How bout if we take a 7.2 MHz local signal modulated with SSB, then divide
it by 10. Now if we demodulate the 720 kHz signal and use the resultant
audio to modulate a new transmitted signal, we would be able to compress our
audio into one tenth the bandwidth -- or could we? We might need to sample
the amplitude of our 7.2 MHz modulating signal and apply that information to
the audio that comes from our divide-by circuit, but what the heck. . .
At the receiving end we just produce the 10th harmonic and demodulate it to
get back where we started.
----- Original Message -----
From: "Dr. Gerald N. Johnson, electrical engineer" <firstname.lastname@example.org>
Sent: Tuesday, April 22, 2003 12:38 PM
Subject: Re: [TenTec] Slightly OT: SSB vs AM
> The carrier and other sideband of AM make for simple detection. E.g. the
> crystal set, just a diode. The carrier does nothing but bias the diode
> detector, the other sideband is just the result of the amplitude
> modulation and has redundant information. When received with a suitably
> locked local BFO and synchronous detection, the carrier shows precisely
> where the local oscillator should be in phase and the simple synchronous
> detector will add information from both sidebands to the output, so long
> as propagation has not changed the phase of the carrier and the
> sidebands. Propagation can destroy the phase consistency necessary for a
> synchronous detector's great reception and much more so for the simple
> diode detector's reception. Then SSB reception works better. Back in the
> 50's John B. Costas proposed that DSB sans carrier gave enough
> information for the phase locked receiver to recreate the local
> oscillator phase with perfect accuracy, at the cost of twice the
> spectral bandwidth. Many times in the past, communications quality has
> been sacrificed to enable narrower spectrum and thus more users.
> The data rate in voice communication is very low. How about close to the
> syllabic rate. For sure much lower than the maximum transmitted
> modulation frequency.
> The voice spectrum has two major groups, one at a few hundred Hz and
> another out maybe three times that frequency with a gap in between.
> There was a proposed scheme in the ARRL handbook a few years back to
> separate those with separate bandpass filters, then heterodyne the upper
> one into the gap at the transmitter, then shift it back to its normal
> range in the receiver. Saved at least a third of the spectrum, maybe
> more, but it went over like a lead balloon.
> For really great bandwidth reduction one might detect phonemes, signal
> their sequence and timing with PSk-31 style signals, then use them to
> run a speech synthesizer in the receiver. Then the RF part would be
> running at Nyquist rate. The major cost would be naturalness. The
> simpler voice synthesizers are being used on NWS weather radio. Igor
> takes a bit of listening practice for good understanding. Then one might
> compress TV a great deal by sending a bit of digital data to establish
> character positions and attitudes, then let the user's TV set do
> animation based on those voice phonemes. Perhaps "live" TV in 100 Hz
> bandwidth... Looses a bit of quality, but talk TV wastes much spectrum
> just showing talking heads. The same information could be transmitted as
> plain audio in 1/1000 th the bandwidth using AM, less using SSB.
> 73, Jerry, K0CQ