Optical switches assist functional test

With the proliferation of fiber-optic products containing interfaces ranging from Gigabit Ethernet to 10-Gbps SONET, the demand for optical testing capabilities is high. Unfortunately, so is the cost for optical test equipment, and delivery times may be long.

You can decrease test time, reduce system operator intervention, and reduce the amount of equipment needed to accomplish product verification by incorporating a well-designed optical switching matrix in your functional test system.

For single-channel optical products, you should, at a minimum, typically perform the following tests on the manufacturing floor:

• TX power—This test requires an optical power meter, which measures the output power of the product's transmit (TX) port.
• Traffic/bit error rate (BER) test—This test employs instruments such as protocol analyzers that provide traffic, using the appropriate data protocol (for example, Gigabit Ethernet or ATM over SONET) for the UUT. These instruments perform BER testing, error detection, and alarm reporting.
• RX sensitivity—This test uses an optical attenuator in conjunction with the traffic instrumentation to test the sensitivity of the UUT receiver (RX) port. The traffic instrument's TX port connects to an optical attenuator. The attenuator reduces the power of the signal arriving at the UUT's RX port to the minimum sensitivity level listed in the product specification. Depending on the product, typical values for RX sensitivity can range from–16 to–46 dBm.
• Eye mask—This test uses an optical-communications analyzer to verify TX port compliance with industry standards. The eye-mask test can verify characteristics such as jitter, data rate, and overshoot.

Testing optical products with multiple channels, or ports, becomes more involved. You need to use many of the same instruments, but you need to find an easy way to use them to exercise each channel. Here, optical switching matrixes provide several benefits:

• Faster test time—Because the various pieces of test equipment are connected instantaneously through the switching matrix, a system operator need not shift fiber connections to various test points.
• Less test equipment—Optical switching can allocate an expensive test instrument between multiple UUT ports.
• Less contamination—The introduction of contaminants into equipment TX and RX ports or onto fiber cable is always a serious issue. Automated switching reduces the number of connections the operator must make, reducing the exposure of the optical connectors to contaminants. In a good design, the user will only have to make connections at the UUT; the switching matrix will make the connections to the test equipment.
• Greater productivity—The test operator can test more products because of reduced test times. In addition, he or she can start a test and then move to another test set or work on repairing failing modules while tests run automatically
• Less exposed cabling—Much of the exposed optical fiber cabling associated with a manual test set can fit inside the switching area on an automated system. Because the user need not access the fiber cabling, you can conceal the fiber or dress it out neatly. This eliminates damages that can result from mishandling.

When specifying equipment for an optical test system, you'll need to include characteristics such as measurement resolution or operating optical wavelength. But to get the most out of an optical switching matrix, you'll also need to pay attention to the equipment's TX power and RX sensitivity. TX power plays a role in instruments with transmitters, such as traffic test sets; RX sensitivity is important in instruments with optical detectors, such as power meters and optical communication analyzers.

Using the TX power and RX sensitivity ratings of both the test equipment and the UUT, you can calculate the optical power budget, or the amount of losses that the switch-matrix design can introduce into the optical test paths. This power budget provides constraints on the number and types of components you can use in an optical switching design.

When specifying equipment for the tester, you must also determine how many instruments of each type you need. That quantity will depend on the switch-matrix layout. Here are some factors that affect the equipment vs. switching choices:

• Cost of equipment vs. switching costs—When testing a UUT with four RX ports, does it make more sense to have four optical attenuators to test RX sensitivity or just one attenuator and switches that allow access to the four RX ports? In the case of an optical communication analyzer, which is extremely expensive, switching is almost always the preferred method.
• Speed of testing—If you test eight RX ports with one optical attenuator, you must test the ports one at a time, potentially leading to extremely long test times. Using a switching design that incorporates eight attenuators would allow concurrent test of the eight ports and could cut the RX sensitivity test time by roughly 85%. This assumes there will be eight instrumentation ports available for traffic testing.
• Power budget limitations—Adding additional switches to share test instrumentation could lead to over-attenuation. Therefore, your optical-switch-design flexibility may be limited.
• Equipment maintenance—If you reduce the amount of equipment, you also reduce the associated maintenance and calibration costs.

Once you have chosen the test equipment, you can finalize the switch-matrix layout. You have three main approaches: design your own matrix; use standard IEEE 488, VXI, or PXI matrixes; or hire a third party to design a customized matrix. To evaluate these methods, you must first define the tests you need to run and the port-to-port (both UUT and instrument) connections they require. In addition, list the components needed to make up the test paths. Most switching matrixes include these types of components:

• optical relays that allow instrumentation to be connected to various UUT ports,
• splitters that allow fixed percentages of power to be taken from the main optical stream to provide access by measurement instruments, and
• optical attenuators, used for UUT RX detector sensitivity testing. In place of the commonly used large, external IEEE 488-controlled attenuators, you can use small and relatively inexpensive manual and programmable attenuators that mount within the switching subsystem. When choosing a manual attenuator, you should note that the attenuator is physically set for a particular attenuation. On encountering a failure, a test program will not be able to step the attenuator back to determine the level at which failure takes place.

Although in-house design isn't usually financially or technically practical unless you plan to reuse the design in numerous test sets, this option does let you dictate the exact configuration of a switching matrix. If you decide to go this route, you'll need to develop

• a method to control any optical relays (a TTL module controlled by a PC or just +5 V),
• a means to mount the components that will make up the switching subsystem, and
• a power source of +5 V or +12 V or both for the optical components and the TTL control circuitry.

