TO: Ken, KC6TEU, and the Microwave Group 04/27/02 From: Dick, K2RIW RE: Directional Couplers, and a VSWR/Power Measurement Procedures.
DIRECTIONAL COUPLERS USED FOR VSWR AND
INTRODUCTION -- Over the years I have heard many engineers, and some smart amateurs, express opinions that reflect a considerable misunderstanding about the operation of Directional Couplers, and how to properly use them in the measurement of Voltage Standing Wave Ratio (VSWR), and power. This memo is intended to give some basic information that may help.
At first, the average electronic technologist is
mystified by at least two of the concepts of how RF behaves within
transmission line structures:
I believe that both of
these principles must be absorbed (and understood), if meaningful DC
measurements are to be properly executed, and believed. Here are my
recommended procedures, with some partial explanations of what is taking
place at each step.
DIRECTIONAL COUPLER USED IN A VSWR OR POWER MEASUREMENT PROCEDURE
(I) DIRECTIONAL COUPLER
CALIBRATION -- The first step in this procedure is to establish the
quality of the Directional Coupler (DC) that you are about to use. I
don't care if the label on the DC says it is a "Cadillac" or "Rolls Royce"
brand, and the calibration sticker says it is traceable to "The Bureau of
Standards" with and accuracy of 0.01 dB; you still have to confirm that it
is good working order right NOW. It is possible that the DC was thrown
onto a concrete floor yesterday, and the internal termination may have
been shattered. If that had happened, it could loose almost all of it's
directional characteristics -- it's "Directivity."
requirement is similar to the proper use of an Ohm Meter. Notice that a
good technologist will always short the two leads together; and the Ohm
meter had better read a small fraction of an ohm, before the technologist
will proceed with the next measurement.
Similarly, a prudent
technologist will measure the Directivity of the DC he is about to use.
It is also useful to know that sometimes the DC can be used far outside
the frequency range it was designed for, as long as the principle of
operation is somewhat understood, and a calibration at the present
frequency is performed. Here is the Directivity Confirmation procedure.
-- Unfortunately, the Directivity Confirmation procedure requires a known
good termination (dummy load), and the procedure will have an accuracy
that rarely is much better than the quality of that termination being
used. First apply an RF signal to the DC "input" port, with a known good
termination connected to the "output" port. Position the DC so that it
favors the Forward flowing signal. Place a power-measuring device at the
directional port. This can be a Power Meter, Spectrum Analyzer,
calibrated Crystal Detector, Scalar Network Analyzer, etc. Measure (and
record) the DC's response to the forward-flowing signal (in dBm units).
If, for instance, you are using a known Directional Coupler (DC) with a
-10 dB Coupling Coefficient, the measured power should be nearly 10 dB
weaker that the power that's being applied from the signal generator. By
the way, "dBm" means Decibels of signal strength with reference to a 1
Next, reverse the DC
"input" and "output" ports, and repeat (and record) the previous
measurement. The difference in the two readings indicates the
Directivity. For instance; if a 0.0 dBm signal generator is applied to a
10 dB coupler, and it measured -10 dBm during the Forward Measurement,
and -30 dBm during the Reverse measurement, that would indicate a
Directivity of 20 dB (the difference in the readings). A DC of "Good"
quality will show a directivity of 20 dB, that is, the apparent reflection
from the termination will appear to be -20 db (an apparent VSWR of
1.22:1), even if the termination is a perfect 50 ohm resistance at the
present frequency. An "Excellent" DC will show a Directivity of 30 dB (an
apparent VSWR of 1.065:1), and there are Instrumentation-type DC's that
can display a Directivity of over 50 dB (an apparent VSWR of 1.006:1).
More on this later; there are ways of improving your DC's Directivity.
Simplistically, you could
say that a DC that displays a Directivity of 20 dB will not be able to
easily resolve the Reflection Coefficient from an unknown load of better
than about -20 dB, there are ways to get around this. Depending on how
well your DC is internally balanced, the finite Directivity (-20 dB for
instance) represents the degree of response it has to a signal that is
flowing in the wrong direction -- this is really it's degree of imbalance.
A modern Network Analyzer uses a complicated "12 point" calibration
procedure to drastically improve the accuracy of a Reflection measurement
it makes with it's "only Good quality" Directional Couplers.
PROCEDURE -- There is an alternate Calibration Procedure that does not
require the inconvenience of reversing the DC to measure it's Directivity.
This is to recognize that a good Short (or Open) circuit has a Reflection
Coefficient of nearly -0.0 dB. In this method, first measure (and record)
the apparent reflected power from a Short (or Open) termination, then
place the Known Good Termination on the "output" port of the DC and repeat
the measurement. The difference (in dB) between the two measurements
represents the DC's Directivity. When using SMA or type N connectors at
10 GHz (and below), an "Open Circuit" will have Reflection Coefficient of
nearly -0.0 dB, and is a good calibration "short/open termination."
