Open Access
Add to BibSonomyAbstract
Optical current sensors based on polarization-rotated reflection interferometry are demonstrated using polymeric integrated optics and various functional optical waveguide devices. Interferometric sensors normally require bias feedback control for maintaining the operating point, which increases the cost. In order to resolve this constraint of feedback control, a multimode interference (MMI) waveguide device is integrated onto the current-sensor optical chip in this work. From the multiple outputs of the MMI, a 90° phase-shifted transfer function is obtained. Using passive quadrature demodulation, we demonstrate that the sensor could maintain the output signal regardless of the drift in the operating bias-point.
© 2016 Optical Society of America
1. Introduction
As the demand for electrical power increases unceasingly, various kinds of power generation plants are being developed and efficient current sensors are needed for smart power-grid management. Optical current sensors have small volumes, wide measurement bandwidths, and are lightweight, when compared to their electrical counterparts. Moreover, optical sensors have other attractive features such as independence from electromagnetic interference (EMI) that causes strong noise in electrical sensors, electrical isolation because of the optical sensor coil without metallic components, and lesser possibility of disastrous damage caused by accidental surge currents [1, 2]. However, against the temperature change and vibration applied on the optical fiber sensor, optical sensors have weaknesses caused by the polarization and phase drift. To increase the endurance of the sensors against environmental changes, optical current sensors based on the polarization-rotated reflection interferometer (PRRI) configuration have been investigated [3, 4].
The PRRI utilizes two orthogonal linear polarizations, and detects the minute phase differences between the two polarizations after they are reflected from a mirror at the end of the sensing coil. In this sensor, the output signal amplitude is dependent on the phase bias during the operation. To control the operating bias point with a feedback circuit, a high-speed phase modulator with a fiber delay line was usually employed. However, this made the sensor complicated and expensive [5, 6]. To reduce the cost by eliminating the bias control, a passive sensor with the initial operating point fixed using a waveplate was reported [7–10]. However, the initially adjusted operating point could drift because of the environmental temperature changes. To avoid such operating-point control issues, optical interference signals with an inherent 90° phase difference were utilized for obtaining a sensor output signal independent of the bias by incorporating N × N fiber-optic couplers [11, 12].
Although the PRRI offers stable operating characteristics, it has a complicated structure consisting of various functional optical components. We have demonstrated an integrated-optic version of the PRRI by integrating various polymeric optical waveguide devices to increase the productivity and stability of the optical sensor [13, 14]. In this work, as a next step towards low-cost, high-yield optical current sensors, we integrate a multimode interference (MMI) device on the chip to eliminate the burden of feedback bias control. Two transfer functions with a phase difference of 90° are obtained from the outputs of a 2 × 4 MMI coupler. An optical current sensor is constructed incorporating the MMI integrated photonic ICs. When the bias point is changed for demonstration purposes, the sensor provides a stable output signal in spite of the bias drift.
2. Operating principle and device design
When a magnetic field is applied along the optical fiber, a circular birefringence is produced because of the Faraday effect. This results in a phase difference between the two circular polarizations. The proposed optical current sensor is composed of two parts: an photonic integrated circuit (PIC) and a sensing coil, as shown in Fig. 1. Low coherence SLED with a center wavelength of 1550 nm was used as a light source. The PIC includes a phase modulator, a TE/TM converter, a TE-pass polarizer, and a 2 × 4 MMI coupler. The sensing coil consists of a quarter-wave plate (QWP), a sensing fiber, and a mirror. The phase modulator integrated on the PIC is used for modulating the phase bias through the thermo-optic (TO) effect of polymers. The polarizer absorbs the TM mode by producing a strong coupling with the surface plasmon mode existing at the boundary between the metal and the polymer [15]. The MMI device splits the input light into multiple outputs because of the interference between multiple modes existing in the multimode region. The 4 outputs of the MMI were detected by the photodiodes for monitoring the total output power. Among them, two ports producing the phase difference closest to 90° were selected for the sensor operation. The transfer function of each output of the MMI exhibits an initial phase difference, depending on the mode interference condition. Therefore, it is possible to find a certain dimension of the MMI, which can cause the output transfer function to bring about a phase difference of 90° [16].
