Directional Coupler is a device that passes signal (RF wave) through to output port and passes a certain amount of the ingput signal (usually a very small portions of the input signal) through another port (Coupled port). (NOTE : Basic functionality of directional coupler is to sample the signal in only one direction, but there are some other types of directional coupler that sample the signal in both directions. This type of directional coupler is called Bi-directional coupler / Dual directional coupler).
On this page I'll first show how the four ports of a coupler work, and then where a coupler sits in a real test setup. After that come the common product types and the specification, read from a measured S-parameter graph. Keep one question in mind throughout: which way is the wave traveling, because that decides which port sees it.
- What Is a Directional Coupler?
- Why we need a Directional Coupler ?
- Types of Directional Couplers
- Typical Specification
What Is a Directional Coupler?
A directional coupler has four ports, and each port has a fixed role for a wave that enters at p1. The main line carries almost all of the power, and a second line placed close to it picks up a small, controlled part. The word directional describes that second line: it responds to a wave traveling one way and ignores a wave traveling the other way.
Operation of a regular directional coupler can be illustrated as below. The signal flows from p1 to p2. A certain portions of the signal can flow out of the p3 (Coupled Port), but cannot flow through p4 (Isolated Port).

Figure 1. The four ports of a directional coupler. A wave entering at p1 leaves mostly at p2, a small part leaves at p3, and ideally nothing leaves at p4.
p1 is the input port : the solid red arrow enters here and runs along the main line.p2 is the output port : almost all of the input power leaves here. This is also called the through port.p3 is the coupled port : the dashed red arrow branches off near p2 and leaves here. Its power is a fixed fraction of the power traveling from p1 to p2.p4 is the isolated port : no arrow leaves it. For a wave traveling from p1 to p2, ideally no power comes out of p4.
One point in Figure 1 needs care. The main line of a passive coupler is a piece of transmission line, so it passes a signal from p2 to p1 just as well as from p1 to p2. What depends on the direction is the coupling. A wave traveling from p1 to p2 couples to p3. A wave traveling from p2 to p1, for example a wave reflected by a badly matched antenna at p2, couples to p4 instead. This is why a coupler can separate the forward power from the reflected power. A device that really blocks the reverse direction is an isolator, which is a different component.
Why we need a Directional Coupler ?
One of the most typical use case for the Directional Coupler is as follows. You may connect the device on TX path (usually between Power Amplifier) and Antenna and measure / monitor the output power. If you use the coupler that output very small amount of the signal through the coupled port, you can monitor the output power without wasting much of the TX energy.
The drawing below shows this setup. The coupler sits between the power amplifier and the antenna, and a measurement instrument is connected to the coupled port p3.

Figure 2. A coupler as a power monitor. The antenna gets almost all of the power, and the instrument on p3 sees a fixed, much smaller copy of it.
In my personal case, I often use this device connected to TX port of the protocol tester and measure the power while the test is running. One thing you need to keep in mind to use the device for this kind of purpose is to make it sure that the input power to the device should be relatively high, otherwise the power coming out of the coupled port is too low and spectrum analyzer would not display it propery.
Let's put numbers on both sides of that trade. Suppose the amplifier delivers 30 dBm, which is 1 W. With a 20 dB coupler, the coupled port gives 30 - 20 = 10 dBm to the instrument. The main line loses only the 1% of the power that went to p3, which is 10 log10(1 - 0.01) = -0.04 dB. With a 10 dB coupler, the instrument gets 20 dBm, but the main line now loses 0.46 dB. So a weaker coupling costs less transmit power. But it also leaves less signal for the instrument, which is the problem described in the paragraph above when the input power is low.
The same setup can also watch the reflected power. A wave reflected by the antenna travels from p2 back to p1 and couples to p4. So a coupler with a second coupled port, or two couplers facing opposite ways, measures the forward and the reflected power at the same time. The ratio of the two gives the return loss of the antenna.
The coupled power is the input power minus the coupling in dB : 30 dBm into a 20 dB coupler gives 10 dBm at p3.The main line pays for the coupled power : a 20 dB coupler costs 0.04 dB, a 10 dB coupler 0.46 dB and a 6 dB coupler 1.26 dB of through loss, even without any dissipation.Pick the coupling from the power you have : a high-power transmitter can use 30 dB or more, while a weak signal needs a stronger coupling for the instrument to see it.A coupler can watch both directions : forward power couples to p3 and reflected power couples to p4.
Types of Directional Couplers
There are so many different types of directional couplers and some of the examples are shown below. (A), (B) are the common types that are frequently used in daily RF testing. (C),(D) are the waveguide types of the directional couples which would be used in very high frequency or high power.