However, if you're using a Wave Guide (WG) type DC, an open circuited WG
flange makes a pretty good transmitting antenna, with a VSWR of about
1.5:1 (reflection coefficient of about -12.9 dB). Therefore, don't use
this as a high reflection termination. Instead, place a sheet of metal
(tightly) across the WG flange as the high reflection termination.
SIGNAL GENERATOR VSWR --
There is an additional danger to the alternate calibration procedure. It
is vulnerable to the VSWR of the signal generator. I would only use this
procedure if there was a 10 dB (or greater) pad between the signal
generator and the DC. Without that pad, the reflected signal could
re-reflect from the signal generator and cause a confusing reading. The
signal-generator-reflected voltage can add to the incident voltage and
create an apparent signal source that would appear as much as 6 dB greater
(or more) in magnitude -- but only during the short/open portion of the
test. Also, if the DUT happens to have a rather high VSWR (reflection of
greater than say -20 dB), I again would recommend the use of a 10 dB pad
at the signal generator.
(II) THE UNKNOWN
MEASUREMENT -- Once you have confirmed that your DC is performing
properly, it is time to place the Unknown Circuit (the Device Under Test
[DUT] ) on your DC to measure, and tune, it's Reflection Coefficient. The
DUT-reflected signal can then be translated into VSWR by using a look-up
table or by performing a two step calculation. Step (1): Convert the
reflection coefficient (in dB) into a reflection Voltage, which is usually
represented by the Greek letter Rho. Step (2): Convert the Rho magnitude
The final dB of Reflection Coefficient in the numerator
must be a negative number that's then divided by 20 and raised to the
power of ten in formula (1). At first, some technologists will understand
that the dB value is negative dB's, they place it into the formula that
has another negative sign in it, that converts it to a positive value (+),
and they come up with answers that are crazy.
CHEAP AND BROAD -- The beauty of using a Directional
Coupler (DC) in VSWR measurement is that, generally, they are rather
inexpensive, and they are rather broadband, therefore a swept frequency
measurement is possible if your power detector is a fast acting one, such
as a calibrated Crystal Detector (and oscilloscope), a Spectrum Analyzer,
or a Scalar Network Analyzer (SNA). As you tune your DUT, it is nice to
know that you are tuning for a broadband match, as opposed to an impedance
match that is only effective across a narrow frequency range.
(III) DC ALTERNATES -- There are a large number of
devices that can serve as the Directional Coupler (DC). They have such
names as Quadrature Hybrid, 90 Degree Hybrid, Branch Hybrid, Branch
Coupler, Magic T, Ring Hybrid, Zero-180 Degree Hybrid, Wave Guide Broad
Wall Coupler, Wave Guide Narrow Wall Coupler, Wave Guide Beth Hole
Coupler, etc. The one kind of hybrid that can't be used this way is a
Wilkinson Half Hybrid, or Zero Degree Hybrid.
(IV) DC EXTENDED FREQUENCY RANGE -- Few technologists
know that a well constructed Directional Coupler (DC) has an operational
frequency range that extends many octaves in the lower-frequency
direction. For instance, if you plotted the Forward Response of a DC
that's rated for operation from 1 to 2 GHz, you would find that it has
useful operation all the way down to 10 MHz (and probably below). The
only thing that changes is it's frequency flatness, and the Coupling
Coefficient decreases -- but that can ba a considerable advantage. Here
is what's happening:
(A) A TEM-type (non Wave Guide type) Directional
Coupler has it's greatest coupling at the frequency where the internal
coupling section is 1/4 wave long. Above (and below) that frequency the
response falls off in a very predictable manner -- it's a SINE wave of
amplitude. In other words, if I was sweeping that DC that's rated for 1
to 2 GHz, and I plotted the Foreword absolute Voltage response versus
frequency at the Coupled Port, the resultant plot would look like a
rectified SINE wave, with the horizontal axis being frequency (instead
of time). There would be a zero response a zero MHz, a broad peak near
1.5 GHz, a second zero near 3 GHz, a second broad peak near 4.5 GHz,
etc. Unfortunately, a DC only has Directivity at the 1/4 wavelength
frequency region, and at lower frequencies -- but that still leaves many
octaves of useful operation.
