Dear LF Group,
Have returned to work this morning after the Christmas break, and have now
waded through most of the 400 or so e-mails that had accumulated. I'm sure
I have accidentally deleted a few, so if you don't get the response you
expected, please e-mail me again.
Thanks for the reports on the 7FSK "measles" mode - well, you have to admit
it livened things up a bit. For the record, I transmitted the following
signals:
25/12, 2230-0712, 135.923k, 30s dots, 0.2Hz tone spacing, 1W ERP
27/12, 0000-0640, 135.923k, 30s dots, 0.2Hz tone spacing, 1W ERP
27/12, 2230-0656, 135.923k, 60s dots, 0.1Hz tone spacing, 0.4W ERP
29/12, 0000-0715, 135.924k, 30s dots, 0.2Hz tone spacing, 1W ERP
29/12, 2245-0715, 135.925k, 30s dots, 0.2Hz tone spacing, 1W ERP
30/12, 2110-0730, 135.923k, 30s dots, 0.2Hz tone spacing, 1W ERP
The message read CQ_M0BMU_M0BMU repeated 4 times an hour (or twice when
sending 60s dots), together with 2 x 6wpm CW ID's per hour
Not being in touch by E-mail over the period meant that I just selected
frequencies which were not in use before I started; sorry if it was a
frequency you were planning to use.
It was nice to see lots of activity on the transatlantic front. The
attached jpeg from last night shows, from the top, G4FTC (60s QRSS),
DJ2LF(60s QRSS), G3YXM(60s QRSS), DF6NM(300s DFCW), IK5ZPV(30s QRSS) and
G3AQC(60s DFCW) - I also copied OM2TW and G3KEV at various times.
The 7FSK generator here is based on a 13MHz TCXO module I found in the junk
box. It has a pin to connect to an external frequency trimming pot.
Applying a variable voltage to this pin gives a reasonably linear frequency
shift up to about +/- 20ppm. Providing the temperature does not change
rapidly, stability of a few parts in 10^8 is obtained. The tuning voltage
is supplied by a simple D/A converter consisting of a string of resistors
and an analogue switch IC, followed by an adjustable gain amplifier to set
the frequency shift. I settled on 0.2Hz spacing for 30s dots, and 0.1Hz for
60s because when viewed on a spectrogram, these spacings gave an easily
identified separation between the tones, allowing for some drift in TX and
RX. The 3 bit digital input at the moment comes from a suitably programmed
EPROM, clocked through it's addresses with a long cascade of digital
dividers. To get to 136kHz, the output is divided down to 1MHz and fed into
the external reference input of my Racal 9084 synthesiser. Some gates allow
the 7FSK output to be overdriven by a CW identification signal obtained via
ON7YD's QRS software. I'm hoping someone will write a 7FSK program soon, so
I don't have to blow a new EPROM every time I want to send a different message!
It seems several people are missing the point of 7FSK - the idea at this
stage of development is not to improve the signal to noise ratio of the
received signal, but simply to increase the speed of signalling that can be
achieved with given signal strengths, as Rik has pointed out. From this
point of view, it seems to work quite well, with the 7FSK covering the same
distances under the same conditions, yet being a lot quicker. Trying to
have even a minimalist QSO with 30 or 60s QRSS takes hours, and is rendered
very difficult by fading. The majority of stations that have so far been
detected across the Atlantic have had to resort to these long dot lengths
in order to be detected at all, so anything that can speed things up has to
be taken seriously. For beacon signals, the speed of signalling is not very
important, since it is only necessary to transmit enough information to
identify the beacon, rather than to have any actual exchange of
information. So for beacons, QRSS is quite satisfactory, while for QSOs it
isn't.
I think those who say that 7FSK is an inefficient use of bandwidth have not
really thought it through. Compared to QRSS, or aural CW for that matter,
7FSK uses 7 tone frequencies, whilst CW uses only 1. Therefore it needs 7
times the bandwidth when using the same dot length. For random
alpha-numeric characters, I reckon morse code averages about 11 dot lengths
per character, whilst 7FSK uses 2 dot lengths for all characters - so for
the same dot length, 7FSK is 5.5 times quicker on average. So the relative
efficiencies of this mode in spectrum usage, measured in
characters/second/Hz or however you want to express it is about 5.5/7, or
about 0.8 times that of CW - not very different, really. If you wanted to
send the same messages using CW and 7FSK in the same length of time, you
would have to use CW dots 5.5 times shorter than the 7FSK dots, so the
bandwidth would be 5.5/7 = 0.8 times that of the 7FSK signal. So from the
viewpoint of spectrum usage, it is swings and roundabouts - QRSS does use
less bandwidth than 7FSK, but only when it is sending information more slowly.
I am slightly amazed that 1.4Hz is considered by some as an excessive use
of bandwidth. Over 1000 stations could operate on 136k simultaneously in
this mode with the same parameters I have been using, even allowing for a
generous guard band between them - many times more than could use, say, SSB
on 80m. Bear in mind that most people are happily using normal hand CW on
136kHz - if you look at this on a spectrogram, you will see it takes up
50Hz or more, but I have never heard the band anywhere close to being full
up - most of last year people complained about there being too few signals
around.
Most of the problem with frequencies seems to be due to the 3 1/2 Hz
bandwidth that Argo has when used in 60s dot mode, which is a bit
restrictive. I have been using Spectrum Lab, which allows you to set any
bandwidth you like within the limits of the FFT. I usually use 10 - 20Hz -
with the window expanded to fill the screen, this gives plenty of
resolution. It also allows you to pan the frequency or alter any of the
other parameters "on the fly", immediately showing the effect on the
previous few minutes' signals, a very useful feature when a number of
different modes are in use. I was actually transmitting 30s dots most of
the time, so the displays people were getting were in many cases not
optimum - with Argo in 60s mode the signal looks very wide, and causes the
dots to overlap (I was not transmitting 2 frequencies at a time as someone
suggested!), which makes the signal harder to copy and gives them the
"measles" appearance. All the same, the improvement in speed over the
neighboring QRSS sigs is obvious.
At the moment, I'm trying to catch up on a bit of sleep, but if anyone
would like to try experiments with this or other modes, please let me know.
Happy new Year to all,
Cheers, Jim Moritz,
73 de M0BMU
|
