Hi Dave,
Yes you are right, I was using a program to decode the CW, although I can now 
manage to read by ear up to about 20 wpm on a good day. Regular listening on LF 
has improved my CW enormously after a 10 year gap, but I still have lapses 
caused by mind wandering, hi. On occasions I seem to get totally out of sync. 
and have to wait until I hear something familiar. Someone on this reflector 
said one time that you have to treat it as a foreign language. I would agree 
with that, to get to any speed you have to recognise complete words. Which is 
why I think one tends to get thrown with unusual groups of punctuation 
characters.

What threw me at first, were the .. .. (II) for repetition. I have not seen 
this before. Is it a popular signal for a repeat? Also of course the @ in the 
email address was an informed guess on my part.

I have been experimenting for some time with software decoders. CW is 
particularly difficult to decode successfully in the presence of noise, for 
several reasons, the variable word length, the variable speed, lack of 
synchronising at either the bit or character level, etc. Most of the decoders I 
have tried are very poor. These include: Hamcom, CWmaestro, PRO-CW, Robocopy, 
and many more. I came to the conclusion that for any degree of success, it is 
essential to use FFT software.

For sending I use a software keyer that I wrote (my hand keying is absolutely 
dreadful), which allows rates from 0.01 to 100 wpm.

Currently I am planning to extend the keyer to send other codes that might be 
easier than CW to decode by software. A fixed length code at a selection of 
fixed rates with a distictive start bit should make it much easier to get into 
phase and stay in phase. 6-bits would be sufficient for 26 letters (case is not 
necessary - I am not planning a word processor), 10 digits and a variety of 
punctuation and procedure signs. 5 bits as in RTTY requires shift characters to 
get enough codes, but shifted codes are just another thing to get wrong, when 
decoding. No stop bit is necessary. Perhaps use bit stuffing for both ones and 
zeros with a bit of opposite polarity after 3 the same to avoid long chains, 
with flag characters 011110 instead of a start bit at a frequency that can be 
adjusted to suit conditions, ie. synchronous transmission of shortish blocks to 
avoid wasteful spaces. Manchester coding is too expensive, twice the frequency. 
Either on/off keying or two frequencies as in DF
CW. Low transmission rates suitable to the band and conditions. Readable by 
ear/eye as well as by program. Code redundancy and FEC are possible additions.

So if you hear something unusual on the band, it just might be me.

73, John, G4CNN


-----Original Message-----
From: "Dave Sergeant"<[email protected]>
To: "rsgb_lf_group"<[email protected]>
Cc: "G4BRA"<[email protected]>
Date: Mon Jul 02 00:24:22 PDT 2001
Subject: LF: SAQ

From Dave G3YMC>

SAQ on 17.2kHz copied 589 both sessions on my 'bit of string' untuned longwire.
Solid copy despite intermittent elecrical QRN at a similar strength (thermostat 
or
similar).  However the signal was many db below GBRs enormous signal.

John Sexton wrote:

The message wasn't very interesting:

I agree.  It seems John's transcript was done with his morse copier software 
rather
than by ear.  What fooled me was the proliferation of not normally used 
puctuation
characters.  Perhaps it shows how inappropriate morse is for sending things 
like web
site addresses, but I was particularly fooled by the + sign (I think) before 
the FAX
telephone number (+46 for Sweden).  Don't think I have ever heard a + on CW 
before!!

Before restarting in the afternoon, I noticed the frequency going up and
down a bit - problems with the regulator?

There was noticeable chirp on the signal at all times, especially on dashes.

SA6Q not heard on 136 but it was probably while I was out in the morning.

73s Dave G3YMC
[email protected]
[email protected]
http://www.dsergeant.btinternet.co.uk

___________________________________________________