From G3PLX:
Some results from the QRPP tests for which I asked for volunteers a few days
ago. The test were to evaluate the performance of the 'Clicklock' technique
which I have been developing. This uses a harmonic-rich 1 pulse-per-second
'click' from a GPS receiver to transform an ordinary SSB receiver into one
which is coherent (i.e. locked in frequency and phase) to GPS. This is done
entirely by sound-card based DSP software, so it could turn out to be a very
easy way to exploit the benefits of using GPS for frequency-locking on LF.
G4JNT volunteered to transmit from a commercial synthesiser with an external
reference locked to GPS. He planned to transmit a few micro-watts. After an
overnight run I had received nothing so I asked him to increase power to a
few watts. I could see him by conventional methods and the reason I hadn't
heard his QRPP signal was clear - he was 0.0063 Hz high of the frequency on
which I was listening and my bandwidth was only +/-0.0022 Hz! After
studying the circuitry of the synthesiser it was clear that although the
main steps were locked to GPS, the 1Hz 'fine-tune' steps were not. He then
connected up one of his DDS kits instead, again locked to GPS. It couldn't
zero exactly on the whole-Hz frequency we had planned to use, but the error
was only half a cycle in 11 months so it wasn't a problem. To speed up this
second test, Andy used a 1 watt transmitter power, giving an estimated 120uW
emrp.
His signal was good, about 12dB/noise over a whole day. There were a few
fades of some 10-15dB, and apart from some random phase jumps which were
probably due to in-shack disturbances, the phase was stable, so the
GPS-locking was working at both ends. The path length between us was 400km.
This 12dB was measured after integrating the signal for 225 seconds at a
time, which is equivalent to a bandwidth of 4.44mHz.
The second volunteer was G3JKV, about 350km away, with 100mW transmitter
power and an estimated emrp of 15uW from what he described as 'as many
strapped-feeder wire dipoles as I could connect in parallel'. He was using a
synthesised laboratory generator fed from a Rubidium standard.
G3JKV was about 2dB down on G4JNT, which doesn't quite add up, but probably
means that it's more difficult to estimate radiated power than we
anticipated and this would make the comparison difficult. His signal was not
quite on frequency but still well within the 2mHz passband. Like the signal
from G4JNT, there were a few fades during the night but the daytime signal
was very stable indeed, and the daytime noise-level was lower. After the run
I estimated the frequency error by counting that his signal had rotated 7
cycles over the 14 hour test. This represents a frequency error of 130uHz.
Later on, Walter told me that it looked like the Rubidium tube had failed
and the system was free-running!
The conclusion of the tests so far as that, given the frequency stability,
it's possible to receive power levels of the order of 15-100uW over
distances of 350-400km with SNR's of 10-12dB, in a bandwidth of 4.4mHz. The
propagation is quite stable enough for daytime tests at this narrow
bandwidth, and indeed I am sure we could have gone lower in power and still
detected a signal. Looking at the log of the G3JKV test, which is
effectively a recording of the signal in the 4.4mHz bandwidth covering the
whole test, I can simulate what would have been the performance of a BPSK
coding system. A 320 bit data sequence would have suffered just 5
corruptions, but of course this is at the rather low data rate of 2.2
milli-baud, that's one bit every 225 secs.
The Clicklock technique is also being evaluated by two other people, one or
both of which may well release software for general use, but it's still
early days and we need to think of the best way to build this idea into
useful communication software, and also think about ways to lock
transmitters to GPS.
73
Peter G3PLX
73
Peter
|