(B) That predictable response outside of the rated frequency range has turned into an advantage for me on many occasions, here are some examples:
(1) For my first published article, "A Stripline
Amplifier/Tripler for 144 and 432 MHz", Ham Radio, February, 1970, I
needed to test the power output, and harmonic content, of the 144 MHz
section and the 432 MHz tripler section of that 4CX250B amplifier. I
needed a 300 watt frequency-indicating power meter, that I didn't
have. A Spectrum Analyzer (SA) can do the job, but it can't tolerate
the 300 watts. If I had a -30 dB DC, the coupled power would be 0.3
watts and the SA could easily make the measurements. But, my
company's Instrumentation Department said they didn't have a -30 dB DC
at that frequency range, and none of their DC's could tolerate 300
I studied what they had and found a solution. They
had a Narda -10 dB type-N Directional Coupler rated for 8 to 12 GHz
and 1 watt maximum. I reasoned that the coupling section was 1/4 wave
long (90 degrees in phase length) at 10 GHz, the center of it's
frequency range. I then divided 144 MHz by 10 GHz, multiplied by 90
degrees, and reasoned that the coupling section was only 1.296 degrees
long at 144 MHz. The SIN of 1.296 degrees is 0.02262. Since this is
a voltage response I took 20*LOG(0.02262) = -32.9 dB. That means that
the coupled response at 144 MHz would be -32.9 dB (weaker) than at 10
GHz, where it was a -10 dB coupler. Therefore it is a -42.9 dB
coupler at 144 MHz. I calibrated it at 144 MHz and found it to be a
-43.1 dB coupler -- close enough. And, since the internal coupled
line is isolated from the main line by -43.1 dB, that means that the
internal 50 ohm termination would never see more than 0.015 watts
when I applied 300 watts of 144 MHz signal to the coupler. I
similarly calibrated it at the harmonic frequencies, applied the 300
watts to it, it worked like a charm, I made all the measurements this
way, and they appeared in the article.
(2) In the low
frequency area of a coupler's response (near 0 degrees of a SIN
function) the response is almost a straight-line response that falls
off at -6 dB per octave (-20 dB per decade) as you go down in
frequency. Therefor the "-43.1 dB coupler" I used at 144 MHz would be
a -63.1 dB coupler at 14.4 MHz. As you are about to see, Directional
Watt Meters use this principle.
(V) BIRD-TYPE WATT METERS -- It is interesting to note
that the slug of a Bird Watt Meter is also a less than 1/4 wave section of
a Directional Coupler. The Bird slug achieves frequency flatness across
its rated frequency range by using a rectifier circuit that has a
low-pass filter action that rises at 6 dB per octave as you go down in
Each slug also has a finite Directivity, depending on
how well it was balanced and calibrated at your favorite frequency.
Therefore, be careful about falling into the trap of using a high power
slug to measure the forward power of your 1 KW XMTR, and then switching to
a low power slug to measure a very low VSWR. Your antenna may be perfect,
and have no reflected power (voltage), but the slugs approximate 20 dB of
Directivity would show an apparent antenna reflection of -20 dB (10
watts). That would lead you into believing that the antenna VSWR was
(VI) COUPLER IMPROVEMENT TECHNIQUES -- AS the above
material shows, a DC that has less than ideal Directivity is really
displaying a slight imbalance that causes it to slightly respond to the
signal that is flowing in the wrong direction on the main line of the
coupler. There are many ways of improving the DC's balance.
(1) Internally, you could re-adjust the accuracy of
it's termination, or you could add a small gimmick capacitor in the
correct location to improve the Directivity balance.
(2) But, an even
better way is to use a Double Slug Tuner, or a Wave Guide E-H Tuner. If
you have a known good termination, you can assume that it has perfect
absorption and essentially no reflection. You then place the tuner
between the DC and the good termination, and adjust it until the DC
shows no reflected power from the termination. You then leave the tuner
connected to the same port of the DC, while you proceed with the VSWR
or power measurements. When you were adjusting the tuner for a null in
the DC's Reflection response, you were really creating a second small
reflected signal that was equal in amplitude and 180 degrees out of
phase at the DC coupled port. That created the improved balance and
made the DC nearly ideal, at that frequency. The bandwidth of this DC
correction technique is dependent on the amount of correction that was
required. When in doubt, recheck the balance at the next frequency.
(VII) TRANSMISSION LINE DIRECTIONALITY -- When I tell a
technologist that a transmission line will keep the two signals completely
separate, that flow in opposite directions on a transmission line, they
often don't believe it -- particularly if the two signals came from the
same source. There are many RF tests that could be performed to prove
this, but I have discovered that a well-informed skeptical person can
always come up with an alternate explanation that supports their point of
view. I have found that the best way is to use visual experiments.
(1) A pool of water is really a radial transmission
medium. If I drop a pebble at the North end of the pool, waves will
travel to the South. Similarly, a pebble dropped into the South end will
create waves that travel to the North. If I drop pebbles at both ends
of the pool, the waves will meet at the middle, and pass right through
each other with no interference, as long as the waves are kept small
enough (use the linear region of wave amplitude -- no white caps).
