K7ITM wrote in news:1186788470.852002.260460
@b79g2000hse.googlegroups.com:

On Aug 10, 2:28 pm, Owen Duffy wrote:
...
I don't think you can compensate for lack of f/b ratio in the coupler,
for example because the coupled lines are too long.
...
I'm curious what you mean by that, Owen...

Tom, I was thinking of several instruments, all of the coupled lines type
of construction, that on a s/c and o/c failed to indicate rho=1, and showed
similar readings when physically reversed, suggesting it was not just a fwd
/ rev matching issue, there was something about the coupler that was too
dependent on the location of the SWR pattern relative to the coupler. Since
they worked better at lower frequencies, the length of the coupler was
likely to be a contribution.

Owen

"Owen Duffy" wrote in message
...
K7ITM wrote in news:1186788470.852002.260460
@b79g2000hse.googlegroups.com:

On Aug 10, 2:28 pm, Owen Duffy wrote:
...
I don't think you can compensate for lack of f/b ratio in the coupler,
for example because the coupled lines are too long.
...
I'm curious what you mean by that, Owen...

Tom, I was thinking of several instruments, all of the coupled lines type
of construction, that on a s/c and o/c failed to indicate rho=1, and
showed
similar readings when physically reversed, suggesting it was not just a
fwd
/ rev matching issue, there was something about the coupler that was too
dependent on the location of the SWR pattern relative to the coupler.
Since
they worked better at lower frequencies, the length of the coupler was
likely to be a contribution.

Owen


Would this be a problem for a directional coupler designed for a specific
frequecy?


Jimmie

"Jimmie D" wrote in
:


"Owen Duffy" wrote in message
...
K7ITM wrote in news:1186788470.852002.260460
@b79g2000hse.googlegroups.com:

On Aug 10, 2:28 pm, Owen Duffy wrote:
...
I don't think you can compensate for lack of f/b ratio in the
coupler, for example because the coupled lines are too long.
...
I'm curious what you mean by that, Owen...

Tom, I was thinking of several instruments, all of the coupled lines
type of construction, that on a s/c and o/c failed to indicate rho=1,
and showed
similar readings when physically reversed, suggesting it was not just
a fwd
/ rev matching issue, there was something about the coupler that was
too dependent on the location of the SWR pattern relative to the
coupler. Since
they worked better at lower frequencies, the length of the coupler
was likely to be a contribution.

Owen


Would this be a problem for a directional coupler designed for a
specific frequecy?


Jimmie

Jimmie, I am talking about the el-cheap inline SWR / Power Meter that is
often sold to hams with unrealistic specs.

You can / should always test the performance of the kit you are using to
determine if you should have confidence in it.

There are a bund of notes on testing a directional wattmeter in the
article at http://www.vk1od.net/VSWR/VSWRMeter.htm .

BTW, for your purposes, if you had a Bird 43 with an element that read
upscale on fwd power (250W element for your application), it is all you
should need to form a reasonable estimate of line loss and set the
transmitter to deliver 100W to the antenna. You might need a smaller slug
to make a measurement of RL on a s/c or o/c termination.

Owen

"Owen Duffy" wrote in message
...
"Jimmie D" wrote in
:


"Owen Duffy" wrote in message
...
K7ITM wrote in news:1186788470.852002.260460
@b79g2000hse.googlegroups.com:

On Aug 10, 2:28 pm, Owen Duffy wrote:
...
I don't think you can compensate for lack of f/b ratio in the
coupler, for example because the coupled lines are too long.
...
I'm curious what you mean by that, Owen...

Tom, I was thinking of several instruments, all of the coupled lines
type of construction, that on a s/c and o/c failed to indicate rho=1,
and showed
similar readings when physically reversed, suggesting it was not just
a fwd
/ rev matching issue, there was something about the coupler that was
too dependent on the location of the SWR pattern relative to the
coupler. Since
they worked better at lower frequencies, the length of the coupler
was likely to be a contribution.

Owen


Would this be a problem for a directional coupler designed for a
specific frequecy?


Jimmie

Jimmie, I am talking about the el-cheap inline SWR / Power Meter that is
often sold to hams with unrealistic specs.

You can / should always test the performance of the kit you are using to
determine if you should have confidence in it.

There are a bund of notes on testing a directional wattmeter in the
article at http://www.vk1od.net/VSWR/VSWRMeter.htm .

BTW, for your purposes, if you had a Bird 43 with an element that read
upscale on fwd power (250W element for your application), it is all you
should need to form a reasonable estimate of line loss and set the
transmitter to deliver 100W to the antenna. You might need a smaller slug
to make a measurement of RL on a s/c or o/c termination.

Owen

Well it a done deal,
Engineering support came out last night and ran the checks for us while Im
on vacation and recovering from minor surgery, Yaaay. They did it the normal
way and by measuring the return loss and they decided the "return loss
method" worked better. Not sure what better means at this point. accurate
enough and easier and faster would constitute better.


