Despite its reputation as a highly accurate measurement device, the OTDR is only one part of a comprehensive strategy for testing fiber links in the premises.
It's common for the customer of a fiber-optic cable installation to require documentation of test results before accepting the job and paying for the work. This obviously leads to certain but often conflicting requirements for the contractor. Testing takes time, so completing all of them in the minimum time means more profit.
Testing, however, needs to be done carefully toensure the measurements are accurate—that is, the measurement made gives a value close to the "real" value. And that can take time. Accurate testing will ensure that no good cables are rejected and no bad ones missed, so the contractor will not have to repair what are really good cables and get callbacks on bad ones.
Lots of time and cost can be saved if you as contractor and your installers know the proper measurements that need to be made, understand how to make those measurements correctly, have the proper tools, keep them in good condition, have them calibrated regularly, and know how to use them efficiently. You also need to convey to the customer that what is being done is in line with industry convention and standards.
Industry committees spend massive amounts of timeand energy developing standards that ensure accurate testing. But those standards are generally written for manufacturers, not users. This tutorial will give you insight into what tests are required, what problems are inherent in testing multimode fiber, how measurement techniques differ, and how to interpret and document the results
How to test premises cables
In premises cabling systems designed for use with LAN backbones, fiber-to-the-desk, closed-circuit television, industrial control signals, etc., there are three tests that may be done: connection verification, insertion loss, and optical time-domain reflection. All cables should betested for continuity with a visual fault locator or fiber tracer, and the connections verified.
In my experience, many fiber-optic cabling problems are caused by poor documentation or confirmation of connections. Since each link consists of two fibers, onefiber must connect a transmitter to a receiver and theother the complementary pair. Documentation andmarkings should allow these connections to bemade simply. This is easily confirmed with avisual light source coupled into the fiber.
End-to-end measurements
The measurement needed for confirming the quality of the installation is the optical loss or insertion loss of each of thefibers in the cable. Loss measurements are made end-to-end on the permanently installed cable plant—the equivalent of the unshielded twisted-pair (UTP) permanent link. Industry standards call for making that measurement with a test source and optical power meter, sometimes called an optical losstest set (OLTS) and reference test cables.
Proposals have been made to also allow for testing of installed cable with just an optical time-domain reflec-tometer (OTDR), but no accepted standard requires this. TIA-568 (both the B version and the soon-to-be-published C version) follows the industry convention, requiring insertion loss testing (called Tier 1 testing in TIA-568) and permitsOTDR testing also (Tier 2) to provide additional information, but does not allow OTDR-only testing in lieu of insertion-loss testing.
OTDR testing of premises cable plants instead of insertion-loss testing causes much confusion among contractors and customers. Hardly a week goes by that I do not get a callregarding this issue. Misinterpretations of these requirements have led to some unhappy instances, including misreadingOTDRs causing the removal and discarding of $100,000worth of good cable and the retesting of 1,100 cables of12 fibers each, as well as several instances of customers return-ing OTDRs to the distributors who sold them the units.
There are five industry-standard ways to test premisesfiber-optic cable (three for insertion loss and two for OTDRs), depending on how you use reference test cables for your setup. Insertion-loss testing can use one, two, or three referencecables to set the "zero dB loss" reference for testing, and each way gives a different loss. Generally, standards prefer the one-reference-cable loss method, but test equipment must use the same fiber-optic connector types as the cables under test. If the cable has different connectors than the test equipment (e.g., LCs on the cable and SCs on the tester), it may be necessary to use a two- or three-cable reference, which will give a lower loss since connector loss is included in the reference and will be subtracted from the total loss measurement.
Any of the three reference methods are acceptable, as long as the method is documented. Be careful, however, since most network link losses assume a one cable reference, which can affect the acceptance of the cable.
OTDRs require a launch cable for the instrument to settle down after reflections from the high-powered test pulse overloads the instrument. OTDRs have traditionally been used with long-distance networks where only a launch cable is used, but this method does not measure the loss of the connector on the far end. Adding a cable at the far end allows measuring the loss of the entire cable, but negates the big advantage of the OTDR—that it makes measurements from only one end of the cable.
All these test methods have seriousissues with accurately testing multimode cable. In the 25 or so years I have been involved with fiber-optic standards, making accurate loss measurements on multimode fiber has been a constant and confusing subject of discussion within the standards committees. We tried to understand how light travels in multimode cable plants and how components, such as connectors, affect the way that light travels. Then we tried to understand how the losses of fiber, connectors, and splices were affected by the methods used for testing.
What follows is how this works, how it affects your measurements, and how you can try to control test conditions toenhance your test accuracy.
No comments:
Post a Comment