coaxpair reflection coeff angle is 'zero'
 

On Wed, 28 Dec 2005 21:37:05 -0800, dansawyeror
wrote:

Owen,

Thanks for the example. I do not have an impedance meter. I do have a vector
voltmeter that will read the phase and value of the reflected signal.

The core problem is: How to measure and improve the performance of a loaded
vertical. The unknowns are the value of ground and antenna resonance.

Setting the coax to a 1/4 wave multiple was a way to remove its phase altering
characteristics from the reflected signal at a frequency of interest. That seems
to be working and is predictable to measure the 1/4 wave odd, open circuit,
resonant points, etc. This simply proves consistent measurement and correct
identification of the electrical cable length.

You're making this way too complex (pun intended).. Why not just
connect the coax to the bridge and put your open-short-load standards
on the far end and do the calibration?

Or move the instrument out in the field?

For example:

http://users.triconet.org/wesandlinda/Field_8405_a.jpg

coaxpair reflection coeff angle is 'zero'
 

On Thu, 29 Dec 2005 00:53:44 -0700, Wes Stewart
wrote:


You're making this way too complex (pun intended).. Why not just
connect the coax to the bridge and put your open-short-load standards
on the far end and do the calibration?

Or move the instrument out in the field?

Wes, I have been guessing that Dan wants to measure the antenna over a
band of frequencies, and doesn't want to be popping up to the
feedpoint for every frequency cal.

No doubt, the process you propose Wes is simpler and more accurate, if
it is physically convenient.

Would calibration against a single s/c standard be accurate enough for
the purpose at hand. Perhaps a coax relay at the antenna feedpoint to
switch between a s/c port and the real load might be accurate enough
for calibration, and a whole lot more convenient. IIRC there is only
around 0.4dB of line loss from the shack (ie the desired VVM location)
to the feedpoint.

Dan, I think you have gotten on a sidetrack about building the
transmission line out to a tuned length. It is not necessary, or even
desirable as far as I can see, but it has the downside of complicating
the calcs and increasing scope for errors when you build out with a
different line type.

Owen
--

coaxpair reflection coeff angle is 'zero'
 

"Owen Duffy" wrote in message
...
On Thu, 29 Dec 2005 00:53:44 -0700, Wes Stewart
wrote:


You're making this way too complex (pun intended).. Why not just
connect the coax to the bridge and put your open-short-load standards
on the far end and do the calibration?

Or move the instrument out in the field?

Wes, I have been guessing that Dan wants to measure the antenna over a
band of frequencies, and doesn't want to be popping up to the
feedpoint for every frequency cal.

No doubt, the process you propose Wes is simpler and more accurate, if
it is physically convenient.

Would calibration against a single s/c standard be accurate enough for
the purpose at hand. Perhaps a coax relay at the antenna feedpoint to
switch between a s/c port and the real load might be accurate enough
for calibration, and a whole lot more convenient. IIRC there is only
around 0.4dB of line loss from the shack (ie the desired VVM location)
to the feedpoint.

Dan, I think you have gotten on a sidetrack about building the
transmission line out to a tuned length. It is not necessary, or even
desirable as far as I can see, but it has the downside of complicating
the calcs and increasing scope for errors when you build out with a
different line type.

Owen

Wes, and Owen, This problem of how to resolve the terminating impedance
seems so simple that I realize that I (again) must be missing something.
Wouldnt it be accurate enough for Dan to record the impedance at his coupler
for two conditions, 1) short circuit at the antenna end of the coax, and 2)
the antenna connected. Plot both impedances on a Smith Chart. Since the
impedance associated with the short is known to be zero, he needs only to
rotate *both* impedances thru the same angle needed to place the short ckt
impedance at zero on the smith Chart.

Jerry

coaxpair reflection coeff angle is 'zero'
 

On Thu, 29 Dec 2005 10:15:50 GMT, Owen Duffy wrote:

On Thu, 29 Dec 2005 00:53:44 -0700, Wes Stewart
wrote:


You're making this way too complex (pun intended).. Why not just
connect the coax to the bridge and put your open-short-load standards
on the far end and do the calibration?

Or move the instrument out in the field?

Wes, I have been guessing that Dan wants to measure the antenna over a
band of frequencies, and doesn't want to be popping up to the
feedpoint for every frequency cal.

