Complex RF Switching Architectures -- Part II


Wireless appliances are increasingly prevalent in our technological world. With the introduction of new parameters and new design considerations, the switching sector is not immune to the upsurge in radio frequency testing needs. Manufacturers therefore demand ever-increasing radio frequency test systems to keep up with technology. This article discusses design considerations for high frquency test systems.
The other tutorials in this series are:


RF Switch Topology Options

General Purpose - Form C relays

With a form C relay, the signal present on the common terminal (COM) needs to be routed to one of the two poles (indicated commonly with Normally Closed -NC- and Normally Open -NO-)

Figure 1. Form C (SPDT) relay



The multiplexer architecture has one common terminal (COM) and N switched terminals. You can close the correct switch and connect the common terminal to any output point. Test engineers typically use this architecture when they have one scope and many channels to acquire or one arbitrary waveform generator and many terminals to distribute the signal. 

Figure 2. Multiplexers connect a single common channel (COM) to various input/output channels. 

Generally, for RF devices, the available multiplexers are in the 4x1 or 6x1 dimensions. Rarely can you find a much higher number of switched terminals in an off-the-shelf RF multiplexer. Next, we analyze how to create more complex multiplexers starting from 4x1 or 6x1 building blocks.

More Complex Architectures

Let's assume a case where you need a 13x1 multiplexer, but only have 4x1 building blocks to use. In this case, the natural solution is to use a cascaded architecture so you can use the specific number of inputs. The following figure, taken from the SCXI-1190/SCXI-1191 user manuals, illustrates the point.

Figure 3. You can cascade multiplexers to create larger multiplexers.

Next you might ask, "What happens to the three parameters discussed in Part I of this article -- insertion loss, VSWR, and isolation?" There is no easy rule or mathematical formula for all the three numbers, because the new combined system has to take into account the cables you use to build the multiplexer and the fact that some parameters do not have a linear relation with the number of stages in the new, more complex multiplexer.

The total insertion loss is generally close to the mathematical addition of the insertion losses of the single stages, including the insertion loss of the transmission cables. VSWR will be largely system dependent, and it would be best to measure these values before implementing any test code where damage may occur, while the overall isolation parameter could remain close to the value of the single module. In this case, the best suggestion is to consider every system as a specific and individual case. Talking with the switch manufacturer should help you determine the characteristics of the overall system.


A traditional switch matrix allows connections between row terminals and column terminals. Modern matrices, such as NI switch matrices, allow connections between any input, including row-to-column, row-to-row, and column-to-column connections. A full matrix topology has relays at every row-column crosspoint. While this topology is flexible and allows as many simultaneous routes as the smaller row or column dimension, it is expensive to provide relays for every crosspoint. Additionally, full matrices are not ideal for carrying high frequency signals because the unused portion of the connected traces adds capacitive load and RF stubs to the transmission lines. This results in reflections that can distort and attenuate the signal.

Figure 4. Traditional matrices are not ideal for high frequency signals due to RF stubs. 


An alternative to a full matrix is a sparse matrix. This topology allows only a limited number of simultaneous row-to-column connections – often only one connection at a time. Sparse matrices are generally made from two multiplexers with their COM terminals tied together. They use fewer relays and are less expensive than full matrices. As shown in the following diagram, a typical sparse matrix can make a single, stub-free connection. 

Figure 5. Sparse matrices remove RF stubs, but only allow one connection at any given time. 

The figure represents one simple way to create a special kind of 4x4 matrix with the SCXI-1190/SCXI-1191, that is to use two of the four multiplexers inside each module. The ComA terminal for instance is connected to the ComB terminal of another module. In this way, you can route any of the four inputs of module A to any of the four outputs on module B. The trick is that this is not a "full fledged" matrix. Only one signal (not multiple signals) can be routed at any given moment because the signal travels only on the coaxial cable present. Even with this limitation, you can use this architecture to potentially solve a good percentage of your needs.


"Dimensionally Flexible" Sparse Matrix

The NI PXI-2593, PXIe-2593, and SCXI-1193 are the first "dimensionally flexible" sparse matrix options for RF signals. A "flexible" sparse matrix offers 16 channels that can be configured into a sparse matrix of any dimension (e.g. 15x1 multiplexer, 2x14 matrix, 3x13 matrix, 4x12 matrix, etc.). Additionally, the "dimensionally flexible" sparse matrix offers some cases where you can connect multiple signals at the same time. To find out more about NI's dimensionally flexible sparse matrix, read the Advanced Signal Routing with the PXI/PXIe-2593 and SCXI-1193 RF Switches white paper. 

How to Make a 3x2 Full Matrix from Individual 4x1 Multiplexers

Sometimes the sparse matrix architecture is not enough to solve your problems. In this case, you need to use a complete matrix architecture. Generally, you can find RF multiplexers (6 to 1 or 4 to 1 are the most common configurations) or simple C-type switches on the market. The following figure is useful when trying to implement the correct connections to create a matrix. In the diagram, we show the case of a 2x3 matrix built starting from multiplexer modules (4 to 1 in this case).

In this schematic, you also can terminate the columns and rows via the RT. In this architecture, you have to take into account the different value that parameters such as crosstalk, isolation, and VSWR assume in the complete system, as we discussed earlier in the article.

Figure 6. You can create an RF matrix from individual RF multiplexers

Terminated Multiplexers Offer Protection for Sensitive RF Sources

Similar to ensuring that the characteristic impedance is matched for the entire system, we need to verify that unused signal paths are properly terminated to ensure that the entire system has the same characteristic impedance. For example, in a typical RF application involving switches, your source, load, transmission line, and switch impedance may be 50 Ω. When the RF relay opens, the source and transmission line impedance will remain at 50 Ω, while your load impedance has jumped to a very large value. This impedance mismatch can cause RF signals to be reflected back into the signal source, which can damage sensitive RF equipment.

When switching sensitive RF equipment, it is sometimes necessary to ensure that the signal is terminated to fulfill the requirement of keeping identical characteristic impedance throughout the system. A terminated switch is essentially a broadband impedance (resistor) to ground that matches the characteristic impedance of the system when the switch is closed.

Figure 7. Terminated RF multiplexers terminate unused channels to ensure impedance matching for all open and closed channels. 


NOTE: A switch that is closed with termination generally cannot handle the full power of the switch. See device documentation for additional information.