Published In October 2001
Testing installed fiber optic cabling or patchcords is not difficult, but sometimes it seems the results you get don’t make sense. After 20 years in the business, I’ve probably seen most of the possible errors. If I share them with you, perhaps you will avoid them. The right tools and skills for the job Most errors are related to using the wrong equipment or using the equipment improperly. Never take a piece of test gear into the field unless you fully understand how to use it. Many times someone calls from the field on a crackling cell phone asking how to use some test gear. The owner never tried to use it before taking it out on the job, and assume we can talk them through it. Sometimes installers have the correct test equipment (fiber optic testing requires a power meter and source to test loss) but are missing required accessories, such as reference patch cables or mating adapters. Like the test equipment, reference patch cables must be compatible with the cable plant being measured and in good condition. Occasionally, people ask us why all the cables have a consistent 20dB loss, but under questioning, they reveal that they are testing single-mode fiber with multimode reference cables. Likewise, 3 to 4 dB loss means testing 50/125 fiber with 62.5/125 reference cables. Just as when you are using a new fiber optic connector, you need to practice in the privacy of your own office to make sure you have everything down pat. If not, it’s easy to call the manufacturer from there for advice. Which standard do you use? Testing loss of fiber optic cable plants requires setting a reference value for zero loss. You do this by connecting your test source and power meter with your reference cable(s) and measuring the power through the reference cable(s). This can be confusing, since various standards call for one, two, or three reference cables. Generally, fiber optic networks like Gigabit Ethernet call for a one-cable reference, but some don’t. As you add more reference cables to your zero reference, you reduce the loss you measure, since you include connectors in the reference. Whichever method you use, it must be documented, and preferably agreed upon by the customer. Some tests also require testing at only one wavelength of light, usually 850nm for multimode fiber and 1,310nm for single-mode. But customers and some standards require testing multimode at both 850 and 1,300nm and single-mode at 1,310nm and 1,550nm. Make certain the customer specifies this also, to avoid retesting every fiber. Learning to interpret measurements The best way to understand measurement errors is through experience. Several common problems crop up consistently in measuring loss. The best way to know a measurement is wrong is to know what the loss should be by first doing a power budget. This is where you calculate the loss of the cable under test by knowing the length of the cable, the loss of the fiber per km, and number of connectors. One you know what the loss should be, erroneous measurements will stick out like a sore thumb. The most common error is measuring loss of a short length of cable and getting a “gain.” Trust me, the fiber cannot amplify the test source input! A gain indicates that the measurement was wrong when you set the zero loss reference or the source has drifted. Usually it means the reference cables or the detector on the power meter was dirty when the reference was set, but afterwards the dirt was cleaned off or blew away on its own. Either way, you need to clean the reference cables and reset the zero reference. Your reference power may have changed. Remember that you must not remove the launch reference cable from the source once you set your zero because it may not return to the same value when reattached. If you find the source has drifted, you may want to buy another one—that’s unacceptable behavior for a test source. Another problem is finding the loss is not consistent at all. This problem can be caused by dirt and solved by cleaning the connectors on the reference cables. It can also be a bad connector on the reference cables or poor connection to the test source. Also, do not bend the cables when you are setting references or testing, because bending causes loss. The loss of the cables should be different at each wavelength. In multimode fiber, the loss is about 3 dB/km at 850nm but only 1dB/km at 1,300nm. In single-mode fiber, it’s 0.4dB/km at 1,310nm and 0.25 dB/km at 1,550nm. But we have often received calls from installers who find the loss at both wavelengths is nearly the same. They are puzzled because it should be much less at the longer wavelength. Fiber optic cable is more sensitive to stress from tension or bending at longer wavelengths. If a cable has higher than expected losses at the longer wavelength, it’s a sure sign it is under stress. If you know your instruments, test methods, and expected results, and how to interpret readings, these measurements won’t fool you. HAYES is the founder of Fotec, the fiber optic test equipment company and the Cable U training programs. He can be contacted at Jh@jimhayes.com.