A high-sensitivity ultrasonic sensing system is proposed and demonstrated. In this system, a phase-shifted fiber Bragg grating (PS-FBG) is used as a sensor to achieve broadband and highly sensitive detection. The PS-FBG modulates the output of a tunable laser to detect the ultrasonic strain directly. Balanced photo-detector (BPD) is used for eliminating the DC component and further amplifying the AC component in the detected signal. Another major function of the BPD is to reject laser intensity noise. As a result, the minimum detectable strain is limited by the BPD’s noise and laser frequency noise. The sensitivity of the system is 9 nε/Hz1/2. Because of its high sensitivity, this system has the potential to be used in acousto-ultrasonic testing without amplifying the input signal and in practical acoustic emission detection.
© 2012 OSA
Ultrasonic testing is a versatile technique, which can be used in applications such as structural health monitoring, flaw detection, dimensional measurements, and material testing. For example, to detect the occurrence of damage, ultrasonic testing detects the acoustic emission (AE) signal that accompanies the appearance of damage. The AE signal always has frequencies from 1 kHz to 1 MHz and strain in the sub-micron range . Therefore, there are two fundamental requirements for detecting ultrasonic waves: broad bandwidth and high sensitivity.
Many researchers have proposed detecting ultrasonic waves by using an optical fiber Bragg grating (FBG) sensor to replace the traditional lead-zirconate-titanate (PZT) sensor due to the advantages of the FBG including flexibility, immunity to electromagnetic interference, corrosion resistance, small size, ability to be embedded, and multiplexing capabilities . Furthermore, another major inherent advantage of FBGs emphasized by published research is that the strain caused by ultrasonic waves is encoded in the Bragg wavelength shift of the FBG, which means that detecting strain is irrelevant to other parameters, such as optical power .
Nowadays, there are two mainstream methods of using FBG sensors for detecting ultrasonic waves: power detection [4–6] and edge filter detection [3, 7, 8]. In the first method, a broadband light source, such as amplified spontaneous emission or super-luminescent diodes, is used to illuminate the FBG sensor. A filter, such as an arrayed waveguide grating  and matched FBG , is used to demodulate the Bragg wavelength shift of the FBG sensor to optical power fluctuation. This method can achieve multiple channels easily, but it sacrifices sensitivity due to the physical nature of the broadband light source. Therefore, this low-cost method is suitable for use in acousto-ultrasonic testing, where the amplitude of the input signal can be controlled and averaging for noise reduction is also acceptable. In one of our previous works , we demonstrated an ultrasonic sensor employing cascaded PS-FBGs. However, due to the phase and amplitude noise in the broadband incoherent light source, the sensitivity of the system was still not sufficient for detecting small AE signals from a long distance, such as 1 m, which is a normal requirement in structural health monitoring.
On the other hand, the second method uses a laser as a light source for high sensitivity. In this system, a tunable laser source (TLS) is adjusted in the linear region of the spectrum of an FBG sensor, which is used as an edge filter. By detecting the reflected power of the FBG, the ultrasonic signal can be analyzed. Usually, a uniform FBG is used as a sensor. However, according to the theory in , the fiber length determines the upper limit of the detectable frequency. Therefore, a FBG with a short grating length is needed, especially for high-frequency signals, but consequently the edge of the FBG has a gentle slope, leading to the low sensitivity. In order to resolve the trade-off between bandwidth and sensitivity, Rosenthal et al. proposed a method using a phase-shifted FBG (PS-FBG) to decrease the effective grating length and improve the sensitivity . Another solution that enhances the sensitivity is to use the transmit and reflect detection method, proposed by Wild et al. . In this design, the reflected and transmitted lights are detected by separated photo-detectors (PDs), and the resulting signals are subtracted from each other to double the strain signal. However, some deformations of the signal are present due to the large DC component in the signal, and the noise level is almost the same according to the results. Sometimes, in order to achieve better performance, another PD is used to monitor the intensity of the laser, which increases the cost and complexity of the system. Compared to the power detection methods, the edge filter detection method has higher sensitivity, but the sensitivity in these methods is still limited and is difficult to use in AE detection, in practice. Recently, pulse interferometry technique was proposed. This technique combines the merits of the power detection method and the edge filter detection method, but high cost is also an issue that must be considered in practical applications . In addition, the sensitivity in this method is comparable to, rather than higher than that of, normal edge filter detection.
