Japanese Journal of Applied Physics Vol. 44, No. 3, 2005, pp. L 117–L 119 #2005 The Japan Society of Applied Physics
High-Reflectance-Resolution Optical Reflectometry with Synthesis of Optical Coherence Function Zuyuan H E, Soshi Y OSHIYAMA, Momoyo ENYAMA1 and Kazuo H OTATE Department of Electronic Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan 1 Department of Frontier Informatics, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan (Received October 8, 2004; accepted November 17, 2004; published December 24, 2004)
A novel high-reflectance-resolution optical reflectometry technique with the synthesis of the optical coherence function for the diagnosis of fiber optic assembly modules is proposed. In this new scheme, the optical frequency of a light source is modulated doubly: one in a sinusoidal wave to synthesize the coherence function into a peak at an arbitrary position within the region under test for a distributed reflection/scattering measurement and the other in a linear sweep to perform the wavelength domain averaging optically for a high reflectance resolution. This approach is free from the numerical averaging required in conventional methods. A simulation shows that the reflectance resolution can be enhanced to the 0.1 dB level. An experimental demonstration is also presented. [DOI: 10.1143/JJAP.44.L117] KEYWORDS: fiber characterization, optical communications, optical reflectometry, optical coherence, Rayleigh scattering
In fiber optic manufacturing, troubleshooting for fiber optic assembly modules is a long-time headache. These modules typically consist of several connected and/or fusion-spliced optical components. After assembly, some of the modules sometimes fail in function or performance tests. In some cases, it is due to component damage or fiber breakage during the assembly process; sometimes, it is due to bad connections or splices. Obviously, the efficient diagnosis of fault points is a critical process in optical manufacturing. Currently, however, there is no efficient technique available to diagnose these modules. In most sites of fiber optic module manufacturing, the only method for troubleshooting is the so-called ‘‘cut-back’’ method, that is, to cut one component back and then to make one measurement, until a fault point is found. Clearly, the development of an effective troubleshooting technique is highly demanded in the industry. On the other hand, optical reflectometry is a powerful technique for distributed optical measurements.1–3) It can produce the profile of reflection/backscattering as a function of linear position along an optical circuit. The reflection occurs at fiber connectors, broken points and bad fusion splices, and the backscattering comes from the micrononuniformity of glass material, i.e., the Rayleigh scattering. Discontinuities due to imperfect fusion splices and microbending in a fiber will cause discontinuing points in the backscattering profile. Therefore, in principle, optical reflectometry can be used to diagnose the fiber assembly modules.4) For this application, a spatial resolution of less than 10 cm and a distance measurement range of 10100 m are required because of the physical dimension of the modules. Moreover, we need a sensitivity higher than 90 dB in order to measure the intensity of Rayleigh scattering accumulated over a fiber section equivalent to the spatial resolution in length, and a 0.1-dB-order reflectance resolution in order to distinguish an out-of-spec fusion splice. Among various available optical reflectometry techniques, optical time domain reflectometry (OTDR) is well known for its application in testing optical fiber communication
E-mail address:
[email protected]
links, but it cannot be used to diagnose an optical fiber assembly module due to its spatial resolution (typically about 10 m) and sensitivity (generally > 60 dB).1) Highly sensitive coherent schemes such as optical frequency domain reflectometry (C-OFDR)3) and coherence domain reflectometry with the synthesis of the optical coherence function (SOCF), which we have developed for optical device tests, optical access network diagnoses, and distributed fiber optic sensing,5–7) are promising in terms of the spatial resolution and the sensitivity, but their reflectance resolution is not satisfactory yet. It has been pointed out that it is possible to enhance the reflectance resolution of C-OFDR by wavelength domain averaging, which involves testing multiple reflectometric profiles with different central wavelengths, and then averaging all the obtained reflection/scattering profiles.8) In this method, however, we need to obtain and then numerically average hundreds of profiles in order to achieve a sufficient reflectance resolution; thus, this method is difficult to be developed as an efficient on-line diagnostic tool for manufacturing circumstances. To date, the major efforts we have made on the SOCF technique are focused on the improvement of the spatial resolution, ranging from several centimeters to several tens of micrometers.5–7) In this letter, we propose a novel reflectometry technique based on the SOCF technique, in which wavelength domain averaging is used in combination with the SOCF to enhance the reflectance resolution. This new approach provides the advantage of a completely optical process, and is free from the timely inefficient numerical averaging process. Simulation and preliminary experimental results are presented. Figure 1 shows the scheme of the optical reflectometer with the SOCF. The device under test (DUT) consists of a mated fiber connector and an open angled physical contact (APC) connector at the end. The lightwave from the laser diode (LD) is divided into a probe wave and a reference wave at coupler 1. The probe wave reflected from the DUT interferes with the reference wave at coupler 2. An acoustooptic modulator (AOM) in the reference path produces a frequency shift; then the interference is detected as the intensity of the heterodyne carrier, which is read out using an electric spectrum analyzer.
