Article
Reference
High-sensitivity-coherent optical frequency-domain reflectometry for characterization of fiber-optic network components
PASSY, Rogerio, GISIN, Nicolas, VON DER WEID, Jean-Pierre
Abstract
The authors have performed high-sensitivity-coherent optical frequency-domain reflectometry (OFDR) measurements of pig-tailed optical devices with centimeter resolution. The oscillations of the Rayleigh backscattering level produced by coherent fading noise was eliminated, strongly improving the capability of OFDR measurements of low level reflections and losses. Sensitivities down to -105 dB were achieved with minute measurement times.
PASSY, Rogerio, GISIN, Nicolas, VON DER WEID, Jean-Pierre. High-sensitivity-coherent optical frequency-domain reflectometry for characterization of fiber-optic network components.
IEEE Photonics Technology Letters , 1995, vol. 7, no. 6, p. 667-669
DOI : 10.1109/68.388759
Available at:
http://archive-ouverte.unige.ch/unige:117879
Disclaimer: layout of this document may differ from the published version.
1 / 1
High sensitivity coherent optical frequency domain reflectometry for characterization of fiber optic network
components
R.Passy, N.Gisin, Group of Applied Physics, Universite de Geneve, 1211 Geneve 4, Switzerland.
J.P. von der Weid, Center for Telecommunications Studies, Pontificia Universidade Cat6lica do Rio de Janeiro Rua Marques de Sao Vicente 225, Rio de Janeiro, 22453, Brazil
Abs~ct-We have performed high sensitivity coherent optical frequency domam reJlectometry (OFDR) measurements of pigtailed optical devices with centimeter resolution. The oscillations of the Rayleigh backscattering level produced by coherent fading noise was eliminated strongly improving the capabitlity of OFDR measurements of low level ~Oections
and losses. Sensitivities down to • 105 dB were achieved with minute measurement times.
I. IN1RODUCTION
With the development of optical fiber communication systems toward to broadband Passive Optical Network, optical return loss of optical components have been reduced down to -50dB in order to improve the responsivity, sensitivity and capacity of these links. Besides the direct effect in duplex transmission (crosstalk), reflection can also affect the modulation properties of the lasers by creating external cavity modes which changes the linearity of the laser optical intensity modulation. Another effect of multiple reflections is the conversion of the laser phase noise to intensity noise. In optical amplifiers, reflections can also cause lasing that drastically restricts the gain, [l].
During the last years several methods have been proposed to characterize optical components. The simplest method is the direct detect.ion of the back reflected light by using a coupler and an optical power meter. However, this method is limited to -75dB and gives only the total amount of reflected light, being impossible to distinguish the component reflect.ion from the Rayleigh backscattering of the fiber. Optical low coherence reflectometry (OLCR) has been currently proposed as submillimeter optical reflectometry [2].
This technique offers a high spatial resolution (tens of microns) with a sensitivity down to -160 dB by using erbium-doped superfluorescent fiber and erbium-doped amplifier. [3]. The main drawback of this technique is that the maximum range is limited by the mirror scanning range which can achieve 0.5 meter.
Optical coherent frequency domain reflectometry OFDR has been also proposed for components diagnosis [4]
with high resolution (millimeter) an high sensitivity (-105 dB). The maximum range of an OFDR is limited by the coherence length of the optical source while the spatial resolution is defined by the total optical frequency deviation and the linearity of the frequency sweep. Distributed
f~dback semiconductor lasers are known to have sharp laser lines and the possibility of optical frequency control via the laser injection current The performance of these lasers received a great improvement with the development of three section· semiconductor lasers with distributed feedback. Here the linewidth decreased down to a few megahertz, with
tun~bility of the order of 2 nanometers free of mode hopping, which are fundamental properties for OFDR applications. In
this letter we present high resolution and dynamic range characterization of optical components using multisect.ion semiconductor lasers.
Il.EXPERIMENTALSETUP
A three section 1.55 µm DFB laser diode is driven by two current sources, one for the central electrode and the other for the side electrodes, which were connected in parallel, and for the temperature control. The laser light was
~unched into a pigtail fiber with a microlens and an optical isolator and the pigtail was connected to the OFDR Michelson interferometer via a FC/PC connector. A shot noise limited photodetector was also linked to the interferometer by a FC/PC terminated optical isolator, and a DC- 300 kHz amplifier was used between the photodetector and the FFf spectrum analyser. In both cases, the distance between the connector and the optical coupler used as beam splitter was the same. The two arms of the interferometer were cut at the same length and also terminated with FC/PC connectors, one of them connected to the system to be measured and the other used as local oscillator, [4].
The optical frequency of the laser diode was swept by modulating the current of the central electrode with a triangular wavefunct.ion. This procedure guaranteed a good linearity of the optical frequency sweep with only a small intensity modulation. Mode hopping in this laser was not observed. The peak to peak current modulation was typically of 20 mA whereas the central electrode bias current was kept at 60 mA. A trigger signal was used to synchronize the data acquisition with a particular part of the laser modulation wavefunction. This allowed a margin of choice of the best
line~ty of the optical frequency sweep, thus minimizing nonhnear sweep thermal effects. The triangular wavefunction half period was typically of 10 ms whereas a 4 ms acquisition time was used in the FFf spectrum analyser.