Fig. 1 Schematic configuration of the optical current sensors consisting of an integrated optics and a fiber sensing coil.
In the PIC device, linearly polarized input light is launched at an angle of 45°, to excite both the TE and TM modes. These then propagate through the PIC and couple into the slow and fast axis components on the output polarization maintaining (PM) fiber connected to the PIC. The two linear polarizations are converted to the respective circular polarizations by the QWP inserted in front of the sensing coil. In the sensing coil, the two circular polarizations accumulate the phase difference by the Faraday effect, which is then doubled in their return path after reflection from the mirror at the end of sensing coil. The reflected light is converted back to their linear polarizations, when they are returning to the PIC. Because of the double-pass through the QWP, the input TE/TM modes are converted to TM/TE modes, respectively.
As the light comes back to the PIC, it is divided into the two paths. In the lower path, the TE/TM mode conversion is caused by a half-wave plate (HWP), and in the upper path, a phase modulator adjusts the bias point of the interferometer. The TE-pass polarizer, made of a thin metal layer covering the MMI, absorbs the TM polarization, and the two TE polarization modes from each path interfere in the 2 × 4 MMI coupler. The four outputs from MMI coupler are detected by photodiodes, and are used for calculating the final sensor output.
For a conventional Mach–Zehnder (MZ) interferometer, the phase difference ϕF(t) produced by the Faraday effect is given by
where AF is the amplitude of the phase change signal with a frequency ω. The intensity of interference signal, including the effect of phase bias, ϕb is expressed aswhere I0 is the maximum output power of interference signal. By using the Bessel function, Eq. (2) can be expanded aswhere J1 and J2 are the first and second kind of the Bessel functions, respectively. The intensity signal produces a time average power signal 〈I〉 and a sinusoidal signal with an amplitude AO, given by Although the output signal amplitude (AO) is proportional to the Faraday effect signal amplitude (AF), it is also dependent on the phase bias (ϕb). Therefore, ϕb must be set to π/2 to maximize AO. This can be explained by the transfer function drawn in Fig. 2(a), in which AO is dependent on ϕb for a signal with a constant AF. To overcome this bias dependence, two transfer functions with 90° phase shift could be used as shown in Fig. 2(b). The output signal amplitude of each transfer function changes in a complementary direction for the change in ϕb. Therefore, the bias dependence can be eliminated by combining these two signals. The output signals from each transfer function with 90° relative phase difference can be expressed as The time average powers 〈I1〉 and 〈I2〉, and the amplitudes A1 and A2 obtainable from the output signals I1 and I2 are expressed as By using Eqs. (10) and (11), the final sensor output AO is found aswhich is proportional to AF with no dependence on ϕb.Fig. 2 (a) Transfer function of a typical optical interferometer, in which the output signal amplitude is depending on the operating bias point, and (b) Two transfer functions with 90° phase difference which are incorporated for the signal processing so as to eliminate the dependence of operating bias.
However, the signal processing method explained above is for an ideal case when there is an exact 90° phase difference between the two transfer functions. In practice, because of fabrication errors, the phase difference bet ween the two transfer functions may not be exactly 90°. By considering a phase error (ϕe), Eqs. (6) and (7) can be rewritten as
which changes the time average signals, and the amplitudes of the sinusoidal signals as Using Eqs. (15)-(18), one can deduce the final output signal asThe final AO has no dependence on ϕb, and has less effect on the phase error ϕe.3. Fabrication and characterization of integrated optic devices
The polymer PIC device was fabricated with ZPU-440 and ZPU-455 polymers produced by Chemoptics with refractive indexes of 1.440 and 1.455, respectively. A single mode waveguide was fabricated as an inverted rib waveguide, with a core-width of 6.0 μm, a height of 6.5 μm, and a lateral core layer thickness of 3.0 μm. The fabrication procedure is illustrated in Fig. 3. ZPU-440 was spin coated on a Si substrate, and cured by UV, as a lower cladding layer. The waveguide pattern made of photoresist was transferred to the lower cladding layer by O2 plasma dry etching in an ICP-RIE machine. Then, a core layer was formed by spin coating a 3-μm ZPU-455 layer. Above the core layer, ZPU-440 of 2-μm thickness was coated, and then the polarizer pattern was fabricated with Cr-Au of 10–100 nm. After that, ZPU-440 was coated once again to increase the thickness of the upper cladding to 8 μm, and then a phase modulator electrode pattern was formed on the surface. For inserting the HWP film, a groove line with a width of 30 μm was formed using a dicing saw, and then the HWP film was inserted at an angle of 45° to the substrate.
