Mono for Talk on FM

[hdsucks]

If you are running a station or format (such as talk) that has all mono programming... do yourself a huge favor - turn off the stereo -- you will have more listeners and a larger coverage area. And it sounds much better.

 

[rfry]

I would pick one nit here, Mr. Fry: the noise does not show up in the L+R signal, but rather, since the noise comes from the subcarrier region, appears out-of-phase in the individual left and right outputs. I'm pretty sure the noise would disappear if the two channels were summed.

However an analog stereo FM receiver in stereo mode does not sum the L & R channels. They remain discrete.

Therefore the noise from the L-R sub-channel region (whether or not there is any sub-channel modulation) appears in both the L & R channels, because the demodulated L-R and L+R waveforms are de-matrixed together to recover L & R.

It is known that the theoretical noise floor of analog FM stereo is worse than monaural FM -- by some 23 dB. This is the reason for the additional range that monaural-only FM transmission achieves for a given S/N, as long as no 19 kHz pseudo-pilot is transmitted to switch stereo receivers into their stereo mode.

//

 

[rrsounds]

I would pick one nit here, Mr. Fry: the noise does not show up in the L+R signal, but rather, since the noise comes from the subcarrier region, appears out-of-phase in the individual left and right outputs. I'm pretty sure the noise would disappear if the two channels were summed.

However an analog stereo FM receiver in stereo mode does not sum the L & R channels. They remain discrete.

Therefore the noise from the L-R sub-channel region (whether or not there is any sub-channel modulation) appears in both the L & R channels, because the demodulated L-R and L+R waveforms are de-matrixed together to recover L & R.

It is known that the theoretical noise floor of analog FM stereo is worse than monaural FM -- by some 23 dB. This is the reason for the additional range that monaural-only FM transmission achieves for a given S/N, as long as no 19 kHz pseudo-pilot is transmitted to switch stereo receivers into their stereo mode.

//

You are basically restating what I said. I only wanted to clarify that the noise is not added to the L+R (as you said in a previous post), but rather is matrixed to it. The noise is out-of-phase in the individual Left and Right channels, which tends to make it stand out from the center-channel (mono) material.

To some readers it may sound like a semantic distinction, but I think it's important.

Kind Regards,
David

 

[dspxscott]

So, a matrixiser on left and right audio, with multiband noise gating on L-R and de-matrixsing would be a very effective way of improving upon stereo noise??!!?

 

[boiseengineer]

Already tried in the 70's.
Wikipeda "Dolby FM used a modified 25 µs pre-emphasis time constant and a frequency selective companding arrangement to reduce noise. "

 

[rfry]

I only wanted to clarify that the noise is not added to the L+R (as you said in a previous post), but rather is matrixed to it.

Considering the difference in the S/N heard by a listener to an analog stereo FM receiver, more noise is present for a monaural transmission that includes a 19 kHz "pseudo-pilot" signal than if that 19 kHz signal is not present. The reason is that de-matrixing the output of the L-R sub-channel demodulator with the L+R waveform that is transmitted adds audible noise to both the L & R channels.

When monaural (L+R, only) analog FM is transmitted without a 19 kHz pilot, then that L+R waveform is heard with equal volume, and with less additional noise than in the case above, from both the L & R speakers of a stereo receiver.

Probably the reason that 73.322(a) does not permit the transmission of 19 kHz modulation except when operating in stereo is to prevent stereo receivers from falsely switching to/from stereo mode based on program modulation.

Of course 73.322(a) also precludes adding a 19 kHz pseudo pilot to the program audio of an FM station not equipped for stereo transmission.

//

 

[cgould]

Maybe I can fill in the missing link between David and Mr. Fry.

What both of you are saying is true, but each of you are talking on two slightly different tracks...

When a radio receiver is decoding stereo programming (or mono programming on an FM signal with FM stereo pilot), whatever noise is present
in the 23-53 kHz subcarrier region of the transmission signal is folded into the decoded Left and Right audio channels on home receivers.