Another issue in both methods is the conversion of the strain signal encoded in the Bragg wavelength shift to power modulation. To a certain extent, this eliminates the major inherent advantage of the FBG sensor. Hence, the minimum detectable strain is limited by the noise generated in the system. For example, in the edge filter detection method, the noise level is mainly determined by the laser intensity noise. Thus, how to improve the sensitivity is a basic and important issue. The Pound-Drever-Hall technique [11–13] is used to eliminate laser intensity noise from the measurements; thus, frequency-noise-limited ultrahigh sensitivity can be achieved. Although the Pound-Drever-Hall technique has the potential to detect ultrasonic waves, until now the highest achieved frequency has been 20 kHz , which is lower than the requirement of AE signal detection. Furthermore, this technique is complicated, expensive, and has weak multiplexing ability.
In this paper, we propose a new system with a simple structure for detecting ultrasonic waves precisely. In our design, a TLS and a PS-FBG are connected by a circulator. The laser power reflected and transmitted by the PS-FBG sensor is input into the two ports of a balanced photo-detector (BPD). The ultrasonic wave can be recorded by monitoring the output of the BPD directly. There are four advantages to this system: 1) high-frequency signals can be detected, 2) the DC component in the detected signal is removed, 3) the amplitude of the signal is amplified by the PS-FBG and the BPD, while the laser intensity noise is rejected by the BPD, allowing a very high sensitivity to be achieved, and 4) a low cost is guaranteed by the simple structure.
2. Experimental setup and principle of operation
2.1 System configuration
Figure 1 is a schematic diagram of the system. An external cavity TLS (Agilent, 81682A) with a 100-kHz linewidth and 0.1-pm tunable resolution is used as a light source. Port 1 and port 2 of a circulator are connected to the laser and the PS-FBG sensor, respectively. The reflected and transmitted light from the PS-FBG are input into the two ports of a BPD (New Focus, 2117). This BPD contains two identical PDs, a two-stage transimpedance amplifier, and a band-pass filter . The gain of the BPD was set to 10 to amplify the small signals. A 5-mm PS-FBG sensor is glued onto the surface of a 250 × 250 × 2 mm3 (l × w × h) aluminum plate. Another 1.5-mm uniform FBG sensor is also glued to the surface in order to compare the different sensitivities among different systems . Both sensors are 50 mm away from the PZT actuator (Fuji Ceramics, M31) in the center of the aluminum plate. A high-acoustic-impedance ultrasonic couplant and a cyanoacrylate adhesive are used to glue the PZT actuator and FBG sensors to the aluminum plate, respectively. A three-cycle sinusoidal burst wave at 0.3 MHz with a Hamming window is generated by a function generator (NF, WF1974), and the peak-to-peak voltage of the wave is amplified to 150 V by a high-speed bipolar amplifier (NF, HSA4012) to drive the PZT actuator.
The basic principle of this system is edge filter detection. Compared to the spectrum of the FBG or the PS-FBG, the linewidth of the TLS is so narrow, that it can be treated as a single wavelength. Before measurement, in order to balance the reflected and transmitted optical powers entering the two ports of the BPD, calibration of the wavelength of the laser to the 3-dB position of the peak in the transmission spectrum of the PS-FBG sensor is needed. When the Bragg wavelength of the PS-FBG is shifted by the strain caused by the ultrasonic wave, the laser power is modulated, which leads the transmitted and reflected power vibration in port 1 and port 2, respectively. By monitoring the output voltage of the BPD, the ultrasonic wave is detected directly.
2.2 Function of the PS-FBG sensor
The PS-FBG sensor has two functions in this design. First, it guarantees the broadband property of this system, which is verified in . Then, due to the steep peak in the spectrum of the PS-FBG, the sensitivity is increased significantly. Similar to the theory presented in , in this system, the AC voltage of the signal can be written as:
Figure 2 shows the spectra of the FBGs we used in the experiment. The FBGs were manufactured by the Fujikura Company according to our design. The sensing range (indicated by the red lines in Fig. 2) in the FBGs is considered to be the region where there is a linear optical response to strain. This region is defined by using the magnitude of residual form. Compared to these two spectra, the slopes of the PS-FBG and the normal FBG in the linear regions are 87 nm−1 and 0.48 nm−1, respectively, which will result in the same input signal yielding detected signals with amplitudes that differ by a factor of about 180 according to Eq. (1). Though the dynamic range is smaller in the PS-FBG because of the narrow peak, for the application of ultrasonic detection, this is not a problem since the strain is always small.
2.3 Function of the balanced photo-detector
The BPD [16, 17] also has three important functions: 1) to remove the large DC component to clearly show the small AC component in a signal, 2) to double the amplitude of the AC signal, and 3) to reject laser intensity noise and common mode noise.
First, in static situations, the reflected and transmitted light powers of the PS-FBG are equal, leading to a 0-V offset in the BPD. This property can effectively prevent large optical power from saturating the output of the PD.
