wrote in news:1187540912.064738.6170
@g4g2000hsf.googlegroups.com:

Scott,

The inconsistency you describe seems to lie in one or more of your
instrument, the coax, or your test setup.

I am not sure whether you are solving a coax problem, or validating the
LP100.

The impedance plot certainly clears away questions about whether you have
properly accounted for velocity factor. The phase plot has a strangely
flattened minimum.

I wonder about your confidence in the 54 ohm load, and your answers to
Richard about choking the cable suggests something non-ideal.

You could try a test with s/c or o/c stub (or both), so eliminating the
54 ohm load. Of course, in such a test, the line loss becomes more
important. It also tests your instrument at extreme mismatch, so you
might want to validate it on o/c, s/c, and some reactive 100:1 loads
using known coax sections.

Back to the loss thing, the line loss calculator at
http://www.vk1od.net/tl/tllc.php uses a more sophisticated line model
than any other calculators I have seen, and contains data for a hundred
or so of the most common commercial lines. It also calculates the
expected impedance transformation for the line.

For example, it suggests that a 6m length of Belden B9204 (RG59/U type)
o/c line at 3.8MHz has an input Z of 0.84-j84.91.

BTW, have you stripped a bit of cable down to see visually if appears
physically sound. Common problems I have encountered have included
unevenly woven braid, unevenly laid strands in centre conductor (though
yours should be solid), un-centred centre conductor, air inclusions in
the dielectric.

As I said, I am not sure that the LP100 has been validated and the focus
is entirely on the particular coax.

Owen