Hi all,

I believe that you have all the elements here to make a system
that can work at "any" SNR, i.e. a QSO could be completed even
when propagation is poor, it would just take longer.

The basic idea is that Laurie transmits a message at fairly
high speed, e.g. so that it takes only a minute to send.
The same message is repeated every minute, and Dex (or John)
has software which sums the received signals until the message
is readable.  Dex then sends his reply (over a reliable and
speedy HF channel), which tells Laurie (who can receive HF
while sending LF) to start sending the next message, or thanks
him for the completed QSO.  Dex resets his integrated message,
and starts accumulating the next one.

The above system is, of course, impractical, because the S/N
of the integrated message only improves as the square root of
the number of transmissions.  The QSO could take months.  Worse,
when propagation is poor, or noise levels are high, the received
garbage degrades the sum, and many episodes of good propagation
are needed to recover.  It could take centuries :)

However, I believe that HBG 75 could serve as a "pilot" carrier.
Laurie would transmit on a very accurately known frequency, e.g.
by locking to GPS or LORAN, or by using a Rubidium oscillator.
Dex, using a receive filter which accommodates both HBG and G3AQC,
has software which measures the apparent phase of HBG, and uses
that angle to (quadrature) demodulate G3AQC.  The integrated S/N
now improves linearly with time (under constant propagation).

As a further refinement, the amplitude of the received HBG and
the noise level are factored in, so that high quality minutes
are weighted more heavily.  The result should be nearly full
utilization of the channel, as if Laurie were always sending at
a speed appropriate to the current conditions.

Mal noted "No QSO should be valid if contact is lost and then the
next day or days later one continues with parts of the callsign
that were missed at the first attempt."  One feature of this
scheme is that contact is never "lost".  The received picture
always keeps getting sharper, albeit very slowly when the signal
is weak.

The above could, in theory, be adapted to almost any signaling
format, CW, DFCW, Jason, etc.

73,

Stewart KK7KA