Morris Hoover Siecor Operations, LLC
For several years now, cable-television operators have used couplers and coupler arrays to make more efficient use of their analog-transmission lasers. These amplitude-modulated (AM) lasers are typically high-powered and expensive, and warrant maximum utilization of the signal.
More recently, the use of dense wavelength-division multiplexing (DWDM) in long-haul routes has created huge demand for erbium-doped fiber amplifiers (EDFAs) and multiplex/demultiplex modules, both of which use tap couplers to internally monitor optical-power levels. Higher time-division and wave-division multiplexing rates-and more high-priority traffic-are increasing carriers' need to assess signal quality and optical power levels. When facilities-based competitive local-exchange carriers (CLECs) are present, the incumbent local-exchange carrier (ILEC) must have a demarcation point for fault location and troubleshooting. This quality-of-service (QoS) monitoring must take place without interruption of traffic.
At SBC Technology Resources Inc., increased high-priority traffic, demarcation requirements due to CLEC unbundled loops, and the use of different transmission protocols led to the decision to add monitoring ports to all optical circuits. Wavelength insensitive, singlemode 1x2 optical coupler products were to be used as test points within SBC's fiber-transport systems, and as demarcation points on CLEC circuits. As a test point, these devices would nonintrusively verify the optical power on the fiber. A test set or protocol analyzer could then be attached to the test point if the received optical power was within the test set's dynamic range. SBC worked with Siecor to develop the 1x2 optical couplers used at the demarcation points to monitor the power levels of the optical signals.
Figure 1 is a representation of a typical 1x2 fused coupler. Light entering the common port is split into two paths. The relative optical power level in each output leg is usually given in percentages. The designation of primary output-always the higher power level-has to do with the way the coupler is made.
Although the nominal insertion loss through either leg of a 1x2 coupler is, by definition, a logarithmic curve, the loss does not begin to sharply increase until the 95% level is reached (see Figure 2). The choice of a 95/5 coupler over a 90/10 coupler would result in a small (~0.2 dB) improvement in loss in the "through" leg, while an added penalty of nearly 4 dB would be found in the "tap" leg. Thus, a 90/10 ratio of output power to monitor power was determined to be optimal for this application. This allows minimal insertion loss through the 90% leg while allowing adequate power out of the 10% leg for most test equipment. Additionally, the use of a 90/10 coupler reduces concerns about reflectance from unterminated ports, compared to typical 50/50 couplers, since any signal reflected from a monitor port would be attenuated approximately 10 dB in both directions. At this point in the SBC application, a number of commercially available couplers could have performed this monitoring function.
Since the goal was to gain information about the signal levels of the "through" leg of the coupler by looking at the "tap" leg, the relationship between the two power levels needed to be well-defined. One way of achieving this would have been to characterize each coupler module once it was installed in the system. However, this approach would have added to the on-site workload during installation and would have complicated record keeping.
In the past, coupler manufacturers sold their devices in 5% power increments, from 50/50 to 95/5, and guaranteed a ±5% tolerance. (In other words, manufacturers sold practically every device made.)
In the case of monitoring optical power and QoS, a new relationship needed to be established for the insertion loss of the two coupler outputs. For typical star and tree couplers, where the nominal output power is equal among all legs, uniformity is the measure that defines how well the output power is actually distributed. For couplers with unequal split ratios, there is no analogous optical parameter.
When team members of Siecor and SBC/TRI met to discuss this need, the following was established. For both the 90% leg and the 10% leg of any coupler, there would be a maximum and minimum allowable insertion loss, across the range of wavelengths listed as a conditional objective in Bellcore 1209 (CO4-12), to ensure compatibility with any wavelengths that might be used in the future. The difference between the highest loss in the 10% and the lowest loss in the 90% could be no more than 2 dB greater than the lowest 10% loss minus the highest 90% loss. That way, the power measured at any monitor port in the SBC system could be used to calculate the power in the through port, within a tolerance of plus or minus 1 dB. This relationship can be shown as (Ls max - Lp min) - ( Ls min - Lp max) < 2, where Lp is loss through the primary port and Ls is loss through the secondary port.
Existing test data for 90/10 couplers suggested that limits of 0.56 to 1.10 dB for the 90% port and 9.40 to 10.80 for the 10% port were achievable, at least in sample quantities. With these values, the loss relationship expression evaluates as the following:
This satisfies the 2.0-dB requirement. In use, then, the mean loss through the secondary port is 10.10, and the mean loss through the primary port is 0.83. The average difference between the two is 9.27 dB, which is added to the observed power from the monitor port to predict the actual power at the "through" port, within 1.0 dB.
