Table 1. Optical loss budget and corresponding tolerable distances for different applications
The results of a controlled experiment demonstrate that optical loss measurements obtained using an overflow injection light source and mandrel winding as a mode-conditioning filter exhibit significant variance depending on the test equipment, while the same tests performed with the Encircled Flux (EF) test method produce close agreement with different test instruments. Encircled Flux is the test method recommended by industry experts for accurate optical loss measurements of both conventional and bend-insensitive multimode fiber, a particularly important feature for high-speed applications with tight optical loss budgets. Encircled Flux is the specified test method for 40G and 100G Ethernet applications.
Table 1. TIA-492AAAD Specifications for OM4 Fiber, September 2009
Optical Loss Testing – Why Is It Important?
Providing an accurate method for conducting optical loss testing of multimode fiber is becoming increasingly important for very high data transfer rate applications that impose more stringent requirements on the maximum tolerable loss in a channel between an optical transmitter and an optical receiver. As an example, Table 1 shows the optical loss budget and corresponding maximum distance for different applications.
The higher the data transfer rate, the tighter the optical loss budget for a channel. The maximum tolerable loss of a 10Gb/s Ethernet channel with OM3 multimode fiber is 2.6 dB, while the maximum permissible loss of a 40Gb/s Ethernet channel and a 100Gb/s Ethernet channel is 1.9 dB with OM3 fiber and 1.5 dB with OM4 fiber, respectively.
What factors affect the accuracy of optical loss testing?
Several important factors can affect the accuracy of optical loss measurement tests:
1. The type and quality of the test reference patch cords.
2. The fiber mismatch between the test reference patch cords and the link under test.
3. The characteristics of the light source and how the light couples to the fiber.
4. The mode-conditioning filter used.
This work will analyze the effects of all these factors on the accuracy of optical loss measurements.
1. Type and Quality of Test Reference Patch Cords:
The type and quality of the test reference patch cord are critical to the accuracy of field optical loss measurements. The end-face geometry of the polished ferrule of the patch cord connector can significantly affect test results and must meet precise parameters, such as bend radius, apex, and fiber protrusion. This is why test reference patch cords must meet end-face geometry requirements such as those specified in Telcordia GR-326. Furthermore, the accuracy of the optical loss test can be greatly reduced if the patch cord connector surface is scratched or simply dirty. Inspect the fiber connector with a fiberscope and, if it is dirty, clean it using a dry cleaning technique.
How do you know your optical loss results are reliable? An easy way to check is to examine a "good, known patch cord" that has already been tested at the factory. Most patch cord manufacturers indicate the actual loss value on their patch cord packaging. Inspect and clean the patch cord connectors. Then, connect this "good, known patch cord" to your equipment as if it were a fiber link and test it using the same reference patch cords you normally use for link testing. Compare the results obtained in the field with the loss value of the factory-tested patch cord. If the result is significantly different (greater than 0.1 dB), the field test is incorrect. The problem could be with the equipment, the testing method, or, most likely, the test patch cords.
2. Disparity Between Two Spliced Optical Fibers.
Fiber disparity is a consequence of inherent fiber characteristics and is unrelated to the splicing techniques used. Intrinsic coupling loss due to fiber disparity is caused by, among other things, the following:
2.1. Different core diameters,
2.2. Core/cladding concentricity error,
2.3. Different numerical apertures.
What are the specifications for these parameters? Table 2 provides a summary of some of the dimensional and optical specifications of OM4 multimode fiber.
What is the intrinsic coupling loss when splicing two fibers that are at the minimum and maximum limits of the TIA-492AAAD specification? Figures 1, 2, and 3 show, respectively, the intrinsic coupling loss due to fiber diameter disparity, core/cladding concentricity error, and numerical aperture disparity.
