In this paper, a new fiber Bragg grating (FBG) sensor exploiting microwave photonics filter technique for transverse load sensing is firstly proposed and experimentally demonstrated. A two-tap incoherent notch microwave photonics filter (MPF) based on a transverse loaded FBG, a polarization beam splitter (PBS), a tunable delay line (TDL) and a length of dispersion compensating fiber (DCF) is demonstrated. The frequency response of the filter with respect to the transverse load is studied. By detecting the resonance frequency shifts of the notch MPF, the transverse load can be determined. The theoretical and experimental results show that the proposed FBG sensor has a higher resolution than traditional methods based on optical spectrum analysis. The sensitivity of the sensor is measured to be as high as 2.5 MHz/N for a sensing fiber with a length of 18mm. Moreover, the sensitivity can be easily adjusted.
© 2016 Optical Society of America
Optical fiber Bragg gratings (FBGs) sensors have been extensively investigated in the last few decades . Their most significant advantages include lightweight, non-obtrusive characteristics, immunity to electromagnetic interference, wavelength-encoded operation and good multiplexing ability. Most FBG sensors are used to measure axial strain because of the higher axial load sensitivity compared to transverse load. However, there has been an interest in measuring the transverse load in addition to the axial strain for many applications such as structural health monitoring , impact and damage detection etc. It has been demonstrated that the spectral characteristics of a FBG under transverse load are quite different from those of FBG under axial strain . In this situation, the different stain distribution along the two perpendicular directions will break down the Bragg condition, and split the reflected Bragg peak into two separate peaks, each relative to a polarization axis. This effect is useful in transverse load sensing since the Bragg wavelength difference between the two polarized directions depends on the transverse load on a fiber . However, a small amount of birefringence induced by transverse load can only cause FBG spectrum broadening rather than a clear spectral splitting because the transverse load sensitivity of FBG is extremely low and the bandwidth of a uniform FBG is much broader than the birefringence induced wavelength shift. Therefore, the load sensing sensitivity and resolution are severely limited. For this reason, FBGs written into polarization maintaining (PM) fibers have been employed as a sensor probe to overcome this drawback . However, using PM fiber devices is rather expensive and complex. In addition, conventional wavelength interrogation schemes is done in the optical domain using an optical spectrum analyzer (OSA). Due to the low-resolution character of the OSA, the resolution of the load sensor is accordingly low. Therefore, a high resolution interrogation technique for uniform FBG-based transverse load sensing is very essential for high-performance sensing applications.
In recent years, with the development of microwave photonic technology , fiber optic sensors combined with microwave photonic technology have attracted great interest [7–10]. Microwave photonics is a field that studies the generation, processing, control, and distributions of microwave signals by means of photonics . For FBG sensing applications, for example, interrogation can be implemented by converting the wavelength shift in the optical domain to the amplitude or frequency change of a microwave signal in the electrical domain [12–14]. Thanks to the high resolution of an electrical spectrum analyzing technique, frequency interrogation in the electrical domain can be precisely measured at a resolution of 1Hz, which leads to a high resolution.
In this paper, we propose a uniform FBG-based transverse load sensor that is implemented in conjunction with a microwave photonics filter (MPF). The MPF is implemented mainly using a polarization beam splitter (PBS), a tunable delay line (TDL) and a length of dispersion compensating fiber (DCF). The fundamental concept of the proposed approach is to convert the separation of Bragg wavelengths for the two polarization modes to the frequency change of the notch MPF. By measuring the frequency shift of the notch, the load applied to the sensing FBG is estimated. Therefore, an electronic spectrum analyzer can be used to replace the OSA. Since a small change in wavelength separation can lead to a large microwave frequency change, the performance in terms of the resolution is significantly increased. The theoretical analysis of the proposed approach is presented and discussed. A transverse load sensor with a sensing sensitivity of 2.5MHz/N for 18mm long fiber under load is experimentally demonstrated. The proposed approach offers several features since it is essentially an interference in the microwave domain, which is more stable and easier to control than optical domain methods. In addition, the sensitivity can be easily adjusted, and could be potentially improved to be extremely high by using various higher dispersion components.