L 117
L 118
Jpn. J. Appl. Phys., Vol. 44, No. 3 (2005)
Z. HE et al.
LD DC in
APC
ISO PC
AC in
DUT
PC
+
Function Generator
Function Generator
Coherence function
AOM
Coupler 2
Scan
Distance
PD
PD +
Computer
+
-
(a)
Spectrum Analyzer
Degree of coherence (dB)
0
(b) -20
-40 -60
-80 0
1
2
Normalized distance Fig. 1. (a) Optical reflectometer with the SOCF. LD, laser diode; ISO, isolator; AOM, acoustooptic modulator; PC, polarization analyzer; PD, photo detector; APC, angled physical contact connector; DUT, device under test. (b) Synthesized coherence function.
The optical frequency of the lightwave from the LD is directly modulated in a sinusoidal wave by modulating the injection current to the LD. From the viewpoint of time averaging, the coherence function is synthesized into a series of periodical peaks, as shown in Fig. 1(b). The period of the peaks is inversely proportional to the frequency of the sinusoidal modulation. We control the modulation frequency to leave the first coherence peak only within the range of the DUT, so that the reflection at the position correspondent to the peak can be extracted exclusively. Moreover, by sweeping the modulation frequency, the coherence peak is scanned along the optical path through the DUT to obtain the reflection/backscattering profile. For coherent optical reflectometry, the reflectance resolution is limited by the coherent speckle effect. Backward Rayleigh scatterings from tiny parts along the optical fiber interfere with each other, resulting in a trace of spatial fluctuation. This type of fluctuation is random with respect to the distance, but fixed with respect to time if the slow variance in fiber length due to environmental disturbance is not considered. Therefore, the spatial fluctuation cannot be removed by averaging the multiple test results, which is generally effective for removing temporal noise. On the other hand, since the coherent speckle is a kind of interference pattern dependent on the light source wavelength, it is possible to reduce the spatial fluctuation by wavelength domain averaging, which involves measuring multiple reflection/scattering profiles with different central
wavelengths, and then averaging all the profiles numerically.8) This method basically requires an off-line process. Here, we propose a novel method that merges the concept of wavelength domain averaging into the SOCF. To do so, the optical frequency of the LD is modulated doubly: one in a sinusoidal wave for the SOCF and the other in a linear sweep for wavelength domain averaging. The repetitive frequency of the linear sweep is much lower than that of the sinusoidal wave. In this way, the wavelength domain averaging is merged into the SOCF and realized optically without involving a numerical process. In Fig. 1, the sinusoidal modulation in ‘‘AC in’’ is for the SOCF, while the central frequency sweeping in ‘‘DC in’’ is for wavelength domain averaging. The reflectance resolution is numerically simulated using a multiple reflector model for Rayleigh scattering, in which the fiber is considered to consist of multiple tiny sections. The lengths of these sections are in the order of wavelength, and the reflectivity of each section is assumed on the basis of the Gaussian distribution. The standard deviation of Rayleigh scattering in SOCF reflectometry versus the number of reflectometric data points taken during the frequency sweep for averaging is calculated for estimation. The result is shown in Fig. 2. The standard deviation represents the reflectance resolution of the system. In Fig. 2, the total width of the frequency sweep for averaging is 50 GHz, and the spatial resolution is set to be 10 cm. From Fig. 2, we can see that the original 5 dB reflectance resolution is improved to the 0.1 dB level by applying 100-point wavelength averaging. This result reveals the applicability of SOCF reflectometry to the diagnosis of fiber optic assembly modules. Figure 2 shows that dividing the 50 GHz total linear sweep into more steps gives a smaller deviation. Therefore, integrating the reflectometric signal continuously during the frequency sweep can improve the reflectance resolution further. The integration can be realized in SOCF reflectometry on photodetection. A preliminary experiment is performed using the setup shown in Fig. 1. The light source is a telecom-grade distributed feedback reflector (DFB) LD. Its current-frequency modulation rate is measured as 0.12 GHz/mA. A sinusoidal modulation for the SOCF with the peak-to-peak amplitude of 1 GHz, corresponding to a spatial resolution of Standard Deviation of Reflectivity (dB)
10m Fiber Coupler 1
10
1
0.1
0.01
0.001 1
10
100
1000
Frequency Shift Averaging Number Fig. 2. Estimation on the improvement of SOCF reflectometry reflectance resolution by wavelength domain averaging. Total frequency shift, 50 GHz; spatial resolution, 10 cm.
