On 11/8/2011 7:29 AM, Owen Duffy wrote:
John wrote in :

I have about 94 inches of RG-142B/U. I am using a Fluke 6061A signal
generator, an HP 8405A Vector Voltmeter, and a Narda dual directional
coupler. I have tried to measure the line characteristics at 434 MHz
but I am not satisfied that the results are accurate. It is very
difficult to get good short and open circuits at this frequency and I
also wonder if the 8405A accuracy suffers since a short is well away
from the nominal system impedance of 50 ohms.

What if I simply calibrate the 8405 with a short on the end of the
line (the measurement plane) then attach my antenna and accept the
readings? Will they be very far from the real value?

I am a little confused about your objective. The subject line seems
inconsistent with your discussion.

If you are trying to measure Z at the reference plane in the simplest
manner, then what you propose in your last par should give you the
magnitude and phase of the reflection relative to a s/c (where
Gamma=-1).

This simple paired measurement of the reflected wave from a s/c and
unknown load depends on the forward wave being constant. That is true if
the Thevenin source impedance of the source at the coupler is equal to
the nominal characteristic impedance of the coupler and the coupler
terminations, cables etc that you use. That would usually be met by a
standard signal generator etc, but some SSGs depart from ideal on their
highest output settings (check the specs). Measuring the forward wave
under significantly different loads will provide an indication as to
whether you can assume that it remains constant with different loads.

I give an explanation of why Vf is constant when Zs=Zo at
http://vk1od.net/transmissionline/VSWR/Zs50.htm .

You will recall that there is an ongoing argument that a ham transmitter
is well represented as a Thevenin source with Zs=50+j0 ohms, as some
accident of design. The article describes a simple test using an
accurate directional wattmeter to demonstrate that under different drive
level and different frequencies, that Vf is often not necessarily
independent of load impedance and that calculations that depend on
constant Vf (such a Mismatch Loss) are in error.

Owen

I think I am the confused one. Do I even need to know the transmission
line characteristics if I am going to short the load end and set the
vector voltmeter for a phase reference of 180 degrees?

I am following the HP app note AN-77 and they do not mention a
transmission line. They say to short the load end of the coupler. I need
to get my antenna away from the test setup, so I add the transmission line.

Has this made any sense?

John

On 11/8/2011 2:13 PM, John S wrote:
I think I am the confused one. Do I even need to know the transmission
line characteristics if I am going to short the load end and set the
vector voltmeter for a phase reference of 180 degrees?

I am following the HP app note AN-77 and they do not mention a
transmission line. They say to short the load end of the coupler. I need
to get my antenna away from the test setup, so I add the transmission line.

The problem is that these measurements are designed to be made
as close to the instrument as possible. Adding a piece of transmission
line adds loss and phase shift to the measurement.

So unless you know how to work backwards from the measure you get to
what you're really measuring at the far end, you won't really have a
valid answer.

Jeff


--
"Everything from Crackers to Coffins"

On 11/8/2011 2:37 PM, Jeffrey Angus wrote:
On 11/8/2011 2:13 PM, John S wrote:
I think I am the confused one. Do I even need to know the transmission
line characteristics if I am going to short the load end and set the
vector voltmeter for a phase reference of 180 degrees?

I am following the HP app note AN-77 and they do not mention a
transmission line. They say to short the load end of the coupler. I need
to get my antenna away from the test setup, so I add the transmission
line.

The problem is that these measurements are designed to be made
as close to the instrument as possible. Adding a piece of transmission
line adds loss and phase shift to the measurement.

So unless you know how to work backwards from the measure you get to
what you're really measuring at the far end, you won't really have a
valid answer.

Jeff

Hi, Jeff -

Yeah, that's what I'm trying to learn by my questions.

Obviously, the antenna can't be sitting in front of the test equipment
so I'm trying to find a way to minimize proximity effects.

I should have used a different subject for the post, I think.