Hence, only the most linear part of the optical frequency sweep was used for the data acquisition. The laser linewidth (4 MHz) and chirp modulation parameter (1.1 GHzJmA) were measured with a Mach-Zender delayed self homodyne technique and we verified that the linewidth was 3.8 MHz in the full range of current and temperature variation used in the experiments. Therefore, although the laser optical frequency was swept over 22 GHz, only about 9 GHz of this range was used for the data acquisition. The calibration of the reflectivity scale was done with the Fresnel reflection of the far end of a fiber connected to the test arm.
ill.RESULTS
Figure 1 (a) shows the reflectogram of a pigtailed optical isolator with its far end connector unplugged. Three main reflectivity peaks are clearly observed. The one near 2 m corresponds to the fiber-air Fresnel reflection at the free end of the isolator pigtail. From its intensity one can evaluate the isolation efficiency of the device: the -14 dB Fresnel reflection is now reduced to a level down to -51 dB, which means an isolation of 37 dB for that particular optical component. Figure 1 (b) shows the same optical device now connected to a short (19 cm) pigtail terminated with an APC connector. The FC/PC reflectivity now decreased 22 dB, as should be expected for a plugged connector of that type. Its reflectivity in the plugged state is then -36 dB (22 dB below a -14 dB fiber-air Fresnel reflection). Comparison between figures 1 (a) and 1 (b) also indicates that the reflectivity peaks beyond 2 m are proportional to the intensity of the reflection at the far end FC/PC connector, so that they arise from beat signals generated by this far end reflection and multiple reflections within the optical system being analysed.
We also note the reduction in the noise floor of figure 1 (b) to -103 dB due to the reduction of the phase noise produced by the Fresnel reflection of the open FC/PC connector.
A reflection peak of -58 dB at 93 cm indicates the position of the isolator, together with a sharp decrease of the reflectivity noise after it. This feature is characteristic of the Rayleigh backscattering reduction after the isolator. Taking into account the 37 dB isolation, the Rayleigh backscattering falls below the detection noise floor, no longer being visible after the isolator. The first peak, at 31.5 cm, corresponds to internal reflections within the interferometer. The local oscillator reflection is back reflected at the laser and detector FC/PC connectors, and again reflected at the LO-air interface. Hence, a reflection interference peak should appear at the length of the interferometer, which was measured to be 31.5 cm. Assuming the same reflectivity R for each of the interferometer's FC/PC connector , a reflection peak corresponding to (-2x14-2x3+R) dB should appear for each input connector, which means a total reflectivity of (-31+R) dB at this site. In addition the test arm connector also contributes with two peaks of (-14-2x3+2R) dB each, so that their contribution is then (-17+2R) dB to the total reflectivity at that point. Solving for R, we obtain a mean reflectivity of -27.1 dB for the plugged interferometer connectors. Of course, the relative phases of the different peaks were not considered here, but the intensity and position of the internal reflection peak are perfectly compatible to what would be expected from typical FC/PC connectors and could, in principle, be used as a reference signal. However, this double internal pathway reflection also means a 65 cm displaced local oscillator with an intensity --40 dB below the main one. This feature actually gives rise to satellite peaks for each reflection of the test arm and is highly undesirable if low reflection peaks are being searched in the presence of more important reflections. For best performance of the set-up, angular polished connectors or high quality splices should be used.
As is it clearly shown in figure l, the reflectogram corresponding to the Rayleigh backscattering signal is formed by a series of peaks and notches, a strongly oscillating rather than the smoothly varying signal typical of the backscattering signal in OTDR reflectograms. Figure 2 (a) shows the Rayleigh backscattering signal generated by a
2 m optical fiber pigtail whose far end Fresnel reflection was eliminated by breaking the fiber extremity and immersing it in an index matching oil.
iil :!:!..
~ ·5
t5 ..m
~
to 15 2-0 2.5 3.0 3.S
Length [m]
Fig I. Reflectogram of a pigtailed optical isolator with the far end FC/PC connector unplugged (a) and conected to a 19 cm pigtail terminated with an unplugged APC connector (b ).
-eo.-~~....-~~--.~~~ ... ~~--.~~~ ... ~~~~~~
co
~·5 ~ -70
-eo
= 0 -eo ..9!
Qi -70
a: -eo
-90 -tlO
-tl)
0.0 0.5 to
(a)
/0.75 dB/m (b) ----
15 2.0 2.5 3.0 3.5
Length [m]
Fig. 2. Rayleigh backscattering measurement of a 1.83 m single mode fiber showing the coherent fading noise (a) and its elimination (b).
