Impact of MFD mismatch on OTDR splice loss measurements

Th 101022


Apparent large losses, "gainers," and real losses can confuse technicians not familiar with the effects of mode field diameter mismatches.

GABOR KISS, Telcordia Technologies

There are three types of fiber in common use: matched clad, depressed clad, and dispersion-shifted (DS). These fiber types are characterized by differences in mode field diameter (MFD). We'll look at the splicing implications of the MFD mismatch that occurs when commingling these fibers.

There are two effects to note. One effect is an apparent splice loss in one direction that is matched by an apparent gain in the other direction, so it has only an aesthetic impact on bidirectional measurements. However, the confusion resulting from this "impossible" occurrence can be distressing. The second effect is a genuine loss that depends on the magnitude of the MFD mismatch. The large MFD mismatch between DS and non-DS (NDS) leads to extremely significant genuine losses that should be avoided if possible.Th 101022

Figure 1. Mode field diameter mismatches can add to loss measurements (a) or reduce them (b). In either case, the unwary technician may end up with inaccurate measurements that must be taken into account when performing splices.

Two types of NDS optical fiber are in common use: depressed clad and matched clad. These terms refer to the relationship between the refractive indices of the fiber core and fiber cladding.

To obtain optical waveguiding, the refractive index of the cladding must be lower than that of the core. A simple way to understand that is to note that the refractive index of a medium is inversely related to the speed of light in that medium, so passing from a high to a low refractive index causes light to speed up (i.e., deflect back toward the core). The refractive index difference at the edge of the core is slightly higher in depressed clad fiber than in matched clad fiber. The magnitude of the refractive index difference affects how tightly light is guided, so that the depressed clad fiber is somewhat less sensitive to bending loss than matched clad.

The differential in core and cladding refractive indices affects the MFD required to keep other optical properties such as cutoff wavelength within desired limits. The MFD of depressed clad fiber is thus somewhat smaller than that of matched clad fiber by several 10ths of a micron. Bellcore GR-20, "Generic Requirements for Optical Fiber and Fiber Optic Cable," allows a median MFD range of 8.8 to 9.3 microns. The tolerance for departure from the mean is a half-micron, hence the actual range of MFD for NDS fibers which conform to GR-20 is 8.3 to 9.8 microns.

Depressed clad fiber will cluster near the bottom of this range and matched clad fiber will cluster near the top of this range. A consequence of this difference is that if other geometrical parameters are equal, then the degree of core overlap when splicing matched clad fiber is slightly greater; hence, average splice loss is slightly lower. Th 101023

Figure 2. The splicing of dispersion-shifted and non-dispersion-shifted fiber can result in significant differences between forward and reverse loss.

The geometry of DS fiber is dramatically different from NDS fiber in that the MFD is on the order of 7 microns. This low value is determined by considerations such as bend resistance and dispersion slope.1 Efforts to create large effective area dispersion-shifted and nonzero dispersion-shifted fibers have been reported.2

Apparent loss
There are two consequences of experiencing an MFD mismatch across a splice: The first is an optical time domain reflectometer (OTDR) artifact and only seen in a one-way OTDR measurement; the second consequence is a genuine loss that is actually present and would be seen by a light source/power meter or a bidirectional OTDR measurement.

An OTDR displays the backscattered power as a function of time. Light emanating from volume elements that are farther away will be detected at later times and is attenuated as it traverses loss events. The scattering ability of fiber is inversely related to its MFD.

If the fibers being spliced are homogeneous in their scattering ability, this indirect measurement works well. But if the fibers are not homogeneous, then artifacts result. If the fiber beyond a splice scatters less than that entering the splice, as will be the case if its MFD is greater, the apparent splice loss will be greater than the true splice loss. Conversely, if the fiber beyond the splice has smaller MFD, thereby scattering more, then more power is detected to be emanating from beyond the splice and displayed as a negative contribution to the splice loss. If the MFD mismatch is great enough, then an apparent "negative splice loss" can be displayed. This apparent violation of conservation of energy can be disconcerting, to say the least (see Figure 1).

Either case can be troublesome to a splicer who is not aware of both possibilities. In the excess loss case, the splicer may repeatedly remake the splice in a vain attempt to reduce it to an acceptable loss. In the case of a negative contribution, the true loss may be large but hidden. Or the splicer may remake the splice until it falls within the "acceptable" range of 0-0.2 dB, which may in reality be an unacceptably lossy splice! The true loss can only be measured with an OTDR by using a bidirectional average.

Real loss
The assumption here is that none of the manifold excess loss contributions in real splices, such as axial, angular, or longitudinal offset, occlusion of the cores by particles, and bending losses, are present. We deal only with the loss due to MFD mismatch.

Felix Kapron, at the NBS Optical Measurement Symposium in 1986, had indicated that when an MFD mismatch is present, there is a small but nonzero contribution to the real loss.3 In other words, even if the splice is perfect, the presence of an MFD mismatch will produce some loss that will not be eliminated by averaging forward and reverse OTDR measurements.

The magnitude of this contribution is given by: dB(R) = 20 log10 (0.5 [(MFD1 /MFD2) + (MFD2 /MFD1)]) ~ 4.343 R2 for |R| <0.8, where R = 2(MFD1 - MFD2)/(MFD1 + MFD2).

The backscatter contribution is given by: B(R) = 5 log(MFD1 /MFD2)2 ~ -4.343 R for |R| <0.35.

The apparent loss is then L(R) = dB(R) + B(R) = 4.343 (R2 - R). We see that for MFD1 = 8.8 microns and MFD2 = 9.3 microns, R = -0.0552, so dB(R) ~ 0.133, B(R) ~ -0.24, and L(R) ~ -0.227.