John

John S wrote in :

On 11/8/2011 2:37 PM, Jeffrey Angus wrote:
....
The problem is that these measurements are designed to be made
as close to the instrument as possible. Adding a piece of
transmission line adds loss and phase shift to the measurement.

The problem is that whilst putting the unknown right at the instrument
might solve some problems, it creates others, particularly when the
unknown is an antenna system.


So unless you know how to work backwards from the measure you get to
what you're really measuring at the far end, you won't really have a
valid answer.

Well, you can work backwards as I explained, though again at the expense
of some uncertainty (error).

....
Obviously, the antenna can't be sitting in front of the test equipment
so I'm trying to find a way to minimize proximity effects.

Don't overlook that the reference plane need not necessarily at the end
of the directional coupler, it could be anywhere that you can
conveniently or inconveniently apply the calibration s/c (eg at the
antenna connector for flange as appropriate).

Try this on the bench with a substantial length of coax to the reference
plane, and measure a 25 ohm load (pair of 50 ohm terms on a Tee). The
reference plane is determined by where you apply the calibration s/c.

It was common practice using slotted lines to make such measurements
with the slotted line a long way from the reference point, which might
be quite close to the antenna, possibly a s/c applied at the antenna
flange, or on a W/G switch near the antenna.

BTW, you have read the AN, and noted no doubt the issue with error due
to the probes loading the test circuit. I would not obsess too much over
perfection unless you are prepared to go to great lengths to try and
allocate the errors and obtain a better result.

Owen

On 11/8/2011 3:58 PM, Owen Duffy wrote:
John wrote in :

On 11/8/2011 2:37 PM, Jeffrey Angus wrote:
...
The problem is that these measurements are designed to be made
as close to the instrument as possible. Adding a piece of
transmission line adds loss and phase shift to the measurement.

The problem is that whilst putting the unknown right at the instrument
might solve some problems, it creates others, particularly when the
unknown is an antenna system.


So unless you know how to work backwards from the measure you get to
what you're really measuring at the far end, you won't really have a
valid answer.

Well, you can work backwards as I explained, though again at the expense
of some uncertainty (error).

...
Obviously, the antenna can't be sitting in front of the test equipment
so I'm trying to find a way to minimize proximity effects.

Don't overlook that the reference plane need not necessarily at the end
of the directional coupler, it could be anywhere that you can
conveniently or inconveniently apply the calibration s/c (eg at the
antenna connector for flange as appropriate).

AHA! I think this is the answer I was needing. So, if I short the end of
the coax and calibrate my vector voltmeter I can then believe the
voltmeter when it says my load is a certain impedance?

I don't know were I got the idea that I had to have the coax
characteristics to be used to modify the readings for accuracy.

Try this on the bench with a substantial length of coax to the reference
plane, and measure a 25 ohm load (pair of 50 ohm terms on a Tee). The
reference plane is determined by where you apply the calibration s/c.

Okay. I'll do that as soon as I can find the instruments. They are
somewhere around here.

It was common practice using slotted lines to make such measurements
with the slotted line a long way from the reference point, which might
be quite close to the antenna, possibly a s/c applied at the antenna
flange, or on a W/G switch near the antenna.

BTW, you have read the AN, and noted no doubt the issue with error due
to the probes loading the test circuit. I would not obsess too much over
perfection unless you are prepared to go to great lengths to try and
allocate the errors and obtain a better result.

Owen

I don't obsess. I'm usually happy with errors of less than 1.5 to 1
(depending on circumstances, of course).

Thanks, Owen.

John

John S wrote in :

....

If you are not concerned with trying to calibrate out the directivity of
the coupler (and if that is greater than the expected / tolerable Return
Loss, you don't need to do so), and you have convinced yourself that Vf
is independent of load impedance (as it will be if Zs=50+j0 and you use
short low loss line, or a large attenuator at the coupler to control
Zs), then the simple approach is to do the following.

