Roy Lewallen, W7EL, wrote:

I measured 80% velocity factor, so if our respective
measurements are good, the velocity factor of your piece should be
around 72%.

Ok, my curiosity got the best of me, and I decided to
to measure the velocity factor. My VNA software has
a "distance to fault" feature, and I "worked backwards",
which is to say that I measured a length of the
BuryFlex with a tape measure (27' 2"), and adjusted
the VF on the distance to fault tool until I obtained the
same physical length. The far end of the cable was
terminated with an open circuit.

I happened to start with the measurement frequency set
to 1 MHz. Lo and behold, the VF needed to compute the
same physical length was 72%, as you suggested.

My own understanding is that VF should be constant with
respect to frequency, so I decided to vary the test frequency.
I should have left well enough alone.

I picked 1, 2, 4, 8, 16, and 32 MHz. For those frequencies,
I measured the following VFs:

1 MHz = 72%
2 MHz = 73%
4 MHz = 75%
8 MHz = 80%
16 MHz = 79%
32 MHz = 79%

I'm rounding the VF to integer values, since I don't think that any more accuracy
can be claimed in this setup.


Since that result was a little surprising, I grabbed some mini 8 (8X)
that was nearby, about 51.25 feet. The published VF is 78%, and
I measured the following, at the same test frequencies:

1 MHz = 78%
2 MHz = 78%
4 MHz = 79%
8 MHz = 79%
16 MHz = 80%
32 MHz = 80%

With this cable, the VF appeared much more constant across the
1 to 32 MHz range.

Is there an explanation that fits with my measurements?

Greg, W8WWV

One thing I've learned is that a (good) short circuit is a better
termination for this kind of measurement than an open circuit. With an
open circuit you get fringing which varies with frequency and cable
diameter.

If you don't have a decent commercially made short circuit, you can do
reasonbly well with a connector and several radial wires from the shell
to the center conductor, or better yet a metal disk. Or you could put a
male connector on a short piece of coax and short circuit the end of the
cable by squeezing the braid into the center and soldering it to the
center conductor, taking its length into account for the measurement of
course. The object is to minimize the series inductance which would be
created with a single-wire or similar connection.

Wes and Ian can probably point you to more possible causes for this
measurement result -- they're the real experts on VNA measurements. I'd
be very surprised if the VF really varies with frequency -- in theory,
it can't, in an ideal coaxial cable.

Roy Lewallen, W7EL

Greg Ordy wrote:
Roy Lewallen, W7EL, wrote:


I measured 80% velocity factor, so if our respective
measurements are good, the velocity factor of your piece should be
around 72%.


Ok, my curiosity got the best of me, and I decided to
to measure the velocity factor. My VNA software has
a "distance to fault" feature, and I "worked backwards",
which is to say that I measured a length of the
BuryFlex with a tape measure (27' 2"), and adjusted
the VF on the distance to fault tool until I obtained the
same physical length. The far end of the cable was
terminated with an open circuit.

I happened to start with the measurement frequency set
to 1 MHz. Lo and behold, the VF needed to compute the
same physical length was 72%, as you suggested.

My own understanding is that VF should be constant with
respect to frequency, so I decided to vary the test frequency.
I should have left well enough alone.

I picked 1, 2, 4, 8, 16, and 32 MHz. For those frequencies,
I measured the following VFs:

1 MHz = 72%
2 MHz = 73%
4 MHz = 75%
8 MHz = 80%
16 MHz = 79%
32 MHz = 79%

I'm rounding the VF to integer values, since I don't think that any more accuracy
can be claimed in this setup.


Since that result was a little surprising, I grabbed some mini 8 (8X)
that was nearby, about 51.25 feet. The published VF is 78%, and
I measured the following, at the same test frequencies:

1 MHz = 78%
2 MHz = 78%
4 MHz = 79%
8 MHz = 79%
16 MHz = 80%
32 MHz = 80%

With this cable, the VF appeared much more constant across the
1 to 32 MHz range.

Is there an explanation that fits with my measurements?

Greg, W8WWV

Roy Lewallen wrote:

Wes and Ian can probably point you to more possible causes for this
measurement result -- they're the real experts on VNA measurements.

Not I... I'm just trying to think acurately about it...

I'd be very surprised if the VF really varies with frequency -- in
theory, it can't, in an ideal coaxial cable.

Greg had written:
adjusted
the VF on the distance to fault tool until I obtained the
same physical length. The far end of the cable was
terminated with an open circuit.
I happened to start with the measurement frequency set
to 1 MHz. Lo and behold, the VF needed to compute the
same physical length was 72%, as you suggested.
My own understanding is that VF should be constant with
respect to frequency, so I decided to vary the test frequency.
I should have left well enough alone.
I picked 1, 2, 4, 8, 16, and 32 MHz. For those frequencies,
I measured the following VFs:
1 MHz = 72%
2 MHz = 73%
4 MHz = 75%
8 MHz = 80%
16 MHz = 79%
32 MHz = 79%
I'm rounding the VF to integer values, since I don't think that any
more accuracy
can be claimed in this setup.
Since that result was a little surprising, I grabbed some mini 8
(8X)
that was nearby, about 51.25 feet. The published VF is 78%, and
I measured the following, at the same test frequencies:
1 MHz = 78%
2 MHz = 78%
4 MHz = 79%
8 MHz = 79%
16 MHz = 80%
32 MHz = 80%
With this cable, the VF appeared much more constant across the
1 to 32 MHz range.
Is there an explanation that fits with my measurements?

Since Greg's test setup measured a much more constant VF for the mini-8,
it doesn't seem that the open-circuit termination is causing much error
(note also that the largest deviation for the BF is at the lowest
frequency, where fringing C would have the least effect).

This seems to bring us back to Greg's method of computing "distance to
fault" from measurements of R and X at one end. Does that method involve
any assumptions about idealized cable behavior that could create an
*apparent* change in the computed VF?


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