Mal
The answer is simple in many cases.
Why not?
I will explain.
You can use a narrower filter if you like, it would have to be 200Hz wide
to
allow the reception of all WSPR signals in the range of the software. CW
filters tend to be too narrow for that, SSB filters are standard on most
rigs, so it is easy to use this mode. It is also mandatory to use SSB mode
for transmitting, and on some rigs it is difficult to transmit on SSB and
have the CW filters in. A good CW 50Hz filter would probably work OK,
given
the very narrow bandwidth of the individual WSPR signals, with a small
number of WSPR transmissions on the band you could use it. It simply isn't
needed though, and it is desirable to allow the software to see the entire
200Hz segment, even if it is just to allow for drift or decoding other
signals.
For some modes the use of narrow conventional filters is bad. They
intrinsically have poor group delay response and cause distortions to the
signal not often directly audible, but that can degrade data is a way that
reduces decoding margins. This effect is worst when the wanted signal is
relatively wideband (occupying the whole bandwidth of the filter) or near
the edge of the passband. DRM, DAB and DTT signals that are OFDM based are
vulnerable.
All the WSPR operation that the MF band would need, could be squeezed in a
segment of the band less than 200Hz wide, probably 100Hz would be more
than
ever needed. WSPR also time division multiplexes, so more than one
station
can occupy the same frequency. No one is proposing WSPR signals all over
the
band, that is self defeating as the software cannot analyse more than a
200Hz chunk of bandwidth.
The key point is that just because the receivers have a wider passband
than
actually needed does not usually render the mode ineffective in the
presence
of signals that are outside the decode passband.
Like any system, reducing the bandwidth of any part of the system, means
that subsequent stages are less affected by adjacent signals.
The antenna itself acts as a filter because it is tuned, your low pass
filter removes many big broadcast signals, but no one is advocating a
tuneable 50Hz wide bandpass filter implemented at 500KHz, or anything
approaching this, for CW or any other mode. It is simply not necessary in
most cases, and a disadvantage in most cases.
Almost all modern conventional receivers have roofing filters that are
quite
wide in early stages, followed by narrow mode specific filters. Just
because
the filter before is wider does not unduly degrade performance of the
receiver after the more narrowband filtering.
Interestingly, my grabber receiver has only very simple roofing filters,
many hundreds of KHz wide and very low Q, they stop almost nothing 'out of
band'
All the rest of the filtering is done in software after an A to D
converter
(soundcard) which has a remarkably high dynamic range. There is no AGC, it
is not needed, and the software filters which operate on the same
principals
as the WSPR decoding system, allow me to listen to a weak signal with an
unwanted signal over 70dB higher, less than 100Hz away. I can actually
still
read the stronger stations on the band whilst transmitting low power on
the
main antenna, only a few tens of feet away, and a few hundred Hz away.
Ironically I get more problem when I transmit on 6m, though the RX antenna
is only about 2 feet away from the 6m beam.
I hope this helps your understanding of the principals behind WSPR and
other
FFT based systems. They are not so different. If that is not what you are
looking for then please clarify.
I have to say that the immunity of WSPR to interference is not as good as
I
would expected, if an unwanted signal comes up in the 6Hz bandwidth even
for
a short time, it can corrupt the decode. It is pretty good at ignoring
strong stable signals that are reasonably spaced away. last night and
this
morning a small number of decodes of Jim's beacon were lost due to my
'wandering sprog'. The presence of the big CW and QRSS signals in my
receive passband (pre WSPR processing) does not affect decode. BHZ and
DI2AM
are often 40dB above noise, and they are only a few hundred Hz away from
Jims WSPR signal, and presented to the WSPR decoder unattenuated by any
filtering.
I find the levels required to decode WSPR to screen about the same as my
ability to read QRSS comfortably, and significantly better than my ability
to read weak CW.
as an EME operator, I can read weak CW quite well. The slow QSB on 600m
is
a problem for both WSPR and QRSS. and indeed there may be instances where
a
quick burst of normal CW would be better than the slow modes. WSPRs big
advantage is that it automatically gathers propagation data. The ability
to
observe and accurately quantify the varying signal strength over time is
extremely useful.
Mark GM4ISM
