Figure 5-3.

Example of a composite fault signature consisting of both scattering and absorption loss. Gaussian probe pulse.

LOG BACKSCATTER POWER

Figure 5-4.

Backscatter signature for a graded-index fiber with a fusion splice at the midpoint. The three other irregularities are unidentified. Fiber J.

Microbending signatures are also absorption-like. The capture fraction associated with radiation loss due to microbending was estimated in the following experiment. We wound a 1 km graded-index fiber loosely in several layers on a felt-covered, 20 cm diameter drum. Ten plastic rods, each 6 mm in diameter, were positioned transverse to the winding. Lateral pressure was applied to the layers of fiber by pressing them against the rods using an adjustable strap concentric to the drum. The effect on the fiber is shown in figure 5-5. The lower curve represents the backscatter response with tension applied to the strap. Increased loss (about 3.1 dB total) occurs at locations where the rods were in contact with the fiber. The upper curve represents the response when the tension was removed. Some residual stress effects can still be observed. It will be noted that, in the stressed state the backscatter signal is a monotonically decreasing function of time. This implies that the capture fraction associated with the removed radiation is very small. A more quantitative estimate of the microbending capture fraction F may be obtained from a comparison of the two backscatter scans at t=0. From eq (1-1), the backscatter power in the stressed state differs from the corresponding power in the unstressed state only in the term anFn. Then, assuming the excess loss in figure 5-5 is a uniformly distributed radiation loss, an is about 7x10-4 m-1, about the same as as. However, the stressed t=0 signal does not increase within an uncertainty of about one percent. This implies that Fn < 0.01 asFs/an, or F. < 5x10-5. This is the basis of the estimate given in table 2.

n

n

A third example of an absorption-like signature is given by the effect of radiation from fiber bends. The magnitude of the capture fraction associated with macrobending of this sort was inferred from the following experiment. A graded-index fiber, which was loosely cabled in a plastic tube, was wound on drums of different diameters. Figure 5-6 shows the resulting signature when the input half of the 570 m cable was wrapped on a drum of 30 cm diameter and the remaining cable wrapped on a drum of 10 cm diameter. There is a small change in slope at the midpoint which implies an excess loss of about 0.9 dB/km. Most of this loss can be attributed to radiation whih is emitted as a result of the decreased bend radius. Figure 5-7 is a similar scan where the final 40 m of the cable is wound on the smaller drum, and a change in slope is also apparent. To obtain an upper limit on the macrobending capture fraction we modeled the backscattering process on a computer using the known parameters for fiber H, and assuming a value of Fn which was judged to produce a barely perceptible change in the backscatter signature. If the observed increase in radiation loss had one photon in 1700 returned in the backward direction, the signature would appear as in figure 5-8. The backscatter response in this case exhibits a signal increase at the location corresponding to the junction of the two drums. Since no such signal can be detected experimentally, we conclude that, for this fiber, Fn < 0.0006. This result is also given in table 2.

Backscatter signatures resulting from a microbending experiment (see text). Lower curve exhibits effect of pressure-induced microbending. Upper curve represents the response in the relaxed state. Fiber K.

Backscatter signature for a fiber wrapped on drums of differing diameter. input half of the fiber is wound on a drum of diameter 30 cm, the output half of the fiber is wound on a drum of 10 cm diameter. Fiber H.

TIME (us)

Backscatter signature under conditions similar to figure 5-6, except the final 7 percent of the fiber is on the smaller drum. Fiber H.

Computer simulation of radiation loss with an excess F value of 0.0006, other parameters similar to fiber H. Excess radiation loss begins at time 345.

5.2 Scatter-Like Signatures

We have seen previously, in figures 2-2 to 2-4, an example of a localized scatter-like fault signature. The scan is also reproduced in figure 5-9 on an expanded scale. The magnitude of the feature depends somewhat on the launching spot size and its location on the input face of the fiber. It is possible to observe a small decrease in the slope of the log backscatter signal following the fault. This could possibly be explained by a redistribution of probe pulse energy into less-lossy modes. The most likely explanation for the origin of the fault is an elongated bubble or dielectric filament of the type described by Rawson [52].

Figure 5-10 represents the scatter-like signature due to a commercial coupler which joins two identical step-index fibers. The signal at the junction interface is off scale. The one-way loss at the connector is about 1.4 dB. This represents a rather extreme example of a scatter-like signature.

6. COMPUTER SIMULATIONS

One of the main motivations for the experimental determination of backscatter parameters is to obtain a realistic data base for purposes of computer modeling. The actual backscatter display signatures can represent a rather complex interaction of many variables. For example, the Rayleigh as well as fault signatures will depend on absorption loss, scattering loss, capture fractions, input pulse shape, input pulse duration, wavelength, SNR, type of backscatter display (direct, logarithmic, differential logarithmic), fiber type (single mode, multimode) and the spatial distribution of any perturbations along the length of the fiber. Computer generated displays which can independently vary these parameters can greatly assist in interpreting and understanding experimental signatures. Also, we may be led to a preferred display scheme for backscatter signatures.

Some of these computer-generated signatures have appeared in this report. comprehensive atlas of backscatter signatures is in preparation for a future report in this

series.

7. CONCLUSIONS

We have described a laboratory OTDR in some detail, indicated its potentialities and limitations, and suggested ways in which the SNR may be improved. We have also described experimental techniques for estimating fiber Rayleigh scattering, capture fractions, and group velocities. The OTDR system was used to examine the backscatter signatures of a number of fibers and fiber perturbations. Some of the main conclusions from the experimental work may be summarized as follows:

1.

Fibers from some manufacturers exhibit much smaller irregularities in background backscatter than similar type fibers from other manufacturers. These signal variations are probably due to diameter fluctuations. It is much easier to observe a perturbation or fault signature on a uniform background. Therefore, in some applications, for example