Choices for equipment include a separate meter and source; an OLTS (optical loss test set), which is a single instrument that includes both; or adapters for copper testers that usually have OLTS functionality for readout on the copper tester. The separate meter and source are usually the most cost-effective, since you need a source on one end of the fiber and a meter on the other, making it possible to use a single set of instruments for testing.
The OLTS or adapter method requires two of the combination instruments, often doubling the cost of the equipment, but also often allowing testing of two fibers simultaneously, saving labor cost.
Next, we need reference test cables. Remember no fiber optic connector has a “loss” by itself; loss is defined by the loss of the connector mated to a reference connector. The reference connector is assumed to be a high-quality connector with low loss, so it does not add to the measurement uncertainty. Our primary concern, therefore, is that the reference cables be in good condition.
The reference cables must mate with the cables to test. That means the connectors must be the same or at least “mateable.” The connectors that have 2.5 mm ceramic ferrules-SC, ST, FC and even the older FDDI and ESCON connectors-can be adapted to each other easily using hybrid mating adapters.
Likewise, the new SFF connectors with 1.25 mm ferrules-the LC, MU and LX-5-are intermateable. But what about others of the 85 or more fiber optic connectors built in the last 25 years? Or more important, what happens when your test equipment does not match the connectors you are testing?
I thought of these questions when I wrote the original standard for testing installed cables many years ago. OFSTP-14 (for multimode and its single-mode equivalent, OFSTP-7) offers three ways to make loss measurements, with the difference being the way the instruments are referenced for “0 dB” loss. If you have ever read OFSTP-14 (officially known as TIA/EIA-526-14), you know there are three methods (A, B and C) listed to make loss measurements, using references with one, two or three launch cables.
If you are testing two connectors such as the ST or SC that are mateable with a simple mating adapter and your test equipment has a compatible connector, you can use a single cable reference (method B), the launch cable. This has the least uncertainty and gives the highest loss value, since no connector is included in the reference set. This method is called for in the TIA-568B standard.
Suppose your test equipment is incompatible with the connector being tested, say test equipment with SC connectors testing LC cables. If the connectors on the cables being tested can be mated to each other with an adapter, you can use method A, which sets the zero reference with two cables, a launch and receive cable. You include one connector loss in your reference, which lowers the loss you measure, but that is not a big issue if you report the method you use.
The big problem comes when the test equipment is set up for connectors different from the cable plant being tested, and the cable plant connectors are plugs and jacks like MT-RJ or OptiJack. You need to have reference cables with plugs but unless your power meter has an adapter for that plug, you cannot do either a one or two cable reference.
You can use method C, which sets the loss reference with three cables-a launch cable with a plug connector, a receive cable that also has a plug and a third cable that has two jacks on each end. The reference measurement has two connector losses included, and it allows making a valid measurement with any combination of test equipment and cables. For that reason, this method has become an international standard.
Whenever you need to test fiber optics, consider what connectors you need to test and what type of fiber is used in the cable plant. With that information, you can decide what method you will need to use and what kinds of reference cables you need. EC
HAYES is a VDV writer and trainer and the president of The Fiber Optic Association. Find him at www.JimHayes.com.