I've provided a spreadsheet that facilitates the calculations over a
range of frequencies.

www.qsl.net/n7ws/8405.zip

So a couple of trips to the end of the cable are all that are required
to calibrate the setup.

(I must confess, I haven't tried this program with a line much over a
few inches in length to determine whether my calibration functions can
handle it, but I think so.)

Without doubt, the chance of errors creeping in using a long cable is
increased, but the alternative of trying to "calibrate" a cable and
subtract its effects mathematically is equally suspect. In fact, the
one-step process is pretty much the same thing; the cable is being
characterized by the calibration process.




No doubt, the process you propose Wes is simpler and more accurate, if
it is physically convenient.

Would calibration against a single s/c standard be accurate enough for
the purpose at hand. Perhaps a coax relay at the antenna feedpoint to
switch between a s/c port and the real load might be accurate enough
for calibration, and a whole lot more convenient. IIRC there is only
around 0.4dB of line loss from the shack (ie the desired VVM location)
to the feedpoint.

Dan, I think you have gotten on a sidetrack about building the
transmission line out to a tuned length. It is not necessary, or even
desirable as far as I can see, but it has the downside of complicating
the calcs and increasing scope for errors when you build out with a
different line type.

Owen

coaxpair reflection coeff angle is 'zero'
 

On Thu, 29 Dec 2005 12:13:17 -0700, Wes Stewart
wrote:


I've provided a spreadsheet that facilitates the calculations over a
range of frequencies.

www.qsl.net/n7ws/8405.zip

So a couple of trips to the end of the cable are all that are required
to calibrate the setup.

(I must confess, I haven't tried this program with a line much over a
few inches in length to determine whether my calibration functions can
handle it, but I think so.)

That's a good idea!

Wes, does it collect enough information to be able to correctly
calculate a phase constant on longer feedline. Some work for the next
version perhaps?

Owen

--

coaxpair reflection coeff angle is 'zero'
 

Owen Duffy wrote:
On Thu, 29 Dec 2005 12:13:17 -0700, Wes Stewart
wrote:



I've provided a spreadsheet that facilitates the calculations over a
range of frequencies.

www.qsl.net/n7ws/8405.zip

So a couple of trips to the end of the cable are all that are required
to calibrate the setup.

(I must confess, I haven't tried this program with a line much over a
few inches in length to determine whether my calibration functions can
handle it, but I think so.)


That's a good idea!

Wes, does it collect enough information to be able to correctly
calculate a phase constant on longer feedline. Some work for the next
version perhaps?

A potential problem is cable loss. When the line Z0 is close to the
impedance being measured, loss doesn't have much effect. But if the two
impedances are very different, a surprisingly small amount of loss can
have a significant effect on the observed input impedance.

Of course, the short circuit measurement will give you the cable loss,
which can then be used in the calibration process. It's just that you
wouldn't be able to do the correction by the simple equivalent of a
Smith chart rotation.

Roy Lewallen, W7EL

coaxpair reflection coeff angle is 'zero'
 

On Thu, 29 Dec 2005 17:50:59 -0800, Roy Lewallen
wrote:

Owen Duffy wrote:
On Thu, 29 Dec 2005 12:13:17 -0700, Wes Stewart
wrote:



I've provided a spreadsheet that facilitates the calculations over a
range of frequencies.

www.qsl.net/n7ws/8405.zip

So a couple of trips to the end of the cable are all that are required
to calibrate the setup.

(I must confess, I haven't tried this program with a line much over a
few inches in length to determine whether my calibration functions can
handle it, but I think so.)


That's a good idea!

Wes, does it collect enough information to be able to correctly
calculate a phase constant on longer feedline. Some work for the next
version perhaps?

A potential problem is cable loss. When the line Z0 is close to the
impedance being measured, loss doesn't have much effect. But if the two
impedances are very different, a surprisingly small amount of loss can
have a significant effect on the observed input impedance.

Wes' procedure calibrates both loss and phase, but my suspicion is
that it does not calculate a correct phase constant for longer lines.
The loss constant is probably simple, assuming a straight line between
the two cal points, but that is probably adequate for the task given
the object being measured.


Of course, the short circuit measurement will give you the cable loss,
which can then be used in the calibration process. It's just that you
wouldn't be able to do the correction by the simple equivalent of a
Smith chart rotation.

Agreed, that was someone else's suggestion, and if the line on the
Smith Chart was a lossless arc rather than a lossy spiral, some more
error creeps in, and the error is larger as VSWR increases.

