[Fwd: LF: The Loran problem]
Sun, 02 Aug 1998 12:27:21 -0400
Paul Keinanen wrote:
> At 19:47 1.8.1998 -0400, Derek, G3GRO wrote:
> > As Paul OH3LWR says, what is needed is a separate
> >receiver and antenna input circuit and a wide band detector to trigger a
> >noise blanker switch in the main RX channel. I think that there are
> >certainly possibilities in this approach. Some years ago the Crawley Club
> >experimented with a similar system designed by G3LNM to suppress severe
> >interference on the 70cm band from the French Syledis position fixing
> >system at our VHF NFD site at Folkestone on the south coast which looked
> >very promising but in the event, it was never actually used because the
> >offending Syledis station conveniently closed down.
> > I think however that it would be useful to aquire some information on the
> >characteristics of the Loran pulse shape etc. to follow this method up as a
> >possible solution.
> After posting my previous message, I located the Norwegian Ministry of
> Fisheries home page, which contains information about the Loran-C system in
> North Sea and North Atlantic up to the Spitzbergen. The page
> contains a basic description of the pulse train. There are 8 (or 9) pulses
> transmitted at 1000 us repetition rate (with 2000 us rate before the 9th
> pulse from the master station). The pulse is _nominally_ 200 us wide.
> However, the actual pulse is shaped to reduce the emitted bandwidth.
> Basically, the beginning of pulse envelope looks like a slightly modified
> sinus curve (0 to about 120 degrees) followed by a linear decay followed by
> something that looks like an exponential decay.
> Since the carrier frequency is 100 kHz or 10 us cycle time, after 3 cycles,
> the amplitude has grown to 50% (-6dB), the peak is reached at 65 us.
> However, after the nominal 200 us pulse length (20 cycles), the amplitude is
> still 16 % (about -15 dB), slowly decaying during several cycles.
> The transmitted power should be 99 % within the band 90 kHz .. 110 kHz and I
> could confirm this by manually digitizing the pulse envelope to a DSP
> simulator (the ancient DSPlay with only 512 point FFT :-), gating a 100 kHz
> "oscillator" and noticed that the strongest out of band spurious were 30 -
> 40 dB down. However, weak spurii were spread well beyond the 136 kHz. I do
> not claim any accuracy in digitizing the pulse envelope, but neither does a
> real Loran-C transmitter exactly follow a theoretical waveform, so it is
> very hard to state anything about the amplitude for the spurii at 136 kHz.
> Due to the gradual attack and decay times in the pulse, the Loran signal to
> noise ratio should be very clean to get a clean blanking pulse and avoid
> false triggering from other sources. The slowly decaying tail seems
> problematic in this sense.
> One solution might be to generate a (digital) pulse train locally and only
> synchronize it with actual Loran transmission. This way the blanking pulse
> could reliably start just prior to the actual transmitted pulse and the end
> of the blanking pulse could be set to a time, when the transmitted pulse has
> decayed to an insignificant amplitude.
> >I have just started to investigate the use of RF signal cancellation
> >techniques using a separate antenna in conjunction with David, G0MRF and
> >Lech, G3KAU.
> If the blanking pulse is generated locally and locked to the transmitted
> signal, what about trying to locally imitate the shape of the transmitted
> pulse, which would contain the same out of band spurii as the transmitted
> signal. After phase shifting, this could be used to cancel the received spurii.
> The problem is how to generate the complex pulse form locally. Since all
> pulses should have the same shape, summing a large number of pulses (100 ..
> 1000) e.g. by using running averages, the pulse signal to noise ratio could
> be increased by 20 .. 30 dB, removing most non-Loran interface. This would
> compensate for any distortion in transmitter and also for any slow changes
> in propagation.
> On one hand this idea might be destroyed by any random path variations
> (selective fading etc), but on the other hand, the frequences are so low
> that quite a lot of processing can be done in a PC.
> Paul OH3LWR
> The file http://odin.dep.no/fid/publ/nels/interfer.html contains a list of
> interface sources to the Loran-C service and it is used for the service
> range predictions. In fact the page contains a large list of signal sources
> from 50 kHz .. 85 kHz and 112 kHz .. 150 kHz listing frequency, location,
> longitude, latitude and radiated power (100 W .. 200 kW). The radiated power
> field is a bit suspicious, some have very rough figures e.g. 100 kW and some
> claim the radiated? power with 5 significant digits :-).
> The list is nearly 2 years old, but should be a useful reference for
> identifying inband (73/136 kHz) and out of band signal sources. If the
> quoted ERPs are reliable, then this could be used to estimate the radiatated
> power from some amateur stations by comparing received signal strengths and
> adjusting for the different distance (at least for over the water paths).