This noise when re-assembled by the DSBSC Demodulator (or switching demodulator as it is done these days) appears out of phase.

The noise is indeed on the left and right channels but the polarity of the noise is 180 degrees out of phase between the left and right channels. If the END USER were to sum the left and right channels together on their radio set, the noise would cancel out, and leave clean audio.

-C

DBSC = Double Sideband Suppressed Carrier

 

[rrsounds]

Maybe I can fill in the missing link between David and Mr. Fry.

What both of you are saying is true, but each of you are talking on two slightly different tracks...

When a radio receiver is decoding stereo programming (or mono programming on an FM signal with FM stereo pilot), whatever noise is present
in the 23-53 kHz subcarrier region of the transmission signal is folded into the decoded Left and Right audio channels on home receivers.

This noise when re-assembled by the DSBSC Demodulator (or switching demodulator as it is done these days) appears out of phase.

The noise is indeed on the left and right channels but the polarity of the noise is 180 degrees out of phase between the left and right channels. If the END USER were to sum the left and right channels together on their radio set, the noise would cancel out, and leave clean audio.

-C

DBSC = Double Sideband Suppressed Carrier

Isn't that what I said?

Kind Regards,
David

 

[rrsounds]

So, a matrixiser on left and right audio, with multiband noise gating on L-R and de-matrixsing would be a very effective way of improving upon stereo noise??!!?

But as this is a receiver-side issue, it is a feature broadcasters themselves have no control over.

Since, IIRC, BW Broadcast is now making receivers, here's your opportunity....

Kind Regards,
David

 

[gtomas]

Can somebody explain why is the noise from the 23-53 kHz spectrum out of phase in L and R audio after de-modulation?

Moreover, since there is redundancy in the DSB-SC modulation (and assuming this would be done in DSP), why not take only lower SB which would be less contaminated with noise than upper SB?


Regards,
Goran Tomas

 

[rfry]

The noise is indeed on the left and right channels but the polarity of the noise is 180 degrees out of phase between the left and right channels. If the END USER were to sum the left and right channels together on their radio set, the noise would cancel out, and leave clean audio.

This was posted earlier by David Reaves, too.

But relatively few analog FM stereo receivers provide a control for the user to sum the L&R outputs so as to reduce the noise on monaural transmissions including a pseudo pilot signal.

So reception of such monaural transmissions on most FM receivers is noisier than it needs to be.

This point seems to be getting missed.

//

 

[cgould]

Richard, and David --

Yes, I did rephrase what David posted. I'm looking at the "picking of nits" from an engineering newbie point of view, and thought I would amplify the points being made. There's a lot of folks who use this list to learn more about the radio engineering side. Doing this helps them to appreciate the nuances of the topics being discussed on the board.



You are absolutely correct, Mr. Fry when you mention how so few radios have a "mono" provision anymore. It is certainly best to handle this on the transmit side of the equation, and just transmit mono (no pilot) if you are doing FM talk.

Goran:

The noise being out of phase has to do with the way the decoding of FM stereo happens in the decoder matrix on the receiver side.

As you no doubt know, what is on the 23-53 kHz is the L-R (or difference between left and right channel) information. Another thing that we know is the higher up in the frequency spectrum you go in FM, the noisier it tends to be. So, when looking to this region for the information, you also get a lot of noise along with the L-R audio.

In the reconstruction of the stereo image, you take the L-R signal, and apply it to the mono L+R signal, which results in L channel only

(L+R) + (L-R) = 2L

You take the L-R channel, flip the polarity and do the same thing as before... R-L and apply it to the L+R signal, which provides right channel only

(R+L) + (R-L) = 2R

(I re-arranged the letters to make it clearer to folks who've always wondered how this process works)

In the process of flipping the polarity of the L-R to get the two audio channels, we've also created a condition where the noise is flipped in polarity with respect to the left and right channels. Just a by-product of the matrix /de-matrix system used for FM stereo. FM stereo is stereo audio with out of phase noise (usually not noticeable unless the signal is pretty weak, or if there is no stereo information to mask the noise).

