In a message dated 4/10/06 9:27:25 AM Central Daylight Time,
amps-request@contesting.com writes:
<< All,
I started reading through the RCA Radiotron Handbook this evening looking
for where the factor of 1.8 is listed for calculating the plate impedance of a
class AB amplifier, and I cannot find it. The edition I have was the older one
back in the 40's with the black cover. The red one was from the 50's and had
more in it if I recall, but I don't have it. I was wanting to find where it
gives this factor, and the ones for Class A, AB1, AB2, B, and C. Bill Orrs
handbook gives the factor of 1.8 on his Pi tank values table, but doesn't
mention
anywhere in the text where it came from. I also looked in a RCA receiving tube
book and could not find it there either, or I am overlooking it somewhere. I
could have sworn it was in the Radiotron Handbook. The Handbook does say that
the plate impedance is twice the peak voltage swing times the peak anode
current. What I'm wanting to know is where are these factors located in print,
as I'd
like to read the whole texts concerning this? Any help
would be deeply appreciated. Thanks to all in advance.
Best,
Will
>>
Will:
I do a great deal of mathematical analyses on high power RF amplifiers. My
favorite resource is the Eimac Care and Feeding of Power Grid Tubes available
on
Eimac's web site. I think Richardson Electronics may also have hard copies
available. You also need the clear plastic overlay and a set of constant
current
curves for the tube you are using. Then you can calculate some pretty close
values for your desired parameters. This is the most accurate mathematical
model that I'm aware of. Some may considered as too tedious, and that's fine.
It
is still the most accurate method of modeling on paper.
My second choice is the mathematical steps described in the RCA Transmitting
Tubes book number TT-5. This book is available from many web resources. There
is a step by step procedure in there for calculation all of the operating
parameters. both RF and DC. The calculations rely on a table of "K" factors
that
are dependent on the plate current conduction angle for the class of service
desired. This is a fairly straight forward mathematical process.
To answer your specific question, the input impedance as seen by the tube of
the tank network is RF plate voltage swing divided by the peak fundamental
component of the plate current. If you don't get this number "right" the effect
is distortion, crappy efficiency and maybe even instability.
Let me give you an example. Let's say that we are designing a class B
amplifier and we know the peak plate current is one amp. Class B suggests 180
degree
plate current. Therefore the DC plate current is going to be the peak plate
current divided by 3.1416. The peak fundamental plate current is going to be
half of the peak plate current or . 5 amp. The tube curves are used to find an
appropriate minimum instantaneous plate voltage based on the desired linearity
and the best value to use is tube dependent. As a generalization, make this
minimum plate voltage (ebmin)
equal to ten percent of your DC plate voltage. Say your plate voltage is 3000
volts, so ebmin is then 300 volts. the plate swing is therefore 3000 - 300 =
2700 volts. So the impedance the tank needs to present to the tube is 2700/.5
= 5400 Ohms. The power output BTW is plate swing times Peak Fundamental Plate
Current times 0.5 and this assumes the average power of a CW signal.
The problem with the RCA K values and any other constant for that matter, is
that they ignore the tube characteristics. Furthermore, often times the basis
for which a multiplying constant is derived is not always known. The RCA K
values and the procedures in Bill Orr's famous works, all assume that the plate
current is going to be a perfect sinusoid over the portion of a cycle for which
it conducts. This is not an accurate assumption and fortunately for
approximation purposes, is usually good enough. The advantage of the K values
and Bill
Orr's calculation process is it gives you a reasonable starting point for a
design. There is always going to be some "lab work" so there is a personal
balance everyone must make individually as to how much time is going to be
spent
calculating and how much time is going to be spent constructing and optimizing.
The first question to ask regarding any mathematical model is "how much
accuracy do I really want or need? The usual response is "as much as possible".
If
you have infinite research dollars and infinite time, this is a reasonable
answer. Who really has that advantage? Obviously there is a practical answer to
this question.
I should point out that the example I gave you only applies to a class B and
any other class of service is going to command a different set of K values. My
example was to illustrate the basic process. When I do an analysis on a
broadcast transmitter, I like to start off with the RCA K values to get an Idea
for
what I'm dealing with. I then follow up using the Eimac method, a drawn
Operating Line and the accompanying calculations to get the final, usually more
accurate results. The Eimac method is restricted BTW, in that the voltage
waveform on the grid must be the same kind of waveform on the plate. IOW, you
cannot
drive a tube (with a resonant plate tank network) with a square wave with a
sinusoidal signal on the plate, and expect meaningful results with the Eimac
Tube Performance Computer. It does not work.
This is probably a more lengthy response than you were looking fore, but I
hope it helps you.
Jeff Glass, BSEE CSRE
Chief Engineer
WNIU WNIJ
Northern Illinois University
WB9ETG
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|