When input power must be split among multiple transistors in an amplifier, the amplifiers themselves, antennas, or in an enormous number of other scenarios, power dividers are the answer. Not surprisingly, they are manufactured by dozens of companies throughout the world and represent a significant portion of the RF and microwave passive-component market. At first glance, RF and microwave power dividers appear simple. But each type has unique characteristics, so choosing the best one for a particular application can be confusing. This article will hopefully help to alleviate that problem.
For anyone not familiar with power dividing and combining, some of the confusion is caused by terms that tend to be used haphazardly over time. For example, power dividers are sometimes referred to as splitters—which is true. But the industry has deemed that power dividers should be called splitters when they are the simplest and least expensive devices, such as those used in home coaxial-cable systems. In addition, although the term coupler and power divider are often used interchangeably, key differences between the two exist.
Coupler or Divider?
Directional couplers are designed to “sample-off” a relatively small amount of power into one port for purposes like monitoring. Another factor to remember is that while the insertion loss of most passive components is simply the total attenuation from input to output, when applied to a power divider, it refers to the additional loss above that created by the splitting process itself.
Furthermore, power dividers can also be power combiners, but this does not mean that their specifications will be the same nor that every type of power divider is equally suited for use as a combiner. As noted, there’s more than meets the eye when it comes to power dividers.
A power divider splits an input signal into two or more outputs that are usually, but not always, equal in amplitude and phase. Regardless of its type, the goal of every power divider is to have the greatest port-to-port isolation, lowest insertion loss and voltage standing wave ratio (VSWR), and least amplitude and phase imbalance over the entire frequency range of the device.
The most basic type of power divider is the T-junction (Fig. 1), which if mechanically symmetrical will divide a signal applied to its input into two outputs that are equal in amplitude and phase. It’s basically just three transmission lines connected at a single point. The T-junction satisfies the basic purpose of dividing an input signal into two separate transmission paths, but it suffers from two very important deficiencies: There’s no way to match impedance at all ports or any means of providing isolation between them. Although these limitations have always been important, the increasing sophistication of multichannel receivers, along with other applications, has relegated them to obscurity.
1. The T-junction represents the basic idea of power division.
Resistive Power Divider
There are two basic types of power dividers: resistive and reactive. The resistive type (Fig. 2) is symmetrical and arguably the least complicated, has the greatest bandwidth, and allows the desired system impedance (typically 50 Ω) to be maintained. It’s a star configuration that has no dedicated input port. Therefore, every path has equal loss and an input at any port will distribute the signal equally to every other port. This allows them to function as “hubs” that are well-suited for connecting multiple receivers, transmitters, or transceivers, as well as for measurement purposes.
The resistive power divider’s simplicity also makes it smaller than other types, since it’s composed only of resistors. In addition, they are the only type that can have a minimum frequency down to dc. They achieve their wide operating bandwidths because there are no frequency-dependent components, i.e., reactive components.
2. The resistive power divider is simple and broadband but lossy.
However, since the resistors used as the dividers absorb power, resistive dividers have comparatively high insertion loss, making them unacceptable for use in many types of systems. For example, the typical insertion loss from one port to another in a 2-way resistive divider is 6 dB, while the loss in a 2-way reactive divider is 3 dB.
The star construction also means there’s no port-to-port isolation. The power-handling capability of resistive power dividers is also limited by the reactance of the resistors; therefore, high-power versions are not practical. Resistive power dividers do not provide isolation between ports, which is a critical factor in applications such as communications systems plagued by crosstalk.
A variant of the resistive type produces an output impedance that’s different from the system impedance. This is typically used in various types of measurement systems. For example, a simple resistive divider is often utilized with network analyzers in which the divided signal is split equally between the network under test and the reference channel. They also can be used to calibrate power sensors and other instruments.
Reactive Power Divider
Reactive power dividers are asymmetrical, essentially quarter-wave transmission lines matched to divide an input signal evenly to multiple output ports. They feature very low insertion loss, provide about 20 dB of port isolation, and are extremely rugged because they don’t include resistors that can burn up, making them well-suited for demanding applications.
