Earth shattering news for GNSS, commercial availability of Chip Scale Atomic Clock (CSAC)

Earth shattering news for GNSS, commercial availability of Chip Scale
Atomic Clock (CSAC)


For 1st generation CSAC, amazing:

16 cc volume (1 cu inch)
35 grams weight
115mW of power
4 orders of magnitude better clock stability than typical frequency
crystals found in typical GPS receivers, that means that a CSAC will
drift in about 3 hours what a typical crystal would drift in a single
second. According to the supplier, Symmetricon, this CSAC is already
as accurate as a Caesium atomic clock !

So cost seems to be the only bottleneck before this component is
everywhere ultra high accuracy is desired. With WAAS receivers costing
around US$ 10k, a US$ 200 atomic clock might be an acceptable extra
cost. Although it probably costs a more than that right now (no price
estimates are to be found).

Currently, WAAS enhanced GPS ranging typically have a worst case
uncertainty of 7-20 meters due to errors that can't be 100% corrected
(ionospheric error, tropospheric error, multipath error and
uncorrected GPS ephemeris and clock errors). In order to calculate a
position, typical WAAS receivers consider the clock an uncertainty as
well, since their internal clocks drift too fast to be considered
stable, so they are being constantly corrected with the received
signal, with an on board atomic clock, the receiver clock becomes
stable enough that they're no longer a full uncertainty, they still
need some clock updates, but only when the sky is full of satellites,
allowing for a very, very accurate time calculation.

So having a CSACs onboard removes the clock uncertainty, requiring one
less satellite for any given performance level, so a basic 3d fix with
3 satellites, RAIM navigation with 4 satellites and FDE RAIM with 5
satellites (fault detection and exclusion). With a CSAC onboard,
LPV200 approach availability would improve to 99.99% of the time in
CONUS, and would become available 98% of the time in places where
today it might be available 70 or 80% of the time (Mexico, Caribean,
western Alaska).

CSACs might also make their way into all DPGS stations, improving
their calculated corrections. All current WAAS stations have a
standard Rb atomic clock (costs US$ 35000 each, lasting less than 10
yrs), replacing each Rb atomic clock with a triple redundant CSAC
facility will save a bundle on WAAS maintenance (with three clocks a
single faulty unit can be detected, excluded and marked for
replacement without stopping the station). Regular atomic clocks also
are temperature sensitive, requiring air conditioning, this CSAC can
handle temperatures from -10C to +50C.

One interesting improvement to the worldwide aviation network that has
been much talked about for a while, but it's still just academic work
is the DME pseudolite. Current DME is a low accuracy (550 ft / 0.1 nm)
solution compared to WAAS, in their current active way (aircraft need
to send a pair of pings, listen to the pair of pongs, before they can
use that distance information). DME pseudolites would enhance DME
stations with a passive (regular GPS like data message, in the same
way ADS-B sends each aircraft position to everybody else regularly
instead of requiring secondary radar inquires to respond to).

DMEs already respond (transmit to aircraft) using the L5 band that is
coming online as the next GPS signal (just one lonely GPS satellite
broadcasting it so far, more coming, plus all WAAS GEOs also have an
L5 signal). Eventually WAAS receivers will listen to GPS signals using
the current L1 band plus L5 as well. So adding a GPS like signal to
each DME station would make a lot of sense, future receivers will be
listening anyways, and the fact that DME stations don't need to go
through 1000 miles of atmosphere to each aircraft can hugely improve
the accuracy of this GPS like signal to each user, as long as each DME
pseudolite (pseudo satellite) has an atomic clock, a CSAC.

Each DME pseudolite transmission reaches all aircraft in view,
allowing DME pseudolites to be used with no congestion concerns. No
ping-pong for new (passive) users. At the same time, current DME
system provides the user with actual ranging to the station. DME
pseudolite provides users with an additional GPS like ranging source,
potentially far more accurate even than normal GPS satellites. It
doesn't directly provides the user with the distance to the station,
GPS methods are needed to calculate that. But most uses of DME today
are in DME-DME systems that determine actual aircraft position, much
like WAAS, but far less accurate.

Like a WAAS reference station, DME pseudolites know precisely where
they are, down to a tenth of an inch (millimeter). To function the
only other information they need is an ultra accurate clock
synchronized to GPS time. By having a local atomic clock, they no
longer need to constantly calculate the current time from GPS/WAAS
signals, they can just compare the local clock with the GPS/WAAS
calculated time, and periodically use the GPS/WAAS calculated time to
make tiny corrections to the local atomic clock, at times when the GPS
signal availability is excellent. A DME pseudolite could also detect
bad GPS signals, since it know where it is and what time is it, it can
validate each ranging signal it receives from each GPS satellite, much
like WAAS and LAAS does.

