Samuel Morse probably isn’t the first name you think of in relation to relay-based switching systems. In fact, his shouldn’t be the first name you think of even when Morse code is the subject. It appears that his associate, Alfred Vail, actually invented the code, but Morse patented it under his own name.
At the heart of the system was the relay. Henry had discovered many of the fundamental laws of electromagnetism, and the relay was a byproduct of that research. He thought it could provide a control function in industry but never actively pursued its development or application.
In contrast, Morse realized the device’s potential for telegraphy. He redesigned Henry’s large and heavy relay by winding coils with many more turns of much finer wire to increase sensitivity.1 Much later, during the development and refinement of the telephone, Western Electric applied materials science and precision manufacturing techniques to the production of tens of millions of relays.
And, development has continued. Solid-state switches increasingly are replacing relays, but high-tech versions of electromechanical relays still handle most switching jobs. It has proven very difficult for semiconductor devices to match the relay’s simplicity, reliability, near-ideal switching characteristics, isolation, cost, ruggedness, and application versatility.
Relay-Based Switching Systems
ATE at all levels of sophistication generally requires signal and power switching. Seldom are separate measuring instruments provided for each signal, so a multiplexer is needed that will sequentially connect the signals to the instrument.Also, it is important to have programmatic control of the power applied to a DUT. A defined order and precise timing for applying the various voltages at device power-on often are required as is the capability to quickly shut down all DUT power supplies if a fault is detected. Switch modules in several configurations are commercially available from a large number of vendors. The comparison chart that accompanies this article lists details of representative products from some of the leading manufacturers. Only a few of these entries are based on solid-state switching. All the rest use relays.
Of course, there is no such thing as a standard relay. The device has become highly differentiated to address specialized applications. For example, coaxial relays perform the basic circuit make/break function but do so while maintaining a constant characteristic impedance. To achieve very low open-circuit capacitance, some coaxial relays feature large contact separation. Armature relays take on power-switching jobs where their robust mechanical design is an advantage. And, very compact reed relays have been developed to handle low-level signal switching, currents in the hundreds of milliamps, and frequencies up to 100 MHz.
Relay coils must be correctly driven, but electrically that is addressed by the module manufacturer. In addition, a switch-module designer will attempt to maintain signal quality.
“Signal ground planes should be isolated from the control circuit grounds, and signal paths should be designed to minimize crosstalk between channels,” said Jon Semancik, business development manager at VXI Technology. “Choosing relays to maximize signal integrity and using multilayer PCBs with extensive ground planes are beneficial. In addition, the control circuitry should go into a quiescent state when not processing commands, making lowlevel signal switching possible.”
On the other hand, it’s the user’s responsibility to ensure that the contacts are not stressed beyond the specified V-I limits and that the relays connect the intended signals to the appropriate test equipment. In fact, because there is such a large number of switch modules available, the user’s first step is to define the test system in sufficient detail that the actual switching needs can be established.
As the comparison chart shows, there are several types of switching- module configurations: single relays, multiplexers, and blocking or nonblocking matrices. If you are controlling separate, simple loads, such as lights or solenoids, single relays may be a good choice. Typically, higher power applications use single relays because more complex switching tends to be limited to lower-level signals. Also, if you’re not dealing with high frequencies and don’t have a large number of signals, single relay modules are a flexible and low-cost solution.
Multiplexers typically are used to select one output from among a large number of signals. Common specifications are, for example, 1×16 or dual 1×4. Using the pole-throw terminology, these would be designated SP16T and DP4T, respectively. To select more outputs, you can use additional multiplexers connected to the same or different groups of signals.
Conversely, a nonblocking matrix allows any input to connect to any output. It uses more relays and is more expensive but provides complete flexibility. An additional consideration in the specification of nonblocking matrices is fanout. If, for example, a 16×16 matrix is required, each input must have sufficient fanout capability to drive all of the 16 outputs.
Large matrices can be developed by grouping several smaller modules, but as the input-output dimensions increase linearly, the number of required relays grows as the product of the dimensions. For this reason, switching-matrix manufacturers usually limit the direct implementation of a square matrix—equal numbers of inputs and outputs—to one that is twice the size of the building-block matrix. In this case, four smaller matrices are required to construct a matrix with twice the number of inputs and outputs. The RF specifications of a 2×2 matrix will be degraded somewhat from those of the four smaller matrices from which it has been constructed.
Going to four times the size requires 16 modules, generally making direct implementation of this and larger sizes impractical. Instead, a three-stage design such as used in the 426K System from Precision Filters provides nearly perfect nonblocking operation but, for example, with only 42% of the number of relays required for a bruteforce construction of a large 162×162 matrix.
The greatest savings are achieved when what seemed to be a matrix application can be treated more simply. For example, if 64 of 256 signals are to be selected as inputs to a 64-channel data acquisition system, it might be possible to switch the signals in blocks of 64. Any reordering or rearranging of the channels can be accomplished in software. This solution requires only a small fraction of the relays used for a completely random selection of 64 from 256.
Today’s Opportunities
The improvements made in solidstate switching have had a significant impact on the cost and size of certain types of switching modules according to Nick Turner, president of Cytec. “There has been a very gradual shift over many years from electromechanical relays to solidstate technology. Some of the application- specific switching ICs now available allow production of large switching systems at one-tenth the cost and size of the relay equivalent.
“The only problem with solidstate is that the devices still tend to be application specific,” he continued. “There is no true replacement for the combination of voltage, current, and low on-resistance you get from bidirectional electromechanical relays in general-purpose switching systems.”