Figure 3. Coaxial and waveguide couplers. The coaxial types cover everyday bench testing, and the waveguide types cover high frequency or high power.
(A) is a small coaxial coupler for bench use : its label reads -6 DB, which is the coupling, and it marks IN and OUT at the two ends of the main line.(B) shows two coaxial couplers with larger connectors : each one has a main line with two connectors at its ends and one coupled port on top. Larger connectors carry more power.(C) is a short waveguide coupler : it has waveguide flanges on the main line and two coaxial coupled ports on top.(D) is a long waveguide coupler : it has waveguide flanges instead of coaxial connectors on all of its ports.
All of these types use the same principle. Two transmission lines, or two waveguides, run side by side over a short length, and energy leaks from one to the other through the field between them. In a coaxial or microstrip coupler, the coupling section is a pair of parallel lines. In a waveguide coupler, it is often a row of holes in the wall shared by the two guides. Because the coupling depends on the length of that section in wavelengths, every coupler has a limited frequency range. The next section shows how that range looks in a measurement.
Typical Specification
I think the most important set of specification for a directional coupler can be summarized in a single s parameter graph as shown below. The important specification can be listed as below.
- How much of the energy pass through from p1 to p2 without unecessary loss. Ideally there shouldn't be any loss except the energy getting out of the couple port. The blue line (s21) indicates property.
- How much of the input energy bounce back. Ideally this should be zero, but in reality it wouldn't be zero. Just the lower, the better. The black line (s11) indicates this property.
- How much of the energy pass through p3 (coupled Port). There is no absolute value (e.g, zero or full power) that are considered to be the best. It just should be as designed. Usually it is designed in such that p3 power is much lower than p1 power. The red line (s31) indicates this property.
- How much of the energy pass through p4 (Isolated Port). Ideally this should be zero, but in reality it wouldn't be zero. Just the lower, the better. The green line (s41) indicates this property

Figure 4. The four signal paths that the specification describes. Each arrow is one S-parameter measured with the input at p1.

Graph Source : Cliick Here
Figure 5. Measured S-parameters of a coupler. Inside the operating range, the through loss stays small, the coupling stays flat and the reflection is lowest.
Let's read the graph inside the operating range, 6 GHz to 7 GHz. The pass through S21 is about -1.1 dB and the coupling S31 is about -8.6 dB. The isolation S41 is about -21.5 dB, and the return loss S11 reaches its best value, about -37 dB, near 6.5 GHz. These four readings are all the numbers a coupler datasheet needs, and the table below shows how each specification is derived from them.
Specification |
Formula |
Value read from the graph |
Coupling |
-S31 in dB |
about 8.6 dB |
Isolation |
-S41 in dB |
about 21.5 dB |
Directivity |
Isolation - Coupling |
about 12.9 dB |
Insertion loss |
-S21 in dB |
about 1.1 dB |
Return loss |
-S11 in dB |
about 37 dB at 6.5 GHz |
Two numbers in this table need a second look. The first is the insertion loss. An 8.6 dB coupler sends 13.8% of the power to p3, so even a lossless coupler would show S21 = 10 log10(1 - 0.138) = -0.65 dB. The measured -1.1 dB means that about 0.5 dB is lost as heat in the device.
The second is the directivity, and it is the specification that decides how well the coupler measures a reflection. Directivity tells you how much stronger the wanted coupled signal is than the leakage from the wrong direction. Suppose you use a coupler with 13 dB directivity to measure an antenna with 20 dB return loss. The leakage has a reflection coefficient of 10-13/20 = 0.22, which is larger than the 0.10 of the antenna itself. Depending on the phase, the reading falls anywhere between 9.8 dB and 18.1 dB of return loss. With 30 dB directivity, the leakage is 0.03 and the reading stays between 17.6 dB and 23.3 dB. So a coupler for reflection measurements needs a directivity well above the return loss it has to measure.
Coupling, isolation and directivity come from S31 and S41 : directivity is the difference between isolation and coupling, 21.5 - 8.6 = 12.9 dB here.Through loss has two parts : the power sent to p3, 0.65 dB for this coupling, and the dissipation in the device, about 0.5 dB here.Return loss is best in the middle of the band : S11 dips to about -37 dB near 6.5 GHz and rises to about -16 dB around 9 GHz.Directivity limits reflection measurements : with 13 dB directivity, a 20 dB return loss can read anywhere from 9.8 dB to 18.1 dB.