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

I was thinking about the discussion on synchronizing a grandfather clock
using a magnet and coil.   Most methods for doing this use an external
power supply or a battery.

What about powering the synchronizing circuit from energy harvested
from the pendulum.   Consider that you can get a quartz watch that will
run for 5 years on a minuscule battery.   Surely there is enough excess
energy to be harvested from the pendulum to exceed the energy available
from such a small battery.

The circuit could use a capacitor for energy storage and eliminate the
battery.   Of course the clock would be only as accurate as the 32KHz
crystal but that should satisfy anyone but a time nut. ;-)

You might have to adjust the pendulum to run a little fast so as to
provide the excess energy for the circuit.   The coil would act to slow
the pendulum each cycle thereby absorbing energy which would be
available to power the circuit.

On 3/10/2021 9:26 AM, Dana Whitlow wrote:

Hi

I suspect that is a function of how the 3805 presents the data.

Sub ns jitter is doing well at 1 second with GPS. Accuracy is different
than jitter. Since the GPS clock is not a direct expression of UTC from
BIH ( nothing is … sorry about that …) there is some back tracking to get
very accurate time.

If you have the $300 to $2000 for a multi band GNSS timing receiver, it will
indeed help a bit. How much will depend a lot on the state of the ionosphere
and the correction process. Troposphere also gets into things. I don’t know of
any receiver that directly estimates Tropo delay.

Best would team the fancy receiver with a fancy standard. An OCXO is better
than a TCXO. Most Rb's beats the OCXO long term. A Cs will beat them both
if you run out long enough. You then get into things like GNSS disciplined
Passive Hydrogen Masers. Properly done they should perform quite well. A
disciplined Active Maser would / could beat a Passive Maser …..

The better the “flywheel” the better the result, at least for frequency / stability.
It will count off seconds quite nicely. Just how far off from “right” those seconds
are is a bit unclear. ( = you still are not accurate)

For accuracy you need a path back to BIH and the “official” definition of UTC.
That’s true even with the brand new fresh from the factory disciplined Active
Maser that you sold the house to buy ….. There are lots of nasty little delays
that creep into the mix …. All of them need to be taken care of to below your
target error level. If you are after < 10 ns accuracy, this could get pretty exciting.

Bob

On Wed, 10 Mar 2021 06:39:46 -0800
"Charlie" charlie@drhabekost.com wrote:

The question is more whether it is good enough. How accurate
do you need your time? How precise does it need to be?
If the answer to both questions is smaller than 100ns, then you
are good with anything you can get your hands on, even the cheapest
GPS receiver. If you need better than that, you need to look
more closely and calculate each contributor to uncertainty,
both random and systematic and see where you can and want to
improve (or compromise).

At what time scales you need that 1µHz stability? Is it for
a few seconds? Or over a few hours? How accurate does it have
to be?
And are you sure it is really 1µHz @ 10MHz? Because that's
a stability better than 1e-13. Which isn't something most people
just have in their lab. You need a an exceptionally good OCXO to
reach down there (an OCXO that costs you as much as a car)
and even those reach it only over a very limited τ between
about 1s and maybe 100s. Beyond a τ of 1000s you have the option
of using a HP5065, possibly with Corby's Super-HP6065 modifications.
Even GPS takes until somwhere like 100ks to reach down to 1e-13.

If you need better than that, either at shorter or longer τ,
then we are clearly deep in atomic clock territory. Either
hydrogen maser. caesium beam standard or cold atomic clock.

		Attila Kinali

--
The driving force behind research is the question: "Why?"
There are things we don't understand and things we always
wonder about. And that's why we do research.
-- Kobayashi Makoto

Hi

I suspect that’s a typo. “larger than 100 ns” ( 1 us > 100 ns ….)
would seem to be the correct way to look at it.

Even crazier if you are talking about your transmit frequency at VHF :) :).
Welcome to why Hz, uHz, etc normally are replaced with ppm, ppb, and
the like … That’s been the case in every place I’v ever worked on this stuff.

Bob

Unless something akin to VLBI or pulsar timing is involved millisecond accuracy will usually suffice for amateur astronomy.

Bruce

Hi,

On 2021-03-10 17:04, Bob kb8tq wrote:

There is means to infer Tropo-delay with single receivers, but it is not
very accurate that I've seen. However, the usual way is to use nearby
receivers as reference. Eventually as the actual position is known,
tropo errors can be inferred more directly for a fixed receiver.

It is worth noting that you do not only want a good multi band GNSS
timing receiver, you also want a good phase-stable and multi-path
rejecting antenna to go with it, such as choke-ring or pin-wheel. Much
of carrier-phase properties is lost in a bad antenna.

Another aspect is that if you care about very accurate time, you need
the antenna, cable and receiver calibrated, as the delay through these
is not fully cancelled in the reception processing. In precise
positioning processing, the antenna should be of known type and properly
orienter such that the phase center calibration is compensated actively.

If you do not need a paper to show how good you are, you can do pretty
good guestimates to roughly your delays and compensate those. At some
point you end up wanting to do a real calibration anyway, if you try to
push the limit downwards.

BIPM these days, not BIH. There is plenty of things and thorns. I've
touched on some.

I've seen some crazy stuff. Temperature stabilized concrete pillars,
temperature stabilized coax. All depends on how deep your pockets goes
and crazyness.

