"Jim Lux" wrote in message
...
Jimmie D wrote:
I assume you're not looking for tenth of a dB precision?

Jim


Actually yes I am..Power must be maintained +- 1db at the antenna.

You've got a bit of a challenge, then.. although +/- 1 dB (is that a 1
sigma or a 3 sigma or a absoulate max min spec?) might not require a tenth
of a dB precision.


1 dB is 25%
1% is 0.04dB

(measuring power at 1 GHz to 0.1dB absolute is moderately challenging,
especially outdoors) For reference, an Agilent E4418 is specified at
+/-0.6% (25C +/- 10 degrees).. plus you have a linearity spec which can
range from 1% to 4% depending on the relative levels of the reference and
unknown.

A good return loss measurement with a decent PNA (like an E8363) should
get you down in the sub 0.1dB transmission measurement with overall loss
in the 0 to 20dB range, so the measurement is clearly feasible at some
level.

The same piece of gear, measuring reflection coefficient (i.e. the put a
short or open at the other end, and measure mag(rho) and work back to
loss)... you said you have about 6dB return loss, so that's a reflection
coefficient (at the analyzer) of about 0.5, and for 2GHz, the uncertainty
would be about 0.01 (out of the 0.5), or, call it 2%... again, about a 0.1
dB uncertainty.

OTOH, that's a $50K piece of test gear, sitting in a lab at 25C +/- 1C

There's also the temperature coefficient of the coax to worry about.
Copper has a temperature coefficient of 0.4%/degree C. A 10 degree change
in temperature is a 4% change in resistance (0.2dB), and the resistance is
a big part of the loss (dielectric loss changes differently, and you'd
have to worry about the dimensional changes too).

In any case, measuring the loss by terminating it in a reflection is
probably the easiest way, and potentially the most precise, because you
can have the source and the measurement at the same location. If you
tried to measure it by transmission loss (put the source at one end and
the detector at the other) you have the problem of the stability of the
source. In a bridge type scheme (which the reflection technique is) you
can essentially compare between the unknown (your cable) and a standard,
and adjust the standard until they match, so the variations in the power
level of the source cancel out (or use something that inherently measures
the ratio of the powers).

Something like the LP-100 wattmeter can probably make the measurement.
It's good to 5% typical, and can do ratioed/match measurements to much
better. I don't know if it can go to 1 GHz, though.


Something like the Anritsu SiteMaster (like the S311D) can do this for
sure(after all, it's what it was designed to do.. measure coax on towers)
http://www.us.anritsu.com/downloads/...1410-00419.pdf

If you need to measure loss on the fly, it's a bit trickier, but one way
is to put a deliberate small mismatch at the end (i.e. you put a 10 dB
directional coupler in the line at the antenna end, with the coupled port
terminated into a short). This reflects a known -20dB back down the line.
You look for changes in the amount of reflected power. Obviously, if the
antenna changes it's reflection, you have to separate that out. There are
clever techniques for this too (like having the coupler terminate in a
switch that is either a load or a short). This kind of thing is pretty
common on antenna measurement ranges, where you need to remove the effects
of the feed cable from the measurement.




Jimmie


Sounds like using my network analyser to measure return loss at the TX in an
envoromentally stabalized building is going to be a lot better than taking
my HP power meter up on the antenna in the middle of the night to measure
the power level at the end of the cable.


Jimmie

"Owen Duffy" wrote in message
...
Jim Lux wrote in news:f9gg4i$7q9$1
@nntp1.jpl.nasa.gov:

...

Jim good points and all noted.

Jimmie hasn't give a lot of detail about the specification he is
apparently trying to meet. Reading between the lines, it might be an
EIRP, and assuming a given antenna gain, he is trying to calculate the
permitted transmitter power output.

Not only is the uncertainty of practical service equipment an issue in
tenth dB accuracy, but no mention has been made of transmission line loss
under mismatch conditions, and mismatch loss.

Jimmie, if you have a plausible story to tell the regulator, then that
might suffice.

If you have assessed the Return Loss of a rho=1 termination, then you
could use that and the measured Forward and Reverse power using say a
Bird 43 at the transmitter end of that known line loss (being half the
return loss) to calculate the power absorbed by the load. The calculator
at http://www.vk1od.net/tl/vswrc.php does just that. The calculator at
http://www.vk1od.net/tl/tllc.php could be used to calculate the expected
RL of the o/c or s/c line section, just specify a load impedance of 1e6
or 1e-6 for each case. For example, at 1GHz, the RL of 200' LDF4-50A with
a 1e-6 load is 8.9dB, and if you got much higher than that, you might
suspect the cable to be faulty.

Tenths of a dB, remember that most service type power meters are probably
good for 6% to 10% of FSD, so I will go with Jim's 1dB accuracy.

