Ok here is my problem.

I would like to build an antenna for the S band. That is near 2500 MHz for
those of us that are not knowing of the letter to frequency correlation. :-)

The difficulty is in testing the antenna to make certain that it will accept
antenna power and not cause the transmitter grief while it is doing it. I
know that I can't go to Radio Shack and buy their SWR bridge to do this
testing. I have a friend that could help me in the 1200 MHz range but he
does not have a spectrum analyzer that will go to 2500 MHz.

He suggested the possibility of taking a signal generator and feeding the
antenna through a directional coupler. The return signal from the antenna
would then be sent to the spectrum analyzer so that it could be monitored
for a relative level and as the antenna was tuned closer to the proper
impedance and the return loss went up the spectrum analyzer would read a
lower and lower level.

Unfortunately we lack a directional coupler and a spectrum analyzer for
these frequencies.

I just got an idea. We still need to come up with a directional coupler
but if we can get a crystal detector that works up there we can monitor
the relative power with an oscilloscope or a DVM.

I have a few more people in the area that I can ask about the directional
coupler.

Does this look like a possible solution to anyone here? Or should I get
a piece of waveguide suitable for this frequency and a crystal detector
and calculate the VSWR by making the detector slide from one end of the
waveguide to the other. I remember in tech school doing this at microwave
frequencies to measure VSWR.

Any suggestions or ideas will be investigated to make this work and will
be greatly appreciated.

Thanks in advance.

Al Butler
ka0ies

Yes, it's quite possible to measure the return loss with a directional
coupler. Yes, if you're dealing with a microwatt or so, it's quite
possible to detect it with a simple diode detector. (Actually not
terribly difficult to get down below 10 nanowatts.) Yes, it's even
reasonable to make your own directional coupler. For example, if you
download and install the free "RFSim99", you'll find in the pulldown
menu under "toolscomponentcoupler" a little window that will show
you the dimensions needed in microstrip or stripline (I believe the
coupling shown assumes 1/4 wavelength of line...), or three other ways
to do it that are more practical for lower frequencies. If you were
really careful, you might even be able to mechanically cut a PC board
to make the microstrip version.

Cheers,
Tom


Allan Butler wrote in message news:lCSEb.601435$Tr4.1574972@attbi_s03...
Ok here is my problem.

I would like to build an antenna for the S band. That is near 2500 MHz for
those of us that are not knowing of the letter to frequency correlation. :-)

The difficulty is in testing the antenna to make certain that it will accept
antenna power and not cause the transmitter grief while it is doing it. I
know that I can't go to Radio Shack and buy their SWR bridge to do this
testing. I have a friend that could help me in the 1200 MHz range but he
does not have a spectrum analyzer that will go to 2500 MHz.

He suggested the possibility of taking a signal generator and feeding the
antenna through a directional coupler. The return signal from the antenna
would then be sent to the spectrum analyzer so that it could be monitored
for a relative level and as the antenna was tuned closer to the proper
impedance and the return loss went up the spectrum analyzer would read a
lower and lower level.

Unfortunately we lack a directional coupler and a spectrum analyzer for
these frequencies.

I just got an idea. We still need to come up with a directional coupler
but if we can get a crystal detector that works up there we can monitor
the relative power with an oscilloscope or a DVM.

I have a few more people in the area that I can ask about the directional
coupler.

Does this look like a possible solution to anyone here? Or should I get
a piece of waveguide suitable for this frequency and a crystal detector
and calculate the VSWR by making the detector slide from one end of the
waveguide to the other. I remember in tech school doing this at microwave
frequencies to measure VSWR.

Any suggestions or ideas will be investigated to make this work and will
be greatly appreciated.

Thanks in advance.

Al Butler
ka0ies

On Sat, 20 Dec 2003 07:11:14 GMT, Allan Butler
wrote:

Ok here is my problem.

I would like to build an antenna for the S band. That is near 2500 MHz for
those of us that are not knowing of the letter to frequency correlation. :-)

The difficulty is in testing the antenna to make certain that it will accept
antenna power and not cause the transmitter grief while it is doing it. I
know that I can't go to Radio Shack and buy their SWR bridge to do this
testing. I have a friend that could help me in the 1200 MHz range but he
does not have a spectrum analyzer that will go to 2500 MHz.

He suggested the possibility of taking a signal generator and feeding the
antenna through a directional coupler. The return signal from the antenna
would then be sent to the spectrum analyzer so that it could be monitored
for a relative level and as the antenna was tuned closer to the proper
impedance and the return loss went up the spectrum analyzer would read a
lower and lower level.

Unfortunately we lack a directional coupler and a spectrum analyzer for
these frequencies.

I just got an idea. We still need to come up with a directional coupler
but if we can get a crystal detector that works up there we can monitor
the relative power with an oscilloscope or a DVM.

I have a few more people in the area that I can ask about the directional
coupler.

Does this look like a possible solution to anyone here? Or should I get
a piece of waveguide suitable for this frequency and a crystal detector
and calculate the VSWR by making the detector slide from one end of the
waveguide to the other. I remember in tech school doing this at microwave
frequencies to measure VSWR.

Any suggestions or ideas will be investigated to make this work and will
be greatly appreciated.

Thanks in advance.

Al Butler
ka0ies

Hi Al,

You are going in the right direction. One problem (or two) is that
the Spectrum Analyzer is wholly unnecessary (overkill, especially when
it doesn't cover the frequency of interest). You have nearly answered
that with the crystal detector, unfortunately that is the second
problem.

