In article , Gary Schafer
writes:

Let's start at the other end and see what happens;

If we have a final amp with 1000 dc volts on the plate and we want to
plate modulate it to 100% or very near so, we need 1000 volts peak to
peak audio to do it.
On positive audio peaks the dc plate voltage and the positive peak
audio voltage will add together to provide 2000 volts plate voltage.

On negative audio peaks the negative audio voltage will subtract from
the dc plate voltage with a net of zero volts left on the plate at
that time. (or very nearly zero volts if we do not quite hit 100%)

How does the tube put out any power (carrier) at the time there is
near zero plate voltage on it?

The negative audio cycle portion is going to be much longer than many
rf cycles so the tank circuit is not going to maintain it on its own.

Why does the carrier stay full?

Gary, you are trying to mix the frequency domain and time domain
information...and then confusing steady-state conditions in the time
domain with repetitive conditions.

The "carrier amplitude is constant" holds true over repetitive audio
modulation. In conventional AM, with repetitive modulation from a
pure tone, there are three RF spectral products. If you deliberately
notch out the carrier component in a receiver, and then reinsert a
steady-state, synchronized carrier frequency component in its
place, you will recover the original modulation audio. The receiver
demodulator sees only a steady, constant-amplitude carrier
frequency component. There is absolutely no carrier amplitude
variation then. But the original modulation audio is demodulated
exactly as if it were the done with the original transmitted carrier.
SSB reception is done all the time that way (except the carrier
amplitude is so low it might as well be zero).

That's a practical test proving only that the carrier amplitude does
not have any change insofar as demodulation is concerned.

As a practical test of just the transmitter, let's consider your basic
old-style AM description...Class-C RF PA with linear plate volts v.
power output characteristic, modulation by the plate voltage. That
plate voltage is 1 KV steady-state. In steady-state, RF output has
a single RF component, the carrier frequency. One.

RF spectral component will follow the general time-domain RF
equations with no modulation. [easy math there]

Apply modulation to the plate voltage with a pure tone. Plate
voltage swings UP as well as DOWN equally. [theoretical perfect
linear situation] Same rate of UP and DOWN. [start thinking dv/dt]

Look at the spectral components with this pure tone modulation.
Now we have THREE, not just one. Any high resolution spectrum
analyzer sampling the RF output will provide practical proof of that.

So, if you want to examine the total RF in a time-domain situation,
you MUST examine it as amplitude versus an infinitely-thin slice of
TIME. You cannot take a finite time chunk out of the RF envelope
and "prove" anything...anymore than you can justify the existance of
three RF components, not just TWO. [if this were the real classroom,
you would have to prove that on the whiteboard and justify it in full
public view...and maybe have to show the class the spectrum
analyzer output]. Remember that the modulation signal also exists
in a time domain and is constantly changing.

If the "carrier sinewave goes to zero and thus power output is zero,"
how do you justify that, a half repetition time of the modulation signal
later, "carrier sinewave goes to twice amplitude and power output is
double"? You are trying an analogy that has a special condition, by
neglecting the RATE of the modulation. It is always changing just as
the carrier frequency sinewave is changing. You want to stop time
for the modulation to show repetitive RF carrier sinusoids and that is
NOT modulation. It is just adjustment of the RF output via plate
voltage. No modulation at all.

The basic equation of an AM RF amplitude holds for those infinitely-
small slices of TIME. The series expansion of that basic equation
will show the spectral components that exist in the frequency domain.
Nothing has been violated in the math and practical measurements
will prove the existance and nature of the spectral components.

For those that like the vector presentation of things, trying to look at
a longer-than-infinitely-small slice of time or just the negative or
positive modulation swings is the SAME as removal of the modulation
signal vector. Such wouldn't exist in that hypothetical situation. It
would be only the RF carrier vector rotating all by itself.

In basic FM or PM, there's NO change in RF envelope amplitude with
a perfect source of FM or PM. "The carrier swings from side to side
with modulation," right? Okay, then how come for why does the
carrier spectral frequency component go to ZERO with a certain
modulation/deviation level and STAY there as long as the modulation
is held at that level? RF envelope amplitude will remain constant.
Good old spectrum analyzer has practical proof of that. [common way
of precise calibration of modulation index with FM] The FM is "just
swinging frequency up and down" is much too simple an explanation,
excellent for quick-training technicians who have to keep ready-
built stuff running, not very good for those who have to use true basics
for design, very bad for those involved with unusual combinations of
modulation.