BTW, directional wattmeters for the ham market are often not capable of
reasonable accuracy on loads other than the nominal 50 ohm load. There
are a range of tests that such an instrument should satisfy, but for
hams, it is usually considered sufficient if the "reflected" reading is
approximately zero on a 50 ohm load.

Owen

I think I have given enough info. But I will try yo expess it in another
way.
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". Power checks are done on a weekly basis. Once a year the line
loss is measured and this value is used to subtract from the power at the
transmitter for the rest of the year. With this in mind it would be most
prudent to measure the cable loss accurately. to establish the annual
benchmark.

Considering the test equipment I have available to use in a temperature
stablized building an Agilent network analyzer or use an old HP power meter
at the top of the tower I am thinking that measuring rho of the cable while
terminated in a short may be the more accurate way to go.


Jimmie

"Jimmie D" wrote in
:

I think I have given enough info. But I will try yo expess it in
another way.
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". Power checks are done on a weekly basis. Once a year the
line loss is measured and this value is used to subtract from the
power at the transmitter for the rest of the year. With this in mind
it would be most prudent to measure the cable loss accurately. to
establish the annual benchmark.

Ok.

You haven't mentioned how you intend to deal with the likely case where
VSWR1.

Considering the test equipment I have available to use in a
temperature stablized building an Agilent network analyzer or use an
old HP power meter at the top of the tower I am thinking that
measuring rho of the cable while terminated in a short may be the more
accurate way to go.

Yes, especially if the NA is calibrated against o/c, s/c and Zo locally.

If for example, you discover that the one way matched line loss is
6.75/2dB (3.375dB), and you measure the fwd and reflected power at the tx
end to be say 200W and 15W, you could use the calculator I mentioned to
determine that the VSWR at the antenna was 1.36. Using that, and setting
forward power at the antenna to 102W for a net power to the antenna of
99.6W, forward and reflected at the tx end of the line should be 222W and
1.1W.

Of course, if the line was perfectly matched, you could just set the tx
end forward power to 100*10^(3.375/10) or 217W and reflected would be
zero... but that is unlikely. You could take the easier, simpler and
conservative way out and just set it for 217W forward irrespective of
mismatch.

It is splitting hairs, but sometimes precision in the method distracts
attention from accuracy!

Owen



using the calculator I mentioned, and setting the line loss to the tx to
3.375dB, loss to antenna to 0, tx power to

BTW, directional wattmeters for the ham market are often not capable of
reasonable accuracy on loads other than the nominal 50 ohm load. There
are a range of tests that such an instrument should satisfy, but for
hams, it is usually considered sufficient if the "reflected" reading is
approximately zero on a 50 ohm load.

I should think, though, that one could calibrate such a
reflectometer/directional wattmeter. That is, you could test it with a
suitable variety of source and load impedances and develop a fairly
simple arithmetic correction that would be accurate.

The interesting question might be whether you could unambiguously take a
particular fwd and rev reading and turn that into a true fwd and true
rev, essentially solving for the mismatch.

Down in the lab here at work we have a whole rack of precision
misterminations (1.1:1, 1.2:1, 1.5:1, etc.) that some talented engineer
built and calibrated some decades ago. They're built on the Maury
bluedot N terminations.



Owen

Sounds like using my network analyser to measure return loss at the TX in an
envoromentally stabalized building is going to be a lot better than taking
my HP power meter up on the antenna in the middle of the night to measure
the power level at the end of the cable.

you betcha..

But you still have the tempco of the cable to agonize about.

Jim Lux wrote in news:f9i1i3$8v5$1
@nntp1.jpl.nasa.gov:


BTW, directional wattmeters for the ham market are often not capable
of
reasonable accuracy on loads other than the nominal 50 ohm load. There
are a range of tests that such an instrument should satisfy, but for
hams, it is usually considered sufficient if the "reflected" reading
is
approximately zero on a 50 ohm load.

I should think, though, that one could calibrate such a
reflectometer/directional wattmeter. That is, you could test it with a
suitable variety of source and load impedances and develop a fairly
simple arithmetic correction that would be accurate.

Yes Jim, some of the deficiencies of the instrument fall to things like
an equal response from the separate forward and reverse couplers. Scale
shape is an issue (especially where the sensitivity is continuously
adjustable using a pot). Phase and amplitude response of the coupler over
the frequency range is another issue not so readily calibrated out. A
coupler that is long will underestimate rho, and some couplers insert
more mismatch than they pretend to measure.

In my experience, many of the instruments that are claimed to work up to
144MHz band might well indicate close to 1:1 on a dummy load, but they do
not indicate rho=1 on a s/c or o/c. Whilst they may serve their purpose
as a null indicator on a 50 ohm load, they are not suited to the loss
measurement such as Jimmie is performing.


