Message-ID: <14313.200103061746@gemini>
From: "James Moritz" <j.r.moritz@herts.ac.uk>
Organization: University of Hertfordshire
To: rsgb_lf_group@blacksheep.org
Date: Tue, 6 Mar 2001 17:50:56 +0000
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII; format=flowed
Content-Transfer-Encoding: 8bit
Subject: LF: <TECH>Transatlantic modes - what next?
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

Dear LF Group,

The winter has seen a fair amount of success in the transatlantic 
tests, and now that the "season" is probably near to an end, it is 
worth thinking about what the next steps could be, in good time for 
next winter. The following ramblings were the result of thinking what 
I could do next, in the way of technical development of my LF 
station, in particular regarding what would be the best mode to 
work on:
-------------------------------------------------------------------------------------
So far, all reported trans-atlantic signals have used some type of 
QRSS. The variable parameter is the dot length - longer dots mean 
better signal-to-noise ratio, with the penalty of slower message 
transmission. Recognisable callsigns have been produced with 60s 
dots or longer, which requires more than an hour to transmit one 
callsign. ON7YD's DFCW roughly halves the time required, at the 
expense of doubling the bandwidth occupied by the transmitted 
signal and being slightly more complex to implement, which seems 
to be a good trade-off. 

The maximum radiated power available to several stations has 
reached the 1W ERP level in Europe, and somewhat more than 
this for the stations active in Canada. The ERP that is feasible 
depends mostly on how big an antenna can be put up, but in any 
case is limited by regulation to 1W in Europe, which eliminates 
further improvements. ERPs of a few 100 milliwatts have been 
sufficient to cross the Atlantic on several occasions.

Under optimum conditions, with favorably sited and equipped 
stations, 3s per dot QRSS signals have been passed both ways 
across the Atlantic, but this is the exception rather than the rule. 
The problem with longer duration dots is that any kind of 2-way 
contact takes an excessively long time, running into several hours. 
This problem is compounded by the fact that propagation "lifts" on 
LF only last an hour or so. So it would seem we are reaching the 
practical limits with QRSS.

What is required for a practical 2 way LF DX mode, capable of 
operating at the extremes of distance and SNR? A while ago, 
G4JNT posted an estimate of what might be theoretically possible 
using different techniques; another way is to look at what is 
needed to serve our purposes. I would suggest the following "wish 
list":

1)Be able to complete a minimal QSO (about 50 characters) in one 
hour. This would give the lowest rate of signalling capable of using 
the propagation lifts to complete a QSO "in one sitting".

2)Be able to transmit/receive all alphanumeric characters and 
essential punctuation/procedure signs, in order to be generally 
usable by any station without special arrangements.

3)Occupy a bandwidth of less than 10Hz - this is neccesary 
because of the very limited spectrum available, and the fact that 
several stations will be operating simultaneously. 

The QRSS modes easily meet 2 and 3; in order to meet 1, a dot 
length of about 7 seconds maximum would be required. With the 
best possible conditions, I guess several stations might manage 
transatlantic QSOs with these dot lengths. However, it would 
probably not be enough to reach the more inland parts of Canada, 
or the USA and further afield. By the way, I reckon about 6dB SNR 
is needed to see a QRSS signal on a spectrogram under 
favourable conditions; if there is much QRN, 10dB is probably 
required. It is possible to see a trace of signal with 0dB or less 
SNR. All this is fairly subjective, however.

One way to use longer dot lengths without increasing overall QSO 
time is to use multiple frequencies - for example, DFCW, VA3LK's 
7-tone scheme, and VK2ZTO's one-tone-per character VFSKCW. 
Taking this to it's logical extreme, it would be possible to assign 
different frequencies to all possible combinations of callsigns and 
signal reports, so each over of a QSO would just consist of a 
single tone. However, it would then be difficult to meet condition 2. 
I guess there must be an optimum trade off between number of 
tones, difficulty of encoding and decoding, redundancy and so on. I 
suspect it might be 2 tone DFCW, but I don't know.

Then there are the "digital" modes, specifically BPSK. Currently, 
most effort has been expended on the MS100, 10 bits per second 
variety of BPSK. This easily meets conditions 1 and 2. However, 
for the same signal levels, QRSS seems to do better with 
acceptable, if much slower, speed. Also, the bandwidth occupied is 
roughly 40Hz, too wide for condition 3. But with the 16 bits per 
character coding scheme normally used for BPSK, 2250 
characters per hour can be transmitted, far higher than is actually 
required. So the bit rate could be greatly reduced, and/or the 
coding altered to a greatly increased number of bits per character, 
hopefully improving the readability of the signal. Reducing the 
overall speed by a factor as much as 45 would still meet condition 
1. To fit into a 10Hz bandwidth, the bit rate would have to be 2.5 
bits/sec (MS400) or less, so you could encode each character with 
up to 180 bits if you wanted to. Or, sticking with 16 bit codes, 0.22 
bits/second (MS4500) would still be OK. What we want is the best 
trade off between bit rate and encoding for very poor signal to 
noise ratio. I don't know a great deal about this subject, but I 
expect some readers of this reflector already know the answer. 

Beacon signals are a bit different; here, the only requirement is to 
positively identify the signal, and make some estimate of the signal 
level. An on-off keyed carrier with a simple repeating pattern and a 
well defined frequency is easily identified with simple equipment, 
and has the advantage of flexibility at the receiving end - you can 
make the bandwidth arbitrarily narrow, or perhaps take advantage 
of the coherent nature of the signal, to improve detection 
capability. You can also monitor several signals at once.

So any suggestions/comments would be welcome -     well, almost 
any! By the way, I now have BPSK at up to 1200W PEP from my 
Decca TX, if anyone would like a sked/tests, etc.

Cheers, Jim Moritz
73 de M0BMU