Gary (and anyone else who cares), since my last post, which responded to
several other posts on the topic of PEP in an AM transmitter, I looked up
some things and cleared up a major misunderstanding in my own mind. I will
add that as comments to the part of your post, below, which is relevant to
the issue. As far as all of your definitions below, PEP wattmeters,
S-metes, SSB signals are concerned, I think you made a lot more mistakes
than you realize. However, I'm going to delete all these irrelevant parts
(most of what you said) and concentrate on the source of the confusion.
I may make comments in a separate post on the parts I deleted fro this
one.

On Wed, 25 Jan 2006, Gary Schafer wrote:


Let's review some definitions to start:

AVERAGE POWER

Average power is found by squaring RMS voltage and dividing by
resistance. Or RMS voltage times RMS current.


PEP

deleted

PEP WATTMETERS

deleted


S METER READINGS

deleted


AM TRANSMITTER

deleted


CONVERTING RMS TO PEAK

deleted


AM LINEAR

deleted


POWER IN SIDE BANDS

deleted


PLATE CURRENT AND VOLTAGE DOUBLING

Here is the crux of the problem. Earlier today I looked in my old RCA
receiving tube manual and transmitting tube manuals at the transfer
characteristics of many dozens of tubes and I looked at them with this
question of PEP for an AM signal. I will incorporate some of what I
learned as comments on your comments. The basic fact that I was not aware
of is that there is an apparent conflict between the relationship
between plate current and plate voltage if you look at the curves that
show plate current as independent of plate voltage and then ask how do you
get, on modulation, a peak input power four times the unmodulated input
power so you can get a peak, on modulation, output power that is four
times the unmodulated output power.

It is easiest to see with a triode tube that is plate modulated.

Nah, "easiest" has nothing to do with it. Triode has nothing to do with
it.

The issue is that all of the triode transfer characteristics curves I saw
showed plate current to be _proportional_ to plate current (but with
offsets and some non-linearities, which are mostly unimportant).
When I looked at all the tetrodes and pentode curves, then, yes, they all
showed plate current independent of plate voltage. However, at any given
plate voltage, plate current was also _proportional_ to screen voltage
(also with and offset and some non-linearities). Now, it makes sense that
if screen voltage is made proportional -- in some fashion (usually a
screen voltage dropping resistor connected to the modulated plate supply)--
to plate voltage, then plate current will increase, or decrease, in
parallel with plate voltage as modulator voltage adds, and subtracts, from
the B+ plate voltage (all as the modulator output signal varies with audio
input waveform)

Doubling the plate voltage will cause the plate current to also
double.

From the curves, the relationship between plate current and plate current
might not always be exactly a 1:1 relationship, but to an approximation
this doubling is an acceptable understanding. And, that is how, on peak
input from modulation one gets four times unmodulated input, and output
will be proportional to input which can be looked at as average or peak,
but the peak output on modulation will also be four times unmodulated
output.

That is if the tube is capable of providing enough emission.

That is a separate issue and anyone designing a circuit and sellecting a
tube for use needs to understand the specifications in the manuals.

This must be a linear function in order to avoid distortion when
modulating.

Almost nothing is perfectly linear. All audio circuits will have
measureable distortion (IM, harmonic, and others). The only criterion is
whether the distortion is acceptable.

Tubes that are weak may not be able to provide this. That is one
reason that PEP may not fully reach 4 times the carrier power with
100% modulation.

I think for this issue one needs at least an oscilloscope to even start
measuring and investigating what is going on (and they need to be
wideband or sampling scopes, too). "Meters" are just indicators.

Screen grid tubes are not linear in this respect. Plate current is
somewhat independent of plate voltage. That is why you must also
partly modulate the screen along with the plate when using a screen
grid tube in the final.

There is an equally important reason why you must, and preferably, fully
modulate the screen voltage as well as the plate voltage (and this is
almost never discussed). If you ever have screen voltage above plate
voltage, then screen current will go up dramatically and so will screen
heat dissipation. You could melt the screen grid with just one word into
the microphone. You can blow the screen grid almost instantly just by
accidentally having screen voltage present without plate voltage.

You want to have a linear plate voltage to
plate current relationship.

This is also why a lot of broadcast transmitters use triodes in the
final. Easier to maintain linear modulation.

I think, if you looked at as many transfer characteristics, as I did
earlier today, for transmitting tubes, you might appreciate that there
is more heterogeneity between triodes than tetrodes or pentodes in terms
of plate I/V relationships. Broadcast AM transmitters never gave us any
kind of high fidelity so linearity was never that much of an issue. In
broadcasst FM transmitters, power and voltage linearity anywhere in the
RF chain was irrelevant.


HANDBOOK

All this can be found in the AM section in some of the older
handbooks.

I was never very satisfied with much in the handbooks, whether early or
late.

The newer ones do not cover AM very well.

They are covering tubes and analog subjects less well, too.
Everything is going digital, solid stae, chips, and software.

Art, W4PON

73
Gary K4FMX