Continuous-wave reflectometer outclasses time-domain types
tim williams
ttc
In checking fiber-optic networks running at 2.5-Gbit/sec data transmission rates, a continuous-wave reflectometer performs more-accurate optical return-loss and reflectance-loss measurements than do optical time-domain reflectometers
In sharp contrast to the measurement performance of an optical time-domain reflectometer (Otdr) when used to check real-time return loss readings in high-speed fiber-optic networks, an optical continuous-wave reflectometer (Ocwr) provides precise results within seconds at either 1310- or 1550-nm wave-
lengths. In addition, the Ocwr is able to process the strong reflections received close to the transmitter end of the network--the so-called dead zone of an Otdr measurement. Used when instal ling fiber-optic cabling, especially in 622-Mbit/sec OC-12 and 2.5-Gbit/sec OC-48 Synchronous Optical Network (Sonet) rings, Ocwr measurements can save test time, effort, cost and resources.
The computed optical return loss (ORL) as measured by an Otdr does not provide sufficient information to detect and resolve the potential reflection problems encountered when configuring high-speed reflection-sensitive fiber-optic networks. Strong reflections at the near or transmitter end of the fiber network under test, in the dead zone, are automatically excluded from the return-loss measurements computed by Otdrs. As a result, these readings might indicate that the return loss is better than it actually is, thereby masking serious reflection problems. Moreover, Otdr return-loss readings are not provided in real-time operation. A full test scan can take several minutes, which is undesirable for most test technicians who prefer test information in seconds.
Key measurements
ORL and reflectance are two key measurements that indicate the magnitude of detrimental reflections within fiber-optic systems. Specifically, reflectance measures the magnitude of individual reflections within the installed fiber. An ORL measurement provides a single number representing the total energy reflected back to the input end of a fiber. This is a relatively new test requirement for many fiber-optic network operators, and it is often motivated by problems encountered when implementing Sonet OC-48 rings.
In practice, return-loss problems might not occur at speeds as high as 622 Mbits/sec. However, when a network capacity upgrade is made to rates of OC-48 or higher, ORL errors generally emerge. Recognizing the potential for such problems, Bell Communications Research in Morristown, NJ, established system return loss and reflectance limits for OC-12 and OC-48 systems (see Table 1). For instance, at OC-48 speeds, Bellcore suggests a -24-dB return loss or better for a Sonet link.
Optical return loss is generally measured by using an Ocwr or Otdr. When trying to determine ORL, users should carefully examine the capabilities of both instruments to determine which proves more efficient and accurate.
The closer a reflective event occurs to the transmitter end of a fiber-optic network, the stronger the reflection is, because reflected light is attenuated over distance. Test analyses demonstrate that typical Otdrs are unable to properly measure these strong and close reflections because they saturate the operation of the Otdr`s receiver.
In contrast, an Ocwr can readily measure return loss made of any reflections. Because of differences in test instrument and measurement capabilities, choosing the proper test instrument for making system return-loss measurements is an important decision, especially when test technicians are trying to ensure that the network is operating within required performance parameters.
An installation and maintenance test tool that has been around since fiber`s infancy, the Otdr operates by launching short, high-energy pulses into an optical fiber. It then measures the pulse transmission time and the amplitude of returned backscatter and reflection signals. From all this information, the instrument calculates and displays a trace and a test summary table, which determines the length of the fiber and the size and location of signal losses and reflections. Depending on the Otdr`s capabilities, the instrument could also make an ORL measurement with each trace.
To make an ORL measurement, the Otdr calculates the area under its trace of the fiber. Mathematically, this area is equal to the total power reflected back toward the transmitter; therefore, it is used to determine the system`s ORL parameter. However, the Otdr`s receiver must be calibrated so that it is sensitive to low levels of reflected power. While this sensitivity calibration enables the Otdr to make measurements over long fiber lengths, it also causes the receiver to saturate and become ineffective when a strong reflection is received.
Most test technicians are aware of the Otdr dead zone--the distance several meters from the instrument where it is ineffective or "blind" and cannot make accurate measurements (see Fig. 1). This dead zone is created by the strong reflection signal (received at the front panel connector on the Otdr) that saturates the instrument`s receiver. Test technicians should note that whenever the Otdr receiver is saturated, the trace is cut off above the saturation point. This cut-off area can be seen when the technician zooms in on a large reflection and observes its peak value being squared off. At that saturation point, the area under the curve cannot be accurately determined.
The Otdr can still function when its receiver is saturated. Therefore, when saturation occurs, the instrument can determine that "at least" an "X" amount of light was received. However, the actual amount of light reflected back could be significantly larger. This measurement deficiency causes many Otdrs to calculate the ORL parameter based on inaccurate data.
For example, if a fiber span requires a 25-dB specification for ORL and the test instrument indicates a 㿈-dB value, the test technician is unable to determine whether the span loss requirement has been met. Furthermore, the reported parameter might be due to a reflection caused by the test jumper cable connector that joins the test instrument to the fiber span. Although this cable connector could saturate the Otdr receiver, the test result might still be within the system loss specification. Consequently, the fiber span could be within specification--which the Otdr is unable to verify.
Test studies confirm the importance of tracking how often ORL problems occur and the magnitude of the reflections. Years of ORL testing reveal that the major source of return-loss problems is dirty connectors. Often, this problem occurs at the jumper cable connection from the transmission equipment into the span. Unfortunately for Otdr measurements, these connections are positioned within the Otdr`s dead zone and cannot be measured accurately.
