Envelope-to: dave@picks.force9.co.uk
Delivery-date: Thu, 23 Feb 2006 20:56:14 +0000
Message-ID: <020a01c638bb$652dfbb0$0300a8c0@LAPTOP>
From: "Peter Martinez" <peter.martinez@btinternet.com>
To: rsgb_lf_group@blacksheep.org
References: <21f.861fa8c.312cfc09@aol.com>
Date: Thu, 23 Feb 2006 20:54:53 -0000
MIME-Version: 1.0
Subject: LF: GPS-locked reception
Content-Type: text/plain; charset=iso-8859-1; format=flowed
Content-Transfer-Encoding: 8bit
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org

>From Peter G3PLX:

Greetings to the group. I was subscribed here some years ago when we first 
got 136kHz and I was transmitting on the band myself but I have been busy 
with other projects since.

One of these has been an idea for the ultimate in frequency stability, but 
using an ordinary SSB receiver, a GPS receiver with a 1PPS output, and some 
soundcard software. Oh, and one more thing is needed - a 47pF capacitor.

I contacted a couple of LFers recently, including Brian CT1DRP, VE7TIL, and 
W1TAG/WD2XES. These stations have been transmitting special signals for me 
that are locked in frequency to GPS, and these signals have enabled me to 
confirm that this new idea is working well. The tests also give some idea of 
how narrow in bandwidth we might be able to go if the propagation will 
permit. I thought it was time I rejoined the reflector and passed on some of 
these ideas.

I first thought of this trick after I connected the 1PPS signal from my GPS 
to the antenna of the main receiver, wondering if a narrow spectrum analyser 
would show this as a comb of harmonics every 1Hz. It sure did, as many 
people will have found, but I also watched the 'click'on a scope connected 
to the receiver audio, and noticed that the waveform had a characteristic 
'beat' to it. I realised that the beat was the difference frequency between 
the receiver dial readout and the nearest multiple of 1Hz, and this could be 
exploited as a drift-cancelling idea.

The technique therefore involves connecting the all-important 47pF capacitor 
from the PPS pulse of a GPS to the LF receiver antenna. The same 1PPS signal 
connects to the right channel of a stereo soundcard. The receiver audio, 
with a mix of the distant weak signals and the RF click, connects to the 
left channel.

The software first uses the 1Hz waveform on the right channel to calibrate 
the soundcard samplerate. The left channel audio is processed through a 
precision software frequency-shifter, like a very fine RIT control, working 
over +/-0.5Hz. The RF click in this shifted audio is examined by the clever 
part of the software and the RIT control is adjusted to zero-beat it. The 
result is that the frequency-shifted audio has the receiver frequency offset 
cancelled. The output is as stable as if it had come from a receiver with 
ALL it's oscillators locked to GPS.

This is a simplified description of course. In fact I don't feed the 
drift-cancelled audio back out again, but process it inside the software, 
and I think this will be the best way to use this idea for weak-signal LF 
work. I have done ultra-narrow waterfalls (+/-2mHz), BPSK demodulation, and 
in particular charts showing phase drift over time to see how well the 
system is working.  Many LF operators have built GPS-locked DDS systems 
already but this simple technique means we can now GPS-lock the LF reception 
process, and it's all done in software (except for the 47pF capacitor of 
course).

One thing that this technique is showing is that the propagation medium may 
be a lot more stable that we thought. I found I can stay within 30 degrees 
of the LORAN line from Rugby (300km away) on 137.57985440499182885 kHz for 
days on end. This means there is no lower limit to how narrow we can try to 
receive extreme QRP over surface-wave paths if both ends are GPS-locked. On 
the 1500km path from CT1DRP to me (54N 3W) the signal is very stable during 
the midday period. Around sunset and sunrise the phase changes more 
rapidly - about 200 degrees/hour. That amounts to a Doppler shift of 
+/-150uHz at sunrise/sunset.  But after the sun rises again, the phase 
returns to the exact same value as the previous day.  I haven't done much 
yet on easy/west paths but I was able to hold WD2XES within 90 degrees over 
2 hours the other night, so QRSS-7200 might be workable over the TA path if 
we had the patience to do it!

There will be no ready-to-fly downloadable software from me. Anyone who 
wants to have a try at this who has already some experience with soundcard 
software, is welcome to try it. There's great scope for experiment and new 
ideas for using this simple idea, perhaps as added front-ends for existing 
software, or for completely new techniques for ulra-narrowband working. This 
reflector would be a fine place to discuss these.

73
Peter