Goran: Yes you can use only one of the sidebands to decode stereo. Not sure how much it saves you in terms of noise, however.

-C

 

[RealityCheckr]

Many years ago I had a Harmon Kardon tube integrated amp. I think the model was A500. It had a number of innovative features including a blend control. This was a really cool concept. You could adjust it to the point where the noise on a particular was low enough in amplitude to sound good, while still having some stereo effect.

It also had separate Bass and Treble for the left and right channel (though no midrange).

Overall a very nice amp, though it was rather sensitive to voltage changes and in the 1920's apartment the voltage drop in summer could cause it to barely put out anything.

I ended up giving to my neighbor when I moved. I have always wanted to find another amp with this feature.

 

[rrsounds]

Many years ago I had a Harmon Kardon tube integrated amp. I think the model was A500. It had a number of innovative features including a blend control. This was a really cool concept. You could adjust it to the point where the noise on a particular was low enough in amplitude to sound good, while still having some stereo effect.

It also had separate Bass and Treble for the left and right channel (though no midrange).

Overall a very nice amp, though it was rather sensitive to voltage changes and in the 1920's apartment the voltage drop in summer could cause it to barely put out anything.

I ended up giving to my neighbor when I moved. I have always wanted to find another amp with this feature.

Interesting how features like this were available on consumer products at one time. I once had an Apt-Holman preamp that had a continuous blend control, too. All the way from L+R , through stereo, to L-R. Very revealing!! Too bad many of the units (mine included) had serious reliability problems a few years down the road. :-(

It was the nexus for my home stereo when I was CE at Z100, and it got a lot of use!

Kind Regards,
David

 

[rrsounds]

<snip>since there is redundancy in the DSB-SC modulation (and assuming this would be done in DSP), why not take only lower SB which would be less contaminated with noise than upper SB?
Regards,
Goran Tomas

Goran, if you band-limit the subcarrier to use just one sideband, your recovered amplitude will be cut in half (6 dB). But the noise will only be cut by 3 dB (because it is not identical/coherent in both sidebands). So actually will you have reduced the signal to noise, not improved it.

Kind Regards,
David

 

[gtomas]

(L+R) + (L-R) = 2L

You take the L-R channel, flip the polarity and do the same thing as before... R-L and apply it to the L+R signal, which provides right channel only

(R+L) + (R-L) = 2R

Of course, silly me... To make it perhaps even more clearer:

(L+R) + (L-R+n) = 2L+n
(L+R) - (L-R+n) = 2R-n

Goran, if you band-limit the subcarrier to use just one sideband, your recovered amplitude will be cut in half (6 dB). But the noise will only be cut by 3 dB (because it is not identical/coherent in both sidebands). So actually will you have reduced the signal to noise, not improved it.

Yes, it occurred to me later that you are losing signal level if taking only one side band. However, the noise in the upper side band would be higher than in lower side band, as it rises with frequency. But it probably won't be nearly as high enough to justify losing 6 dB of signal in taking only the lower side band...


Regards,
Goran Tomas

 

[fm-engineer]

I worked for a small corporation of stations in the mid to late 80's that required the pilot to be turned off during Paul Harvey news. 12:06 pm IIRC.

 

[rrsounds]

With DSP techniques, it should be possible to test for dissimilarities in the two sidebands. If the receiving system is linear, then anything that is not the same in both sidebands could be rejected as noise.

I can't think of a more effective way to reduce the stereo 'penalty'. And probably greatly reduce the effects of multipath interference, too.

Kind Regards,
David

 

[dspxscott]

With DSP techniques, ......
Kind Regards,
David

The new BW broadcast tuners will include a range of DSP techniques to significantly reduce the noise of both stereo and mono reception. Watch this space!