I knocked up a small spreadsheet solution myself, it uses the gamma
(with a small g) figure returned by my line loss calculator at a
frequency, and the line length to calculate the impedance
transformation. (Gee Excel is ugly with complex numbers.)

A more general solution would be one that calculates the fundamental
RLGC model from k1, k2, vf, and Zo, and can calculate the impedance
transformation as a function of Gamma and freq. I have a Perl library
that I use for such things, but it won't port to Excel very easily.
(If only Microsoft would extend Excel's capabilities instead of
renaming and relocating functions from version to version.)

It highlights the convenience of a direct reading impedance meter!
Still, I can see the advantages of the VVM over an impedance bridge,
and they are both in a different class to the MFJ.

Owen
--

coaxpair reflection coeff angle is 'zero'
 

On Thu, 29 Dec 2005 23:40:50 GMT, Owen Duffy wrote:

On Thu, 29 Dec 2005 12:13:17 -0700, Wes Stewart
wrote:


I've provided a spreadsheet that facilitates the calculations over a
range of frequencies.

www.qsl.net/n7ws/8405.zip

So a couple of trips to the end of the cable are all that are required
to calibrate the setup.

(I must confess, I haven't tried this program with a line much over a
few inches in length to determine whether my calibration functions can
handle it, but I think so.)

That's a good idea!

Wes, does it collect enough information to be able to correctly
calculate a phase constant on longer feedline. Some work for the next
version perhaps?

No next version. I've got an n2pk analyzer now. Much easier to use,
at least over its frequency range.

www.n2pk.com

coaxpair reflection coeff angle is 'zero'
 

On Fri, 30 Dec 2005 02:50:52 GMT, Owen Duffy wrote:

On Thu, 29 Dec 2005 17:50:59 -0800, Roy Lewallen
wrote:

Owen Duffy wrote:
On Thu, 29 Dec 2005 12:13:17 -0700, Wes Stewart
wrote:



I've provided a spreadsheet that facilitates the calculations over a
range of frequencies.

www.qsl.net/n7ws/8405.zip

So a couple of trips to the end of the cable are all that are required
to calibrate the setup.

(I must confess, I haven't tried this program with a line much over a
few inches in length to determine whether my calibration functions can
handle it, but I think so.)


That's a good idea!

Wes, does it collect enough information to be able to correctly
calculate a phase constant on longer feedline. Some work for the next
version perhaps?

A potential problem is cable loss. When the line Z0 is close to the
impedance being measured, loss doesn't have much effect. But if the two
impedances are very different, a surprisingly small amount of loss can
have a significant effect on the observed input impedance.

Wes' procedure calibrates both loss and phase, but my suspicion is
that it does not calculate a correct phase constant for longer lines.
The loss constant is probably simple, assuming a straight line between
the two cal points, but that is probably adequate for the task given
the object being measured.


Of course, the short circuit measurement will give you the cable loss,
which can then be used in the calibration process. It's just that you
wouldn't be able to do the correction by the simple equivalent of a
Smith chart rotation.

Agreed, that was someone else's suggestion, and if the line on the
Smith Chart was a lossless arc rather than a lossy spiral, some more
error creeps in, and the error is larger as VSWR increases.

I knocked up a small spreadsheet solution myself, it uses the gamma
(with a small g) figure returned by my line loss calculator at a
frequency, and the line length to calculate the impedance
transformation. (Gee Excel is ugly with complex numbers.)

A more general solution would be one that calculates the fundamental
RLGC model from k1, k2, vf, and Zo, and can calculate the impedance
transformation as a function of Gamma and freq. I have a Perl library
that I use for such things, but it won't port to Excel very easily.
(If only Microsoft would extend Excel's capabilities instead of
renaming and relocating functions from version to version.)

It highlights the convenience of a direct reading impedance meter!
Still, I can see the advantages of the VVM over an impedance bridge,
and they are both in a different class to the MFJ.

All good points and while not thinking too much about it (or at all
really) the original problem was measuring the parameters of a loaded
vertical. This sounds like a pretty narrowband problem. IF the line
isn't too many wavelengths long and the frequency sweep isn't too
wide, the phase rotation from one end of the frequency band to the
other should be much less than one trip around the Smith chart and the
loss variation should be minimal and for all intents linear.