Jimmie

On Sat, 11 Aug 2007 07:58:36 -0400, "Jimmie D"
wrote:

They did it the normal way

Hi Jimmie,

Given the long and winding road to this point, it would give me pause
that suddenly something became "normal." The remainder of your post
is in contradiction to your earlier statement:
On Thu, 9 Aug 2007 08:13:45 -0400, "Jimmie D" wrote:
The normal procedure for doing this is to inject a signal at one end and
measure the power out at the other.
For the sake of clarity (normality aside), what you originally
described Thursday is called "insertion loss."

On Fri, 10 Aug 2007 01:17:31 -0400, "Jimmie D" wrote:
Power delivered to the antenna but be maintained with in +- 1 db in this
case that power is 100 watts. Power is normally
checked at the TX and recorded after allowing for line loss as "power at
the antenna".
This again defines "insertion loss."

and by measuring the return loss and they decided the "return loss
method" worked better.

A description of the classic self-fulfilling prophecy.

I presume you mean this to be "the normal way," but it doesn't really
describe a method or procedure (a "way"); instead, it describes an
outcome. There are many "ways" to measure a characteristic called
"return loss." Some "ways" are more accurate than others.

Having introduced this term, "insertion loss," there remains one more
term to consider: "reflection loss." This and "return loss" can be
found scaled on the common form of the Smith Chart.

The distinction to these terms are that "return loss" and "reflection
loss" are a single port characteristic (that port being the "load"
which, of course, is NOT the antenna, but rather the line and the
antenna). "Insertion loss" is a two port characteristic that properly
conforms to your original question.

ALL such losses are defined by the system within which they reside.
This means you have to also characterize the impedances of BOTH the
load and the source. This last requirement is often dismissed in this
forum where the determination of the source's Z is frequently rejected
as being an impossibility (even when it is specified by the equipment
designer).

When Zsource = Zline = Zload, then many complexities are removed. I
have seen others ask you the characteristic Z of the load with no
response by you; and I am certain you have no comfortable assurance
about the Zsource of your transmitter. However, to this last, it
would be immaterial if Zline = Zload.

Not sure what better means at this point. accurate
enough and easier and faster would constitute better.

This, too, simplifies what is an exceedingly difficult determination
(of "return loss," "reflection loss," or "insertion loss") for the
accuracy you originally suggested. Accurate, easy, and fast are not
normally words used in conjunction except in advertising promotions.

The accuracy of any power determination is related to the known Z of
1. The load;
2. The source;
3. The detector.

At 1 GHz, these determinations are not so easily dismissed as trivial,
nor confirmed by dragging a $20,000 analyzer into the shop. The
analyzer answers the problem of knowing its own source Z, but it does
not answer what that source Z is of the transmitter (again, only a
necessity in the face of returned power).

Now, given no one has actually correlated accuracy to any metric here,
and given that accuracy is determined in large part by the three Zs
above; then a little more discussion is in order. Using only two (the
detector and the load could be interchanged for the simpler analysis):
Zsource = 100 Ohms
Zload = 33.3 Ohms

view in fixed font:

1 - Gammaload²
Error = ------------------------------
(1 ± Gammasource · Gammaload)²

Error = +0.42dB to -0.78 dB

These errors are independant of other errors such as instrumentation
error (meter linearity, conversion problems, ...) or operator errors
(reading the meter - a mirrored scale is required to keep this below
5%). Modern instrumentation (if you have the $$$$) solves some of
this, others dismiss it as a trivial concern and rely on name brand
(Bird is frequently uttered to achieve perfection).

Now, as to the variability in the error wholly associated with just
the Zs (providing you can accurately determine them - yes, a game of
infinite regress). The allowable error of 1dB is nearly wiped out
with some very possible characteristics and you haven't even begun
balancing the error budget. With luck (a fictional village where
every armchair technician resides) the error induced by mismatches
could be 0. That luck demands you know the length of the line (again,
with some accuracy - I enjoy the irony here too). The variation built
into the Error computation is from not knowing that length (as is
common, few know this with enough precision in wavelengths). At 1
Ghz, the characteristic of
aproximately 200ft of coax
is apocryphal.

73's
Richard Clark, KB7QHC

On Aug 10, 5:29 pm, Owen Duffy wrote:
K7ITM wrote in news:1186788470.852002.260460
@b79g2000hse.googlegroups.com:

On Aug 10, 2:28 pm, Owen Duffy wrote:
...
I don't think you can compensate for lack of f/b ratio in the coupler,
for example because the coupled lines are too long.
...
I'm curious what you mean by that, Owen...

Tom, I was thinking of several instruments, all of the coupled lines type
of construction, that on a s/c and o/c failed to indicate rho=1, and showed
similar readings when physically reversed, suggesting it was not just a fwd
/ rev matching issue, there was something about the coupler that was too
dependent on the location of the SWR pattern relative to the coupler. Since
they worked better at lower frequencies, the length of the coupler was
likely to be a contribution.

