On Sun, 13 Jun 2010 10:03:40 -0700 (PDT), Keith Dysart
wrote:

'Unusual'? Certainly not unexpected once one thinks about it.

Hi Keith,

The "unusual" was lost in the ellipsis that I will reveal:
"... The highest efficiency operating point does not exactly coincide
with the lowest plate current because the power output continues to
rise for a while on the inductive side of resonance coming out of the
dip in the plate current. ..."

Continuing, Mendenhall presents the problems of trade-offs between
what would seem to be maximum power for worsening characteristics in
performance - the goal is what he calls minimizing synchronous AM
versus Efficiency.

Synchronous AM is a problem that can be introduced by measurement
equipment, as Mendenhall relates: "The input impedance of the
envelope detector must provide a nearly perfect match ... 30dB return
loss ... to the sampling line."

This sidebar relates to what Owen characterized as "Usability" where I
have recited the objective technical specification to reduce the
subjectivity of the term.

There's more that could be said, but I am waiting to see if Richard is
willing to subscribe to his own reference's writings. If not, and
Mendenhall doesn't mince words on the topic, then as Perry Mason would
observe "The D.A. is impeaching his own witness!"

I've always loved Perry Mason.

73's
Richard Clark, KB7QHC

On Jun 13, 12:35*pm, Richard Clark wrote:

I am waiting to see if Richard is willing to subscribe to his own reference's writings

I responded to you some 20 minutes before you posted, and now await
your response.

Also please comment on whether or not a Class C amplifier operating on
a linear portion of its transfer curve will function as a linear
amplifier.

RF

On Sun, 13 Jun 2010 11:04:50 -0700 (PDT), Richard Fry
wrote:

I am waiting to see if Richard is willing to subscribe to his own reference's writings

I responded to you some 20 minutes before you posted, and now await
your response.

So your subscription to your own authority is contingent upon me?

I'm flattered.

Also please comment on whether or not a Class C amplifier operating on
a linear portion of its transfer curve will function as a linear
amplifier.

"If the amplifier is tuned exactly to resonance, the plate load
impedance will be purely resistive and the load line will be linear."

I would note that the bulk of his paper concerns frequency products -
something the bears quite intimately upon linear operation.

Your question is rather opaque in that the appearance of subtext is
undoubtedly bound up in the term "linear" which Mendenhall has already
employed within the norms of amplifier design.

How linear is linear? Give me a technical criteria (subjectivity
abounds and is not very informative except for statements of fashion).
Express the least departure from a line response in percent that you
accept as being linear. If it exceeds Mendenhall's design or
commercial product, then I am afraid you won't be satisfied with any
answer.

73's
Richard Clark, KB7QHC

On Jun 13, 1:20*pm, Richard Clark wrote:

How linear is linear? *Give me a technical criteria...

A truly linear device does not produce r-f intermodulation products
when signals of different frequencies are applied to it at the same
time.

For many practical uses a linear device does need to be perfectly
linear. Let's say that the IM products need to be only - 20 dB w.r.t
the lowest amplitude value of the various input frequencies. The
Mendenhall paper I quoted conclusively shows that even such nominally
linear performance is not an attribute of Class C vacuum tube r-f
amplifiers -- especially at, and near the center frequency of the PA
output tuning/matching network.

That Mendenhall paper also shows that such r-f amplifiers do not
present a functional, 50+j0 termination to r-f energy entering via the
tx output connector. That is the reason why such energy is present in
the PA plate circuit, giving rise to r-f intermodulation and/or
changing the power dissipation there.

RF

On Sun, 13 Jun 2010 12:16:40 -0700 (PDT), Richard Fry
wrote:

On Jun 13, 1:20*pm, Richard Clark wrote:

How linear is linear? *Give me a technical criteria...

Let's say that the IM products need to be only - 20 dB w.r.t
the lowest amplitude value of the various input frequencies.

Medenhall's paper shows "RF spectrum showing distortion products" that
are 30dB down.

His statement: "If the amplifier is tuned exactly to resonance, the
plate load impedance will be purely resistive and the load line will
be linear." is attached to that linearity which is greater than you
expect.

