Jim Lux wrote:

The circuits I am thinking of sample a length of line (NOT A
POINT) so the sample records average voltage (or current) from a
period of time.
Many simple reflectometer designs do indeed sample the line over a
short length of line, and that short length may be 100mm or more.
Ideally, they would take the sample at a point. (Since a point has
zero length, I can't quickly think of a sampling technique that truly
takes a point sample.)

The voltage sample is easy... measure the voltage using an infinitely
thin probe.

The current sample is measured in a similar way by measuring the
magnetic field over a infinitely small segment of the conductor. There
are sensitivity issues or bandwidth issues, but there are lots of very,
very small magnetic field probe schemes around.

Agreed; we're discussing principles here, and the issue of single-point
sampling is mostly a practical one.

In principle, we can always choose a method of sampling that doesn't
require a finite length of line. Within the limits of our skill and
imagination, we can also make the current and voltage pickups physically
smaller, so that they occupy less length along the line. Or if skill and
imagination fail, we can shift the whole discussion to longer and longer
wavelengths, to make the error as small as we like. It may not be
practical, but no general principles are being broken.

The issue of single-point sampling is interesting in its own right, but
in this much wider discussion it is only a minor detail. In order to
move on with the wider discussion, let's agree to assume that
single-point sampling always *can* be achieved, within the accuracy that
we require.



If one says, "point sample" == "less than 1/1000 wavelength), I think
it's actually pretty straight forward, certainly for 100 MHz or less.
(3mm is 1/1000 lambda).

Even for practical instruments, this particular source of error is
usually quite small. At any one frequency, it is always possible to null
the bridge in the reverse direction, so that the voltage and current
samples (as described by Cecil) will cancel. How well the cancellation
holds over a wider frequency band will depend on the choice of bridge
circuit and the way it is constructed.


--

73 from Ian GM3SEK 'In Practice' columnist for RadCom (RSGB)
http://www.ifwtech.co.uk/g3sek