I think I am the confused one. Do I even need to know the transmission
line characteristics if I am going to short the load end and set the
vector voltmeter for a phase reference of 180 degrees?

I am following the HP app note AN-77 and they do not mention a
transmission line. They say to short the load end of the coupler. I
need to get my antenna away from the test setup, so I add the
transmission line.

And you understand that the Gamma found is at the reference plane (the
plane of the calibrating s/c), and you can adjust it, or the calculated
impedance to another point on a known feedline using the well known
Telegrapher's Equation (http://www.vk1od.net/calc/tl/tllc.php solves
this problem for a range of popular lines), albeit subject to error due
to uncertainty about the known line.

(I did consider at one stage extending TLLC to allow specification of
mismatch in terms of Gamma, rectangular and polar, but no one ever asked
for it and I thought it not in demand. The complication is that finding
Z from Gamma needs to use the nominal Zo of the test equipment, not the
actual Zo of the lossy transmission line. I usually use a spreadsheet to
perform the calcs, Excel can handle complex numbers using the COMPLEX
and IM* functions either in the Analysis Tookpak in earlier versions, or
built in to the later versions.)

An important thing to keep in mind is that while the measurements you
make are of the TL in differential mode, it may be carrying significiant
common mode components which will affect the differential currents. In
making your measurements, if you change the common mode current path
from the normal system configuration, you are measuring a different
system and the results might not apply. There seems an unwarranted
assumption in most discussion of such measurement projects that there is
inisignificant common mode current.

Has this made any sense?

Perhaps it is my turn to ask.

Owen

Owen Duffy wrote in news:Xns9F98507D98D82nonenowhere@
88.198.244.100:

....
(I did consider at one stage extending TLLC to allow specification of
mismatch in terms of Gamma, rectangular and polar, but no one ever asked
for it and I thought it not in demand. The complication is that finding
Z from Gamma needs to use the nominal Zo of the test equipment, not the
actual Zo of the lossy transmission line. I usually use a spreadsheet to
perform the calcs, Excel can handle complex numbers using the COMPLEX
and IM* functions either in the Analysis Tookpak in earlier versions, or
built in to the later versions.)

Agilen't AppCad can be a convenient tool for one-off calcs.

Nevertheless, there is value in see the results of measurement as you make
them, it helps to minimise the chance of leaving the task with errored
data.

Owen

On 11/8/2011 2:54 PM, Owen Duffy wrote:
John wrote in :

...

If you are not concerned with trying to calibrate out the directivity of
the coupler (and if that is greater than the expected / tolerable Return
Loss, you don't need to do so), and you have convinced yourself that Vf
is independent of load impedance (as it will be if Zs=50+j0 and you use
short low loss line, or a large attenuator at the coupler to control
Zs), then the simple approach is to do the following.

I have convinced myself of nothing. Hence, my questions here.

I measured the coupler several years ago:

Narda Dual Directional Coupler...

435 MHz

Forward direction:
Coupling = -33.1 dB
Directivity = -56.8 dB

reverse direction:
Coupling = -33.5 dB
Directivity = -74.7 dB

The manual for the Fluke generator says it is 50 ohms output Z. I used
your Line Loss Calculator to find that my 2.4m coax is 50-j0.1 and
0.671dB loss.

Nice numbers, but I don't know what to do with them.

And you understand that the Gamma found is at the reference plane (the
plane of the calibrating s/c), and you can adjust it, or the calculated
impedance to another point on a known feedline using the well known
Telegrapher's Equation (http://www.vk1od.net/calc/tl/tllc.php solves
this problem for a range of popular lines), albeit subject to error due
to uncertainty about the known line.

I understand very little.

The Fluke gen feeds the Narda dual directional coupler. The coupler
output has the 94 inches of RG-142B/U attached. The coax has all the
ferrite cores in my possession slipped onto it to moderate common mode
current. The vector voltmeter A input is attached to the forward coupler
sampling port and the voltmeter B input is attached to the reverse
coupler sampling port.