73's
Richard Clark, KB7QHC

On Jun 13, 3:31*pm, Richard Clark wrote:
On Sun, 13 Jun 2010 12:16:40 -0700 (PDT), Richard Fry
wrote:

On Jun 13, 1:20*pm, Richard Clark wrote:

How linear is linear? *Give me a technical criteria...

Let's say that the IM products need to be only - 20 dB w.r.t
the lowest amplitude value of the various input frequencies. *

Medenhall's paper shows "RF spectrum showing distortion products" that
are 30dB down.

His statement: "If the amplifier is tuned exactly to resonance, the
plate load impedance will be purely resistive and the load line will
be linear." is attached to that linearity which is greater than you
expect.

73's
Richard Clark, KB7QHC

To Richard Fry,

Richard, it appears that you have ignored my post where I reported
data that proves the output resistance of Class B and C RF amps is non-
dissipative. Have you reviewed or ignored my reference to Chapter 19
in Reflections 3, or the combined sections of that chapter that appear
in my web page at www.w2du.com that uses an example from Terman's
Radio Engineers Handbook? If you haven't yet reviewed it I urge you to
do it, and then if you disagree with my measured data I'd like for you
to explain the reason for your disagreement.

Walt, W2DU

On Jun 13, 2:58*pm, walt wrote:

To Richard Fry,

Richard, it appears that you have ignored my post where I reported
data that proves the output resistance of Class B and C RF amps
is non-dissipative.

I know your conclusion, thanks Walt.

The impedance of a perfect transmission line also is non-dissipative,
but that does not change the power dissipated in a termination of that
transmission line equaling its Zo.

Do you agree that the Mendenhall paper I have quoted shows that the
source impedance of an operating Class C vacuum tube PA and its output
tuning/matching network designed for a 50 ohm load is not, by itself,
a functional 50+j0 ohms?

RF

On Jun 13, 2:31*pm, Richard Clark wrote:


Medenhall's paper shows "RF spectrum showing distortion products"
that are 30dB down.

Not near the resonant frequency of the output tuning/matching network.

He shows as little as 5 dB at 0.8 MHz offset of the external signal
from the carrier of that tx. It would be less than 5 dB for lesser
separations.

The reason that the IM performance improves for greater frequency
separations is related to the r-f bandwidth of the output network --
not the "linear" operating characteristics of the PA, or that a 50+j0
ohm load is seen by the external energy reaching the tx output
connector.

RF

On Sun, 13 Jun 2010 13:36:53 -0700 (PDT), Richard Fry
wrote:


Mendenhall's paper shows "RF spectrum showing distortion products"
that are 30dB down.

Not near the resonant frequency of the output tuning/matching network.

Another subjective term. How near is near? All products that
Mendenhall shows in the illustration titled "RF spectrum showing
distortion products" are 30dB down and those distortion products fall
between modulation sidebands (unless you are speaking of doubling and
trippling of the carrier frequency).

Returning to the amplifier of his own design, in his classic WREK
engineering report [you seem shy of acknowledging your authority's
work here] submitted to the FCC. Now, when I say "of his own design,"
I mean by him personally, at the bench, using tools, assembling
components against his own schematic. He explicitly states:
"Distortion at 100% = 1%." At "frequency removed from carrier" from
-35dB to -73dB out to "600 Khz and beyond."
"Harmonic and spurious radiation" at least -70dB.
"Harmonic distortion" at 100% worst case 2.3%
"Plate efficiency of 70% to 80%"
"Total efficiency of 68%" [plate power in / RF power out]
operating Class C.

And of course his statement that sends minds reeling:

"It was thus necessary to determine the plate load impedance
= (Eb - Emin) / I1
~= Eb/ Idc
~= 1000 Ohms
[Not very far from Walt's determination of 1400 Ohms from his own
bench.]
"Since this was to be coupled into a Z output of 50 Ohms, a impedance
transformation of 20:1 was needed. This was accomplished by using a
voltage divider of two series capacitors to series tune the plate
circuit."

His EIMAC reference found in other, later published papers conform to
every detail in his engineering report to the FCC and in those same
later papers. In fact, they offer a multitude of operational
parameters for a variety of tubes at many power levels, across many
bands and classes of operation.

73's
Richard Clark, KB7QHC