[time-nuts] Moving GPSDO

Keith Loiselle keith.loiselle at gmail.com
Tue Oct 25 17:09:34 EDT 2016

Hal Murray wrote:

I think Jackson Labs makes some non-survey GPSDOs.


The FAQ says there is a non-survey mode.

You might find something interesting in time-nuts archives from about 2 years


Here are some comments we put together regarding mobile GPSDO's:

Using GPSDOs in mobile applications is a fascinating and very complex
topic. There are three main issues affecting 3D mobile mode performance: 1)
the oscillator sensitivity to acceleration, 2) the GNSS receiver
sensitivities, and Kalman time-constants wrt motion and multipath, and its
performance difference in 3D versus Position Hold mode, 3) the products’
sensitivity to possible airflow or temperature changes during motion (for
example when operated in a UAV that has no airflow shielding).

Jackson Labs Technologies, Inc. indeed makes a large number of
mobile-capable GPSDO’s, some with vibration- and acceleration-hardened
oscillators. This typically involves hardening of the crystal mounts and
low-g sensitivity operation, and/or some sort of electronic g-compensation,
and proper “steering” of the GPS receivers’ Kalman filter modes (time
constants) and Doppler versus Carrier Phase tracking modes, but all of this
depends on what “mobile” really means, e.g. what type of dynamics can be
expected. There is a huge difference in expected operation between walking
and a jet pulling 5g maneuvers.

In our GPSDO’s and GNSS receivers starting with the low-cost $220 M12M
Replacement Receiver the Kalman and tracking mode can be auto-set and
auto-sensed via the GPS:DYMANIC command. The software will recognize
acceleration, velocity, and change of altitude and set the GPS receiver to
the proper mode that is optimal for a particular environment. There is a
huge difference between operating while walking, or the aircraft is taxiing
on the ground and getting all this multipath from the angled adjacent
hangar roofs, versus what needs to happen once the aircraft is 200+ feet
off the ground for example.

In terms of sensitivity to acceleration, and GPS performance while in
motion, we continuously do flight testing as well as drive testing and have
found that the modern GPS receivers are not very sensitive at all to
high-velocity airborne operation. Driving on freeways does present
challenges due to overpass shading, as well as multipath caused by highway
signs passing by. Usually this is not an issue because modern receivers are
able to track such a large number of satellites. Also, it is impressive to
see how quickly the modern GNSS receivers recover from their TCXO’s being
accelerated in aerobatic flight for example.

GPSDO Oscillator sensitivity to acceleration depends on the type of crystal
used. TCXO’s typically have the highest g-sensitivity (2 to 3ppb or more
per g per axis), while SC-cut DOCXOs we use have sensitivities of 1 to
1.5ppb per g typically, and low-g ordering options have less to much less
than 0.3ppb per g per axis sensitivity. During our flight testing in jets
and propeller aircraft for example we found that our “standard” DOCXOs as
used on the Fury DOCXO and other products exhibit negligible effects on
frequency stability and timing stability even when flying acrobatics (zero
g, 2g steep turns, etc). These maneuvers are typically short (less than 30
seconds) and the phase/frequency deviation experienced does not cause a lot
of drift, and once you pull out of the maneuver the crystal will go back to
its nominal frequency. These events are also short enough typically that
they fall well within the time constant of the loop filter, so the filter
does not “chase” the frequency offset. Thermal and airflow sensitivity
seems to be much more of an issue for TCXO’s and single-oven oscillators
than acceleration sensitivity during driving/flying, while warmup and
retrace errors seem to trump the g-sensitivity in DOCXO products in
airborne applications.

However if you are talking about dismounted applications (someone walking
and wearing a unit in a back pack etc) it is a totally different ballgame
because this backpack could be set onto the ground and turn over, the
person could be walking or lying down, etc etc. The change of direction for
dismounted use is much higher than for vehicle or aircraft use (if a
vehicle turns upside down you are in big trouble anyways, and aircraft
usually fly coordinated meaning the highest acceleration is always down the
z-axis. This allows users to mount the GPSDO in the aircraft with the
lowest crystal g-sensitivity pointing toward the z-axis to minimize
errors). Thus for accurate operation in these types of applications usually
low-g-sensitivity oscillators are used (less than 0.3ppb per g). Vibration
due to turbulence etc is nominally sinusoidal and thus will only cause
phase noise and short-term perturbations that average out over time.