(2) I can tap the 1/4 inch guy wire on my 200 foot
Rohn-55 tower and watch the wave travel up the guy wire, strike the
tower, reverse in polarity, and propagate back down to me (it hit a
"short circuit"). I can wait until the wave has struck the tower, and
started back to me, then I can strike the wire again (with any polarity)
to start a second wave going up the guy wire. As the two waves meet in
the center, they pass right through each other with no interference, as
long as the waves are small enough that I don't get into non-linear
stretch (deflection) of the steel.
(3) I say that most
linear transmission mediums obey this property -- even RF in free space.
Those waves that meet in free space pass through each other with no
real interference. When you move your Handy Talky Radio around a room
that is reflective, you will find what you think are signal nulls. This
is because you are using an antenna that has no Directivity, and it is
responding to at least two waves that are out of phase. Similarly, the
probe that is used on a Slotted Line VSWR setup has no directivity, and
it displays the Standing Wave Ratio that is caused by the signals that
flow in both directions through the Slotted Line. This measurement
technique has become the classic way of specifying the Reflection
Coefficient of an RF device -- it's VSWR.
(VIII) LETS DO AWAY WITH VSWR -- If you took the
directional probe from the slug of a Bird Watt Meter and operated it on
that Slotted Line, you would discover that the Standing Wave has
disappeared, and you could now independently measure the amount of power
(or voltage) that is flowing in each direction (by reversing the slug) --
that's really what you wanted to know in the first place.
In the past, that Slotted Line measurement was the only
way you could conveniently measure the reflected voltage -- by using an
interferometry technique to indirectly measure it as VSWR. It really is
time that we abandon "VSWR measurements" because we don't do it that way
any more. We should only discuss the Reflection Coefficient -- in watts
ratio, volts ratio or dB ratio (choose your favorite units), because we
now directly measure the reflected signal. We RF mavens seem to spend
half our life converting back and forth between VSWR, Voltage Reflection
Coefficient (S11, S22) or Power Reflection Coefficient, just so that we
can communicate with a technologist (or the data sheet) that uses the
other system of units.
"VSWR" is now a "coded message," it's really time that
we "Break the Code" or stop using that code when we're training the new RF
recruits. I'll admit that we will have to keep mentioning it, for
(IX) TROMBONE IMPROVEMENT -- I'll warn you that these
last three paragraph will only be appreciated by a person with a rather
exacting-type of personality.
Once you accept the fact that RF power can
independently flow in two directions on a transmission line, you then
realize that changing the length of a lossless transmission line does not
change the Reflection Coefficient; thus it doesn't change the true VSWR of
your antenna. However, if the Directional Coupler (DC) device your using
(coupler or a Bird) has less than ideal Directivity, than the Reflection
Coefficient, and VSWR, will appear to change. This is because there is a
small amount of Forward-flowing signal (I'll call it the Leakage Signal)
that's mistakenly being picked up by your coupling device, that beats
against the real Reflected Signal that your coupler is now measuring (from
your antenna, for instance). As you change the length of the transmission
line (with a Trombone Line), the two signals go in and out of phase with
each other. This will show up as a cyclicity of the apparent Reflected
Signal Power, as the Trombone is operated. This assumes that your
trombone can move about one wavelength at your frequency -- you not going
to do this at 80 meters, Hi. Although, there you could insert fixed
lengths of low loss cable to get the same effect.
Knowing the operation of the system, and its
shortcomings can allow you to gain a higher accuracy in the Reflection
Coefficient measurement. A perfect DC or Bird would show no change in
reading as the Trombone (on the antenna side) is operated. The magnitude
of the "ripple" is an interferometry effect that is telling you exactly
how strong is the Leakage Signal into your coupling device. Once you know
the strength of the Leakage, you can subtract it out of your measurement.
This is exactly the accuracy improvement procedure that is done in the
microprocessor of a modern Network Analyzer. You can convert the Ripple
into a Leakage Magnitude by using the following formulae:
Here is a measurement example. Assume I'm measuring the Reflection Coefficient of my UHF antenna system and my DC says that the Reflection is around -19.5 dB. As I operate the Trombone after the Coupler, I see a Peak reading of -19 dB, and a valley reading of -21 dB. That's a Peak-to-Peak reading of 2 dB. The formula tells me that my Leakage Signal is 0.1146, or -18.81 dB (weaker) than the Peak and Valley measurements I have made. That relative Leakage voltage was in-phase at the -19 dB reading, and out-of-phase at the -21 db reading. I can choose to subtract the voltage from the -19 dB, or add it to the -21 dB reading. This relative voltage will thus be 1.1146, or 0.9954 (as a voltage), and I can take 20*LOG of these voltages. Thus, I can either add 0.94 dB (in absolute terms) to the -19 reading, or subtract 1.06 dB (in absolute terms) from the -21 dB reading. In either case the corrected reading will be an antenna Reflection Coefficient of -19.94 dB.
I hope this information is useful to those who could read this far. Feel free to correct the mistakes.