The same problem exists at the far or receiver end of a fiber. For example, when a reflection occurs at a short distance in front of the far-end connector, the Otdr cannot reliably measure its value because Otdr resolution degrades over long distances and, therefore, groups the end reflection parameters together. Consequently, precise ORL measurements cannot be made at both ends of a long fiber-optic network with an Otdr.
In contrast, precision return-loss measurements can be accomplished using an Ocwr. This reflectometer differs markedly from an Otdr in that it uses a continuous-wave light source rather than a pulsed light source. A pulsed source requires the instrument`s receiver to be set at a high gain level so that pulses reflected from long distances can be properly measured for accurate Otdr traces. When a continuous-wave source is used, the amount of light reflected back is constant (not a short pulse). The Ocwr receiver gain can, therefore, be adjusted for optimal return-loss measurements.
Test approach
The Ocwr monitors its continuous output power and simultaneously measures the amount of power reflected back to its transmitter. This capability allows the receiver to be optimized for maximum ORL range, thereby preventing saturation. To make the measurement, the technician plugs the Ocwr into the system under test, selects return loss and obtains a true return-loss reading within three seconds. In this test approach, the Ocwr detects the same light as does the transmission equipment. It then makes its measurement of ORL from the true total amount of light reflected, rather than by making a calculation based on potentially incorrect trace data.
To illustrate the difference between Otdr and Ocwr return-loss measurements, a comparison test was made using a fiber-optic setup representative of a typical interoffice Sonet link (see Fig. 2). Strong reflections were intentionally introduced into the link to represent a situation where a test technician needs to identify and locate optical reflection problems.
In the test link, a 9-m-long test jumper cable was connected from the Sonet transmitter to the fiber distribution frame within a central office. A pigtail from the frame was fusion-spliced to 10.5 km of outside-plant fiber-optic cable. A midspan reflective connection was made, as was a defective splice (0.5-dB loss) 70 m away. After the splice location, an additional 2.2-km length of outside-plant fiber was terminated at the far-end or receiver fiber distribution frame. This frame was connected to the Sonet receiver by a 15-m-long test jumper cable.
During the test, the fiber-optic link was initially scanned without the Sonet transmitter and receiver equipment in place; this setup represented the test conditions for a new fiber installation (see Fig. 3). The central office jumpers were left in place because they designated the likely sources of reflection problems. A flat-polished connector was installed on the 9-m jumper at the link to the 10.5-km cable. This connection was done intentionally to create an unacceptable reflection that would be detected and reported during the reflectance and return-loss testing.
Next, the fiber-optic cables were scanned using an Ocwr and two different Otdrs. When the scan was performed with the Ocwr, test results were obtained in less than five seconds, with no need to disconnect and reconnect to another test set (see Table 2). The ORL measurements at 1310 and 1550 nm were 10.4 and 26.4 dB, respectively. These loss results were graded as unacceptable for a Sonet OC-12 or OC-48 link. To ensure that the receiver connector was not contributing significantly to the ORL, the receiver end was terminated with a mandrel wrap.
Then, the same measurement was taken using Otdr-1 (see Fig. 4). This instrument was manually configured to use optimal scan parameters to detect all events because automatic operation incorrectly scanned to 50 km using a 4000-nsec pulsewidth. Instrument parameters were, therefore, set to scan 25 km using a 50-nsec pulsewidth. After testing, the Otdr did not detect the ends of the 9- and 15-m jumpers. In addition, the receiver end was incorrectly reported at the connection between the 2.2-km fiber and the 15-m jumper. This Otdr automatically computed the ORL when the trace was made. The total return loss equaled less than 21.3 dB, an unacceptable level for either OC-12 or OC-48 system operation.
Upon reviewing the Otdr trace, the test technician might conclude that the open end is the source of the problem and would either accept or would terminate the far-end and rescan the link. Note that the computed ORL value-- less than 21.3 dB--is a best-case value. Instead, the test technician needs to know the worst-case value. The system under test should not be accepted under these conditions.
To give the Otdr another measurement test, the fiber was rescanned with a narrow 20-nsec pulsewidth in an effort to improve resolution to the point where all events could be detected. Unfortunately, the test results proved worse. Event 1 at 9 m was still not found, the 0.5-dB loss at the 70-m point past the end of the 10.5-km fiber was not detected, and total return loss was reported as not measurable. Based on these results, the test technician might accept the link as presented with no indication that an unacceptable reflection existed within the central office.
To corroborate these measurement findings, a second and different type of Otdr was used. This instrument was set to automatically scan to 21 km using a 300-nsec pulsewidth. It disclosed the following test results: The event at 9 m was missed; the 0.5-dB loss event was missed; the end of the 15-m jumper was missed; the open end was incorrectly reported as being more than 1000 m beyond the actual end; and the return loss was reported as less than 30.9, which was judged as an inconclusive measurement.
For further analysis, the second Otdr was then reconfigured to scan to 15 km using its narrowest (50-nsec) pulse width. The latest tests results were better than those obtained by the first Otdr. The second Otdr failed to detect the end of the 9-m jumper but did find all of the other events. In addition, it reported a return loss of 28.8 dB, more than a factor of two different from the actual value. By accepting the 28.8-dB value, a technician would wrongly approve a system for OC-12 or OC-48 traffic. u
Tim Williams is product marketing manager at TTC, Germantown, MD.