Maker Pro
Maker Pro

FM Stereo (the truth?)

A

Anthony Fremont

Jan 1, 1970
0
Can anyone verify that stuff that phil stated? For those that haven't
seen it, he claims that rapidly switching the baseband output from an FM
discriminator rapidly (38kHz) between the L and R channels (speakers,
preamp in, or whatever floats your boat) will separate the L+R signal
into its L and R components. Phil provided one link making that claim,
but it is severely lacking in any technical explanation of that type of
decoding. I've searched and searched and I can't find anyone else
claiming the same thing. Seriously, this is no troll, I've never heard
of this before. Is it really true? Can anyone elaborate on how/why
that would work? Wouldn't Nyquist (sp?) limits apply here?
 
A

Anthony Fremont

Jan 1, 1970
0
Anthony Fremont said:
Can anyone verify that stuff that phil stated? For those that haven't
seen it, he claims that rapidly switching the baseband output from an FM
discriminator rapidly (38kHz) between the L and R channels (speakers,
preamp in, or whatever floats your boat) will separate the L+R signal
into its L and R components. Phil provided one link making that claim,
but it is severely lacking in any technical explanation of that type of
decoding. I've searched and searched and I can't find anyone else
claiming the same thing. Seriously, this is no troll, I've never heard
of this before. Is it really true? Can anyone elaborate on how/why
that would work? Wouldn't Nyquist (sp?) limits apply here?

After doing some more digging, I've found that there is a (somewhat
cheesy IMO) method to GENERATE the baseband signal by "chopping" the L
and R channels at 38kHz. You then have to filter all the harmonic trash
above 53kHz that is created by this "alternative" process, and finally
insert a 19kHz pilot tone. You can then feed this to the equivalent of
a wideband (allowing freqs to 53kHz) mono FM transmitter to send out a
reasonable facsimile of a "stereo" signal.

I could find no such method being used to demodulate (detect,
discriminate or whatever you wish to call it) an FM stereo signal.
After what is required to create the baseband signal using the method
above, I am not sure that you can run this procedure in reverse and
achieve any significant channel separation. Can anyone elaborate on
this?
 
J

Jasen Betts

Jan 1, 1970
0
Can anyone verify that stuff that phil stated? For those that haven't
seen it, he claims that rapidly switching the baseband output from an FM
discriminator rapidly (38kHz) between the L and R channels (speakers,
preamp in, or whatever floats your boat) will separate the L+R signal
into its L and R components. Phil provided one link making that claim,
but it is severely lacking in any technical explanation of that type of
decoding. I've searched and searched and I can't find anyone else
claiming the same thing. Seriously, this is no troll, I've never heard
of this before. Is it really true? Can anyone elaborate on how/why
that would work? Wouldn't Nyquist (sp?) limits apply here?

consider the L-R carrier it's an AM signal

when it is at it's peak the total signal is (L+R)+(L-R) = 2L

but when it is in the trough the total is (L+R)-(L-R) = 2R

simple eh?

Nyquist limits apply, yes, that's why FM quality is what it is.

Bye.
Jasen
 
Anthony said:
After doing some more digging, I've found that there is a (somewhat
cheesy IMO) method to GENERATE the baseband signal by "chopping" the L
and R channels at 38kHz. You then have to filter all the harmonic trash
above 53kHz that is created by this "alternative" process, and finally
insert a 19kHz pilot tone. You can then feed this to the equivalent of
a wideband (allowing freqs to 53kHz) mono FM transmitter to send out a
reasonable facsimile of a "stereo" signal.

I could find no such method being used to demodulate (detect,
discriminate or whatever you wish to call it) an FM stereo signal.
After what is required to create the baseband signal using the method
above, I am not sure that you can run this procedure in reverse and
achieve any significant channel separation. Can anyone elaborate on
this?

Yes, it it a "reversible" process. Good separation depends on good
phase lock of the demodulating switching signal to the 19kHz pilot.

-michael

Parallel computing for Apple II's
http://members.aol.com/mjmahon
 
B

Bob Eldred

Jan 1, 1970
0
Anthony Fremont said:
Can anyone verify that stuff that phil stated? For those that haven't
seen it, he claims that rapidly switching the baseband output from an FM
discriminator rapidly (38kHz) between the L and R channels (speakers,
preamp in, or whatever floats your boat) will separate the L+R signal
into its L and R components. Phil provided one link making that claim,
but it is severely lacking in any technical explanation of that type of
decoding. I've searched and searched and I can't find anyone else
claiming the same thing. Seriously, this is no troll, I've never heard
of this before. Is it really true? Can anyone elaborate on how/why
that would work? Wouldn't Nyquist (sp?) limits apply here?

I think it is true. Is not the base band signal L + R + (L - R)sinwt where w
is 2*Pi*38KHz? Now, when the sine is positive you have to a first
approximation L + R + L - R = 2L and when the sine is negative you have L +
R - L + R = 2R So if you switch phase coherently at 38KHz you should get 2L
and 2R separated assuming the relative amplitudes are correct. The fact the
the sine is not a square wave for switching but only the first harmonic of
the square wave produces an ampitude term that affects the depth of
separation. For this to work you have to have the full base band including
the 38KHz sub carrier side bands. Obviously it won't work on the L + R audio
signal alone.
Bob
 
A

Anthony Fremont

Jan 1, 1970
0
Bob Eldred said:
:

I think it is true. Is not the base band signal L + R + (L - R)sinwt where w
is 2*Pi*38KHz? Now, when the sine is positive you have to a first

Now see, I don't know. I guess that's the problem with being a
hobbiest. ;-)
approximation L + R + L - R = 2L and when the sine is negative you have L +
R - L + R = 2R So if you switch phase coherently at 38KHz you should get 2L
and 2R separated assuming the relative amplitudes are correct. The fact the
the sine is not a square wave for switching but only the first harmonic of
the square wave produces an ampitude term that affects the depth of
separation. For this to work you have to have the full base band including
the 38KHz sub carrier side bands. Obviously it won't work on the L + R audio
signal alone.