As shown in Figures 1, 2, and 3, fiber disparity losses can be, in the worst case, quite significant compared to the loss budgets allocated to data transfer rate applications of 10 Gigabit and higher. So what happens? We generally don't observe such high losses in field tests for two reasons: 1) most current fibers are manufactured with tighter tolerances than those specified, and 2) not all modes are excited to the same extent; that is, there is less power for higher-order modes than for lower-order modes due to the mode filtering used (a topic explained later in this paper).
3. Light Source Characteristics and How Light Couples to the Fiber
Injection conditions and how light couples to the fiber can greatly affect optical loss measurements. In the case of multimode fibers, different injection power distributions (injection conditions) can result in different attenuation measurements.
The power distribution between the fiber modes is achieved by controlling the injection spot size and the angular distribution using an optical system that couples light to the fiber, as shown in Figure 4.
In graded-index fiber, the lower refractive index of the core near the cladding causes higher-order modes to follow a curved path that is longer than the axial beam (the "zero-order mode"). Because higher-order modes travel farther into the glass core and are more likely to be scattered or absorbed, they have greater attenuation than lower-order modes. Therefore, a long, fully filled fiber (all modes injected into the fiber at the same power level) will have less power in the higher-order modes than a short length of the same fiber. The term "equilibrium mode distribution" (EMD) describes the mode distribution in a long fiber that has lost its higher-order modes. This change in the mode distribution between long and short fibers can make a big difference in measurements made with the fiber. Figure 5 illustrates the relative mode power distribution for multimode fiber.
Figure 5 illustrates the equilibrium mode distribution (EMD) for a long multimode fiber segment and the mode power distribution for a short multimode fiber segment when a mode filter is used. Mode filtering removes higher-order modes but provides only a rough approximation of the EMD.
4. Mode Conditioning Filter
4.1. Mandrel-Wound Mode Conditioning Filter
A mandrel-wound injection patch cord is generally used as a mode conditioning filter. Mandrel winding tenses the fiber and increases the loss of higher-order modes when light is injected into the fiber under test.
The use of mandrel winding as a mode filter is described in Annex A.1.3.1.2 of TIA-455-78B. A far-field measurement is used to compare the outgoing power distribution of a long (greater than 1 km) section of test fiber, excited with a uniform overflow light source, with the outgoing power distribution of a short section of the fiber with the mandrel applied. The mandrel diameter is selected to produce a far-field power distribution in the short section that approximates the far-field power distribution of the long section.
The mandrel-wound method is widely used for field fiber optic testing. This method prescribes the use of a Category 1 overflow light source and a reference test patch cord wound around a mandrel. The specification for the Category 1 light source is determined by the Coupled Power Ratio (CPR) test, illustrated in Figures 6 and 2. Clause 6.4.2.1 of ANSI/TIA-568-C.0 provides acceptable mandrel diameters for common multimode cable types (five turns).
The problem with the mandrel-wound test method is that it can lead to significant measurement variability. This variability is due to 1) differences in the power distribution of the modes of the coupled light source depending on the equipment used, since the specification of Category 1 is quite lax** and 2) differences in the characteristics of the multimode fibers used in the test reference patch cords, since the effects of bending can vary depending on the fiber manufacturer.
4.2. Mode Conditioning Filter - Encircled Flux.
Encircled flux (EF) is a novel method for controlling the power distribution of light injected from the end of a test reference cable (TRC) when performing optical loss measurements.
The upper and lower limits of the light power distribution are specified as a function of the core radius and for wavelengths of 850 nm and 1300 nm, as shown in Figure 7 [3]. This EF method was developed by optics experts to reduce the variability of link loss measurements compared to other optical loss measurement methods that use an overflow injection light source and mode conditioning filters (e.g., mandrel winding). Encircled flux is also the test method specified in the IEEE 802.3ba standard for 40 Gb/s Ethernet (40GBASE-SR4) and 100 Gb/s Ethernet (100GBASE-SR10).
TIA Adopts Encircled Flux in New Fiber Testing Standard.