Beam propagation method (BPM) simulation was performed to design the 2 × 4 MMI coupler [17, 18]. When the output waveguide pitch of the MMI was larger than 14 μm, the power coupling between the output ports was less than 1%. For reducing the insertion loss and increasing the output power uniformity, the width and pitch of the MMI were optimized as 70 μm and 23 μm, respectively.
To characterize the phase difference between the transfer functions of the MMI coupler outputs, the spectral response was measured by using a delayed MZ interferometer as shown in Fig. 4(a) [16]. The phase difference (Δϕ) induced by the delay line is expressed as
where k0 is the propagation constant in air, Δλ is the wavelength difference from λ0, neff is the effective index of the waveguide, and Δl is the path length difference because of the delay line. For , the wavelength difference (Δλ) becomesFig. 4 (a) Delayed Mach-Zehnder interferometer connected to the MMI device for the purpose of phase delay characterization, (b) Simulation results of the MMI output spectrum for a MMI length of 5150 μm, and (c) Spectral response of the fabricated MMI with a length of 5230 μm.
For a Δl of 106 μm, λ0 of 1550 nm, and neff of 1.449, Δλ becomes 15.8 nm. In the simulation, an MMI with a length of 5150 μm exhibited optimum results as shown in Fig. 4(b). The spectral response obtained from the MMI device with a length of 5230 μm exhibited a spectral response as Fig. 4(c), which is similar to the design result. Experimental result exhibited an increased free spectral range and a higher insertion loss which arises from the fabrication error. The phase differences between the output transfer functions are calculated as shown in Table 1. The phase difference of the fabricated device had a deviation of less than 2° from the ideal values.
Table 1. Comparison of the phase difference (Δϕ) and the phase error (ϕe) of the output transfer curves obtained from simulation and measurement data.
MMI devices with a multimode region length from 4950 μm to 5350 μm were fabricated. For inputs of I1 and I2, the output powers from the 4 ports were measured as shown in Fig. 5(a). Then the relative phase difference was calculated as shown in Fig. 5(b). The uniformity was less than 1 dB and the phase difference error was less than 5° for the MMI of 5230 μm. The simulation results are also shown in Figs. 5(c) and 5(d) for comparison. Because of the interference of many modes, the MMI coupler is highly sensitive to the device dimension and the refractive index of the polymer material. Hence, the fabrication yield has to be improved by a careful control of the fabrication process as well as the polymer material synthesis.
Fig. 5 (a) The relative optical output power of MMI with various multimode region length, and (b) the relative phase difference between the output signals, (c) and (d) show the BPM simulation results.
In order to measure the insertion loss of the device, light was launched from the output port connected to the sensing coil, and the total power transmitted to the other side was measured. The insertion loss was measured to be 14.2 dB, which comprised a Y-branch loss of 3 dB, a fiber-waveguide mode-mismatch loss of 2.3 dB, a propagation loss of 3.7 dB, a groove line loss of 0.9 dB, and a 3-dB loss due to the TM mode absorption by the polarizer. The extinction ratio of the polarizer was over 20 dB, and the resistance of the TO phase modulator was 28 Ω.
4. Current sensing experiments
The current sensing fiber coil was prepared using a highy birefringent (Hi-Bi) spun fiber produced by Fibercore Co. Five turns of the Hi-Bi spun fiber were wound on a cylindrical sensing coil. The QWP placed in front of the sensing coil was made of photonic crystal fiber for reducing the temperature dependence; the detailed fabrication process can be found in [14].