If you go back to your original situation and have this theoretical
power meter working with conventional AM, prove there are ANY
sidebands generated from the modulation of plant voltage...or one
or two or more. :-) Going to be a difficult task doing that, yet there
obviously ARE sidebands generated with conventional AM and each
set has the same information. Lose one and modulation continues.
Prove it solely from the time-domain modulation envelope. Prove
the carrier component amplitude varies or remains constant.

Hint: You will wind up doing as another Johnny Carson did way
back in 1922 (or thereabouts) when the basic modulation equations
were presented on paper. [John R. Carson, I'm not going to argue
the year, that's in good textbooks for the persnickety] With
conventional AM the CARRIER FREQUENCY COMPONENT
amplitude remains the same for any modulation percentage less
than 100. Period. I not gonna argue this anymore. :-)

Len Anderson
retired (from regular hours) electornic engineer person

Avery Fineman wrote:
So, if you want to examine the total RF in a time-domain situation,
you MUST examine it as amplitude versus an infinitely-thin slice of
TIME.

You might want to remind everyone that the mathematical Fourier transform of
a signal is an integral that extends from time=minus infinity to plus
infinity. Since Real Spectrum Analyzers (or network analyzer) need to
produce results in something, oh, less than infinite time (probably less
than the time between now and the next donut break), they're necessarily
limited in the low frequency detail they can provide. True, if Gary's
transmitter is transmitting a zero at the moment he connects a spectrum
analyzer, he won't see anything at all on the display, but as you point
out -- this is an equipment problem, not a mathematical one.

I'm still a believer in SSB-FM, BTW. :-) But I have enough respect for you
that I won't attempt to argue it further without first finding the time to
prepare a few drawings to demonsrate why!

---Joel Kolstad

In article , "Joel Kolstad"
writes:

Avery Fineman wrote:
So, if you want to examine the total RF in a time-domain situation,
you MUST examine it as amplitude versus an infinitely-thin slice of
TIME.

You might want to remind everyone that the mathematical Fourier transform of
a signal is an integral that extends from time=minus infinity to plus
infinity.

The series expansions of the basic modulation type RF time-domain
expression don't use Fourier series. The series expansions show
the sprectral content are still equivalent to that infinitely-thin slice
of time as the function of amplitude.

I think that Panters "Signals, Modulation, and Noise" text has it
worked out in there (by memory, don't have that one handy here).
The "Landee" text I mentioned is an old one and not that familiar
to most.

Since Real Spectrum Analyzers (or network analyzer) need to
produce results in something, oh, less than infinite time (probably less
than the time between now and the next donut break), they're necessarily
limited in the low frequency detail they can provide. True, if Gary's
transmitter is transmitting a zero at the moment he connects a spectrum
analyzer, he won't see anything at all on the display, but as you point
out -- this is an equipment problem, not a mathematical one.

No, it's an argument problem. :-)

There's no infinitely-fast RF power meter in existance. Yet.


I'm still a believer in SSB-FM, BTW. :-) But I have enough respect for you
that I won't attempt to argue it further without first finding the time to
prepare a few drawings to demonsrate why!

I'm not arguing that "single-sideband FM" won't work. I just don't like
the name. The technique DOES work from all the explanations of
experiments, is reproducible. [it isn't the "cold fusion" thing. :-) ]

I've not seen any convincing case that the single-whatever FM thingy
has any practical applications.

For narrowband voice, SSB AM is just dandy and a phasing system
using the Gingell-Yoshida polyphase network is quite easy and
error-tolerant to make a good phasing exciter. It can be used in
"reverse" to get an easy-to-select sideband demod or an ordinary
AM detector that yields false stereo (one sideband to each ear),
already done with simple CW receivers.

Len Anderson
retired (from regular hours) electronic engineer person

Hi Len,

Avery Fineman wrote:
For narrowband voice, SSB AM is just dandy and a phasing system
using the Gingell-Yoshida polyphase network is quite easy and
error-tolerant to make a good phasing exciter. It can be used in
"reverse" to get an easy-to-select sideband demod or an ordinary
AM detector that yields false stereo (one sideband to each ear),
already done with simple CW receivers.

....or real stereo! The Kahn/Hazeltine AM stereo sysem did this -- L in the
lower sideband, R in the upper. Hence envelope detectors recovered L+R, and
AM radios built back to the beginning of (radio) time kept working.

On the other hand (and I know this is just asking for abuse), the Motorola
C-QUAM AM stereo system could be applied to SSB modulation and still work,
which obviously Kahn/Hazeltine can't. I don't imagine C-QUAM's designers
were considering this, however.

