On 6 Dec 2006 03:15:54 -0800, "4nec2" wrote:

Somehow, and using Nec2, I seem not always able to get expected
impedance values when using the results produced by L.B. Cebik's stub
matching software at http://www.cebik.com/trans/ant-match.html


Arie,

The versions of NEC that I have used all treat transmission line
elements defined as such as lossless, and so the results will be in
error to some extent, depending on the scenario. Allowing for that
error, models I have built and checked appear correct.

The spreadsheet by L B Cebik linked from the page above depends on an
assumption that Z is real for some of the match types, and again
assumes that the line sections are lossless (implying Zo is real).

The assumption of lossless lines might often be acceptable for open
wire lines, but less so for common commercial coaxial lines.

My experience with stub matching is that significant trimming of the
theoretical design is usually necessary. I suspect tolerances of
transmission line parameters is a large contributor to the error, and
some stub designs become extremely sensitive to small changes in
components. Assuming that stub matching is 100% efficient (ie that
line sections are lossless) is another significant contributor.

Does anybody know a source for (similar) stub matching software
(freeware or evaluation), so I can check where things go wrong ?


The line loss calculator at http://www.vk1od.net/tl/tllc.php will
calculate the transformation in series line sections of about 100
common line types, having regard for line loss. Impedances are dealt
with as complex quantities. This also allows calculation of the input
impedance of an o/c or s/c stub by specifying an appropriate
termination impedance. The line loss model used is described on the
page above.

For example, what is the impedance of a quarter wave stub of RG58C/U
at 14MHz? You know it cannot be infinite, but that is what lossless
estimates will produce. The calculator above suggests that 3.53492m of
Belden 8262 s/c stub has an input impedance of about 2300+j0 ohms. At
7MHz, the same stub would have an input impedance of about 2.3+j49.96
(a quite lossy equivalent inductor) where lossless estimates would
suggest 0+j50.

A useful feature of a "Smith chart program" would be a decent
transmission line loss model for transmission line segments. Most that
I have seen seem to ignore it, Winsmith allows specifications of line
loss, but it is tedious and nevertheless only allows specification of
real values for Zo.

XLZIZL and the derived TLDetails are applications that consider the
line loss in calculations (but whilst estimating Xo, assume that Ro is
equal to nominal Zo). There are other programs that use a similar
approximation but seem to have unrealistic built in line loss data.

Owen
--