[time-nuts] Loran C returning to a station near you...
Poul-Henning Kamp
phk at phk.freebsd.dk
Tue Jul 14 19:18:02 UTC 2015
--------
In message <55A53A67.7010902 at rubidium.dyndns.org>, Magnus Danielson writes:
>> If the goal is a reliable backup for GPS, there are smarter ways to
>> use the 100kHz band than Loran-C pulses, and there really isn't much
>> reason to stay compatible with Loran-C receivers.
>
>True. I would look at PRN-codes if I where to do such a system today.
>What may be an issue is the amount of sidebands allowed, as it would put
>limits on the chipping-rate of PRN-codes or for that matter other forms
>of modulations.
I would probably stay with the pulses, they have some very desirable
properties in terms of transmitter and antenna design and bandwidth.
But I would get rid of the current spread-spectrum design and do
something like this:
We pick a basic period as a prime number of microseconds, for
instance 262139µs (just below 2^18) and we define an epoch for this.
This means that all transmitters are autonomous based on a local
TAI reference.
Each transmitter emits an individual PRN-spaced pattern of 32 pulses
in the basic period. The exact pulse patterns for a transmitter
will be picked based on vectors to neighboring transmitters.
The pulse polarity is "plus, minus, data" where every 3rd pulse is
used to implement a serial data-channel which communicates
chain-configuration data, TAI/UTC info with some bits left
over for civil defence warnings.
Using one global "GRI" means that there no longer any "chains".
This eliminates a host of failure-modes on the transmitter side and
the receivers will automatically be "all-in-view".
With all transmitters autonomous and independent, RAIM is possible.
The basic period is relatively long to attenuate any CW interference
for time/frequency purposes but the higher pulse-per-period count
compensates this for location purposes.
Making the pulse-pattern per transmitter and PRN-like eliminates
all the "shadow" effects ("baseline extension" etc) and makes for
quick (re)acquisition based on pattern-matching.
The PRN pattern will also dramatically attenuate the "loran-lines"
which polluted nearby VLF and LF bands.
The +/-/data pulse polarity makes it possible to detect the start
of the period by tracking where in the potential basic period pulses
do not change polarity: 3 doesn't divide 32, so there is a + - + -
sequence from all transmitters around the start of the period.
But then again, I have spent far more of my life on Loran-C than
can ever be justified :-)
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.
More information about the Time-nuts_lists.febo.com
mailing list