Owen

Hi Owen,

I've recently done at least a cursory study of the coupled-line
hybrid, and I found nothing to indicate that directionality is
affected by the line length. In fact, the usual length where it's
practical is 1/4 wave, since that's the length that provides maximum
coupling, and the coupling near that frequency changes only gently
with changes in frequency (falling off on either side). I was
particularly interested in finding that the directionality is
independent of the length, assuming uniform cross-section at least.
If this is in error, I'd really like to know about it, because it
affects something I'm working on.

I'm not sure exactly what sort of bridge is used in microwave network
analyzers; I do know that the ones we build out to a few hundred MHz
use resistive bridges, which are relatively frequency insensitive. (A
key trick is how to read the bridge imbalance without introducing
errors...)

Cheers,
Tom

K7ITM wrote in
ups.com:

....
If this is in error, I'd really like to know about it, because it
affects something I'm working on.

Interesting findings Tom.

The way I think of these couplers is that you are trying to sample V and
I at a point on the main line, and a longish coupler of that type departs
from that ideal.

The effect I observed, and in several instruments, was obvious and
repeatable. I wonder that if the length of the lines is not the cause, if
it was the untidiness of the way in which the detector circuit was
implemented at each end of the coupler section. Of relevance also, is
that insertion of the instruments also caused significant SWR (1.2 in
the case of one of them) at the extreme uppoer end of their specified
range. IIRC two of the instruments had no equalisation / compensation,
they had a resistor at one end of the coupled line and a cap/diode at the
other end.

I still have one of the things that did this, and I have since nulled it
for 75 ohms, but I will have a play with it when I get home next week.

Owen

K7ITM wrote in
ups.com:

....

Tom, for avoidance of doubt, I am not talking about the type of directional
coupler that uses a couple line and that you would terminate with matching
load. I am talking about the cheap VSWR meters that have about 100mm long
coupled line, that is quite tightly coupled, and the resistor at one end of
the line is adjusted to balance the electric field sample with the magnetic
field sample for a null reading with V/I=Zn.

Owen

On Aug 11, 1:37 am, Owen Duffy wrote:
K7ITM wrote roups.com:

...

Tom, for avoidance of doubt, I am not talking about the type of directional
coupler that uses a couple line and that you would terminate with matching
load. I am talking about the cheap VSWR meters that have about 100mm long
coupled line, that is quite tightly coupled, and the resistor at one end of
the line is adjusted to balance the electric field sample with the magnetic
field sample for a null reading with V/I=Zn.

Owen

Hi Owen,

I'm not sure I see the difference. The load on the cheapie you
describe is just the load required to terminate that line. I have a
freely redistributable field solver program that will calculate the
even and odd mode impedances for you from the geometry and the
dielectric's permittivity, and from those impedances and the length
you can predict the proper termination impedance of both the "through"
and the "coupled" lines, and the coupling at any particular
frequency. It IS a problem if you try to do it in microstrip because
the propagation velocity for the even and odd modes is different, but
in true TEM configurations, I believe the directionality is fine if
you maintain uniform cross-section. Actually, a way that they make
broadband coupled lines is to have a central section tightly coupled,
and another section on each end of that which is less tightly
coupled. You can extend it to 5 sections or more, to get even broader
bandwidth.

Info about them is out there, but it wasn't as easy for me to find as
I figured it would be. ;-)

Cheers,
Tom

I haven't read all the posts in this thread, so I'm not sure that the method I'm advancing for measuring line
loss has not already been discussed.

The method I'm suggesting measures the input impedance (R and jX) of the line in question with the opposite
end of the line terminated first with an open circuit and then with a short, at a frequency where the line
will be close to lambda/8. With low-loss lines the R component will be very small, requiring an RF impedance
bridge that can produce accurate values of R in the 0.2 to 2-ohm range. (The General Radio GR-1606A is a
typical example, which, while using this procedure, will yield answers of greater accuracy than the methods
described in the current posts.) With the 1/8wl line the + and - reactances appearing in the measurements will
be approximately the value of the line Zo.

The open and short circuit values are then plugged into a BASIC program that I wrote years ago, which appears
in Chapter 15 of both Reflections 1 and 2. It also appears on my web page at www.w2du.com. The program outputs
the line attenuation, the complex Zo, and the electrical length of the line. The program solves the equations
appearing in Chipman's "Theory and Problems of Transmission Lines," Page 135.

On my web page go to 'View Chapters of Reflections 2' and click on Chapter 15 to see the detailed explanation
of the procedure.

The BASIC program TRANSCON (Transmission-line Constants) is listed there, but to save your having to load the
program from the typed list I will email a copy of the actual operable program to anyone who requests it,
addressing your request to .

I hope this suggestion will prove to be of value.

Walt, W2DU