I have just finished a sequence of updates/enhancements to the Pi Netwotk
analyzer program which I have developing. The response to the initial version
was far better than I anticipated, encouraging me to put in a little
additional effort.
I took care of a few minor computational glitches which pop up ocassionally,
and hope I caught all of them. I console my guilt for this with the knowledge
that Mr. Gates's operating system contained well known non-trivial errors for
over 25 years, and he has enough support staff to populate a medium size city.
The new version contains three primary additions: an expansion of the circuit
model to 2 full PI-sections, addition of file librarian for saving and
recalling designs, and explicit calculation of phase delays. By going to 2
sections, it is possible to analyze a wider class of networks. I have already
used it to see where a 2-section PI may be advantageous in certain cases of
amplifier tank circuits. Also, by just omitting the output capacitor, it
reverts to the standard Pi-L, whose superior harmonic suppression can be
viewed in great detail.
This last week I happened to see another interesting but unusual application
for this type of analyzer (this is happening regularly) which is close to
home, as far as this discussion group is concerned. It relates to the
necessity of keeping the cathode of a GG amp as close to ground as possible
(electrically.) There have been numerous comments and much discussion related
to this, because the potential for positive feedback is high at unexpected
frequencies. But how can you actually estimate the actual effective Z from
cathode to gnd, especially if there is a tuned grid network inline? Answer:
you look at the Pi-network ( I am assuming the typical grid matching circuit)
backwards, with the load resistance representing your driver. Knowing what
that driver's 'output' terminals represents is not so easy, especially over a
wide band. But you can easily make a number of guesses, at least covering the
extremes, and see what actually get reflected over to the cathode of the amp
through the grid network. This is the impedance that will control the main,
big feedback loop originating from the feedback capacitance, and you will be
able to detect where potential trouble frequencies might lie.
The librarian simply lets you save and recall your designs in files which can
name, and recall from a file menu on demand. This beats the hell out of
trying to keep a lot of data in spreadsheet format.
The phase delay calculations are very useful (mandatory, in fact) for
designing things like antenna phasing networks, line stretchers, and delay
networks. I actually included this feature because of an ongoing project
which includes a somewhat complex exercise in phasing and delay network
designs. So now you can get an instant readout of the phase delay at any
point from input to output.
NOTE: Many people (myself included, up to about 6 months ago) rely on the
data and formulas from ON4UN's LF DXing book, for the design of phasing
networks. Well, there are some significant errors in that reference and I
have approached both the author and ARRL about making corrections - neither
seemed at all concerned. If you need to design a line stretcher, some of his
formulas WILL give you the wrong answers. My program will NOT.
For those who received a copy from my initial emailing last month, you will
automatically receive the new version via email, and it is now named PINET4.
Any new requests are welcomed and the new mailing will go out in the next few
days.
Eric von Valtier K8LV
|