# Re: Roberts Gemini 49

• From: "DAB sounds worse than FM" <dab.is@xxxxxxxxxx>
• Date: Mon, 24 Aug 2009 01:06:49 +0100

"jamie" <jamie_p84@xxxxxxxxxx> wrote in message
news:54ce4f47-2a69-4a03-a28b-51eef9759bf2@xxxxxxxxxxxxxxxxxxxxxxxxxxx
On Aug 23, 11:15 pm, Peter Watson <pe...@xxxxxxxxxxx> wrote:

Algebraically, Jamie is correct - The final equation should read

Right =0.5 * [(L + R + n1) - (L - R + n2)] = R + (n1 - n2)/2

You seem to be going to extreme lengths to prove that the sign is
unimportant, ie if (and only if) the following is true:

<Quote>

"The n1 and n2 noise signals are random processes (i.e. a time
signal
of a random variable), and random processes have to be described by
their probability distributions / statistics. In this case, n1 and
n2
are Gaussian random, zero-mean signals with the same variance
(variance = power). And if you add two such signals together the
resultant signal will have exactly the same statistics as if you
subtract the signals from one another. Therefore, it doesn't matter
at
all whether I put a '+' or a '-' in front of n2 in the 2nd
equation."

</Quote>

I freely admit that I don't know enough to know if this is true
*but*
common sense says it is not. Is it likely that the noise *in the
real
world* on the Mono FM component and the noise Stereo DSBSC
component has
the same spectrum, amplitude and variance?

*However* all this is irrelevant as you quoted the equations in
a post about combining L and R to make Mono (despite your
protestations
to the contrary)

<Quote>

">> You're assuming the hissing from the L & R channels will be
exact
opposites in
phase and cancel each other out when recombined.

They are and they do.

The signals that are used to reconstruct the Left and Right stereo
channels are as follows:

L + R and L - R

But these signals have noise added by the RF front-end, so in
reality
the signals used to reconstruct the Left and Right channels are:

L + R + n1
L - R + n2

where n1 and n2 are different noise signals (the bandwidth is
different, so the noise is obviously different as well) added to
the
sum and difference signals respectively.

The Left and Right stereo channels are reconstructed as follows:

Left = 0.5 * [(L + R + n1) + (L - R + n2)] = L + (n1 + n2)/2

Right =0.5 * [(L + R + n1) - (L - R + n2)] = R + (n1 + n2)/2

The noise on the Left and Right channels is therefore (n1 + n2)/2.

In comparison, the noise added to the mono L + R signal is just n1
(see above).

n1 != (n1 + n2)/2

QED."

</Quote>

So, *in the context* of the discussion Jamie was having before you
tried
to change the subject, using your expansion:

Mono = Left + Right

Mono = L + (n1 + n2)/2 + R + (n1 + n2)/2 = L + R + n1 + n2

This is *wrong*, irrespective of the spectrum or characteristics of
n2,
as actually:

Mono = L + (n1 + n2)/2 + R + (n1 - n2)/2 = L + R + n1

ie Summing the outputs of a stereo FM receiver is the same as just
demodulating the Mono signal (ie switching the receiver to 'Mono')

So, returning to Jamie's original point, just why did you say:

<Quote>

"Actually, although I previously said that Roberts radios should be
a
*single speaker* but for FM they were set by default to receive in
stereo mode, so they required higher SNR to receive a hiss-free
PR
company, the cheeky bugger who replied tried to suggest that this
was
by design. Er, incompetent design then, pal."

</Quote>

A voice of knowledge and sanity at last, besides my own!
Thanks Peter :-)

Awwwwwwwwwwwwwwwwwwwwwwwwwwwwwwww, bless.

--
Steve - www.savefm.org - stop the BBC bullies switching off FM

"It is the sheer volume of online audio content available via
internet-connected devices which terrifies the UK radio industry. I
come, offering very local, unregulated content, as well as opening a
window to the radio stations of the world." - from the Myers Report

.

## Relevant Pages

• Re: Roberts Gemini 49
... zero-mean random noise signals. ... world* on the Mono FM component and the noise Stereo DSBSC component ...