Reducing Backscatter Echoes in OTDR Measurements

As instrument range and performance increase, so does susceptibility to error.

By Josef Beller, Hewlett-Packard GmbH, Germany

Optical time domain reflectometers (OTDRs) are well-accepted measurement tools for fiber installation, maintenance, and operation. During the last decade the most obvious trend has been the ever-increasing dynamic range performance caused by expanding link lengths and a tendency towards shorter measurement times. However, to maintain full advantage of the increasing dynamic range figures in the order of 40dB, the instruments' linearity must be preserved even at the lowest signal power levels ⁽¹⁾. This topic has been addressed in Bellcore's generic requirements for OTDRs GORE 196 with the addition of a measurement range specification ⁽²⁾. Improved linearity behavior is certainly a desirable characteristic for a measurement instrument, but in some cases it can generate a side effect, which can be expressed as susceptibility to second order echoes. This means that a signal appears beyond where one would expect to see the end of a fiber under test.

Echoes
It is common knowledge that reflections in optical links cause echoes. An echo's strength depends on the refractive index difference at a particular location and on the angle at which the reflected light is coupled back into the fiber. Such echoes can be extremely disturbing in high-speed transmission systems. Therefore, maximum reflectance and return loss figures for a fiber optic link are specified in reference ⁽³⁾. For example, limits for OC-48 links are -27dB reflectance at distinct locations, and 24dB for the total return loss.

By the same token, such reflections can generate ghost signals that superimpose on the results of OTDR backscatter measurements. These ambiguous effects are more likely with modern high end OTDRs due to improved low level signal sensitivity and linearity. This is because second order effects in OTDR measurements show up only as tiny signals. A more detailed investigation of backscatter measurements with OTDRs, which is beyond the scope of this article, can be found in reference ⁽⁴⁾.

What are the most reflective locations in a typical OTDR measurement arrangement, and how do they affect the measurement results? Figure 1 shows the four dominant sources of reflections in the head-end of almost every OTDR. In general, most attention is given to the front connector reflection. It is the user's responsibility to keep this connector clean to guarantee a low-loss and low-reflectance coupling to the fiber-under-test. By the way, this is the only front-reflection that, under normal circumstances, can directly be seen on the OTDR display. Only an extremely small amount of this visible front-reflection is contributed by the finite directivity of the directional 3dB-coupler. A small part (typically <10^-5...10^-6) of the outgoing laser pulse is guided unintentionally to the APD (Avalanche-Photo-Diode) port. With an angled front connector having a reflectance of <-60dB, the coupler is the ultimate limitation to the achievable front-end reflectance. The two other sources of (internal) reflections are usually not taken into account. Most likely this is because they don't appear clearly on the screen. Measurements have shown that the reflections from the laser chip dominate the APD reflections. However, this depends on the actual chip-to-fiber coupling arrangement. The effect of these internal reflections will be discussed later on in Figure 4.

In this article we look at a simple arrangement of two fiber spools, fiber 1 with length L₁ (~16km) and fiber 2 with length L₂ (~8km), connected by a splice with an insertion-loss of ca. 1.7dB. The splice-loss was intentionally chosen to be this high, as the splice pattern can then be seen more easily in the 2^nd order echoes further away. All backscatter measurements shown were conducted at a wavelength of 1550nm and a pulsewidth of 10µs. To be able to isolate the echo effects the results of some simulations, based on the same parameters, are given in Figures 3 and 4.

It is important to understand that Figure 2 gives an ideal representation of an OTDR trace with a rms-SNR (root-mean-square Signal-to-Noise-Ratio) close to 40dB. A high quality OTDR with real world optical components would disclose more or less some of these 2^nd order phenomena that are depicted in the following figures.

Besides the sources of reflection in the OTDR's front-end that have already been mentioned, in general there are external reflections on the fiber-under-test as well. To simplify matters only a reflection at the fiber-end will be taken into consideration. The effect of multiple reflections on the fiber link can be determined as superposition of individual events.

All signals generated from the original backscatter signal, by being reflected at any part of the test set-up or generated after the laser pulse reached the fiber end, are referred to as 2^nd order signals. Of course these 2^nd order signals will produce 3^rd order echoes and so on, for the same physical reasons. However each new generation will considerably be attenuated further, so it is sufficient and appropriate to look only at the 2^nd generation signals at this point of time to get the basic understanding.