So the 19kHz pilot tone doesn't need to be removed before the "chopping"
begins? I have a decent grasp of heterodyning signals and the results,
but this 38kHz chopping and its consequences are very new to me. So
what you are saying is that the receiver can employ the same chopping
technique (as long as it is phase locked to the pilot tone) and the L-R
sidebands centered around 38kHz will demodulate and restore the
separation? Man that's just too weird. ;-) If that's the case, I
guess I owe philth an apology. I still haven't found any web sites that
demonstrate this technique being used within a receiver. I guess the
key piece that leads me to believe this may be true (that it's a
reversable process) is that I've seen several references to the
importance of the phase of the pilot tone. In SSB reception, the phase
of the injected carrier is not important for recovering the information,
so there must be some reason for its importance in FM stereo reception.
Being a ham, I never got much chance to experiment with FM stereo. :)
 
B

Bob Masta

Jan 1, 1970
0
On Sun, 15 Jan 2006 16:30:53 GMT, "Anthony Fremont"

So the 19kHz pilot tone doesn't need to be removed before the "chopping"
begins? I have a decent grasp of heterodyning signals and the results,
but this 38kHz chopping and its consequences are very new to me. So
what you are saying is that the receiver can employ the same chopping
technique (as long as it is phase locked to the pilot tone) and the L-R
sidebands centered around 38kHz will demodulate and restore the
separation?

I don't know the details of broadcast FM, but you can consider a
chopper to be multiplication by a square wave, as opposed to
the more typical multiplication by a sine wave we think of in
heterodyning. So you would get the desired spectrum, plus
images at 3x, 5x, etc with decreasing amplitude. The trick
would be removing the unwanted images, or using some
clever arrangement to move them out of the desired band.

As a historical note, a laboratory instrument called a "lock-in
amplifier" used choppers for synchronous detection of
extremely faint signals. A reference signal provided the
chopper drive, and the output of the chopper was then
simply low-pass filtered to give a DC value proportional
to the amount of reference-frequency component in the
input signal. They typically had 2 quadrature channels,
since one channel can't resolve a quadrature input, and
they were known to give false responses to harmonics.
But a lot of real-world lab work didn't have problems
with that. The typical application I recall was with a
light stimulus chopped by a slotted wheel to provide
pulsed stimulus to the experiment, and the lock-in
used to determine the response (be it photodetector,
physiological, etc.) The main virtue of lock-ins was
that they could resolve signals buried in copious
amount of noise. The same process is now done
digitally with synchronous averaging, where the
entire response waveform can be recovered.

Best regards,





Bob Masta
dqatechATdaqartaDOTcom

D A Q A R T A
Data AcQuisition And Real-Time Analysis
www.daqarta.com
Home of DaqGen, the FREEWARE signal generator
 
R

Rich Grise

Jan 1, 1970
0
After doing some more digging, I've found that there is a (somewhat
cheesy IMO) method to GENERATE the baseband signal by "chopping" the L
and R channels at 38kHz. You then have to filter all the harmonic trash
above 53kHz that is created by this "alternative" process, and finally
insert a 19kHz pilot tone. You can then feed this to the equivalent of
a wideband (allowing freqs to 53kHz) mono FM transmitter to send out a
reasonable facsimile of a "stereo" signal.

I could find no such method being used to demodulate (detect,
discriminate or whatever you wish to call it) an FM stereo signal.
After what is required to create the baseband signal using the method
above, I am not sure that you can run this procedure in reverse and
achieve any significant channel separation. Can anyone elaborate on
this?


You say you've done ham radio. Look at it this way. You've already
demodulated the FM, so you've got your L+R on the "baseband", the
19KHz "pilot tone", and L-R DSBSC modulated on a 38KHz subcarrier.
(double-sideband, suppressed carrier).

Well, imagine you've reinjected the 38KHz subcarrier in this DSB
signal, and you've got straight AM, just like an AM radio signal,
but at 38 KHZ.

Now look at the instantaneous value of that. 38,000 times a second,
it goes negative or positive, by more or less, which is the amplitude
of the L-R signal.

But that's if you filter it out from the baseband L+R and are looking
at it all by itself. You don't do that - you chop just what's come
out of the FM discriminator, baseband, modulated subcarrier and
all, (you might want to suppress the 19KHz subcarrier), and if you
did a time-domain display and took the amplitude at each excursion
of the subcarrier, you'd find that their sum (which the demodulated
FM is, after all) is 2R on one half-cycle and 2L on the other.

Hope This Helps!
Rich
 
F

Fred Abse

Jan 1, 1970
0
In SSB reception, the phase
of the injected carrier is not important for recovering the information,
so there must be some reason for its importance in FM stereo reception.

Reinserted carrier phase doesn't matter in the case of SSB, but it does in
the case of DSB, which is what the Zenith-GE system uses to modulate the
(L-R) component. It's probably 30 years since I had to do the math. I
shall have to go dig it out, if I can still find it.

I've spent a little time during a very busy couple of weeks trying to find
definitive references to Zenith-GE, and come up with not much beyond
superficial descriptions. Anybody know if it was ever patented? Searching
classification 179 produces quite a few "interesting" ideas ;-). The
nearest I can find is the Crosby patent #2851532, but that's not the whole
nine yards, using as it does, a frequency-modulated subcarrier.
 
Top