I put the best short circuit I can muster on the far end of the coax and
set the vector voltmeter to read 180 degrees. I record the A and B
voltage inputs.

(I did consider at one stage extending TLLC to allow specification of
mismatch in terms of Gamma, rectangular and polar, but no one ever asked
for it and I thought it not in demand. The complication is that finding
Z from Gamma needs to use the nominal Zo of the test equipment, not the
actual Zo of the lossy transmission line. I usually use a spreadsheet to
perform the calcs, Excel can handle complex numbers using the COMPLEX
and IM* functions either in the Analysis Tookpak in earlier versions, or
built in to the later versions.)

I put the antenna on the far end of the coax. I read the A and B
voltages and the angle between them. I use Excel to calculate the
results as indicated in the HP AN-77 app note.

An important thing to keep in mind is that while the measurements you
make are of the TL in differential mode, it may be carrying significiant
common mode components which will affect the differential currents. In
making your measurements, if you change the common mode current path
from the normal system configuration, you are measuring a different
system and the results might not apply. There seems an unwarranted
assumption in most discussion of such measurement projects that there is
inisignificant common mode current.

Has this made any sense?

Perhaps it is my turn to ask.

Owen

I am aware of the common mode current problems and I do everything I can
to minimize them. I don't know if I am successful, but I test the
effectiveness of my efforts by running my hand up and down the coax and
watching the voltmeter. I haven't been able to get all variation out,
but some of the variation is due to my hand proximity to the antenna itself.

Back to you.

John

John S wrote in :

....

The manual for the Fluke generator says it is 50 ohms output Z. I used

Ok, be aware that sometimes that impedance is gauranteed to less than
full output.

If it is 50+j0, then your Vf reading should not change in magnitude with
load variation. If it does change, you have to factor it into the calcs
as in the AN.

your Line Loss Calculator to find that my 2.4m coax is 50-j0.1 and
0.671dB loss.

Since it seems you are applying the cal short at the load end of that
line section, then its loss is not so important (so long as it is
stable).


Nice numbers, but I don't know what to do with them.

....
The Fluke gen feeds the Narda dual directional coupler. The coupler
output has the 94 inches of RG-142B/U attached. The coax has all the
ferrite cores in my possession slipped onto it to moderate common mode
current. The vector voltmeter A input is attached to the forward
coupler sampling port and the voltmeter B input is attached to the
reverse coupler sampling port.

I put the best short circuit I can muster on the far end of the coax
and set the vector voltmeter to read 180 degrees. I record the A and B
voltage inputs.

Ok, that sounds fine. If find it helps to measure something that is
known. Do you have a pair of 50 ohm terms and a T that you can measure
and convince yourself that it is working.


I put the antenna on the far end of the coax. I read the A and B
voltages and the angle between them. I use Excel to calculate the
results as indicated in the HP AN-77 app note.

Sounds fine.


I am aware of the common mode current problems and I do everything I
can to minimize them. I don't know if I am successful, but I test the
effectiveness of my efforts by running my hand up and down the coax
and watching the voltmeter. I haven't been able to get all variation
out, but some of the variation is due to my hand proximity to the
antenna itself.

Ok, but it remains a potential problem.

Owen

On 11/8/2011 4:15 PM, Owen Duffy wrote:
John wrote in :

...

The manual for the Fluke generator says it is 50 ohms output Z. I used

Ok, be aware that sometimes that impedance is gauranteed to less than
full output.

If it is 50+j0, then your Vf reading should not change in magnitude with
load variation. If it does change, you have to factor it into the calcs
as in the AN.

your Line Loss Calculator to find that my 2.4m coax is 50-j0.1 and
0.671dB loss.

Since it seems you are applying the cal short at the load end of that
line section, then its loss is not so important (so long as it is
stable).

Okay! This is another answer for which I was hoping. This simplifies
everything.

Thanks again, Owen.

John