A CSAC enhanced DME pseudolites with GBAS aviation receivers with an
integrated CSAC and a IMU (inertial navigation unit) might just make
LAAS obsolete, since both CSAC and IMU both provide with means to
verify the validity of the GBAS augmented navigational fix, detecting
both jamming, spoofing and solar storms. DME pseudolites could perform
basic localized integrity monitoring just flagging bad satellite
signals instead of transmitting a full fledged correction signal. It
could also transmit ionospheric corrections observed as part of the
ranging calculation process, there are a lot more DME stations in the
world than there will ever be GBAS reference stations. The receiver
could use the pseudolite iono correction say if it's less than 50nm
from the station, otherwise it would use the GBAS iono corrections.
You might ask why a dual frequency GBAS receiver that can calculate
its own iono corrections need iono corrections for, the answer is, it
will take a long time until all GPS satellites are broadcasting an L5
signal and without L1+L5 signal from a satellite, the user can't
calculate it's iono free pseudo range by itself, it might take another
20 or more years realistically until all active GPS satellites are
broadcasting L5 signals, having great iono corrections allow for end
users to calculate higher accuracy ranging calculations from all
healthy GPS satellites, not just from satellites with an L5 signal.
One would hope DME pseudolites will be able to use current semi-
codeless GPS techniques, but will also be able to use L1, L2C and L5
to perform ultra accurate triple frequency iono corrections, to
calculate iono free pseudo ranges with just 1 mm uncertainty instead
of the usual 1-2 inch uncertainty with a stationary receiver, and in
GPS time is distance and vice-versa (conversion via the speed of
light). This might even allow for survey free DME pseudolite
installations, the pseudolite can easily determine its own precise
location within one day of test activity.

Ultimately, we could have far more pseudolite only DME stations than
we currently have regular DME stations, since a single DME channel can
be shared between many passive DME stations (CDMA based PRN codes,
like GPS uses today). Areas with CAT III and CAT II approaches could
benefit from multiple DME pseudolites, allowing VPL and HPL in the 1
meter range (CAT IIIb using satellite requirement).

Marcelo Pacheco

On Tue, 18 Jan 2011 19:36:20 -0800 (PST), macpacheco


It seems to be $1500 each for small quantities:
http://classic.cnbc.com/id/41131245
--
Muzaffer Kal

DSPIA INC.
ASIC/FPGA Design Services

http://www.dspia.com

/41131245

Great info, thanks Muzaffer. Given typical advancements in this kind
of technology, should mean $500 CSACs for high volume purchasers in
2-3 yrs, good enough for mass usage.

On 2011.01.18 22:36 , macpacheco wrote:


The CSAC is not a deploy-able product. It has to be integrated into a=20
receiver. That, with design, integration, testing and certification=20
will drive up the cost considerably. The $1500 will go to about $10 -=20
$15K as an educated guess.

For a fielded WAAS station, the -10=B0C may be unacceptable. (There is=20
the military version which is likely more expensive, that goes lower,=20
-40=B0C).


--=20
gmail originated posts filtered due to spam.

wrote:

From what I've seen all WAAS stations today are co-located with Air
traffic control centers. Perhaps one or two in alaska are stand alone.
Even -40C is probably not cold enough for Alaska, there it will need
some climatization.

Anything new will take years before the FAA is willing to touch it.
They are a lot more conservative even than the military. The current
WAAS network is operational, what's the need to replace things out of
the blue. By the time they start considering it, price will have
dropped considerably, probably when they perform the migration from
semi codeless to L1 CA+L5, probably circa 2020 (they stated that their
plan is an all or nothing thing, once they migrate, all WAAS reference
stations won't receive nor P(Y) nor L2C, L1 CA+L5 only, so they can't
migrate before L5 reaches FOC anyways, its a step backwards in my
opinion, but I digress) ... But as far as integration, atomic clocks
are a misnomer anyways, they are actually frequency sources, a timing
signal, or am I wrong ? I mean... they don't really keep time
(nanoseconds since a given date), they make a 10 MHz timing output
plus one or two other formats, that in turn is used not only to track
time, but also used for all kinds of very important receive/transmit
actual RF hardware (the more accurate clock available for the GPS RF
receive can improve receive SNR margins for instance and can transmit
a more precise signal for instance when generating the C band uplink
for the GEOs). A 10 MHz clock is a 10MHz clock, sure it will require a
ton of testing, but making it work, is it that expensive ?

Marcelo Pacheco - Not an electronics expert.