New technology improves fiber-optic component testing
New technology improves fiber-optic component testing
Optical reflection discrimination uses an optical pulse technique to eliminate the mandrel wrapping associated with ocwr test methods.
Michael A. Brown rifocs Corp.
The migration of fiber-optic devices and technologies into the data and telecommunications industries has resulted in a growing demand for components--connectors, cables, patch cords, and adapters, among others. The burgeoning demand for these products has created a need for improved manufacturing and testing techniques to meet customer expectations. Increasingly, a manufacturer`s test and measurement savvy can be a key factor for success: Increasing manufacturing throughput while reducing cycle time and costs can "make or break" the product`s acceptance in the fiber-optics market. Yet fiber-optic testing can be a daunting process to newcomers and experienced hands alike.
Each of the components mentioned above can degrade the performance of an optical system due to power losses and backreflections associated with splices, connector interfaces, adapters, and detector surfaces. Two primary sources of performance degradation are Fresnel reflections and Rayleigh scattering. Fresnel reflections occur at the boundary interface of two materials with dissimilar refractive indices, as caused, for example, by an air gap in a poorly designed and specified connector. Rayleigh scattering is a common source of optical attenuation due to molecular impurities in the fiber itself.
Backreflections from a splice or component can cause deviations in the spectral width of a given laser diode, injecting noise into the system or even shifting the operating wavelength--an unacceptable condition for wavelength-division multiplexing (wdm) and dense wdm products. Where fiber is used for duplex operations, crosstalk between the transmitter and receiver can occur. Excessive reflections can damage higher-power lasers.
These issues are not trivial and loss measurements are critical, so standards have been developed over the years to help take the guesswork out of fiber-optic testing. Since any measurement is only as good as the equipment used to perform the test, the National Institute of Standards and Technology has developed traceability and calibration standards that many manufacturers of test equipment rely on. Moreover, the robust fiber-optic working group of the Telecommunications Industry Association and Electronic Industries Association, both of Arlington, VA, has developed fiber-optic test procedures (fotps) that have proven to be of significant value. These fotps include tia/eia 455-34, fotp-34, for measurement of insertion loss and tia/eia 455-107, fotp-107, for return-loss measurement.
Key performance measurements for product and quality assurance are taken for two primary operating criteria. The first, optical power loss--called insertion loss, which equals
10log (powerin/powerout)
--is the most common measurement. Insertion loss of a splice or connector pair is specified as the ratio between the input and output power as expressed in units of decibels. Today`s connectors typically exhibit insertion loss readings in the 0.2- to 0.5-dB range.
Insertion-loss measurements can be accomplished using a precision laser source and optical power meter for direct power measurements. A factor here is the selection of an optical power meter that has a detector optimized for the operating wavelength under test. An optical power meter with an indium gallium arsenide detector is preferable for most of today`s applications in the 1300- and 1550-nm wavelengths, due to its spectral flatness and measurement accuracy across each of these operating windows. These measurements are direct and simple.
The second performance measure is that of return-loss measurement for the component in question. Here we are measuring the ratio between the optical power into a component or system and its reflected optical power (backreflection) in units of decibels:
return loss = 10log (powerin/powerback)
Unlike insertion loss, where a low number is good, for return-loss measurements higher numbers are better. For example, a good-quality fiber-optic connector exhibits return loss in excess of 50 dB. In fact, some connectors, such as Diamond`s angled-physical-contact connector, routinely exceed 60 dB.
ocwr and otdr methods
Until recently, two technologies--each with its own strengths and weaknesses--were usually used to measure optical return loss. The most widely used method remains the optical continuous-wave reflectometry (ocwr) method. In this technique, a continuous wavelength is passed through an interface, connector, or device under test, and return loss is measured directly. Given a calibrated light source and detector-optimized optical power meter, return-loss measurements can be accurately accomplished.
Although the ocwr method is, indeed, direct and straightforward, its application in a volume production environment is problematic. Figure 1 shows a simple ocwr test configuration. Note that reflections (R) occur at all the interfaces shown. Our goal, however, is to isolate reflection R2--the device under test. The return-loss test set will be calibrated and referenced, and R1 reflections will be accounted for. To eliminate reflections R3, we apply a mandrel wrap just beyond the device under test. This is accomplished by manually wrapping five or six turns of the optical fiber around a mandrel tool. Alternatively, a gel, oil, or matching block can be applied to the connector endface. Reflection R2 can then be measured.
However, mandrel wrapping proves to be problematic: Not only is it labor-intensive, but for a number of fibers and applications it is not practical. High-numerical-aperture pay-out fibers, ribbon, some multistrand cables, and other cables with built-in stiffeners defy this technique. Oil and gel alternatives are not desirable in most cases, because the last thing an optical connection needs is the application of a substance likely to attract dust and dirt. It is important to remember that a dirty fiber-optic interface is the leading cause of out-of-specification performance readings during testing. This factor more than any other affects measurement repeatability.
