Mandrel wraps unravel return-loss measurements

Mandrel wraps unravel return-loss measurements

Comparing return-loss readings with and without a mandrel wrap helps locate the reflection source

Michael Scholten

Telecommunications techniques Corp.

Because digital data rates are steadily increasing and analog modulation schemes are becoming more complex as more information is pumped through fiber-optic networks, system performance is increasingly troubled by reflection losses. Fiber-plant managers must therefore have an effective strategy for detecting the presence of unacceptable reflections and locating their source. Testing for reflections before system turn-up enables the detection and correction of potential problems, ensuring reliable in-service operation. Recommended test tools include optical continuous-wave reflectometers, optical time-domain reflectometers (OTDRs) and the strategic use of mandrel wraps.

Telecommunications carriers and customers are especially concerned about reflections that occur in such high-speed fiber-optic transmission systems as synchronous optical networks. The Sonet infrastructure, with its easy access to tributaries, powerful overhead capability for network control and monitoring, and standard optical interface, is becoming the transmission system of choice for heavy payload users. For example, Sonet optical carrier, level 48 (2.4-gigabit-per-second) systems are common, and Sonet OC-192 (10-Gbit/sec) transmission has been demonstrated. However, the higher the Sonet rate, the more opportunities exist for reflections to degrade system performance. Therefore, before the fiber-optic cable is commissioned, the negative effects of reflections must be isolated and minimized to prevent future maintenance and to ensure network performance and quality of service.

Optical reflection measurements become important, therefore, because as data rates increase, optical pulsewidths narrow. At OC-48, optical pulses are less than one nanosecond wide. Transmission systems must reliably detect these pulses to receive reliable data. Multiple optical reflections caused by dirty connectors, defective mechanical splices or fiber flaws can generate multiple, reduced images of the transmission signals at the receiver. These received signals add to the transmitted signals and generate noise. Excessive noise causes bit errors in digital transmission, corrupting user data or delaying receipt of data because of retransmission.

For reliable OC-48 data transmission, a system return-loss specification of at least 24 decibels must be maintained, and no individual reflectance can exceed -27 dB. If testing indicates that system return loss is lower than 24 dB, the excessive sources of reflection must be found and eliminated to ensure reliable communications. In addition, hybrid fiber/coaxial-cable systems that employ analog transmission over fiber are highly intolerant of reflections. Multiple reflections directly degrade the carrier-to-noise ratio. For analog video systems, reflections larger than -55 dB cannot be tolerated if required carrier-to-noise ratio levels are to be met.

Sources of optical reflection are well-known. Glass-air interfaces at mechanical splices, dirty or mis-keyed connectors and air-gap attenuators typically create the worst reflections. The deleterious effects of reflections are widely known to fiber-optic equipment vendors; therefore, today`s optical components--if clean and undamaged--should not generate excessive reflections. Unfortunately, cleanliness and perfection are not guaranteed. The steady increase in bandwidth demand often requires that previously installed systems be upgraded to operate at higher data rates. In many installations, these older systems contain unacceptable reflections because of flat-polished or poorly cleaned connectors.

Detecting undesirable reflections

The presence of unacceptable reflections can be measured by two common methods--reflectance testing and return-loss testing. Reflectance testing measures the ratio of reflected light at a discrete point to light incident on that point. The reflectance of individual components after installation in a system can be measured only by using an OTDR. Mini-OTDRs typically provide a table that indicates the location and reflectance of reflective components. The user can set a threshold so that unacceptable reflections are highlighted in the table.

Return-loss measurements provide a single number, signifying the ratio of incident (or input) light to total reflected light, usually expressed in dB. Because reflections are undesirable in optical transmission systems, lower values of reflected power are preferred. Higher and more desirable return-loss values result when reflections are reduced.

Return loss and reflectance testing--which are often confused--can be differentiated by noting that reflectance is a measure of individual reflective defects, whereas return loss registers the combined effects of all reflections in an installed system. By way of analogy, compare the two methods to optical loss of a component (for example, connector or splice loss) versus end-to-end system loss.