The Encircled Flux test method is specified in TIA-526-14-B, "Optical Power Loss Measurements of Installed Multimode Fiber Cable," and is harmonized with IEC 61280-1-4. The TIA is currently working on a new Appendix 2 to ANSI/TIA-568-C.0, which addresses various fiber optic issues. One section of this appendix replaces the mandrel winding test method specified in TIA-455-78B and TIA-526-14-A with the following:
For multimode cabling, cabling standards stipulate that attenuation measurements must be taken in accordance with TIA-526-14-B, which requires that the injection patch cord outlet meet specific injection conditions determined by the spectral width and power distribution of the modes. These injection conditions can be provided by the field test instrument itself or by using an external mode conditioner.
Comparison of optical loss measurements obtained by overflow injection (OFL)/mandrel winding according to TIA-426-14-A, Method B, in contrast to Encircled Flux of TIA-426-14-B, Annex A
To better understand the variability in optical loss measurements, we designed a controlled experiment in which all variables were kept constant except for the injection conditions.
In the first series of tests, the same permanent bonds PL a), PL b), and PL c) were tested using the same test reference cables (with conventional MMF), TRC #1 and TRC #2, wound around a 22 mm mandrel, using two different test apparatuses, OLTS #1 and OLTS #2. In the second series of tests, the test reference cables were replaced with cables containing an integrated EF module, EF #1 and EF #2, instead of the mandrel-wound version. Optical loss was measured under the different injection conditions described above in the three 20-meter permanent links with the following 2 connectors:
Permanent link a) with conventional multimode fiber
Permanent link b) with bend-insensitive multimode fiber from supplier A
Permanent link c) with bend-insensitive multimode fiber from supplier B
Figure 8 shows the test data.
**Note: A 4 dB difference in the CPR value of a Category 1 light source represents a 150% increase in the power of higher-order modes.
Measured Results
The results of the controlled experiment are presented in Figure 9. The first series of tests was conducted under the same conditions, with an overflow injection light source and a mandrel-wound mode conditioning filter. There is an approximately 0.5 dB difference in optical loss when the tests are performed with OLTS #1 compared to OLTS #2, a difference that affects all permanent links regardless of fiber type. What is the cause of this discrepancy?
The difference in loss measurements can be attributed to the amount of light coupled to the multimode fiber by each of the two test instruments. Although both light sources are Category 1 specified, the specifications are insufficient to guarantee that a uniform power distribution of the modes is injected from the end of a test reference patch cord when using the mandrel-wound mode filter. A difference of 0.5 dB in measurements can mean the difference between passing and failing the link when working with loss budgets on the order of 1.5 dB to 2.6 dB.
In the second series of tests, the optical loss measurements matched (within a tolerance of +/- 0.1 dB) with the two test instruments (OLTS #1 and OLTS #2) when using an Encircled Flux mode-conditioning filter.
Good news!
The good news is that there is general agreement within the industry on test procedures that can be relied upon to provide accurate and repeatable optical loss measurements. The Encircled Flux test method reduces the variation in loss measurements to as little as +/- 10% for link attenuation measurements greater than 1 dB. Compliance with IEC 61280-4-1 can be achieved using a mode controller, which can be integrated into the test equipment or be an external device inserted between the light source and the fiber under test.
The mode controller is a passive device that ensures injection conditions meet IEC 61280-4-1 requirements regardless of the light source used (LED or laser).
Conclusion:
Field optical loss testing is not as straightforward as it seems and can be affected by many variables, such as fiber disparity, the type and quality of the test reference patch cords, and injection conditions (OFL/mandrel wound versus Encircled Flux).
The stringent optical loss requirements for high-speed applications necessitate a precise testing method for field link examination. Industry experts recommend the Encircled Flux testing method for its accurate optical loss measurements for both conventional MMF and BIMMF.
Encircled Flux is also the specified testing method for 40G and 100G Ethernet applications.
Author:
Author: Paul Kish, Director of Systems and Standards at Belden
About the author:
Paul Kish, Director of Systems and Standards at Belden, is a key contributor to the development of TIA, ISO, and IEEE cabling standards and is also a member of the BICSI Technical Information and Methods Committee.