In the following figures, examples of how second order echoes can disturb OTDR measurement results are given. Figure 3 depicts the backscatter echo due to a strong end-reflection. This echo is generated by the laser pulse that travels down the fiber, and then, reflected at the fiber end, travels back towards the OTDR again. Naturally its backscatter takes the opposite direction to the end reflection first, is turned around there, and then is guided back to the OTDR.

Main vs. Reflected Signals
An important point to mention is that the backscatter echo occurring from the end reflection is mirror-faced compared to the main backscatter signal. Any event in a distance d on the fiber under test with length L (L = L1+L2) can now be seen at a position 2L-d. The splice between fiber 1 and fiber 2 illustrates this in particular. Still the backscatter looks linear, meaning the slope of the signal is a straight line that gives the fiber attenuation. However, it is now generated by a laser pulse traveling in the opposite direction towards the OTDR. Some further ideas about the fiber end reflection and its echo are described in reference ⁽⁵⁾. The actual trace that can be seen on the OTDR display is the sum of the dotted curve plus the echo parts drawn in solid lines. The upper curve in Figure 3 showing the largest echo signal corresponds to an open fiber end with approximately -14dB reflectance.

Conversely, the echo signal generated by the internal reflections in the OTDR's head-end that can be seen in Figure 4, is an integral function of the first order backscatter. That means it starts at zero as soon as the original backscatter signal hits the laser and APD and then is reflected back into the fiber-under-test again. This way an image of the whole backscatter serves as a second ongoing probe signal. The backscatter of which cumulates to a maximum, and declines until it reaches eventually zero again at a distance 2L, corresponding to twice the length of the fiber under test.

This echo demonstrates the surprising fact that even without any reflections on the fiber-under-test a ghost signal can be generated that extends the original backscatter far beyond the fiber end. As these signals normally are very weak, they can only be discovered with high dynamic range measurements. Please notice that the strength of the internal reflections at the laser diode or at the APD can't be derived from the height of the front reflection on the OTDR display. These reflections do not appear on the trace directly.

In cases where considerable internal and fiber-end reflectance is present as well, all echo signals superimpose and lead to a measurement result pictured in Figure 5. The two arrows point to the echo patterns of the splice (originally between fiber 1 and fiber 2) and illustrate that the signal past the fiber end consists of two parts. The mirror-faced part originates from the end reflection; the other part stems from the backscatter that is reflected back at the laser and APD into the fiber-under-test. The step size of both events is lower because two echo signals superimpose at this part of the trace.

Clearly, by taking care of the fiber end reflection the user can considerably reduce the effect of 2^nd order echoes. However, there's still that part arising from the internal reflections in the OTDR's head-end that one can't get rid of. This effect has actually increased over the last years. As the market has demanded laser diodes with increasing output power for high performance OTDRs, laser manufacturers have optimized the coupling efficiency between chip and the fiber to minimize power loss. Today a coupling efficiency better than 50% is quite common, which means that the mirrored laser chip surface represents a strongly reflecting plane for incident light.

Fortunately the 3dB-coupler in Figure 1 combining laser diode, APD, and fiber-under-test attenuates light that is directed to, and reflected from, the laser diode. This doesn't remove the disturbing echo effect, but helps to lessen it. A much more expensive solution, namely replacing the 3dB-coupler by a circulator (as shown in Figure 6), would help to further weaken the effect of higher order echoes. Returning light from the fiber-under-test would hit only the APD and not APD plus laser at the same time as is the case with the 3dB-coupler arrangement. Only that part reflected from the APD is directed to the laser diode resulting in a considerably smaller amount of light being finally guided back to the front connector and into the fiber-under-test again.

Conclusion
Obviously, OTDR users should avoid reflections whenever possible in every measurement setup. Careful interpretation of high dynamic OTDR measurement results in case there are still strong reflections is mandatory. It's also important to keep in mind that the two major sources for second order signals are the head-end (including internal) reflections and any other reflection on the link under test. And last but not least, the simplest way to avoid trouble with higher order echoes is to choose a pulsewidth as small as possible and only as wide as necessary.

Hewlett-Packard Company, European Centre, P.O. Box 999, 1180 AZ Amstelveen, The Netherlands, Phone: (31 20) 547 9900.

References:
1. Josef Beller: Measurement range: A new performance specification for OTDRs. Lightwave Oct./1997.
2. Bellcore GORE 196 Generic Requirements 1995/1997.
3. Bellcore TR-NWT-00253.
4. Dennis Derickson (editor): Fiber Optic Measurement Handbook, Prentice Hall 1997.
5. Ronald J. Larrick: Improving OTDR Measurements for Return Loss and Splice Loss Using End-of-Fiber Reflections. NFOEC'97.