The interesting question might be whether you could unambiguously take
a
particular fwd and rev reading and turn that into a true fwd and true
rev, essentially solving for the mismatch.

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.


Down in the lab here at work we have a whole rack of precision
misterminations (1.1:1, 1.2:1, 1.5:1, etc.) that some talented engineer
built and calibrated some decades ago. They're built on the Maury
bluedot N terminations.

I have always though that a budget priced set of mismatches would be real
handy, and have wondered why MFJ (or someone else for that matter) don't
offer a set for checking / calibration of the MFJ259B etc.

Owen

On Aug 9, 10:17 pm, "Jimmie D" wrote:
"Owen Duffy" wrote in message

...



Jim Lux wrote in news:f9gg4i$7q9$1
@nntp1.jpl.nasa.gov:

...

Jim good points and all noted.

Jimmie hasn't give a lot of detail about the specification he is
apparently trying to meet. Reading between the lines, it might be an
EIRP, and assuming a given antenna gain, he is trying to calculate the
permitted transmitter power output.

Not only is the uncertainty of practical service equipment an issue in
tenth dB accuracy, but no mention has been made of transmission line loss
under mismatch conditions, and mismatch loss.

Jimmie, if you have a plausible story to tell the regulator, then that
might suffice.

If you have assessed the Return Loss of a rho=1 termination, then you
could use that and the measured Forward and Reverse power using say a
Bird 43 at the transmitter end of that known line loss (being half the
return loss) to calculate the power absorbed by the load. The calculator
athttp://www.vk1od.net/tl/vswrc.phpdoes just that. The calculator at
http://www.vk1od.net/tl/tllc.phpcould be used to calculate the expected
RL of the o/c or s/c line section, just specify a load impedance of 1e6
or 1e-6 for each case. For example, at 1GHz, the RL of 200' LDF4-50A with
a 1e-6 load is 8.9dB, and if you got much higher than that, you might
suspect the cable to be faulty.

Tenths of a dB, remember that most service type power meters are probably
good for 6% to 10% of FSD, so I will go with Jim's 1dB accuracy.

BTW, directional wattmeters for the ham market are often not capable of
reasonable accuracy on loads other than the nominal 50 ohm load. There
are a range of tests that such an instrument should satisfy, but for
hams, it is usually considered sufficient if the "reflected" reading is
approximately zero on a 50 ohm load.

Owen

I think I have given enough info. But I will try yo expess it in another
way.
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". Power checks are done on a weekly basis. Once a year the line
loss is measured and this value is used to subtract from the power at the
transmitter for the rest of the year. With this in mind it would be most
prudent to measure the cable loss accurately. to establish the annual
benchmark.

Considering the test equipment I have available to use in a temperature
stablized building an Agilent network analyzer or use an old HP power meter
at the top of the tower I am thinking that measuring rho of the cable while
terminated in a short may be the more accurate way to go.

Jimmie

As I mentioned before, be sure the cable is really 50 ohm (assuming
your instruments are calibrated to 50 ohms), or at least determine
what it is. Make your rho measurement; at that length of line, you
can adjust the frequency of measurement over a small range and get
values for rho at angles of 0 degrees and at 180 degrees. I will
assume that the cable is 50 ohms and the cable attenuation changes
practically none between the two readings, so the readings will be the
same. Now without changing anything, measure an attenuator with
nearly the same attenuation your think the cable has, also open-
circuited/shorted at the output. If the attenuator has the same
attenuation as the line, you should get the same value. You can then
have that attenuator calibrated at 1GHz to make sure it's correct.
Because your measured attenuation is twice the line attenuation, you
will get the line within 1dB if the measurement is within 2dB. It
shouldn't be very expensive to get a couple attenuators that would
bracket the line loss, and have them calibrated, and expect that they
would hold the calibration for a relatively long time if they aren't
mistreated. Seems like we never see much variation from one cal to
the next of decent attenuators.

As Jim noted, beware of environmental changes. I don't think that
dimensional changes will much matter, but the copper resistance will,
some. The effect, though, is not nearly as much as Jim suggested,
because of skin effect: a 1 degree C change causes the DC resistance
to change by 0.4%, but the AC resistance changes by only 0.2%. Since
the dB attenuation due to copper resistance is linear with resistance,
if the line attenuation is about 3.5dB, you'd need a 10% change in AC
resistance to see an 0.35dB change in attenuation. That's a 50 degree
C change, perhaps worth worrying about if you're in an extreme
climate. Looking at it another way, it's about 0.007dB/degree C.
It's probably worth making a point to measure the line loss at or near
the temperature extremes it experiences, though that would mean
climbing the tower at a couple times you might least like to. Be sure
moisture doesn't get into the line!

Cheers,
Tom

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...

Cheers,
Tom

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