Content-Transfer-Encoding: 8bit
Message-ID: <3617AC3245C2D1118A840000F805359C01AB8C83@pdw-mercury-1.dera.gov.uk>
From: "Talbot Andrew" <ACTALBOT@dera.gov.uk>
To: rsgb_lf_group@blacksheep.org
Subject: LF: RE Slow CW vs. BPSK etc.
Date: Thu, 1 Jun 2000 13:48:32 +0100
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1; format=flowed
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

> Kevin Ravenhill wrote :
> 
> >> The general consensus of opinion seems to be that if you are going
> to use
> >> "machine" modes, BPSK has considerable superiority over any form of
> slow CW
> >> for the kind of very weak signals often encountered on the LF
> bands.
> 
>When you say "machine" modes, are you meaning CW decoded by software or
>by the brain (be it aurally or visually) ? There is a profound
difference...


Actually, there probably isn't that much difference in practice, again
comapring like with like.
The ear(or eye) / brain combination serves as a very good error
correction mechanism where the mode being transmitted contains
sufficient redundancy to allow this.  

Examples are CW with it's long and short elements, Hellschreiber with
its pattern of letter shapes and, of course, the enormous redundancy
given by the context and spelling of words which means we can usually
understand text    whxre eveqy wbrd h6a i smeluing misyaki in it .

If this is contrasted with machine generated codes, such as
convolutional and block coding superimposed onto plain ASCII coded text,
then I guess that the fully machine based system can produce very much
better results if you are prepared to accept the overheads.   One major
drawback being that the better an error correction scheme is, the longer
it takes to process all the interleaved data.   One amateur-developed
data mode using 64 parallel tones (MT63) has 10 second overhead on data
at normal typing speed.  Whilst being extremally good at coping with QRM
(it is designed primarily for the HF environment, not white noise) the
10s between typing a character and it's appearing on the receiver's
screen does make for enforced leisurely QSOs. 

For a decent real time communications system, with a reasonable
trade-off of delay against error correction, the machine system probably
is about on a par with a (very good) human ear brain combination.   The
real advantage comes in that the machine can cover a much wider range of
data rate vs. error correction trade off values, instantly, and can
automatically adapt itself to changing propagation condidtions.  Just
compare PSK31 with 20 - 30 WPM CW, it is considerably better.

CW operators do constantly trade off time delay vs S/N.   A few years
ago I once sat on a (windy and rainswept as usual) hilltop taking over
20 minutes to receive a report and locator on 10 GHz using CW - with the
other station sending repeatedly one letter or number at a time in
conjunction with a 144MHz talkback link where I called whenever the
character had been copied.   That equates to something like 0.05 Bits /
second using a protocol which in machine terms would be a multiple
redundant error correcting Automatic reQuest Repeat system. And all just
to set a new personal best distance on this band.   10GHz does not lend
itself to low data rate narrow bandwidth communications so CW is still
of considerably more use there than it is on the LF bands.   So much for
the class A/B licence debate and access to HF bands !

The QSO would have gone something like this :

G4JNT on 2m:		"Send the first number of the report"
G8### on 10GHZ	"1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  "
'jnt			"OK Got that, now the second letter"
G8###			"2  2  2  2  2  2  2  2  2  2  2  2  2  2   2  2
2  2  2  2  2  2  2   2  2  2  2  "
'jnt			"Negative again, again"
G8###			"2  2  2  2  2  2  2  2  2  2  2  2  2  2  2   2
2  2  2  "
etc. etc. 

How much easier if we could have just wound the software down to 0.05
B/s and told it to get on with the job.  Even QRSS can't work on the
higher microwave bands, drift of most systems is a few tens or hundreds
of Hz taking the signal well outside the tuning range of Spect,,,,
Incidently, QSOs like this are common on the uW bands and, before anyone
asks, contest rules have allowances for this dual band operation built
in.

Whilst on this matter, If there are any communication specialists out
there who know of a scheme - is there a way of coding and getting the
full benefit of (say) 0.05 B/s using a spread spectrum waveform 1 - 3kHz
wide (SSB transceiver) that is immune to several 100 Hz of frequency
shift.  The moonbounce community would leap at such a data scheme as
well as the uW operators

Not a very LF orientated ramble for this reflector, but ..........

Andy  G4JNT


-- 
The Information contained in this E-Mail and any subsequent correspondence
is private and is intended solely for the intended recipient(s).
For those other than the recipient any disclosure, copying, distribution, 
or any action taken or omitted to be taken in reliance on such information is
prohibited and may be unlawful.