At the end of this month a long term writing task will come to an end and I can
begin to look at a number of various shelved projects again. One of these is
GPS locked BPSK for LF.
With a low cost GPS module (and they are becoming very cheap now !) anyone can
get a 1 pulse per second signal accurate to within a microsecond of UTC. Which
makes all the usual requirements for decoding coherent transmissions such as
clock and carrier recovery redundant. I've had one or two ideas about simple
equipment to demodulate, and not using soundcards, and will be looking for
someone rto do some experimenting with.
Can anyone can generate synchronised PSK with symbols that are an integral
number of seconds in length and timed to within a few milliseconds of UTC ?
Frequency needs to be accurate to within about a tenth of the symbol period,
ie for 1 second symbols an accuracy of about 0.1Hz will suffice, for 30 second
signalling 3mHz is needed.
I have already made a start, using a PIC based two channel A/D converter to
read the output of an I/Q detector working on the 1kHz tone from a
converntional receiver. This reads the level from the I and Q mixers every
second synchronised to the 1 PPS from a GPS module. The integrated I and Q
voltages are sent every second to a PC which in turn integrates over a longer
period and calculates the phase, plotting to a graphical display.
At the moment all I have for a test signal is the GPS locked DDS source (see
RadCom last year) sending a locked BPSK waveform with a sufficiently low signal
leakage from a 60mm piece of wire as the antenna, to make the signal virtually
undectable on ARGO with 10s dots and just about visible with 30 second dot mode
and probably too weak to resolve any SlowCW signalling information. Using 30
second BPSK symbols a very rough and ready quick test last night indicated a
very definite square wave phase plot, mirroring exactly the transmitted
waveform that would have given reliable decoding.
That's as far as I am going for now until the writing task is complete and sent
off, but I also have an outline idea for a even simpler PIC based interface
that does away with the I/Q detector, taking in the 1kHz directly and
generating a GPS synchronised quadrature LO internally, but that will have to
wait a few weeks. Meanwhile, if there are any Windows programmers who would
be interested in taking this on board with a view to producing a soundcard
version can you get in touch. In the Chirpview software by G0TJZ, Andrew
locks the Soundcard to GPS by feeding the pulse into one of the audio channels
which gives a resolution equal to the sampling rate.
For slow signalling at LF a few tens of milliseconds timing accuracy is
adequate for symbol timing, and given an accurate external 1 PPS signal, GPS
itself may not even be essential if this can be set with reference to another
timing reference - such as MSF, DCF etc.