It's almost painful to look at the (complex) envelope of AM and notice that
the quadrature signal is completely unused. Sending stereo over I and Q
strikes me as a 'interesting,' (does anyone know of a commercial system that
does this? Surely somebody's must...) but of course it would break
compatibility with current receivers and I imagine someone who's more
knowledgeable than I could point out some pitfalls as well.

---Joel

Hi Len,

Avery Fineman wrote:
For narrowband voice, SSB AM is just dandy and a phasing system
using the Gingell-Yoshida polyphase network is quite easy and
error-tolerant to make a good phasing exciter. It can be used in
"reverse" to get an easy-to-select sideband demod or an ordinary
AM detector that yields false stereo (one sideband to each ear),
already done with simple CW receivers.

....or real stereo! The Kahn/Hazeltine AM stereo sysem did this -- L in the
lower sideband, R in the upper. Hence envelope detectors recovered L+R, and
AM radios built back to the beginning of (radio) time kept working.

On the other hand (and I know this is just asking for abuse), the Motorola
C-QUAM AM stereo system could be applied to SSB modulation and still work,
which obviously Kahn/Hazeltine can't. I don't imagine C-QUAM's designers
were considering this, however.

It's almost painful to look at the (complex) envelope of AM and notice that
the quadrature signal is completely unused. Sending stereo over I and Q
strikes me as a 'interesting,' (does anyone know of a commercial system that
does this? Surely somebody's must...) but of course it would break
compatibility with current receivers and I imagine someone who's more
knowledgeable than I could point out some pitfalls as well.

---Joel

In article , "Joel Kolstad"
writes:

Avery Fineman wrote:
So, if you want to examine the total RF in a time-domain situation,
you MUST examine it as amplitude versus an infinitely-thin slice of
TIME.

You might want to remind everyone that the mathematical Fourier transform of
a signal is an integral that extends from time=minus infinity to plus
infinity.

The series expansions of the basic modulation type RF time-domain
expression don't use Fourier series. The series expansions show
the sprectral content are still equivalent to that infinitely-thin slice
of time as the function of amplitude.

I think that Panters "Signals, Modulation, and Noise" text has it
worked out in there (by memory, don't have that one handy here).
The "Landee" text I mentioned is an old one and not that familiar
to most.

Since Real Spectrum Analyzers (or network analyzer) need to
produce results in something, oh, less than infinite time (probably less
than the time between now and the next donut break), they're necessarily
limited in the low frequency detail they can provide. True, if Gary's
transmitter is transmitting a zero at the moment he connects a spectrum
analyzer, he won't see anything at all on the display, but as you point
out -- this is an equipment problem, not a mathematical one.

No, it's an argument problem. :-)

There's no infinitely-fast RF power meter in existance. Yet.


I'm still a believer in SSB-FM, BTW. :-) But I have enough respect for you
that I won't attempt to argue it further without first finding the time to
prepare a few drawings to demonsrate why!

I'm not arguing that "single-sideband FM" won't work. I just don't like
the name. The technique DOES work from all the explanations of
experiments, is reproducible. [it isn't the "cold fusion" thing. :-) ]

I've not seen any convincing case that the single-whatever FM thingy
has any practical applications.

For narrowband voice, SSB AM is just dandy and a phasing system
using the Gingell-Yoshida polyphase network is quite easy and
error-tolerant to make a good phasing exciter. It can be used in
"reverse" to get an easy-to-select sideband demod or an ordinary
AM detector that yields false stereo (one sideband to each ear),
already done with simple CW receivers.

Len Anderson
retired (from regular hours) electronic engineer person

Avery Fineman wrote:
So, if you want to examine the total RF in a time-domain situation,
you MUST examine it as amplitude versus an infinitely-thin slice of
TIME.

You might want to remind everyone that the mathematical Fourier transform of
a signal is an integral that extends from time=minus infinity to plus
infinity. Since Real Spectrum Analyzers (or network analyzer) need to
produce results in something, oh, less than infinite time (probably less
than the time between now and the next donut break), they're necessarily
limited in the low frequency detail they can provide. True, if Gary's
transmitter is transmitting a zero at the moment he connects a spectrum
analyzer, he won't see anything at all on the display, but as you point
out -- this is an equipment problem, not a mathematical one.

I'm still a believer in SSB-FM, BTW. :-) But I have enough respect for you
that I won't attempt to argue it further without first finding the time to
prepare a few drawings to demonsrate why!

---Joel Kolstad