Two effects contribute to the decay of the backscattering signal in OFDR measurements: the optical attenuation of the system under measure and the loss of interference contrast due to limited coherence of the laser source. In this case, only this latter effect is present, because of the extremely low attenuation of the fiber in the meter range. Then, the coherence length of the laser, as well as its linewidth, can be evaluated from the decay of the backscattering signal. In our case, the decay rate was 0.75 dB/m, corresponding to a laser linewidtb of 5 MHz, relativelyin good agreement with the 3.8 MHz measured with the delayed self homodyne technique. Indeed, we could easily observe Rayleigh backscattering in fibers up to 10 m, but when a far reflection is present the laser phase noise is converted into intensity noise which smears out the backscattering signal. Therefore, although we observe reflection peaks at distances as far as far as 20 m, low level precision measurements can be done for distances only up to -5 m.
The oscillating behaviour of the backscattering signal arises because of the interference of sinusoidal waveforms with random phases coming from each neighbouring part of a certain section of the test fiber.
3.0
Length [m]
Fig. 3. Reflectogram of a 2x2 optical coupler. The peak near 3 m is due to multiple reflections inside the interferometer.
Because of this random character, the resulting total intensity fluctuates, giving rise to the noisy feature of the reflectogram. This feature is also observed in optical low coherence reflectometry [5], and contrary to reported in [6], the Rayleigh backscattering of an OFDR reflectogram indeed fluctuates although the laser wavelength is being swept during the measurement. This random phase interference signal, however, can be eliminated if the phase are varied during successive measurements of an averaging process.
This can be done either by slightly varying the test fiber geometrical configuration ( inducing strain for phase variation ) or, in a best way, by varying the laser diode central wavelength and making use of the chromatic dispersion of the fiber. The best way to do this is to sweep the laser temperature whilst averaging the OFDR signal.
Figure 2 (b) shows the averaged signal for a total laser temperature variation sweep of 5.4°C. Considering the laser frequency/temperature coefficient of 10 GHz!°C (at constant current control), we see that a strong reduction of the fading noise was obtained by sweeping the laser the laser optical frequency by 54 GHz over 2500 reflectograms. Of course, the chirp parameter slightly varies with temperature causing a small loss in spatial resolution. This effect was evaluated by measuring the peak width increase of the Fresnel reflection of a pigtail connected to the interferometer. The resolution of the OFDR decreased from 2 cm to 9 cm.
As an application of the fading noise reduction method we measured the insertion loss of an optical 2x2 coupler as shown in figure 3. Here again we recognize the internal reflection peak at 65 cm, the reduced far end reflections and a drop of 3.15 dB of the Rayleigh backscattering level at the coupler. Considering the 3 dB loss due to the 2x2 feature of the coupler and the fact that the light pass twice over the coupler, we obtain a mean insertion loss of 0.075 dB for that particular optical component, which was compatible with the manufacturer's specification and measured in a very simple way.
IV. CONCLUSION
In conclusion, we presented results indicating that fading noise is indeed present in high sensitivity OFDR systems, and that it can be easily eliminated by sweeping the laser temperature during the FFT averaging process. For
high resolution measurements, and where low level reflection are not being considered, the laser temperature sweep can be avoided, and centimeter resolution scale over 10 m are easily obtained with an acquisition time of a few seconds, still keeping a sensitivity of -100 dB. When reflection of the order of the Rayleigh backscattering level are considered, the temperature sweeping easily reduces the fading noise from 20 to less than 3 dB, strongly increasing the performance of the OFDR system whilst keeping minute acquisition times and -100 dB sensitivity over 5 m. The resolution, however, is poorer in that case.
ACKNOWLEDGEMENT
The authors are grateful to the Swiss telecom and Telebras for financial support. One of us (JPW) also acknowledge the support of the Brazilian agency CNPq.
They would also like to thank Dr. B. Broberg of the IMC - Swedeen for several helpful discussions.
REFERENCES
(l] K. Miyamoto "Return-loss measurements improve system reliability" Laser Focus World, 30, 8, p. 81, August 1994.
[2) H. H. Gilgen, R. P. Novak, R. P. Salathe, W. Hodel and P. Beaud
"Submillimeter optical reflectometry" J. Lightwave Technol. 7, 8, p. 1225, august 1989.
[3] K. Takada, T. Kitagawa, M. Shimisu and M. Horiguchi "High sensitivity low coherence reflectometer using erbium doped superfluorescent fiber and erbium doped amplifier" Electron. Len. 29, p. 365, 1993
[4] R. Passy, N. Gisin, J.P. von der Weid and H. H. Gilgen "Experimental and theoretical investigation of coherent OFDR with semiconductor laser sources", J. Lightwave Technol. (in press)
[5) K. Takada, A. Himeno and K. Yukimatsu "Jagged appearance of Rayleigh backscatter signal in ultrahigh resolution optical time domain reflectometry based on low coherence interference", Optics Len. 16, 18, p. 1433, september 1991.
[ 6] K. Shimitsu and T. Horiguchi "Characteristics and reduction of coherent fading noise in Rayleigh backscattering measurement for optical fibers and components", J. Lightwave Technol. 10, 7, p. 982, july 1992.
