[BC] Mono Expands on the FM Band

[Robert Orban]

At 08:38 AM 1/18/2006, you wrote:
>From: Robert Meuser <Robertm at broadcast.net>
>Subject: Re: [BC] Mono Expands on the FM Band
>To: "Broadcasters' Mailing List" <broadcast at radiolists.net>
>Message-ID: <43CDF2AC.30409 at broadcast.net>
>Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>
>Dave
>
>You should discuss the vsb ssb FM stereo issue with Bob Orban as I know he 
>has
>run the numbers and they are not really favorable. As for the implementation
>part, it sounds like Motorola AM Stereo.
>
>R
>
>Hershberger, Dave wrote:
> >>  A old time engineer told me that the FCC made a major
> >>mistake in choosing
> >>the GE/Zenith FM stereo system over the Murray Crosby system.
> >>  The Murray
> >>Crosby system worked better and had much less hiss at low
> >>signal levels but
> >>it was opposed by broadcasters because it interfered with
> >>67-Khz SCA music
> >>services which was a major revenue source for FM broadcasters
> >>at the time.
> >>A short-sighted decision I think in retrospect.
> >>
> >>-Nat
> >>

I computed the composite waveform of a single-channel-only FM stereo 
modulation that assumed single sideband operation of the subchannel. It 
turns out that you lose the left/right interleaving property that the 
current DSB system has (the composite is not longer equivalent to 
alternately sampling the L and R channels at 38 kHz), and the peak 
modulation increases if you hold the level of one channel constant while 
decreasing the level of the other channel. To constrain the composite peak 
modulation to a maximum value, this would require turning down the peak 
levels of the L and R channels and would reduce the S/N advantage obtained 
by not using the 38-53 kHz region. I didn't compute whether S/N is gained 
or lost when both of these effects are taken into account.

Another practical problem is that the Hilbert transform filter needed to do 
this in a maximally phase-linear way on the TX end would end up adding 
considerable delay to the transmission, assuming that the HT pair consists 
of an FIR filter and a delay line. The length of the HT filter required to 
cover 30-15,000 Hz with acceptably low error is thousands of taps, and if 
you used FFT techniques to compute it, this would reduce the computational 
load at the expense of even more throughput delay. This would make off-air 
monitoring impractical through talent headphones. At the receiver, you 
could use a nonlinear phase HT pair to reduce cost, but this would build 
phase distortion into the system. In a digiceiver, you could prevent phase 
distortion at the cost of burning considerable DSP power, the minimization 
of which would add yet more throughput delay to the system.

One of the big advantages of the current system is that the DSB suppressed 
carrier modulator requires no filtering at all at the TX and only modest 
filtering at the RX to remove high frequency images. This makes the system 
capable of very low throughput delay.

Bob Orban