Even FFT-based tools like ARGO and
JASON, amazing as they are
in their present form, could be better optimized if the receiving
computer knew
exactly when each bit was due to start.
Suggestion : "synchronized" JASON-like mode at 30 seconds
dotlenght
- Data rate would be 1 character per second. Sounds slow, but taking
into account that we mainly transmit callsigns in QRSS the average
character takes about 10 dots so this mode would be as fast as QRSS at 6
sec/dot (or half the data-rate of the traditional QRSS3)
- If any character would always start exactly at the full minute (lower
'nibble' first 30 sec., higher 'nibble' last 30 sec.) and we stick to the
character-set of JASON (64 characters) only 6 bits are needed (vs. 8 bits
for JASON). Thus only 3 bits per 'nibble' are needed or only 9 'channels'
(vs. 17 for JASON).
- Some timing ('synchronisation') would be required, but an error up to
0.1 seconds can be easily tolerated (see below)
- At 30 sec/dot the channels can be separated as little as 0.1Hz apart,
so the whole signal would fit in a 1Hz window (vs. 4.3Hz for Jason). This
small bandwidth would make it easier to keep away from LORAN lines.
- A typical QRSS-like QSO would take only 1 hour
- If LORAN lines (or other carriers) within the 1Hz bandwidth can be
avoided SNR would be almost as good as QRSS30, but at a 5 times faster
data-rate.
- Or otherwise said : it would allow a transatlantic QSO within 1 hour,
using only moderate power levels. Recent test have shown that 100mW ERP
signal can cross the pond regulary in QRSS30, when condx are right it
might work with even far less ERP.
At a samplerate of 4410Hz (22050Hz / 5) and an FFT-size of 131072
points (2exp17) would give a convenient FFT-duration of 29.72 seconds, so
if the FFT starts at 0.14 seconds after each full 30 seconds it will end
0.14 seconds before the end of the 30 seconds period, allowing some
flexibility in the timing.
73, Rik ON7YD
|
