Test methods cut fiber manufacturing costs
Test methods cut fiber manufacturing costs
In a quest to cut manufacturing costs, fiber producers are streamlining their test processes using OTDRs.
David Kritler PK Technology Inc.
When one reflects on recent developments in optical-fiber testing, several topics come to mind: spectrum analysis of dense wavelength-division multiplexing systems, the evaluation of polarization-mode-dispersion effects in optical links, and out-of-band network monitoring--not to mention the characterization of new link components such as optical amplifiers and fiber gratings. While these important developments certainly warrant their share of attention, they make it easy to overlook developments of similar stature that are affecting the world of optical-fiber manufacturing test processes. New testing strategies now being adopted by some fiber manufacturers are not only improving product quality, but they are also helping to nudge along the expansion of the world`s optical-fiber networks by cutting production costs and ensuring that fiber and cable prices continue their slow but steady descent.
Market forces and cost-control methods
In today`s highly competitive optical-fiber market, where product differentiation is sometimes obscure and price pressure is intense, the difference between winners and losers is often their relative costs of production. Higher production costs usually mean higher prices, and higher prices impair a manufacturer`s ability to compete effectively in an open market. Given that testing and handling expenses can account for a significant portion of the production cost of optical fiber, most successful manufacturers continually scrutinize their fiber testing and handling methods to identify potential targets for cost reduction.
One of the more obvious cost-reduction strategies, which has been widely implemented over the last few years, is the minimization of fiber test regimes--i.gif., the suite of tests routinely performed on fibers or cables. This has been accomplished by two primary means. First, if production processes are reasonably well controlled, manufacturers can perform only sample testing of a given fiber performance parameter rather than screen the entire production. A second method, which has also been available to manufacturers for some time, is the use of "alternate" test methods versus "reference" test methods to measure a specific fiber parameter. Alternate test methods often require less costly test equipment, are often faster, and usually deliver measurements with accuracy similar to those produced by the reference methods. An example of the alternate cost-reduction strategy in action is the use of an optical time-domain reflectometer (otdr) rather than a cutback measurement system to perform singlemode attenuation testing.
New cost control methods for fiber test
While minimization of test regimes has reduced production costs, possibly the most revolutionary cost-reduction tool has only recently made its appearance on the production floor: statistical/theoretical test methods. The result of maturing production and measurement technologies, statistical/theoretical test methods combine production statistics and fiber transmission theory with relatively minor extensions of existing measurement systems to predict critical fiber performance parameters.
Since the predictions resulting from these techniques basically duplicate measurement data previously obtained from other dedicated test systems, the net result is the elimination of the redundant testing performed by these systems. Not only are the expenses related to test time eliminated, but also the expenses related to fiber handling, alignment, and other steps. Examples of these new statistical/theoretical test methods that have either been recently released or are in progress as drafts in the tia, itu, and iec standards organizations are briefly described below.
Spectral attenuation modeling by otdr (tia fotp-120, itu-t G.652, iec 60793-1): Spectral attenuation modeling relies on the fact that the shape of a singlemode fiber`s spectral attenuation curve is highly consistent, and is predictable if the attenuation at a few key wavelengths is known. Modeling begins with the collection of a large body of spectral attenuation data for a particular fiber type using the cutback reference test method. This body of data is then statistically reduced to an N ¥ M matrix of coefficients, where N is the number of key, "predictive" wavelengths and M is the number of "prediction" wavelengths at which the attenuation will be estimated.
For example, the "predictive" wavelengths might be 1310 nm, 1550 nm, plus one or two key wavelengths around the water peak region, such as 1360 and 1410 nm. The "prediction" wavelengths might be all the wavelengths between 1200 and 1600 nm in steps of 10 nm. Note that manufacturers must develop a "characteristic" prediction matrix for each particular fiber type that they produce (e.g., singlemode fiber, dispersion-shifted, non-zero dispersion fiber) given the unique shapes of their spectral attenuation curves. Once this matrix is developed, the spectral attenuation data for a fiber can be determined quickly by acquiring attenuation data at the three to four predictive wavelengths and then performing matrix multiplication of this data and the characteristic matrix.
The predictive attenuation data for spectral modeling may be acquired using any standard attenuation technique, and manufacturers will realize some benefit simply from the reduced data collection time. However, it is clear that maximum cost reductions will only be obtained if the data is acquired using an otdr. This is because otdr testing already needs to be performed on all fibers for point defect screening and length measurements, and the unit is capable of providing accurate attenuation data (when data is acquired bidirectionally). Also, since otdrs capable of acquiring data at multiple (up to four) predictive wavelengths are commercially available, the otdr is the obvious choice for this measurement technique. The use of an otdr to collect modeling data virtually eliminates the need to perform time-consuming and costly cutback attenuation testing (and its associated fiber handling) and, in some cases, may also eliminate the rewinding processes often required for this testing. A typical spectral attenuation curve predicted from four-wavelength otdr attenuation data is shown in Fig. 1.