Rubidium and Cesium oscillators have the best g-sensitivity specs, in fact
the Microsemi Rb’s and Cs’ we use typically have sensitivities below our
measurable noise floor when doing a 2-g holdover test. There will be a
quick frequency offset due to the TCXO g-sensitivity, then the atomic loop
quickly pulls in the frequency error to near zero again.

In terms of time-nuts applications, most of our units can operate in mobile
mode without any large reduction in stability or accuracy. The only product
that is not recommended for mobile operation is the Motorola M12M and M12+
receiver as used in the Fury for example. While there is a command to put
the M12M GPS receiver into mobile mode, the performance when doing so is
not optimal. The solution to that is to retrofit the unit with the M12M
Replacement GNSS receiver which uses a u-blox M8T GNSS receiver that has
absolutely charming timing performance in 3D mobile mode.

That said, there is a real issue if you operate in challenged environments
such as under foliage or indoors. In those environments the Position Hold
mode operation with Auto Survey can increase stability by 10x or more. We
had posted some example plots here some years ago showing mobile mode
operation versus Position Hold mode operation when the GPS antenna location
is extremely challenged and is only seeing about 10% of the sky from a
window flanked by walls. The stability differences between 3D mobile and
Position Hold mode were eye-opening. For those type of challenged
applications using stationary Position Hold mode makes all the difference.

So in summary: most of the higher end (SC cut) type oscillators will work
very well in 3D mobile mode assuming that they are shielded from airflow
(not necessary for the DOCXO oscillators). For the most stringent
environments one should choose a ruggedized, low-g type oscillator. All our
GPSDO’s have either mobile mode operation by default, or can be set to
mobile versus Position Hold mode. Lower-end TCXOs can have significant
g-sensitivities which can really cause issues if used in dismounted
applications where the unit can be easily turned over, but in aircraft and
vehicle usage these accelerations are typically temporary and can be easily
compensated for by the loop or entirely fall inside of the loop time
constant. Multipath presents significant challenges when driving on
freeways or taxiing on an airfield. Modern GNSS receivers perform very well
in mobile applications if the user sets the Kalman filter parameters and
Doppler versus Carrier Phase tracking modes properly according to
environmental dynamics. On most JLT products this setting can be chosen to
be fully automated via the GPS:DYNAMIC SCPI commands. If the Kalman filter
is set incorrectly (setting pedestrian mode in a vehicle for example) then
there will be a very significant delay in the position versus actual
position due to the low-pass effect of the Kalman filter. Some older GPS
receivers do not perform well at all in mobile mode, one example is the
M12+ and M12M receiver.


On Sun, Oct 16, 2016 at 11:56 PM, Hal Murray <hmurray at megapathdsl.net>

> jgray at zianet.com said:
> > What about this scenario - the GPSDO has a fixed position on initial
> > powerup, but then it is moved periodically or even continuously, without
> > updating the GPS coordinates. The distance from the original coordinates
> > could be anywhere from close by, to across the state. Periodically, a new
> > survey could be run to keep things from getting too far out in left
> field.
> Most commercial GPSDOs assume they are running at a fixed location and
> won't
> work very well unless they are near their surveyed location.  Near is
> measured in feet, not miles.  That allows them to work with fewer
> satellites
> which was important with the GPS receivers in old gear like the TBolt and
> Z3801A.
> But that's all software/firmware.  If you build your own GPSDO you can use
> whatever GPS receiver you like as long as it puts out a PPS or something
> that
> you can lock to.
> I think Jackson Labs makes some non-survey GPSDOs.
>     http://www.jackson-labs.com/index.php/products/lte_lite
> The FAQ says there is a non-survey mode.
> You might find something interesting in time-nuts archives from about 2
> years
> ago.
> --
> These are my opinions.  I hate spam.
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/
> mailman/listinfo/time-nuts
> and follow the instructions there.

More information about the time-nuts mailing list