Return loss can also be measured using an optical time-domain reflectometer (otdr), although with less precision. otdrs are most frequently used for characterizing relatively long fiber links. Unlike the ocwr method, otdrs do not directly measure backreflections. While insertion-loss testers provide a discrete loss measurement, the otdr technology provides the user with a visual, graphic display of the optical "signature" as a function of time, over a typically long (5- to 10-km) fiber run. A high-intensity optical pulse is launched into the fiber, and a high-speed optical detector records and graphically displays the observed reflections. Looking at the optical scan, the operator can observe losses due to splices, breaks, connectors, and other attenuations and can calculate return loss.
The otdr method is widely used for locating events across a fiber run and for reflectance testing of splices and connectors in Synchronous Optical Network and cable-TV systems. However, the otdr test technique and the cost of the otdr itself render this method less widely used for component test purposes than the ocwr method. Additionally, otdrs exhibit an undesirable characteristic intrinsic to their design: Measurements cannot be taken within their "dead zone"--typically within 5 to 15 m of the device. This shortcoming limits the practical use of otdrs for component test purposes.
An integrated solution
With the limitations of both techniques in mind, manufacturers and customers alike have looked for new testing solutions that have the potential to meet increasing product demand while reducing costs and improving quality levels. An ideal solution would provide all of the benefits of the ocwr method but without the time-consuming mandrel wrapping or blocking techniques; and it would incorporate some of the attributes of the otdr technique. It also would address a key concern in testing circles--that of repeatability and accuracy of measurement.
A new test method that uses a combination of fiber-optic test technology, application-specific software control, and test process methodology provides all of the accuracy of a direct return-loss measurement with the advantages of using a pulsed optical signal. The new technique, called optical reflection discrimination (ord), performs return-loss tests without the need for mandrel wrapping for all devices under test that have leads in excess of 3 m in length.
ord is based on the same principle as otdr technology. A laser pulse or series of pulses is launched through an optical splitter into the fiber under test. Each disturbance along the light path causes some level of reflection. The portion of the light that is reflected backwards eventually reaches the optical coupler and, in turn, the optical power meter. Like the otdr, the respective power meter and optical receiver must be capable of accurately sampling the reflected light pulses. The bandwidth of the optical receiver must be of the same order of magnitude as the second-order frequency component of the launch pulse. The microcontroller must discriminate between reflections caused by the measurement setup as well as those caused by the device under test and reflections that precede and follow it. These signal levels are sampled and digitized. A software algorithm controls the sample intervals and positions so that individual reflections can be isolated. The digital signal is scanned and the reflection amplitude is quantified. To achieve accurate reflectivity or return loss measurements, this signal is compared against a precise, internal reference.
Direct amplitude measurement
ord differs from the ocwr method in that a very stable laser source injects pulses into the device under test (see Fig. 2). Just as in the previous example, we seek to isolate reflection R2. This is accomplished during the calibration step; a high-speed pulse is launched into the fiber, and the amplitude of the backreflections is measured. This allows for both direct measurement of the reflection(s) and specific identification of the reflection under test. Once this is accomplished, no further calibration is required for that run. Under normal testing operations, reflections from these pulses are detected by a fast optical receiver, and their amplitude is measured and further discriminated via software control. Note that we are directly measuring the amplitude of the reflection in question. In fact, reflections occurring at or beyond R3 can be accomplished as well. In this manner, we have eliminated the testing problems associated with mandrel wrapping, while providing an effective return-loss measurement tool for those products with which mandrel wraps simply do not work.
As incorporated in a new test instrument--the Component Test Set--the ord technique can be partnered with application-specific software control. In this scenario, tests are performed via mouse-activated commands (see Fig. 3). The screen in the figure shows the top-level data for the specific test run or job. The user can pre-specify any number of test configurations dependent upon the devices to be tested and the requisite upper and lower specification limits, store them in the test instrument, and call them up as required. The test instrument "remembers" the number of re-tests allowed and documents all measurements taken, providing data for post-test reduction.
With the elimination of mandrel wrapping, a reduction in testing time--as much as 40% per component--has been observed. Figure 4 shows data taken in a time study that compares the speed of the new technology with that of a rifocs ocwr return-loss tester as well as that of a competitive brand ocwr tester. As an additional data point, we also ran the new test instrument in manual mode, which most closely approximates the ocwr technique. The results, accomplished by an experienced operator, show that the new technique provides a per-termination time savings of 16% to 40% compared to the ocwr testers. The amount of throughput improvement achieved will depend on several factors unique to a given manufacturing process. Moreover, repeatability curves for the new method show that its measurement repeatability is equal to or better than that of the ocwr method.
The repeatability issue and the effect that operator error can have on the process underscores the importance of pairing ord with software control for process improvement and data capture and analysis. The universal trend in fiber-optics production is to do more with less while improving quality levels and, presumably, customer satisfaction. The key to enabling this transformation on the plant floor must involve computers and software control. The testing of fiber-optic components--be they active or passive--is no exception.u
Michael A. Brown is director of sales and marketing for rifocs Corp., Camarillo, CA. He can be reached at Michael.A.Brown@ rifocs.com.