On Aug 30, 2010, at 8/30 4:39 AM, Bill, W6WRT wrote:
> I was (and am) most interested in decoding signals right at the noise
> level and partly covered with QRM and QRN, as opposed to stronger
> signals which were distorted by multipath.
To decode weak signals under good band conditions, a good matched
filter should help when there is no QRM within 500 Hz or so of both
Mark and Space tones. Open up any filter that is ahead of the matched
filter up, let the matched filter do the work instead of the I.F.
filter in the rig. Do not use any spectral shaping (such as "dual
passband" filters) -- those filters and narrow I.F. filters will ruin
the shape of the matched filter in the modem.
Brian, K6STI used to emphasize the use of matched filtering in RITTY.
More recently, Rob, KL7NA and one of his students have also done a
study which shows that matched filtering helps with RTTY. You won't
notice its presence except with very, very weak signals in a quiet
band (what Alex, VE3NEA, had plotted in his "RTTY Compare" plots as
the AWGN condition).
When selective fading is absent, turn off the ATC -- since ATC will
add a little noise component while not helping anything when the ideal
slicer threshold is a constant anyway. You can visually see the
presence of selective fading on a crossed ellipse indicator -- the two
bananas do not fade synchronously when there is selective fades.
When selective fading is present, either turn on the ATC or select
what both the ST-8000 and the KAM Plus call the "FM" mode. The FM
mode basically precedes the demodulator by a hard limiter. When there
is some selective fading, the limiter equalizes the mark and space
amplitudes; thus keeping the slicing level constant and therefore not
requiring an ATC.
When selective fading is very strong however, for example taking out
the space signal completely, there is nothing for a limiter to
equalize :-). That is where a good ATC circuit can be better since
the circuit will bias itself towards more of a Mark-only or Space-only
reception. ATC circuits are more complex to implement and much more
complex to get the parameters matched to the fading characteristics.
Selective fading is not a multipath effect, by the way. On HF, you
can have no multipath whatsoever and yet selective fades can be present.
(HF selective fading is the result of the channel being what is called
"Rayleigh." Rayleigh channels have Gaussian properties for the
signal's in-phase and quadrature components. The amplitude of these
two components has, through mathematics, a Rayleigh distribution. So,
just because there is no multipath does not mean there is no selective
fading.)
When a limiter is used, you do loose a couple of dB in sensitivity
compared to a linear detector. But you typically lose 10 dB due to
the selective fading itself, so a limiter helps instead of hurt.
Which brings an interesting segue... the presence of multipath also
does not mean there is selective fading present :-). Very often, when
you hear a signal from across the pole, turning off ATC (yes,
counterintuitive) will help instead of hurt. Again, the noise from
the ATC circuit is greater than what it contributes in the form of
moving the slider level automatically to track the mark and space
amplitudes.
73
Chen, W7AY
_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty
|