Mode-field diameter by otdr (tia itm-6, iec and itu-t committee drafts): Another cost-reduction measure being implemented or considered by many fiber manufacturers is the determination of mode-field diameter (mfd) by otdr. While mfd has long been associated with the otdr as being responsible for otdr signature artifacts such as splice "gainers" and non-uniformities in unidirectional backscatter, the theoretical relationship between otdr backscatter and mfd has found new life with mfd prediction. Like spectral attenuation modeling, the motivating factor is again the reduction of measurement costs.
Historically, mfd measurements have been performed using some type of far-field or near-field scanning technique. These techniques offer reasonable measurement accuracy, but measurements are relatively slow (approximately 30 sec per wavelength). Furthermore, these scanning measurements yield mfd measurements only at the ends of the fiber length, since only 2-m end samples are typically tested. The mfd uniformity along the length of the fiber remains unknown, which is problematic given that the average production lengths of optical fibers are increasing and are likely to continue to do so given the economies of scale associated with long fiber lengths.
mfd prediction by otdr offers manufacturers an ideal solution to all of these issues. First, the otdr-based method is fast. Like spectral attenuation modeling, the information required for the prediction can be acquired as part of normal otdr defect and attenuation screening using essentially the same measurement setup. Second, the otdr yields mfd nonuniformity data (from the "bidirectional sum" signature), which solves the "end sample" dilemma of long fiber lengths. Finally, the otdr-based mfd measurement can be extremely repeatable and is relatively simple. Both ends of the fiber under test are temporarily spliced to buffer fibers having known mfds. When the fiber is measured with the otdr, the splice losses at the two ends of the fiber are acquired along with the normal point defect and attenuation data. This splice loss data is acquired bidirectionally, and the mode fields at the two ends of the test fiber are determined using relatively simple formulas, which relate the difference in the two unidirectional splice losses to the ratio between the known buffer fiber mfd and the unknown test fiber mfd. A typical otdr test station setup, which is capable of mfd prediction, is shown in Fig. 2.
Cutoff wavelength by otdr (tia itm-6, iec and itu-t committee drafts): Given the fact that an otdr can be used to predict both spectral attenuation and mfd, cutoff wavelength is left as the only routinely measured fiber parameter that still has to be measured by a traditional transmission measurement system. But this may not be the case for long. Some manufacturers are already considering exploiting the theoretical relationship between cutoff wavelength and the ratio of the mfds measured at two wavelengths (by an otdr or other methods) to predict cutoff wavelength.
Like spectral modeling and mfd by otdr, cutoff by otdr would produce testing cost reductions by moving the measurement from one test station, a cutback measurement system that is relatively slow and requires additional fiber handling, to the otdr, where the measurement can be obtained essentially for free. Cutoff by otdr would provide the added benefit of cutoff non-uniformity data.
While all of this may sound like the OTDR is on the fast track to replacing yet another transmission measurement system, it should be noted that cutoff by otdr has challenges. One must consider that when used with mfds predicted by otdr this technique is essentially "a prediction of a prediction," and great care must be taken if the accuracy of the otdr predicted cutoff data is to meet customer expectations. The key problem areas are temporary splice loss stability and splice loss algorithm accuracy, both of which can add errors to the critical bidirectional splice loss data.
However, both these problems are receiving significant attention from fiber manufacturers and equipment suppliers, and they appear tractable. New fiber alignment products have been developed specifically for otdr prediction applications and work continues towards more sophisticated otdr data analysis techniques.
The future?
Special attenuation modeling by OTDR, MFD by OTDR, cutoff wavelength by OTDR--where will manufacturing test go from here? Some would argue that the exploitation of the otdr measurement technique is not finished. This viewpoint is supported by the considerable amount of press describing experiments with even more otdr-based test methods, such as chromatic dispersion by otdr and polarization mode dispersion by otdr. However, given the fact that most of the world`s fiber manufacturers sample only these two parameters, it could be that telecommunications providers might be more interested in these new measurement techniques. Interest on the part of telecommunications providers appears to stem from the concern that the uniformity of both of these parameters is important to the proper functioning of some next-generation transmission systems, and sections of any installed cables that are found by otdr methods to exceed system specifications may need to be replaced.
Beyond test reduction, it seems likely that both fiber and cable manufacturers will try to follow the cost-reduction path set by other high-technology manufacturers, such as those in microelectronics, and move as close as possible to full automation of both manufacturing and post-manufacturing test processes. While this is clearly a formidable task for manufacturers, some cost-intensive production activities, such as fiber preparation, fiber alignment, and data analysis, are already experiencing some degree of automation. As a result, some opportunistic manufacturers are already receiving the benefits of both optimized testing strategies and fiber-handling automation: lower production costs and higher profit margins. In today`s highly competitive fiber and cable market, this may mean survival. u
David Kritler is marketing manager for prodtion and laboratoy test systems at PK Technology Inc., an ifr company (Beaverton, OR).