Long the choice for use in the distribution of signals in some antenna arrays, they also find homes in coax-based in-building systems such as legacy distributed antenna systems (DASs). In short, reactive power dividers are efficient, cover reasonably wide bandwidths, handle high RF power levels, and can be fabricated using waveguide, stripline, microstrip, transformers, and other technologies.
The basic advantage of reactive power dividers over other types is that they have very low loss, operate at high frequencies, and provide significant isolation between output ports. They are often fabricated coaxially with an air dielectric in which a 50-Ω input impedance is changed to 25 Ω at the outputs by varying the ratio of inner and outer conductors, with each output in parallel at 50 Ω.
Wilkinson Power Divider
This is a good place to mention one of the most important contributors to the power-division domain: Ernest Wilkinson, whose “N-way hybrid power divider,” details of which were published in “Institute of Radio Engineers (IRE, now the IEEE) Transactions on Microwave Theory and Techniques” in 1960. This achievement ensured Wilkinson a place in RF engineering history because it solves basic problems associated with the T-junction, which has made the reactive Wilkinson power divider a staple of RF and microwave design ever since they first became commercially available.
3. Wilkinson power dividers, such as this coaxially fabricated version, have unique properties that make them useful for a wide array of applications.
The Wilkinson power divider (Fig. 3) was created in part as a solution to the problems of matching and isolation that are inherent in a typical T-junction divider, in which a large amount of the power reflected from port 2 enters port 3 and thus provides little isolation. The Wilkinson power divider employs quarter-wavelength transformers to divide the input signal. But to understand why Wilkinson’s creation is important, remember that network theory dictates no three-port device can be simultaneously matched, lossless, and reciprocal.
The Wilkinson power divider covers two of these bases, as it’s inherently matched and reciprocal, but not lossless. So, its creator placed a resistor between the output ports, which not only dramatically reduces loss (if both of these ports are also well-matched) but provides isolation as well. That is, assuming both arms of the divider are the same, the signals will also be in phase. Therefore, the resistor will draw no current and thus add very little loss (ideally none, but this is not an ideal world).
So, while loss is not eliminated, it can be very low, reciprocity is maintained, and isolation is provided—all through the addition of one component. A good example of a Wilkinson power divider is Pasternack’s PE2021 (Fig. 4) that operates from 7 to 12.4 GHz. The PE2021 can handle as much as 10 W of continuous-wave (CW) input power. It has 1 dB of insertion loss and achieves 16 dB of isolation. Furthermore, the PE2021 maintains an amplitude balance of ±0.3 dB and a phase balance of ±6 deg.
4. Pasternack’s PE2021 4-way Wilkinson power divider operates from 7 to 12.4 GHz and handles 10 W of input power (CW).
Wilkinson’s creation has other benefits as well. For example, it’s comparatively simple and can be implemented using printed components as well as with lumped inductors and capacitors. When fabricated in microstrip on a board, its cost is extremely low, too.
The initial implementations of the Wilkinson power divider were limited in bandwidth, but an enormous amount of work has been conducted over the years to improve on this metric, and today much greater bandwidths are achievable. It can be configured as an N-way (more than two output) device, although it’s primarily used for 2-way power division. Common division ratios are 2- through 8-way, although other division ratios, such as 5- and 10-ways, can be realized as well.
Countless variations of the Wilkinson power divider have been designed and constructed, particularly in recent years because of the availability of simulation software and improved fabrication techniques. Some Wilkinson designs allow for unequal division with large division ratios, as well as dual and broadband frequency operation. Designs that take advantage of coupling between the quarter-wave transmission lines to reduce layout size while maintaining a good bandwidth have also been developed.
Furthermore, it’s possible to create Wilkinson power dividers with unequal (i.e., asymmetric) splitting ratios. These dividers, which are often used with array antennas and several types of receivers, consist of two quarter-wave segments of different impedances that realize the desired power division.
This discussion covers some—but far from all—of the fundamental information about coaxial power dividers. A more thorough discussion requires considerable editorial real estate. Fortunately, many textbooks are available that discuss power division and power combining from the fundamentals through various configurations (of which there are many), as well as couplers and implementations in waveguide.
Steven Pong is Product Manager at Pasternack.