Second, when an AC signal is caused by the Bragg wavelength shift of the PS-FBG sensor, the voltages in port 1 and port 2 will simultaneously experience changes with opposite phases and the same amplitude, which leads to the final output AC signal having twice the amplitude that would be in a single port. Therefore, the final voltage can be written as:
Third, the noise level will decrease because the laser intensity noise and common mode noise will be rejected by the BPD. In a traditional system, there are four main kinds of noise sources. Those are the noise in the PD itself, common mode noise, laser intensity noise, and laser frequency noise. The measured noise can be written as Eq. (3), because all of those noise sources are uncorrelated.
2.4 Minimum detectable strain
Furthermore, according to the theory in , the Bragg wavelength shift is proportional to the applied strain, which is expressed as:Eq. (2), Eq. (4), and Eq. (5) into Eq. (6), the minimum detectable strain is expressed as:Equation (7) indicates that the gain setting in the PD has no influence on the minimum detectable strain. Furthermore, as we know, is irrelevant to , but is proportional to . Thus, increasing the input laser power partially functions to improve the minimum detectable strain, but the improvement is not linear with the laser power increase. In addition, because of the narrow linewidth of the TLS, is enough small to compare with , as is demonstrated in the following experiment. For better analyzing the sensitivity of the system, Eq. (7) also can be written in the form of power spectral densities (PSDs), given by:
3. Results and discussions
3.1 Performance of BPD
In this experiment, the performance of the BPD was verified. Four different sets of experimental conditions were tested: 1) the BPD without laser input, 2) laser power of 0 dBm input into port 1 of the BPD, 3) laser powers of 0.13 dBm and −0.3 dBm input into port 1 and port 2 of the BPD, respectively, and 4) laser power of 0 dBm input into each of the ports of the BPD. The band-pass filter in the BPD was not used in this experiment. The DC voltages in each of these conditions were 0.0057 V, 6.3499 V, 0.8588 V, and −0.0008 V, respectively. The voltage in condition 1 means that this BPD was not perfectly balanced. However, we believe this low offset voltage barely influenced the experimental results. The voltages in other conditions show the ability of this BPD to remove the DC voltage in the signal, and the more balanced the optical powers are, the better the elimination of DC voltages is.
The PSDs of the AC voltage signals are shown in Fig. 3 to illustrate the noise rejection ability. The PSDs are calculated by the Fast Fourier Transform of the temporal response, which is the method applied in all of the following PSD figures. In Fig. 3, condition 2 indicates the largest noise level, while conditions 3 and 4 have lower noise levels, especially when the power in both ports of the BPD is matched. This phenomenon means that the BPD has efficient noise rejection ability, as suggested by Eq. (4). Though the DC component in condition 3 is small, the noise level is 3 dB larger at high frequencies and 8 dB larger at 1.5 MHz than that in condition 4, which means that in order to reject laser intensity noise, critical balancing of the powers of the two ports is very important. Thus, when the DC voltage is removed, the BPD will achieve the best performance. Moreover, the noise level in condition 4 is slightly higher than that in condition 1. The possible causes for this difference are the interferometric noise  because the end of the fiber used in this experiment has a PC connector.
3.2 High sensitivity of the system
In the second experiment, a comparison of three different FBG sensing systems was performed to determine the signal-to-noise ratio (SNR) improvement of our system. Those systems are: 1) normal FBG sensing system , 2) PS-FBG sensing system , and 3) our novel PS-FBG balanced sensing system. The laser power was set to 0 dBm. In this part of the experiment, the band-pass filter in the BPD was set to transmit 10 kHz–12 MHz to filter out the low-frequency variation of the Bragg wavelength of the PS-FBG caused by temperature fluctuations. Figure 4 shows the temporal responses and PSD curves of the experimental results. The temporal responses in condition 2, condition 3 and the inset of condition 1 are real-time waveforms, and the temporal response in condition 1 is the averaged waveform. All the systems can detect the ultrasonic wave correctly. The AC amplitudes in each condition are 0.0044 V, 0.36 V, and 0.7552 V, respectively. The AC amplitude of condition 3 is larger than that of condition 2 due to the property that this system uses both the reflected and transmitted light modulated by the PS-FBG. However, the amplification is not precisely a factor of two because the laser is not located in the 3-dB position of the transmission spectrum of the PS-FBG due to the 2-dB insertion loss from port 2 to port 3 of the circulator used in this experiment. Therefore, in order to compensate for the insertion loss of the circulator, the wavelength of the TLS is positioned at about the 5-dB position. Furthermore, the AC amplitude of condition 2 is 82 times larger than that of condition 1 due to the different slopes, as can be seen from Eq. (1). The actual amplification is also smaller than the prediction because the laser wavelength offset at the 5-dB position causes the slope to decrease. We believe that the performance of this system can be further developed by using a circulator with a small insertion loss. Another explanation is that this is caused by the variation in strain coupling between the plate and sensor due to bonding variations. From Fig. 4(b), it is obvious that our novel system has the best SNR, as high as 28 dB, which is 14 dB better than the SNR in the PS-FBG sensor and about 30 dB better than that in a traditional FBG sensor. Moreover, the FBG sensor has the worst SNR, which requires averaging over 1024 samples to decrease the noise level for manifesting the wave enough to prevent it from being buried under the noise.