The additional requirement of operating over the wavelength ranges of 1,260 to 1,360 nm and 1,430 to 1,580 nm called for the use of couplers that were wavelength-flattened. However, even wavelength-flattened couplers exhibit some "curvature" below 1,290 nm and above 1,570 nm. For this reason, the allowable insertion loss was "opened up" somewhat for the wavelengths outside the two 40-nm-wide windows around 1,310 and 1,550 nm. Since the plan was to package these devices in modules with connector ports (adapters) for all inputs and outputs, an additional "uniformity" value had to be determined, to account for insertion loss due to optical connectors and the variability in that loss due to manufacturing tolerances.
The result was an even tighter specification for the raw couplers. The specified insertion loss limits are shown in the Table. These insertion loss limits needed to be met over a temperature range of -40 to +70°C.
The unique requirements of this application also made it necessary for Siecor to increase its measurement and test capabilities. Up until this point, all coupler products were specified to function at 1,310 nm, 1,550 nm, or both. To assure that the new monitor modules were compliant with SBC's specification, test equipment had to be procured which was capable of performing insertion loss, directivity, and polarization dependent loss (PDL) measurements over the wide range of wavelengths. Return loss was measured for each connector during the polishing process, and again for the completed module. This was done at 1,550 nm only.
For insertion loss measurement, an optical spectrum analyzer (OSA) was first tried, but the received power levels were too low to make accurate measurements. The decision was then made to purchase two tunable laser sources (TLS), one for the 1,260-to 1,360-nm range, and the other for the 1,430- to 1,580-nm range (see Figure 3).
An alternative method for insertion loss testing is to use an incoherent light source, which eliminates any resonance effects that can be found with laser sources. Such a test setup includes a rotatable diffraction grating and a slit, but allows a wider range of wavelengths than an OSA. This addresses the signal-to-noise problems of the OSA. Finally, an optical power meter with capability for PDL and return loss measurement was procured.
In a two-way circuit, four monitoring couplers can provide appropriate points for testing and fault isolation (see Figure 4). The transmit and receive path couplers are typically combined in one dual-path connector module.
In a dual module, the light paths indicate the "through" circuits (see photo on page 140). The monitor ports are either connected permanently or on an "as needed" basis to monitoring equipment. The value these modules have in the system greatly depends on installation instructions, which must facilitate proper connections.Quickly incorporating these monitoring modules into the existing SBC network required packaging them in a manner that was compatible with existing termination hardware. Southwestern Bell and Pacific Bell use two different styles of rack-mounted hardware, and Siecor offers units that are plug-compatible with both.
The product development consisted of making modifications to existing coupler module products to conform to SBC's particular needs. Vertical and horizontal 90/10 monitor modules were produced for SBC. Each dual and single module had a common language equipment identifier (CLEI) code and test data labels. Similarly, each product was packaged with its "birth certificate," which included a unique serial number and a copy of the final test data for each port on the module. That test data is also recorded on labels on the rear of the module. The serial number is recorded on the CLEI code label.
The need for high reliability in telecommunications systems is well understood. However, the act of adding components to the system to help insure quality could have just the opposite effect, if those components were not equally reliable. At Siecor, the successful testing of couplers to Telcordia specifications is an essential part of supplier qualification. The GR-1209-CORE specification controls basic optical performance in different environmental conditions, while GR-1221-CORE is designed to test for long-term reliability.
The battery of tests used on the coupler modules includes high temperature storage, both in dry and high-humidity conditions. To insure quality over the life of the products, both sets of tests should be repeated periodically, on a sampling basis.
The monitor modules that Siecor developed for SBC Communications demonstrate a co-operative effort between a customer and a supplier, which resulted in improved test and measurement capabilities for both parties. The project led to a new family of products that use the output power of a monitor port to predict the output power of a "through" port. In addition, the capabilities of coupler manufacturers were expanded because these suppliers were challenged to produce products meeting upper and lower bounds for insertion loss over a broad range of wavelengths.
Morris Hoover is staff engineer, product development at Siecor Operations, LLC (Keller, TX). Hoover received help on this project from Eddy Barker, senior member of the technical staff at SBC Internet Services (Dallas).