The proposed transverse load sensing system is illustrated in Fig. 1. The output from a broad-bandwidth amplified spontaneous emission (ASE) is coupled into a Mach-Zehnder modulator by which a microwave signal generated from a vector network analyzer (VNA) is imposed onto the optical beam. The optical microwave output is then sent to a sensing FBG through a 3-dB coupler (OC1). If a transverse load is applied to the FBG, the FBG will be birefringent and two Bragg reflection bands corresponding to two orthogonal polarization modes will be induced. As mentioned, it is often difficult to distinguish the slight wavelength differences by simple optical spectrum analysis. If the spectral information is converted from the optical domain to the microwave domain, the difficulty could be overcome. To do so, the reflected signals from the FBG are subsequently fed into a fiber PBS via a polarization controller (PC) to be split into two orthogonal linear polarization components corresponding to the fast axis and slow axis, with x denoting the slow axis and y denoting the fast axis respectively in Fig. 1. The behavior of the sensing system from the point of view of polarization in the traveling path from the FBG to the PBS can be explained as follows. As mentioned above, two Bragg reflection bands corresponding to two orthogonal polarization modes will be induced when a transverse load is applied. Generally, since the optical path is made of standard single mode fiber, the output polarization of the reflected light waves will tend to drift as the light propagates. However, for short lengths of single-mode fiber (mainly pigtails of fiber optic component in our experiment, 2m or so), the orthogonality of the two modes has no obvious change. The reflected light waves are then sent to the PBS through the PC. By tuning the PC to have an angle of 45° to one principal axis of the PBS, the reflected optical signal will be split into two orthogonal polarization components having equal power. The two signals then travel along separate fiber paths and finally come together in a 3dB coupler (OC2), forming a Mach–Zehnder (MZI) interferometer. In the interferometer, a TDL is introduced in one arm to change the length difference of the MZI. For the other arm, a tunable attenuator (TA) is used to balance the optical powers between the two channels to obtain a maximum notch depth. The optical signal at the output of the MZI is then sent to a dispersive device which is a length of DCF in our demonstration, and then detected at a high-speed photodetector (PD). Finally, a two-tap microwave filter is thus realized and its frequency response is observed by a VNA.
The operation of the proposed transverse load sensor using the notch MPF is described as follows: The key component in the MPF is the PBS, which split the FBG reflected signal into two orthogonally optical microwave signals. Since the two signals undergo a differential time delay, a two-tap notch MPF is thus realized. The delay line of the MPF consists of two sections: one is the preset time delay difference (TDD) introduced by the TDL in the MZI, and the other is the tunable TDD due to the wavelength separation variation of the two Bragg reflection bands for a fixed length of the DCF. The total TDD, ΔTd, between the two taps of the filter is given by:
The electrical frequency response H(f) of the proposed photonic microwave notch filter can be expressed as :16]. Denoting the reflected optical powers from the two FBGs as P1 and P2, the total incident power on the PD which performs the summing operation is given by:Eq. (3), the MPF is essentially an optical carrier based microwave interference. As expected, the filter response correspond to the well-known two-taps transversal filter behavior , providing deep notches in the spectrum, whose frequencies are given by:Eq. (1) and (4), obviously when the transverse load changes the wavelength separation, thus changing the TDD between the two signals, the notch frequency will vary with the applied load. If we track the first notch frequency, it will be a function of the wavelength separation of the sensing FBG, thus the applied load. Considering the relationship between the wavelength spacing and the applied load,Eq. (5) into Eq. (1), the final TDD is given by
As shown in Eq. (8), provided the fixed delay difference ΔT0 is much larger than the tunable time delay, the shift of the notch frequencies can be tuned linearly by the applied load, offering good characteristic for sensing applications. By varying the value of the load, the time delay between the signals reflected from different gratings is changed, and thus, tunability of the filter frequency response is achieved.
3. Numerical simulation and experimental investigation
A numerical simulation based on the theoretical model described in section 2 is adopted to simulate the frequency response spectrum of the notch filter. Since the load measurement is performed by tracking the changes in the notch frequency, therefore in order to get better accuracy, a large notch depth is required. As is well known, the maximum notch depth appears only when the two reflected signals are equal in power . It should be noted that the notch depth could be infinite in theory by exactly matching the power. However, the maximum notch depth of a practical filter is usually limited and decided by the noise floor of the detection electronics. Figure. 2 shows the calculated FSR values for different values of the fixed time delay difference ΔT0. It can be observed in Fig. 2 that the larger TDD, the smaller FSR, and accordingly, the lower notch frequency. Actually, the notch frequency is inversely proportional to the TDD, as shown in Eq. (4). Since the sensing mechanism of our proposedmethod is based on tracking the frequency shift of the first notch. Given the fixed TDD introduced by the TDL is unchanged, the variation of the applied load will cause an extra TDD and then result in a frequency shift of the notch. Thus, by monitoring the frequency shift of the notch, the transverse load applied to the FBG can be measured.
