All,

The system is a vector network analyzer and a coax feedline. The
analyzer does not have built in 'electronic line length' compensation.
So to compensate for 'near' frequency measurements I have created coil
of coax on the floor. The theory is when actual measurements are made on
the antenna the phase can be normalized by zeroing the phase to the
reference coax.

The issue is: Attempts measure the electrical length yield inconsistent
results.

The first step was to determine the approximate length. This was done by
setting the sweeper to a frequency and counting the phase inversions to
the antenna feed - open ended. Lengths of coax were laid out on the
floor until the phase inversion pattern was 'very close' visually.

Next the the frequency was set to 200 kc sweep at one meg on the
antenna. A near phase inversion was centered on the phase display.
Lengths were added or subtracted to the reference coax to get the same
center inversion. (The inversion baseline was verified by looking at a
larger sweep for reference.) So far so good.

Next the same thing was done at 10 meg. A slight modification to the
baseline coax.

Finally the system was retested at 1 meg. Whoops. It was off.

The differences were small, 6 to 8 inches out of an estimated 125 feet.
However it is not possible to get a common match between 1 meg and 10
meg - hence the anomaly.

The reference coax is coiled on the floor. Is it possible this created
an inductance which affects phase?

Thanks - Dan

dansawyeror wrote:
The reference coax is coiled on the floor. Is it possible this created
an inductance which affects phase?

Is it possible that common-mode currents are upsetting
your measurements?
--
73, Cecil http://www.w5dxp.com

1. The coax has loss, which is different at each frequency.

2. The coax's characteristic impedance is slightly different at each
frequency.

3. Fringing at the open end is different at each frequency, so the
capacitance at the end and hence the effective length of the coax is
different at each frequency.

All of these make it impossible to use a simple model of a lossless coax
with fixed impedance for very accurate results.

Coiling it doesn't create inductance or have any other effect on the
inside of the coax where, hopefully, the signal from your network
analyzer is going.

Roy Lewallen, W7EL

dansawyeror wrote:
All,

The system is a vector network analyzer and a coax feedline. The
analyzer does not have built in 'electronic line length' compensation.
So to compensate for 'near' frequency measurements I have created coil
of coax on the floor. The theory is when actual measurements are made on
the antenna the phase can be normalized by zeroing the phase to the
reference coax.

The issue is: Attempts measure the electrical length yield inconsistent
results.

The first step was to determine the approximate length. This was done by
setting the sweeper to a frequency and counting the phase inversions to
the antenna feed - open ended. Lengths of coax were laid out on the
floor until the phase inversion pattern was 'very close' visually.

Next the the frequency was set to 200 kc sweep at one meg on the
antenna. A near phase inversion was centered on the phase display.
Lengths were added or subtracted to the reference coax to get the same
center inversion. (The inversion baseline was verified by looking at a
larger sweep for reference.) So far so good.

Next the same thing was done at 10 meg. A slight modification to the
baseline coax.

Finally the system was retested at 1 meg. Whoops. It was off.

The differences were small, 6 to 8 inches out of an estimated 125 feet.
However it is not possible to get a common match between 1 meg and 10
meg - hence the anomaly.

The reference coax is coiled on the floor. Is it possible this created
an inductance which affects phase?

Thanks - Dan

On Thu, 15 Mar 2007 14:37:28 -0700, dansawyeror
wrote:

The reference coax is coiled on the floor. Is it possible this created
an inductance which affects phase?

Hi Dan,

There is a vastly simpler way:
1. Measure to a short circuit.

73's
Richard Clark, KB7QHC

Richard,

It would seem a short might reduce some end capacitive effects when
compared with an open circuit. However it would not help in the case of
'minor' frequency dependent changes in coax inductance.

Thanks,
Dan

Richard Clark wrote:
On Thu, 15 Mar 2007 14:37:28 -0700, dansawyeror
wrote:

The reference coax is coiled on the floor. Is it possible this created
an inductance which affects phase?

Hi Dan,

There is a vastly simpler way:
1. Measure to a short circuit.

73's
Richard Clark, KB7QHC

Hi Dan

What would happen if you closed the coax shield around the open end of the
coax?

Jerry


"dansawyeror" wrote in message
...
Richard,

It would seem a short might reduce some end capacitive effects when
compared with an open circuit. However it would not help in the case of
'minor' frequency dependent changes in coax inductance.

Thanks,
Dan

Richard Clark wrote:
On Thu, 15 Mar 2007 14:37:28 -0700, dansawyeror
wrote:

The reference coax is coiled on the floor. Is it possible this created
an inductance which affects phase?

Hi Dan,

There is a vastly simpler way:
1. Measure to a short circuit.

73's
Richard Clark, KB7QHC

On Fri, 16 Mar 2007 06:45:48 -0700, dansawyeror
wrote:

It would seem a short might reduce some end capacitive effects when
compared with an open circuit. However it would not help in the case of
'minor' frequency dependent changes in coax inductance.

Hi Dan,

A short "might reduce some?" That is rather an understatement.

"It would not help... 'minor' changes?" That is even worse.

Trying a short in place of an open would have taken less time, and
proven these two statements are wrong. A short is also an excellent
way to measure Ohmic loss in the cable. If you read your user's
manual for your vector network analyzer (assuming it has one), you
would find that a short is the preferred method. Also, for vector
network analyzers that do not have built in correction, it is also
expected you repeat the measurement with a Zc load. Then, and only
then do you proceed to doing it with an open (which will reveal the
extent of fringing).

I wouldn't do an open anyway. Instead, I would measure a load
constructed to mimic the measurable (you must have some idea what it
is you are going to be measuring, don't you?) and put it in a shielded
enclosure. Then move this text fixture to the vector network analyzer
and repeat to see the effects of the line.

73's
Richard Clark, KB7QHC