The sensor was operated by applying 60-Hz high current signal across the sensing coil, and the output signal was obtained after electrical signal processing using a personal computer with an analog to digital converter. The bandwidth of the optical current senosr is dependent on the electronic circuits, and then the sensor could measure a faster signal (~100 MHz) caused by the surge current. The difference of the optical power from each output channel was calibrated by adjusting the gain of the photodetector. In order to observe the bias-point dependence of the sensor output, the phase bias was modulated by operating the phase modulator of the PIC. It produced a phase change of over 2π for a frequency less than 1 Hz. The time average powers 〈I1〉 and 〈I2〉 were calculated as shown in Fig. 6(a). The phase difference between the transfer functions was close to 90°. The optical amplitude signals, A1 and A2, which were proportional to the 60 Hz current signal, were measured as shown in Fig. 6(b). Although A1 and A2 were dependent on the bias point change, they changed in complementary ways. In order to measure the phase difference between the two transfer functions, a Lissajous curve was drawn with 〈I1〉 and 〈I2〉 as shown in Fig. 6(c). The phase deviated by 11.5° from 90°.
Fig. 6 For a change of ϕb over 2π, (a) the time average power signals 〈I1〉 and 〈I2〉, and (b) the amplitude signals A1 and A2 were calculated from the sensor output signal. (c) a Lissajous curve drawn by the 〈I1〉 and 〈I2〉 for calculating the phase error.
To obtain the final sensor output signal, 〈I1〉, 〈I2〉, A1, and A2 values were inserted in Eq. (12), and the calculation result is shown in Fig. 7 with green dots. The result contained more than 10% error depending on the phase bias (ϕb). This was caused by the relative phase error (ϕe). To reduce the effect of ϕe, Eq. (19) was used with a ϕe of 11.5° to obtain the output signal as shown with black dots in Fig. 7. Even after the phase error compensation, the output still had a dependence on ϕb. By careful examination, we found that the ϕb dependence occurred because of the fluctuation of the total output power of the MMI. Therefore, we incorporated an additional calibration of the signal in terms of the total output power as a function of ϕb. As the result, the sensor output exhibited much less fluctuation (within the RMS error range of 2%) as shown by the red dots in Fig. 7.
Fig. 7 The sensor output signals as a function of ϕb obtained by the simple magnitude of orthogonal vectors (green), the phase error compensation (black), and the power change calibration (red).
In order to observe the bias independent characteristics of the current-sensor operation, the sensor output signal was monitored under an open loop condition. The applied current was changed from 80 A to 900 A as shown in Fig. 8. The optical sensor exhibited an excellent linear response, proportional to the applied current. By comparing the output signal with an electrical current sensor, the optical sensor was found to exhibit an RMS error of 0.2% and a peak error of ± 0.5%. The error was increasing significantly for the current less than 100 A, and the sensor had a current sensing resolution of about 1 A, which was limited by the electronics.
Fig. 8 Bias-free OCT sensor output measured without the feedback bias control, in which the output signal exhibited good linearity with a peak error of ± 0.5%, and an RMS error of 0.2%.
Temperature dependence of the proposed sensor may be influenced by the two parts, the PIC and the fiber sensor coil. The temperature of PIC has to be maintained by using a thermo-electric cooler unit included inside the device package. However, the sensing coil with the QWP connected at the input could produce a temperature dependent phase change. To reduce the temperature dependence of the fiber-optic QWP, a photonic crystal fiber with low temperature dependence was incorporated in our previous experiment [14]. Then, the temperature dependence was reduced to be less than ± 1% for a temperature change of 80°C.
4. Conclusion
An optical current sensor, operating without bias feedback control, fabricated by integrating an MMI device on a polymeric PIC was proposed in this work. From the MMI output signals, two transfer functions with a phase difference of 90° were obtained. To confirm the bias independent operation of the sensor, the operating bias point was changed during the sensing experiment using the integrated TO phase modulator. For a phase bias change of over 2π, the sensor maintained a stable output within the RMS range of 2%. Without applying any bias control, the sensor was operated in an open loop. For an applied current of 900 A, a peak error of ± 0.5% and an RMS error of 0.2% were obtained. The PIC incorporated in this work enables passive optical current sensing with the advantages of low cost, high yield, and enhanced stability.
Acknowledgments
This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (2014R1A2A1A10051994).
References and links
1. K. Bohnert, P. Gabus, J. Nehring, H. Brändle, and M. G. Brunzel, “Fiber-optic current sensor for electrowinning of metals,” J. Lightwave Technol. 25(11), 3602–3609 (2007). [CrossRef]
2. J. D. P. Hrabliuk, “Optical current sensors eliminate CT saturation,” in Proceedings of IEEE conference on PES Winter Meeting (IEEE, 2002) 2, pp.1478–2481.