Numerous vendors offer IEEE 488-, VXI-, or PXI-controlled optical-switch instruments. These instruments typically include individual relays packaged in a box or on a PCB. Some instruments allow for more than one relay configuration. Typical configurations include 1x4, 1x8, 1x16, 2x8, and 2x16. When defining these instruments, you must define the fiber mode, the operating wavelength, the configuration, and the interface connectors.

These switch implementations are limited because of the minimal number of configurations that vendors offer. In addition, these modules usually do not offer integrated splitters or attenuators, so you might need additional equipment.

For simple applications (requiring a 1x4 matrix, for instance), an approach based on off-the-shelf equipment may be appropriate, but you must weigh several factors:

• If you need more than a couple of modules, this approach may become too cumbersome.
• Test times typically are longer because all practical layouts allow only one channel to be verified at a time.
• If only one channel can be accessed at a time, maximum traffic-flow capabilities cannot be verified on multiport UUTs.
• Splitters may need to be remotely mounted near the switching modules.
• A lot of fiber cable will be exposed.

Figure 1 shows a block diagram of a design using a standard switching matrix that could accomplish the four major optical tests.
 

Third-party customized switch

The most effective—although the most expensive—approach for developing a complex switching matrix relies on a third-party supplier. Companies that offer customized switch matrixes will work with you to design a complete layout that includes all necessary splitters and attenuators. The vendor will also ensure that issues such as component interaction and power budget are handled properly. The vendor will also fabricate all mechanical assemblies and provide power supplies and onsite support. Most customized systems include software drivers, so you can integrate the hardware into an automated manufacturing process.

Figure 2 demonstrates two custom approaches employed to test optical products. The Figure 2a setup tests a cross-connect switch that has its own internal traffic generator. The Figure 2b setup verifies an eight-port Gigabit Ethernet product using external traffic devices. Both designs allow concurrent traffic testing of all optical ports as well as providing TX power, RX sensitivity, and eye-pattern testing.

Figure 2. These custom switching matrix layouts for (a) a cross-connect device and (b) an eight-port Gigabit Ethernet device employ optical scopes, power meters, and attenuators, which the switching matrix connects to the respective UUTs.

When choosing components for a switching matrix, take into account these options:

• Wavelength operation—Specs for relays, attenuators, and fiber typically list a wide bandwidth of operation, while splitters and couplers usually have one or two small windows of operating bandwidth.
• Insertion loss—This spec describes the amount of power lost when a signal passes through a particular component or fiber. You'll need this value for power budget calculations.
• Repeatability—This spec describes how much variation you can expect within the rated component. For a relay, the repeatability may represent the maximum variation in insertion loss over 100 consecutive closures.
• Connector type —For a switching system, FC connectors offer the best stability and consistency. Other connector types (such as ST, SC, and LC) allow more movement, which can affect repeatability.
• Fiber core—Match the fiber core type (such as 50µm or 62.5µm) between UUTs and test instrumentation. If the fiber will be exposed in a damaging environment, opt for a protective jacket. Also, ensure that the fiber is long enough to avoid sharp bends, which can attenuate signals.
• Straight or angled fiber—Fiber can come with the ends straight or angled. Angled cable provides better return loss, while straight fiber provides lower insertion loss. In manufacturing test, the short fiber lengths usually make the insertion loss more of an issue, so the straight fiber is usually the better choice.
• Specialized fibers—Many specialized fibers are available to overcome dispersion or polarization problems. For switching matrixes used to test non-DWDM products, standard, SMF-28 fiber cabling will usually work with no noticeable degradation.
• Splitter loss—This specification describes the combination of the splitter's insertion loss plus the ratio of power split per leg. Splitter loss can be calculated as 10[log(splitter percentage)] + rated insertion loss.
• Control—Some components work with simple voltage values while others require more sophisticated timing circuitry.
• Power requirements—Most switching matrixes typically require +5 or +12 V with current requirements up to 350 mA.

When laying out an optical switching matrix, you must determine the optical power budget for all optical paths. You can determine the available power by looking at the power difference between the TX source power and the RX detector sensitivity for each path. The total losses introduced by the equipment in the path obviously must not exceed the available power and should leave as much available power as possible.

Table 1 lists the power budget for the Figure 2a circuit. The loss figures are "worst-case scenarios," and you can expect better results on an actual system. You can use fusion splicing in place of the manual connectors to reduce connector losses between components.

In addition to reducing the amount of equipment you need, a well-designed switching matrix can significantly decrease test times. Table 2 shows estimated test times for an eight-port Gigabit Ethernet system. When running these tests using the custom setup, an operator must insert 16 fiber cables and can then walk away for three minutes. When running the tests using the manual setup, the user must make 40 fiber cables insertions over the course of the program. The user must also use the PC keyboard to respond to prompts roughly 25 times. So, the manual approach takes significantly longer, and the operator must stay with the test system throughout.

The key to designing a switching matrix is to develop a plan that clearly defines the tests to be run, the equipment needed to accomplish the testing, and the associated software and calibration packages. You can extend the basics presented here to other types of optical tests, whether you develop custom equipment yourself, integrate standard instrument modules, or commission a complete system from a third-party supplier.

Kaminow, Ivan P., and Thomas L. Koch, eds., Optical Fiber Telecommunications IIIA, Academic Press, San Diego, 1997.

Kartalopoulos, S.V., Introduction to DWDM Technology: Data in a Rainbow, IEEE Press, Piscataway, NJ, 1999.