To verify the effectiveness of the proposed approach, a proof of concept experiment is carried out based on the system shown in Fig. 1. An ASE source with a spectral width of 30 nm is used as the broadband optical source. The broadband light wave is intensity modulated using a MZM modulated by a swept microwave signals from the radio frequency port of a vector network analyzer (VNA, Agilent E8364A). The modulated light wave is then sent to the sensing FBG which has an initial Bragg wavelength of 1557.1nm with a length of 6mm. The compression device are two polished aluminum plates with a width of 18mm and the FBG is placed parallel to a support reference fiber, as shown in the inset of Fig. 1. The reflected signals from the FBG are split into two polarization components which travel along different paths after the PBS. The original time delay difference introduced by the TDL between the two paths is controlled to be 320 ps, corresponding to the time delay of a 6.50-cm single mode fiber. According to Eq. (4), the original FSR of the MPF is calculated to be 3.125GHz, and therefore, the frequency of the first notch is 1.563GHz. The TA is used tobalance the optical powers between the two channels to obtain the maximum notch depth. To translate the wavelength difference into a time delay difference, the output signal from the OC2 is sent to a DCF which has a total dispersion parameter of 124 ps/nm, and then detected at a high-speed PD and observed by the VNA. Finally, a transverse load sensing is evaluated. When the load is increased, the frequency response at different load levels is recorded. Figure 3 shows the measured frequency response of the MPF at different load of 0N, 10N and 20N. As shown in Fig. 3, all of the response spectra are clean and have a maximum notch depth exceeding 40 dB, indicating good resolution for sensing applications. When no load is applied, the first-notch frequency is located around 1.558GHz, which agrees well with the theoretical value of 1.563 GHz calculated based on Eq. (4). The measured FSR for 0N, 10N and 20N are 3.116GHz, 3.066GHz and 3. 018GHz respectively. Again, the experimental results agree well with the theoretical prediction. Figure 4 shows the zoomed spectra of the first notch at different applied load from 0N to 50N with a step of 10N. It is obvious that the notch move towards the lower frequency as the load is increased. This phenomenon indicates that the TDD is increased as the applied load is increased, which leads to the change in the FSR, and thus, the change in the first-notch frequency.
Figure 5 shows the measured first-notch frequency shift as a function of the applied load from 0N to 50N with an interval of 5N. The results in Fig. 5 confirm the expected linear relationship between the applied load and the frequency of the notch, as predicted by Eq. (8). The sensitivity is estimated by linearly fitting the measurements in Fig. 5, which is 2.5 MHz/N when the length of the fiber under load is 18mm. It should be noted that in our experiment, a negative slope is obtained. This is because that a DCF with a positive dispersion coefficient is employed and the reflected light corresponding to the slow axis is delayed. In this case, when the applied load is increased, the TDD between the signals reflected from the two gratings corresponding to the slow axis and fast axis will also be increased. As a result, the notch will move towards the lower frequency linearly in response to the increasing load and thus a negative slope is obtained. Accordingly, if a fiber optic component with a negative dispersion coefficient is employed, or the position of the TDL after the PBS is exchanged, a positive slope will be obtained.
It should also worth noting that that, due to the periodicity of the MPF, there exist many notches whose frequencies are multiple times of the first-notch frequency. As can be seen from Fig. 3, for the case of no load, multiple notches at 1.558 GHz, 4.674 GHz, and 7.790
GHz are generated. In principle, according to Eq. (4), the higher the notch frequency is, the larger change against the applied load will occur. If the kth notch is chosen to track, its sensitivity will be (2k + 1) times of that of the first notch frequency. However, considering the super high frequency measurement resolution (~1Hz) of the VNA, the frequency sensitivity is not the dominant factor affecting the sensing performance. On the other hand, due to the long distance DCF, the carrier suppression effect arises at the PD, which reduces the frequency response and increases the fluctuation at higher frequency in the filter. Therefore, in our experiment, the first notch is chosen as the frequency to be measured since it offers the best signal-to-noise ratio and stability. Choosing the first notch also provide the benefit that the interrogation system can be implemented using lower frequency components at a lower cost. Moreover, it is helpful to avoid the measurement ambiguity since the first notch have the lowest frequencies among all the notches. The resolution of the sensing system is limited by the frequency measurement resolution (~1Hz) of the VNA. In this system, the sensitivity is 2.5 MHz/N for a sensing fiber with a length of 18mm. Therefore, the resolution is calculated to be 7.2 × 10−6 N/mm. In order to avoid measurement ambiguity, the measurement range is limited by the FSR. In the experiment, the microwave photonics filter has an initial FSR of 3.116GHz. Considering the sensing sensitivity is 2.5 MHz/N, the maximum load that the system can measure is 1246 N. This measurement range is high enough for most applications.
The sensitivity of the proposed sensor system can be adjusted. An adjustable sensitivity is important for a sensor since an optimal trade-off between the sensitivity and measurement range can be achieved by controlling the sensitivity. In the proposed sensing system, the sensitivity can be adjusted in two ways, by adopting longer DCF or chirp fiber grating (CFBG) which has a higher dispersion value, or by reducing the fixed TDD introduced by the TDL. In the first approach, if the dispersion value is increased, for a given wavelength separation of the two FBGs induced by the transverse load, a larger time delay change will be generated. Thus, a smaller load will cause a greater notch frequency shift, leading to an increased sensitivity. The major disadvantage in implementing this approach is that the use of the high dispersion components making the system costly and amplify the carrier suppression effect. A simple method to achieve an adjustable resolution is to reduce the fixed TDD, which can be done by simply adjusting the TDL. However, based on Eq. (8), choosing lower fixed TDD may weaken the linearity of the sensor system, which is important for sensing application. In general, an increased sensitivity will make the measurement range reduced. Thus, there is a trade-off between the sensitivity and measurement range.