3. A. Enokihara, M. Izutsu, and T. Sueta, “Optical fiber sensors using the method of polarization-rotated reflection,” J. Lightwave Technol. 5(11), 1584–1590 (1987). [CrossRef]
4. K. Bohnert, P. Gabus, J. Nehring, and H. Brandle, “Temperature and vibration insensive fiber-optic current sensor,” J. Lightwave Technol. 20(2), 267–276 (2002). [CrossRef]
5. T. Masao, K. Sasaki, and K. Terai, “Optical current sensor for DC measurement,” in Proceedings of IEEE conference on Asia Pacific IEEE/PES Transmiss. Distrib. Conf. Exhibit. (IEEE, 2002) 1, pp.440–443.
6. R. M. Silva, H. Martins, I. Nascimento, J. M. Baptista, A. L. Ribeiro, J. L. Santos, P. Jorge, and O. Frazao, “Optical current sensors for high power systems: a review,” Appl. Sci. 2(4), 602–628 (2012). [CrossRef]
7. F. Briffod, L. Thevenaz, P.-A. Nicati, A. Kung, and P. A. Robert, “Polarimetric current sensor using an in-line faraday rotator,” IEICE Trans. Electron. E83-C(3), 331–335 (2000).
8. J. Zubia, L. Casado, G. Aldabaldetreku, A. Montero, E. Zubia, and G. Durana, “Design and development of a low-cost optical current sensor,” Sensors (Basel) 13(10), 13584–13595 (2013). [CrossRef] [PubMed]
9. K. Kurosawa, “Development of fiber-optic current sensing technique and its applications in electric power systerms,” Photon. Sensors 4(1), 12–20 (2014). [CrossRef]
10. G. M. Muler, L. Yang, A. Frank, and K. Bohnert, “Simple Fiber-optic current sensor with integrated-optics polarization splitter for interrogation,” in Applied Industrial Optics: Spectroscopy, Imaging and Metrology Conference, 2014 OSA Technical Digest Series (Optical Society of America, 2014), paper AM4A.3. [CrossRef]
11. D. W. Stowe and T.-Y. Hsu, “Demodulation of interferometric sensors using a fiber-optic passive quadrature demodulator,” J. Lightwave Technol. 1(3), 519–523 (1983). [CrossRef]
12. Y. Li and R. Baets, “Homodyne laser Doppler vibrometer on silicon-on-insulator with integrated 90 degree optical hybrids,” Opt. Express 21(11), 13342–13350 (2013). [CrossRef] [PubMed]
13. M.-C. Oh, W.-S. Chu, K.-J. Kim, and J.-W. Kim, “Polymer waveguide integrated-optic current transducers,” Opt. Express 19(10), 9392–9400 (2011). [CrossRef] [PubMed]
14. W.-S. Chu, S.-M. Kim, and M.-C. Oh, “Integrated optic current transducers incorporating photonic crystal fiber for reduced temperature dependence,” Opt. Express 23(17), 22816–22825 (2015). [CrossRef] [PubMed]
15. M.-C. Oh, J.-K. Seo, K.-J. Kim, H. Kim, J.-W. Kim, and W.-S. Chu, “Optical current sensors consisting of polymeric waveguide components,” J. Lightwave Technol. 28(12), 1851–1857 (2010). [CrossRef]
16. L. Zimmermann, K. Voigt, G. Winzer, K. Petermann, and C. M. Weinert, “C-Band optical 90-hybrids based on silicon-on-insulator 4x4 waveguide couplers,” IEEE Photon. Technol. Lett. 21(3), 143–145 (2009). [CrossRef]
17. Y. Sun, X. Jiang, and M. Wang, “Analysis of imaging properties in multimode interference couplers,” Proc. SPIE 5279, 192–198 (2004). [CrossRef]
18. L. B. Soldano and C. M. Pennings, “Optical multi-mode interference devices based on self-imaging: principles and applications,” J. Lightwave Technol. 13(4), 615–627 (1995). [CrossRef]