Jpn. J. Appl. Phys., Vol. 44, No. 3 (2005)
Z. HE et al.
-30
(a)
Output (dBm)
-35 -40 -45 -50 -55
-60 13.0 13.5 14.0 14.5 15.0 15.5 16.0 16.5 17.0
Distance (m) -40
Output (dBm)
(b) -45 -50 -55 -60 13.0 13.5 14.0 14.5 15.0 15.5 16.0 16.5 17.0
Distance (m) Fig. 3. Distributed reflection signal profiles measured by SOCF reflectometry. The mated PC connector at 14.27 m and the open APC connector at 15.45 m in DUT can be observed. (a) Time averaging for 50 times; (b) 100-point wavelength averaging.
10 cm, is carried out. The modulation frequency is swept centered around 6 MHz to scan the first-order coherence peak for distributed detection. The central frequency is swept over a range of 50 GHz at a rate of 1 Hz for wavelength domain averaging. The frequency of AOM is 40 MHz. In this preliminary demonstration, we employ the average function of the spectrum analyzer to carry out 100point averaging instead of the continuous integration ultimately required. Figure 3 shows the distributed reflection profiles of the
L 119
DUT. Figure 3(a) is the trace of the reflection profile obtained by performing time averaging for 50 times. Although we can recognize the reflection peaks at 14.27 m and 15.45 m, which correspond to the mated connector and the APC connector, respectively, the trace exhibits a severe spatial fluctuation due to the interference between multiple reflections and scattering. This proves that the interferenceinduced spatial fluctuation cannot be removed by time averaging. Figure 3(b) shows the result of 100-point wavelength domain averaging, which was obtained by averaging (integration) while the center wavelength was sweeping. It is clear that the spatial fluctuation is suppressed, and the reflectance resolution is improved. As to the regular subpeaks, they are side lobes featured in SOCF reflectometry. These side lobes form a limit factor to the dynamic range of the reflectometry, which can be improved by introducing an appropriate window function in the modulation for the SOCF.6) In conclusion, we proposed an SOCF-based high-reflectance-resolution reflectometry technique by merging the SOCF technique with wavelength domain averaging. It was demonstrated that the 0.1-dB-level reflectance resolution can be realized with the new reflectometry technique. Ultimately, this new technique can provide the advantage of an online optical process. This result made this technique a promising tool for the diagnosis of fiber optic assembly modules. Further experimental studies are being performed to implement continuous integration and improve the dynamic range together with the reflectance resolution. 1) A. L. Lacaita, P. A. Francese, S. D. Cova and G. Riparmonti: Opt. Lett. 18 (1993) 1110. 2) R. C. Youngquist, S. Carr and D. E. N. Davis: Opt. Lett. 12 (1987) 158. 3) J. P. von der Weid, R. Passy and N. Gisin: J. Lightwave Technol. 13 (1995) 954. 4) Z. He, E. Sahinci, C.-C. Chang, Y. Yang and W. Mahmood: Proc. Int. Conf. Systemics, Cybernetics and Informatics (The International Institute of Informatics and Systematics, Orlando, 2002) Vol. X, p. 119. 5) Z. He and K. Hotate: Opt. Lett. 24 (1999) 1502. 6) Z. He and K. Hotate: J. Lightwave Technol. 20 (2002) 1715. 7) K. Hotate: Meas. Sci. Technol. 13 (2002) 1746. 8) K. Shimizu, T. Horiguchi and Y. Koyamada: J. Lightwave Technol. 10 (1992) 982.