In summarize, a FBG-based transverse load sensor by utilizing a notch MPF has been proposed and experimentally demonstrated. The fundamental principle of the work is to translate the wavelength separation changes of the FBGs induced by the transverse load to the notch frequency shift of the MPF. The experiment results showed that a high sensitivity of 2.5 MHz/N was achieved for a sensing fiber with a length of 18mm. The proposed sensor system exhibits several important features, including high resolution, immunity to coherent interference and adjustable sensitivity.
This work was supported by the National Natural Science Foundation of China (NSFC) (61307108); Natural Science Foundation of Jiangsu Province (BK20161562); Major Project of Nature Science Research for Colleges and Universities in Jiangsu Province (15KJA140002). Y. Wang acknowledges the Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions.
References and links
1. A. D. Kerdey, M. A. Davis, H. J. Patrick, K. P. Koo, C. G. Askins, M. A. Putnam, and E. J. Friebele, “Fiber Grating Sensors,” J. Lightwave Technol. 15(8), 1442–1463 (1997). [CrossRef]
2. Y. Wang, M. Wang, and X. Huang, “High sensitivity fiber Bragg grating transversal force sensor based on centroid measurement of polarization dependent loss,” Meas. Sci. Technol. 21(6), 065304 (2010). [CrossRef]
3. R. B. Wagreich, W. A. Atia, H. Singh, and J. S. Sirkis, “Effects of diametric force on fiber Bragg gratings fabricated in low birefringent fiber,” Electron. Lett. 32(13), 1223–1224 (1996). [CrossRef]
4. C. M. Lawrence, D. V. Nelson, and E. Udd, “Measurement of transversal strains with fiber gratings,” Proc. SPIE 3042, 218–228 (1997). [CrossRef]
5. T. Mawatari and D. Nelson, “A multi-parameter Bragg grating fiber optic sensor and triaxial strain measurement,” Smart Mater. Struct. 17(3), 035033 (2008). [CrossRef]
6. J. Capmany and D. Novak, “Microwave photonic combines two words,” Nat. Photonics 1(6), 319–330 (2007). [CrossRef]
7. X. Dong, L. Y. Shao, H. Y. Fu, H. Y. Tam, and C. Lu, “Intensity-modulated fiber Bragg grating sensor system based on radio-frequency signal measurement,” Opt. Lett. 33(5), 482–484 (2008). [CrossRef] [PubMed]
9. A. L. Ricchiuti, D. Barrera, S. Sales, L. Thevenaz, and J. Capmany, “Long fiber Bragg grating sensor interrogation using discrete-time microwave photonic filtering techniques,” Opt. Express 21(23), 28175–28181 (2013). [CrossRef] [PubMed]
10. Y. Wang, J. Zhang, and J. Yao, “An Optoelectronic Oscillator for High Sensitivity Temperature Sensing,” IEEE Photonics Technol. Lett. 28(13), 1458–1461 (2016). [CrossRef]
11. J. Yao, “Microwave photonics,” J. Lightwave Technol. 27(3), 314–335 (2009). [CrossRef]
12. W. Liu, W. Li, and J. P. Yao, “Real-time interrogation of a linearly chirped fiber Bragg grating sensor for simultaneous measurement of strain and temperature,” IEEE Photonics Technol. Lett. 23(18), 1340–1342 (2011). [CrossRef]
14. Y. Wang, J. Zhang, O. Coutinho, and J. Yao, “Interrogation of a linearly chirped fiber Bragg grating sensor with high resolution using a linearly chirped optical waveform,” Opt. Lett. 40(21), 4923–4926 (2015). [CrossRef] [PubMed]
16. J. Capmany, B. Ortega, and D. Pastor, “A tutorial on microwave photonic filters,” J. Lightwave Technol. 24(1), 201–229 (2006). [CrossRef]
17. J. Capmany, J. Cascon, J. L. Martin, S. Sales, D. Pastor, and J. Marti, “Synthesis of fiber-optic delay line filters,” J. Lightwave Technol. 13(10), 2003–2012 (1995). [CrossRef]
18. D. B. Hunter and R. A. Minasian, “Reflectively tapped fibre optic transversal filter using in-fiber Bragg gratings,” Electron. Lett. 31(12), 1010–1012 (1995). [CrossRef]