Unacceptably low return loss points to the presence of one or more unacceptable reflections. Return-loss testing is quick, requiring only a few seconds to complete. It is generally made using an optical continuous-wave reflectometer. Reflectance of optical components can also be characterized by using an optical continuous-wave reflectometer before installation.

Measurement error

To measure return loss, connect an optical continuous-wave reflectometer to the fiber or system under test. Often, users want to make sure reflection at the connection between the test equipment and the fiber is not contributing an "error term" to the system return-loss measurement. The return-loss value measured by this reflectometer includes reflections from the system under test, as well as any reflection at the connection between the optical continuous-wave reflectometer and the system being tested.

If the reflection from the connection to the system is as large as the return loss of the system, the total return power is doubled, which decreases the return loss by 3 dB and introduces a significant error into the system return-loss measurement.

Increasing use of optical continuous-wave reflectometers and OTDRs for reflection testing leads to confusion about when and why to use "mandrel wraps." Fiber-optic test technicians know that a tight bend in a fiber causes optical loss. At the tight bend, light escapes from the fiber`s core into the cladding, and from there, into the outside world. A mandrel wrap is an effective way to introduce temporary high loss into a fiber-optic path. It is fabricated by tightly wrapping singlemode fiber several times around a mandrel (a small-diameter rod such as a pen). The mandrel wrap causes a significant bending loss at typical transmission wavelengths, such as 1310 and 1550 nanometers. The loss increases with more wraps or a smaller wrap diameter, and is larger at 1550 nm than at 1310 nm. By placing a mandrel wrap in the fiber immediately past the connection in the system under test, it is possible to measure the return loss of that initial connection. Be careful when you are bending fibers, especially when the fibers are not protected by a buffer or jacketing; otherwise, lightwave losses may be introduced.

Reducing light

The mandrel wrap markedly reduces the amount of light that reaches the system under test. Any reflections from the fiber under test are further attenuated as they return back through the mandrel wrap. Five wraps of a jumper around a 5/16-inch mandrel (a typical diameter of a ball-point pen) are usually enough to attenuate system return loss by more than 50 dB at 1310 and 1550 nm. With the wrap in place, the measured return loss is due entirely to the return loss of the connection before the mandrel wrap.

Comparing return-loss readings with and without the wrap in place determines whether reflection at the test connection is introducing appreciable error or dominating the system return-loss reading. Guidelines are available for interpreting system return-loss readings by comparing readings with and without the mandrel wrap.

If an optical continuous-wave reflectometer provides a reference return-loss capability, the reading obtained with the mandrel wrap in place may be stored and automatically subtracted from the reading with the wrap removed. Thus, a corrected system return-loss reading is automatically provided.

Locating strong reflections

Common sources of excess reflections are glass-air interfaces at open fiber ends, air-gap attenuators, poorly mated connectors (because of debris that keeps endfaces apart or connectors that are improperly keyed), overpolished connectors and dirty connectors. These problems typically occur at either end of fiber systems or at mid-span fiber crossconnects. An OTDR may be used to detect strong mid-span reflections. However, multiple, closely spaced connections might appear as single reflective events on the OTDR trace.

A problem can be isolated by taking return loss or OTDR measurements with mandrel wraps strategically placed just before and then after potentially high-reflectance connections. If return loss significantly increases (improves) when a mandrel wrap is placed at a point in the fiber, the source of strong reflection is beyond the mandrel wrap.

Some optical continuous-wave reflectometers are available with a low-reflectance connector, such as an angle-polished FC/APC or newer ultra-polished FC/PC. Reflection from such connectors is typically less than 0.0001% (return loss is greater than 60 dB); therefore, the contribution of the connector is insignificant and can be ignored.

Unfortunately, FC/APC connectors are not compatible with widely used FC/PC connectors. Mating an FC/APC to an FC/PC connector will not only result in poor loss and reflection characteristics, but might also damage one or both connectors. For convenience in the field, some customers order their test instruments equipped with FC/PC connectors. Unfortunately, return loss from even a clean, mated FC/super-PC connection is typically 45 dB. If either connector ferrule is dirty or damaged, much lower (worse) return-loss readings might result.u

Michael Scholten is systems engineer at Telecommunications Techniques Corp. in Germantown, MD.

More in Components