On Sat, 14 Feb 2004 16:39:52 -0000, "Richard"
wrote:

Do we
have technology these days to do what we thought was impossible in this
case?

While the sidebands above the wanted nominal carrier are mixed with
the lower sidebands of the unwanted signal, but if the lower sidebands
of the wanted signal and the upper sidebands of the unwanted signal
are free from any interference, it might be possible to determine
which sideband belongs to which signal and then demodulate them
separately.

This would at least require that all IF sidebands of both the wanted
and unwanted sidebands are digitised (prior to the limiter) and then
do some frequency domain processing after FFT.

When looking at an ordinary FM signal in frequency domain, there are
an infinite number of sidebands on each side of the nominal carrier
frequency (while in AM there is exactly one pair of sidebands) at
integer multiples of the baseband (audio) frequency. However, the
amplitude for each sideband is determined by the Bessel function of
the modulation index of each baseband signal. While the modulation
index for most audio tones is in practice quite low (below 5), there
are only going to be a few sidebands with any significant amplitude
around the nominal carrier. Thus, the bandwidth of an FM signal is not
infinite in practice.

For instance, in a stereo transmission, there are quite a lot of
signal amplitude around the baseband 38 kHz subcarrier frequency in
the area 36-40 kHz, thus there are going to be strong FM sidebands
around 36-40, 76-80 and possible 118-120 kHz around the carrier on
both sides.

Since there is some symmetry between the upper and lower sidebands, so
this may help reconstruct the missing sidebands if the other side of
the carrier is clean, while the other side is contaminated.

However, I have no idea if this has been attempted in practice.

Paul OH3LWR