Being able to solve a customer’s switching problem in a smaller package was a capability stressed by Loofie Gutterman, president of Geotest-Marvin Test Systems. “Relays have become smaller. We often design products in the 6U PXI form factor, which increases the available real estate by 250% compared to a 3U module. Combining small relays and larger modules allows us to offer large matrix products with a high switch density and, because the number of I/O pins increases only linearly with matrix size, compact connectors (Figure 1).”
In addition to high switch density, look for a module to have the capability to be quickly reconfigured. For example, the Geotest-Marvin Model GX6264 is a 128-point differential scanner/multiplexer but can be used as eight separate one-ofeight scanning groups. In addition, because of the product’s design, Geotest was able to satisfy a customer’s request for differential scanning with groups of individually addressable relays simply by making a firmware change.
Better software packages also are available to make developing and operating a switching system easier. Rather than programming at the register level, there are many advantages to a more abstract approach. Register-level programs can be very difficult to maintain. It’s far simpler to understand a function call that amounts to “connect group A” instead of a hard-coded sequence of relay closures that suits a particular test setup.
Common User Problems
Unfortunately, neither advanced hardware nor software will solve an application problem if the user doesn’t have a good understanding of switching systems and their limitations. For example, make-before-break (MBB) and break-before- make (BBM) contact timing are important in multiplexers and matrices that can connect a large number of signals. In a BBM product, the last relay to close will open before the next relay closes. In an MBB product, the next relay will close before the last one has opened, creating a momentary short.
The comparison chart shows two groups of detailed electrical specifications that must match or exceed the application’s requirements. The black specs describe the basic voltage, current, and power capabilities of the contacts. Note that a relay typically cannot switch as high a current as it can carry. Also, because switching AC voltages produces less arcing than does DC, the voltage limits are higher for AC than for DC.
To View Switching Systems Comparison Chart1 Click Here
To View Switching Systems Comparison Chart1b Click Here
To View Switching Systems Comparison Chart2 Click Here
To View Switching Systems Comparison Chart2b Click Here
The red specs relate to a relay’s RF performance. Only the fundamental specifications are included. In addition to considering these characteristics, choosing the most suitable switch topography is critical. Some designs exhibit wide bandwidth variation depending on the selected switch path, and others produce transmission-line stubs when certain paths are switched off. The use of additional isolation relays and special-purpose RF star relays may be required to achieve the best overall RF switching performance.2
Accounting for possible future requirements in addition to well-defined present needs should be part of switching-system selection. Norton Alderson, vice president of marketing at Universal Switching, said, “Many clients leave switching-system definition as the last part of their planning. Automated switching should be considered an integral part of the test-system design because so much relies on it.
“Once an automated system has been installed at a location where one previously was not available,” he observed, “other users may wish to connect to it as well. Additional capacity provisioned in the design will greatly facilitate future changes and system growth.”
Keithley’s Charles Cimino, business development manager for DMM and switching products, agreed. “The user’s test and measurement goals, strategies, and error budgets must be understood before selecting a switching solution. For example, the need to remote sense for low resistance or in high-current applications requires four-pole systems while twoand single-pole systems provide higher density for mainstream applications,” he explained.
“The biggest mistake we probably see is simply a reluctance to call and ask for advice,” said Cytec’s Mr. Turner. “There are so many diverse applications for switching systems that it is difficult to convey all the details in a generic switching catalog. We tell people that if they don’t see what they need, just ask. Chances are we can build it or have built it in the past,” he continued. “It often is much more cost-effective to buy a system tailored to your specific needs than to make do with an off-the-shelf solution.”
In general, these comments are true, but especially apply to RF and microwave switching systems. If you have not previously developed such a system, don’t attempt a design without expert help. Many of the vendors listed in the comparison chart offer design consultancy services and have the experience to guarantee the performance your application requires.
Future Developments
If current trends continue, even more RF and microwave switching solutions will be available in the future. “This is without doubt the fastest growing area for switching systems,” said Kevin Leduc, strategic sales manager at Racal Instruments. “Switching systems are used in rigorous testing to make sure a satellite is ready for launch, and satellite ground stations often include a switching system to route signals between multiple antennas and receivers. In addition, the radio frequency ICs that are fueling satellite and other wireless communications growth need to be tested on ATE systems that include microwave switching. Of course, military and homeland-security programs also are heavy users of microwave switching systems and components.”
Bob Stasonis, the sales and marketing manager at Pickering Interfaces, attributed part of his company’s recent business to strong growth in high-frequency ATE applications. “We are seeing many requirements for large-scale crosspoint matrices for RF and microwave signals. The increased use of wireless communications as well as microwave security systems and automotive options calls for more high-frequency switching,” he said.
In contrast, Keithley’s Mr. Cimino cited high growth in functional production testing of precision active and passive electronic devices such as LEDs, lasers, and giant magnetoresistive (GMR) heads. Mr. Alderson from Universal Switching commented on the market growth caused by the demands of ATE for higher density, lower cost, higher bandwidth, and faster control. He also sees increased activity in replacing older manual patching systems with automated switching equipment.
Growth in the market for switching systems is good because it broadens choice and reduces cost. To take advantage of both effects, start from a clear definition of your test problem and discuss it with a number of switching-system vendors. Sure, these people have a product to sell, but they also are switching experts. It shouldn’t take too many conversations before you have a good idea of which products you want to buy and from whom.
References
1. “Long-Distance Legacy,” Stewart
Fists’s History of Technology Columns,
The Australian, 1999,
http://www.electric-words.com/hist/
991102history.html
2. Lum, J., “The Importance of
Switching,” EE-Evaluation Engineering,
November 2001, pp. 72-79.
November 2004