On Feb 11, 2009, at 11:53 AM, Ian White GM3SEK wrote:
> he only ones that seem to make sense are
> mid-way between the two tones at 2125 + 85 = 2210Hz, so I'm presuming
> that with 1275Hz mark all the notches should be positioned at 1360Hz?
I'm not so sure that ad-hoc notching of the center of the mark and
space tones helps in general -- depending on the Q of the notch, it
could actually hurt.
If the notch is too wide, it will start reducing the important keying
sidebands of the mark and space signals themselves. And that will
hurt the ultimate SNR of the demodulated signal.
The proper way to demodulate RTTY is not through using an arbitrary
notch but to design a proper Matched Filter for the Mark and Space
tones to start with. For relatively wide shifts like 170 Hz shift
RTTY, the match filters usually depresses the response in the region
between Mark and Space tones -- but not in an arbitrary manner. The
response between the mark and space tones should maximize the SNR.
See the explanation here:
http://en.wikipedia.org/wiki/Matched_filter
I.e., for an AWGN channel, the impulse response of the matched filter
is the time reversal of the received pulse -- this is so that
mathematically, the filter's convolution is just a simple cross-
correlation and therefore maximizes the SNR. The received pulse need
not be a rectangle -- but will depend on the propagation conditions.
An arbitrary notch will almost for certain not give you the response
described above.
I believe RITTY was the first Amateur modem to use a Matched Filter
which adapts to a 45.45 baud pulse under different propagation
conditions.
The output of RTTY Matched Filters for AWGN channels typically have
responses that look like triangular waves, rather than sinusoidal
profiles. You can see this triangular shape alluded to in the figures
in the above wikipedia page, and an actual Matched Filter output can
be seen in Figure 1 here (about a quarter way down in the "RTTY
Monitor" section):
http://homepage.mac.com/chen/w7ay/cocoaModem/More/Contents/part2.html
The slightly rounded corners in Figure 1 are due to a band pass filter
that is placed around the RTTY signal -- you can tell from this figure
that when you use a filter that is too tight, when there is no QRM, is
detrimental to weak signal copy since the peaks of the triangle will
be more rounded and you therefore lose SNR since the noise is not
proportionally reduced.
73
Chen, W7AY
_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty
|