"Steve Nosko" wrote in message ...

Side bar: I have one of the first directional power meters, the Micro
match. It uses a resistor, yes, resistor, for the current sense rather than
the now common current transformer toroid. circa 1945 -50.

Not really all that uncommon. A Wheatstone bridge works fine for
monitoring SWR, and transformer-coupled versions of essentially a
Wheatstone bridge are commonly used in S-parameter test sets used with
network analyzers. Nice for flat response over lots of octaves, if
you are verrrry careful about the construction.


You might think of microstrip or stripline as "hammered flat coax."
:-)

The accepted terminology is "Stripline" (think "strip transmission
line") for the line with two flat ground planes on either side of the
"center conductor" which can be thought of as this flattened coax....and
"micro strip" ( I have no memory aid) for the one that is one-sided where
the "center conductor" is on the top of a PCB with a ground plane on the
bottom..

(And though stripline is true TEM, microstrip is quasi-TEM because of
the different dielectric above and below the microstrip.)

It is just a transmission line on a PCB, the characteristic impedance
dependent on trace line width, thickness of the foil, dielectric
constant
of the PCB material and, to some extent, the thickness of the PCB.

I think the PCB material (thickness & properties) has a larger
contribution to the Zo that the runner thickness...at least at the freqs I
worked at (1GHz.). Good ole' Wheeler & Sobol equations for the Zo.

Yep, by far. The free "RFSim99" program will help with the dimensions
for five different kinds of couplers, you pick the coupling, including
stripline and microstrip. The one thing it fails to mention is the
1/4 wave line length for the given coupling.


Directional couplers are simply a quarter wavelength of transmission
line (coax or microstrip or stripline) that runs parallel to the main
line
connecting to the antenna.

I believe a 1/4 wave is not important here.

Right...but beware of NULLS in the response at integer multiples of
1/2 wavelength, and the response returns to the same value at 3/4 wave
as it was at 1/4 wave. They are broadly rounded peaks, as Len says
useful over perhaps an octave for non-precision power measurements.
This can be used to advantage for a 146-440MHz coupler: make it 1/4
wave at 146, and it will respond fine at 3/4 wave at 440. But 1/4
wave (even with the 0.5 VF of glass-epoxy circuit board mtrl) is a bit
long. But...you can fold that in a "U" and get it down to ten inches
or so board length.

For broader bandwidth, it's common to make the stripline or microstrip
in three (or five, or...) sections with varying coupling. The center
section is highest coupling, and the outer sections are lower. You
can find design info on the web about this. The broadening is quite
significant.

....
Typical directional coupler coupling is 20 db down from the main line.

This is a matter of choice by the manufacturer construction. What this
means is that the power coming out the (correct end of the) secondary line
is 20dB down (1/100 th) from the power on the main line.

In fact, -20dB with 100 watts input gives you a WATT at the "FWD"
output, which is WAY more than I want to deal with! Simple diode
detectors let me "see" down to -50dBm, so even a milliwatt is really
more than I need. Even with 10mW of excitation, a -30dB coupler is
quite useful. DESIGN the coupler to give you what you want at the
output!


While not perfect, directional
coupling differences of about 20 db are good enough to warrant the
name "directional."

This is a matter of primarily the mechanical design.

AND the loading on the "other" end of the coupled line. Steve touched
on this, but I think it deserves more emphasis. Consider: if you have
a -20dB coupler and 100 watts forward and zero return on the "through"
line, you have a watt trying to come out the "forward" coupled port.
If you don't terminate that perfectly, some of that watt goes back the
other way, and spoils the directionality. SO...if you screwed up the
trace width (perhaps because you didn't know the dielectric constant
or thickness of the board material exactly, or your etching wasn't
perfect), you can still adjust for very nearly perfect directivity by
adjusting the load on the coupled line. This is also why it's good,
especially at microwaves, to NOT try to use two detectors on opposite
ends of the same coupled line, because the detectors are more
difficult to have be really good resistances and not introduce
parasitic reactance. One way to make the load adjustable (assuming
low enough coupled power that you can use SMT resistors) would be to
provide pads for multiple termination resistors (two or three) and use
a parallel combination that gives you just the right termination.
This assumes you have a really good load (including any line and
fittings) you can put on the through-line to insure near-zero
reflections there! Calibration-quality loads for GHz frequencies are
NOT cheap, and garden-variety coax is practically guaranteed to NOT be
50 ohms exactly. (45-55 is about the best reasonable expectation you
should have for RG-58.)

....

Cheers,
Tom