I have yet to go that crazy, but at times I wonder.

Cheers,
Magnus

On Wed, Mar 10, 2021 at 2:21 PM Bruce Griffiths
bruce.griffiths@xtra.co.nz wrote:

I think some of the most demanding requirements for optical astronomy
by amateur/semi-pros are set by observing Near Earth Object (NEO)
asteroids or observing outer solar system objects occultations (when
these objects pass in front of a star). NEOs can have motion rates up
to several hundred arcseconds per minute at close approach and you
don't want to degrade the measured position by any more than about
0.2". Similarly outer solar system objects such as Kuiper Belt Objects
will have a velocity across the line of sight during the occultation
of around 10-15 km/s and size in the few hundred meters to few hundred
km range. Both of these work out to wanting time accuracies around the
few millisecond regime.

Cheers,
Tim

Hi

If you have a “Tropo Observatory” that gives you anything close to
24 / 7 / 365 data you are very lucky. Indeed there are people doing
Tropo by looking at GPS and saying “what’s left is Tropo” ….. If
that’s the approach your local observatory is using …. hmmmm …..

Bob

Hi,

On 2021-03-11 00:38, Bob kb8tq wrote:

Well, there is a high correlation with near-by stations on the tropo.
Post-processing pushes the errors way down.

It used to be that you had local sensory package to provide input for
tropo estimation. That faded out because it was not providing much
improved result than nearby observations.

Also, tropo has a different function for how it adds up in the
pseudo-range measurements to that of other sources, such as ionsphere.
So, separation can be done once the ionsphere shift is gone, which
double-frequency allows. Higher-order parts of ionsphere can be observed
if you like, and that gives a finer detail about the wavefront angle
into the tropo, if you want to play that game. Yeah, the signal does not
go line-of-sight through any of those, that's an approximation in
itself. There are people scratching their heads and refining it over time.

Kind of pitty that tropo has the same sign on code and carrier and seems
relative insensitive to frequency. Ah well, at least we can cancel the
ionspheric shift.

Cheers,
Magnus

Sent from my iPhone

Dana,

The Tbolt is different from most GPSDOs in that the 10MHz clocks the whole receiver. So there is no sawtooth error to correct for. Read the manual.

It is less sensitive, as was receivers from that manufacturer and age. Look in the manual what gain is expected at the antenna port.

AMU - “a mostly meaningless unit” is how I was described it from an industry insider. Note that you can change the unit to SNR if you want. Then SNR is also not always comparable  between manufacturers and models.

Kind regards,

  Björn 

Amplitude Measurement Units? There was a discussion about this some
years ago. The TBolt (and other Trimble receivers) give a user the
choice of reported GPS signal level in either AMU or dBc units. Mark
Sims (Heather author) collected data using both units to create a
cross-reference. I've attached his data (file:
2010-Sims-TBolt-AMU-dBc.zip).

His reply from 2010 is here:

mode).

The entire thread is worth reading from the archives. [1] [2]

See also table 7.3 on page 124 of:

ftp://geodesy.noaa.gov/pub/abilich/papers/ABilichThesis.pdf

/tvb

[1] https://www.febo.com/pipermail/time-nuts/2010-February/044968.html

[2]
http://lists.febo.com/pipermail/time-nuts_lists.febo.com/2010-February/027397.html

Bob, Tom and others-

It is 1 uHz. My error, not putting in PPM or PPB /24 hours, typical transmission/reception time frame for (very) vlf.

Thanks all,

Charlie
N6CFH

-----Original Message-----
From: time-nuts [mailto:time-nuts-bounces@lists.febo.com] On Behalf Of Bob kb8tq
Sent: Wednesday, March 10, 2021 8:05 AM
To: Discussion of precise time and frequency measurement time-nuts@lists.febo.com
Subject: Re: [time-nuts] World's most precise.... wall clock

Hi

I suspect that is a function of how the 3805 presents the data.

Sub ns jitter is doing well at 1 second with GPS. Accuracy is different
than jitter. Since the GPS clock is not a direct expression of UTC from
BIH ( nothing is … sorry about that …) there is some back tracking to get
very accurate time.

If you have the $300 to $2000 for a multi band GNSS timing receiver, it will
indeed help a bit. How much will depend a lot on the state of the ionosphere
and the correction process. Troposphere also gets into things. I don’t know of
any receiver that directly estimates Tropo delay.

Best would team the fancy receiver with a fancy standard. An OCXO is better
than a TCXO. Most Rb's beats the OCXO long term. A Cs will beat them both
if you run out long enough. You then get into things like GNSS disciplined
Passive Hydrogen Masers. Properly done they should perform quite well. A
disciplined Active Maser would / could beat a Passive Maser …..

The better the “flywheel” the better the result, at least for frequency / stability.
It will count off seconds quite nicely. Just how far off from “right” those seconds
are is a bit unclear. ( = you still are not accurate)

For accuracy you need a path back to BIH and the “official” definition of UTC.
That’s true even with the brand new fresh from the factory disciplined Active
Maser that you sold the house to buy ….. There are lots of nasty little delays
that creep into the mix …. All of them need to be taken care of to below your
target error level. If you are after < 10 ns accuracy, this could get pretty exciting.

Bob

time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.