A slotted line is the classic solution, which is both cheaper and more
effective. It allows you to control the variables and move in the
direction you are only vaguely outlining above (using a detector).
You may choose to go another path (directional couplers and what-not),
but the following issues remain to be acknowledged and answered:

A crystal detector is going to be wholly useless in measuring return
power, you would have to couple too much energy into the probe and
that would considerably upset the VSWR (or return loss) you were
attempting to measure. In short, it would turn out to be a massive
mismatch in its own right; basically you would be measuring the
problem you introduced instead of the load.

The correct method, and it is vastly more sensitive AND accurate, is
to use the crystal detector "lightly" coupled to the load (through a
slot or between the lines) to detect the standard 1 KHz tone that is
specifically used for this type of measurement (HP or Boonton have a
wide range of 1 KHz meters with range switches for exactly this
application). You then standardize the maximum reading along the
track of the line where the maxima occurs, shift the detector to the
minimum and read the return loss in dB directly from the meter and
range switch setting.

The ability to confirm you have correctly determined the VSWR (or
return loss) is by reducing the detector loading and taking another
measurement. If the determination has changed, you were too heavily
loaded in the former determination. You repeat this process until you
either run out of sensitivity, or you duplicate the last
determination.

There are, of course, third and fourth and... let's just agree there
will be problems: taking instrumentation for granted especially
directional couplers and slotted lines. These components at these
frequencies can easily appear to be working, and yet through slight
connector mating problems force considerable error into measurements.
It is easy to measure a value, but it takes more effort to determine
its accuracy. Connector problems and device mismatches (like the
naive use of the crystal detector to brute force measure power)
inevitably lead to gross errors that present seemingly pristine
determinations (if you simply stop at the first measurement). You
should ALWAYS replace your unknown with a standard and measure it.
The standard should also be not only a 50 Ohm load (and good luck
proving it is 50 Ohms), but also a pure short and a pure open.

One of the classic "gotchas" is source purity. If your source, at
that frequency, exhibits any distortion, you could likely be measuring
the return loss at a harmonic. Another classic problem that is widely
denied in this group (like death and taxes) is source mismatch. If
you have any pretense to accuracy (say measuring or hoping to measure
a VSWR less than 1.5:1) then your source MUST match the anticipated
LOAD Z (even if the actual load Z does not) and all must match the
transmission line Z.

A lot of these "gotchas" come with the illusion that once having made
a measurement (however accomplished) that its accuracy is somehow
validated by the instrument (the common crowing of $300 invested in
the Bird Wattmeter is a hoot, jacking that up to a $60000 Spectrum
Analyzer is even more mocking). It goes without saying that there is
a lot of slack allowed in actually pumping RF out into the Æther, but
the claims of accuracy of how much for such an itinerant
accomplishment are unwarranted.

73's
Richard Clark, KB7QHC

Thanks Richard;

I can understand the need to calibrate and verify everything as you said.

One of the things that popped up was your comment on making certain that the
source matches the intended load Z. When I worked on receivers in the past
we used to use what is called "hard microvolts". This was done by putting
a 3dB pad in the line between the receiver and the signal generator. That
way the impedance of the generator was more likely to be 50 ohms even though
the generator was specified to be 50 ohms.

I always remember that it was hard microvolts because it was harder to make
specifications on the radio with the attenuator in place.

Thanks for all the pointers. It is amazing that some people don't
understand that the equipment may be up to snuff, but the procedures are
blowing the results out of the ballpark.

When we used network analyzers we always tried to make the device we were
developing so it could plug into the face of the analyzer. If that was not
possible we would use cables made by the factory for that analyzer, clean
all the contact surfaces thouroughly, make certain that connectors were
torqued properly for the series and then do a full calibration at the end
of the cables so the analyzer would be able to give us results as good as
possible.

Unfortunately I don't have access to that type of equipment anymore
otherwise I would be using that.

Thanks again for the information.

Al Butler
ka0ies

On Sun, 21 Dec 2003 17:27:21 GMT, Allan Butler
wrote:

Thanks Richard;

I can understand the need to calibrate and verify everything as you said.

One of the things that popped up was your comment on making certain that the
source matches the intended load Z. When I worked on receivers in the past
we used to use what is called "hard microvolts". This was done by putting
a 3dB pad in the line between the receiver and the signal generator. That
way the impedance of the generator was more likely to be 50 ohms even though
the generator was specified to be 50 ohms.

I always remember that it was hard microvolts because it was harder to make
specifications on the radio with the attenuator in place.

Thanks for all the pointers. It is amazing that some people don't
understand that the equipment may be up to snuff, but the procedures are
blowing the results out of the ballpark.

When we used network analyzers we always tried to make the device we were
developing so it could plug into the face of the analyzer. If that was not
possible we would use cables made by the factory for that analyzer, clean
all the contact surfaces thouroughly, make certain that connectors were
torqued properly for the series and then do a full calibration at the end
of the cables so the analyzer would be able to give us results as good as
possible.

Unfortunately I don't have access to that type of equipment anymore
otherwise I would be using that.

Thanks again for the information.

Al Butler
ka0ies

Hi Al,

You are welcome. Visits to some of the Hamfests should reveal a lot
of microwave plumbing that could do all of this for you. Few who
traffic in that trade know the original cost of that stuff, and it
goes for literal pennies on the kilobuck. Then there are those who
are savvy enough to charge a healthy hunk of change - I would trust
them more for at least knowing how to preserve their investment.

Example: I recall seeing an HP Coaxial slotted line for sale at $20
and the fellow offering that the detector probe stripped from it would
make a neat depth gauge. How my heart sank....

As for your discussion about the 3 dB pad, its application is to serve
as isolation from source Z variation. It usually came in the form of
a T resistance network if I recall correctly.

Hammering out all these variables in the SHF lends perspective to what
matters, and how much, at a manageable scale that can be handled all
in one place on the bench (unlike spreading out over 10's of meters
for HF).

73's
Richard Clark, KB7QHC