Thus, for fibers at the two extremes of the GR-20 median criteria, a perfect heterogeneous splice will appear to display an OTDR "gainer" -0.227 dB when seen from the large MFD side and a loss of 0.253 dB when seen from the small MFD side.

Table 1 indicates the magnitude of the different quantities of interest for fibers spliced perfectly to a depressed clad fiber with MFD1 = 8.8 microns or MFD1 = 7 microns, with the real loss contribution highlighted.Th 101025

Since the real loss contribution is always positive, regardless of which direction a non-homogeneous splice is measured, the apparent loss will be different in the forward and reverse directions. Table 2 gives some examples of interest.

Clearly, the contribution to real loss from MFD mismatch is exceedingly small when splicing matched clad to depressed clad fiber. It is only at the extremes of the GR-20 tolerance that a real loss contribution of >0.1 dB is incurred. By comparison, the loss requirements for fusion splicers in GR-765, "Generic Requirements for Single Fiber Single-Mode Optical Splices and Splicing Systems," are 0.2 dB for passive alignment and 0.1 dB for active alignment. But when the inhomogeneity involves commingling of DS and NDS fiber, extremely large real losses will be incurred, at least 0.125 dB and up to nearly 0.5 dB.

Real world impact
In showing that the real world impact of commingling matched clad and depressed clad fiber is negligible, a couple of simple suggestions would be:

  • A rule can be instituted to correct the OTDR estimates: "When splicing small (depressed clad) to large (matched clad), subtract 0.25; when splicing large to small, add 0.25."
  • Rather than measuring each splice as it is made, measure the loss of the completed span by setting the OTDR markers at the two ends of the span. If the average loss falls within requirements, there is no need to look further. The impact of MFD mismatch on the average loss of many splices decreases with number, since that is not an additive phenomenon.
Th 101026

For the case where splicing is DS to NDS and both ends are accessible, a field study of many hundreds of splices gave the histogram in Figure 2 for differences between forward and reverse loss. In some cases, differences as high as 2.5 dB were found, and they corresponded to OTDR traces exhibiting a dramatic "stair step" appearance (see Figure 3). In this case, both the forward-reverse difference and two-way average were known. Assuming an MFD1 of 7 microns for the DS fiber and varying the MFD2 of the other fiber to which it is spliced, we obtained the numbers in Table 3. The fitted values for MFD2 indicated the presence of NDS fiber. In addition, values of differences between 0.86 and 1.25 dB in the histogram indicated that DS fiber was spliced to both depressed clad NDS and matched clad NDS fiber. Clearly, this run was built with whatever scrap fiber happened to be available!

Meanwhile, a good example of the effects of MFD mismatch between DS and NDS fiber when only one end was accessible was observed in an undersea communications system recently installed off western Australia. The system consists of eight individually procured lengths of commercial undersea cable spliced together to make a 50-km span. Each of the procured segments was individually tested with an OTDR prior to splicing, while after splicing, the entire span was tested using a power meter. Since power meters only measure end-to-end loss, the MFD mismatch of one segment was not apparent until the power-meter measurement was followed up with an OTDR measurement after the installation.Th 101024

Figure 3. With differences between forward and reverse loss as great as 2.5 dB, mode field diameter mismatches can create dramatic optical-time-domain-reflectometer readings.

The OTDR revealed a large gaining event at 24.2 km of 1.4 dB and a large loss at 36 km of 2.2 dB. That was observed from an onshore station following the installation of the cable. Knowing MFD mismatches generate OTDR artifacts, a measurement in the reverse direction would confirm the mismatch. Unfortunately, a reverse measurement was not possible, since the sea end of the cable was now inaccessible. As a result, cable factory records had to be researched to find a possible explanation for the two events.

Measuring the distance between the spliced cable joints and comparing those lengths to the factory-procured lengths, the span between the gaining and losing event was easily matched to a specific factory segment. Factory records did indeed confirm DS fibers for the segment in question and NDS fibers for the remaining seven segments in the 50-km span.Th 101027

It should be noted that the simple equation given above to estimate the backscatter contribution is not able to predict a "gainer" as large as the 1.42 dB seen above. Indeed, as shown in Table 2 (the last three rows), even applying the equation to unrealistically large and small MFDs fails to generate gainers larger than 1.05 dB, as the real loss contribution dominates. Undoubtedly, the equation given by Kapron is breaking down at these extremes.

This particular fiber communications system permits the use of DS and NDS fiber intermixed without compromise to system performance. However, OTDR evaluation of the splice loss can be a challenge to a splicing technician who is unprepared for the MFD mismatch.

Gabor Kiss is a research scientist at the Advanced Systems Reseach Laboratory of Telcordia Technologies (Morristown, NJ), and Christopher Littlejohn is a manufacturing engineering manager at Maripro Inc. (Goleta, CA), a wholly owned subsidiary of the Nautronix Group.


  1. K. Ohsono et al., 2000, "Low Non-linear Non-zero Dispersion-shifted Fiber for DWDM Transmission," Hitachi Cable Review, No. 19, August 2000, p. 19,
  2. C. Weinstein, "Fiber design improves long-haul performance," Laser Focus World, Vol. 33, No. 5, 1997, p. 215; http://www.corning
  3. F. Kapron, C. Kozikowski, and R. Crotts, "Mode Field Diameter Effects on OTDR Splice Measurements," NBS Optical Measurement Symposium 1986, p. 81.

LIGHTWAVE Editorial Mission
Lightwave is a monthly international publication focusing on fiber optics and optoelectronics, the technologies driving the growth, convergence, and improved performance of telephony, computer communications, and video. Lightwave provides technology news as well as applications and product information for corporate and technical managers and staff engineers. Lightwave's editors emphasize analysis and interpretation in their reports on the technological impact of fiber-optic components, systems, and networks in these markets.

More in Network Test