The performance of the system under different input optical powers was also tested, as shown in Fig. 5 . The input laser power was changed to 3 dBm, 0 dBm, and −3 dBm, while other experimental conditions were the same as in the above experiment. Figure 5 shows that the received signals also have 3-dB differences in each case, matching the differences of input laser power. Although the noise generated by the 3-dBm laser input is the largest and the noise generated by −3 dBm is the smallest, the difference between them is much less than 6 dB. This phenomenon matches the theory in Eq. (7) very closely, which means that both the PD’s noise and the laser frequency noise contribute to the final detected noise and that these two types of noise are comparable. Increasing the input laser power can partially improve the performance of the system. When 3 dBm of laser power is used, an SNR of about 30 dBm can be achieved in this experiment.
3.3 Applications of the system
Finally, two practical applications were also experimented. A high-speed amplifier is always necessary for amplifying an ultrasonic wave to a large amplitude in other experiments [4, 7]. Thus, every PZT actuator needs an amplifier, leading to an expensive system. However, due to the high sensitivity in our system, a high-speed amplifier is no longer necessary. The PZT actuator was connected to the function generator directly in this experiment. Signals with peak-to-peak voltages of 10 V, 1 V, and 0.1 V were input into the PZT actuator, respectively. The band-pass filter in the BPD was set to transmit 10 kHz–1 MHz to filter out the high frequencies in order to improve the SNR further. The input laser power was 3 dBm. Figure 6 shows the PSDs of the system for signals of different input levels. From Fig. 6, the signals for each of the experimental conditions are separated by differences of about 10 dB, which is proportional to the input amplitude of the input signal.
Under a separation of 5 cm between the PZT actuator and the PS-FBG sensor and a laser input power of 3 dBm, the strain generated by a 0.1-V signal is considered to be the minimum detectable strain because the signal level is almost the same as the noise level. At the 5-dBm position, the slope of the PS-FBG is = 70 nm−1. The PD’s response factor is 1 V/mW. The detected noise level is = 0.03 V/Hz1/2. According to Eq. (8), the corresponding minimum detected sensitivity in this system is calculated as: = 9 nε/Hz1/2.
A simulated AE signal generated by pencil lead break test was also performed to verify the practical application of this system. In this experiment, the testing subject was changed to a long, thin steel plate with dimensions of 1000 × 35 × 1.5 mm3 (l × w × h). The AE signal source was generated by breaking a pencil lead at one end of the plate, and detected by our sensing system at the other end. A laser input power of 0 dBm was used. Figure 7(a) shows the wave detected without a filter, which contains both low- and high-frequency signals. In order to obtain only the AE signal at high frequency as measured by the traditional PZT sensor, the band-pass electrical filter was set to a low cut-off frequency of 1 kHz. In the AE signal detected after the filter, as shown in Fig. 7(b), the S0 and A0 modes of the Lamb wave are clearly separated. We believe this result shows strong potential for application to non-destructive testing.
A high-sensitivity ultrasonic PS-FBG balanced sensing system was proposed and demonstrated in this study. By adjusting the wavelength of the TLS, the reflected and transmitted optical powers of the PS-FBG sensor were balanced in two ports of a BPD. The BPD was used to detect the optical power modulated by the Bragg wavelength shift. Using the strain encoded in the Bragg wavelength shift, an ultrasonic wave was detected. The requirement of high sensitivity in ultrasonic testing is satisfied by two aspects of this approach: 1) the steep slope of the peak in the PS-FBG spectrum and 2) the ability of the BPD to eliminate the DC signal, double the AC signal, and reject the laser intensity noise and the common mode noise. Both theory and experimental results show that the system is immune to laser intensity noise, leading to the achievement of high sensitivity, which is limited by BPD’s noise and laser frequency noise. The experimental results show that the sensitivity of this novel system is 14 dB and 30 dB higher than that of the PS-FBG sensor and the FBG sensor, respectively. The detectable sensitivity in this system is 9 nε/Hz1/2. Moreover, a 0.1-V ultrasonic signal can be detected without the help of an amplifier, and a simulated AE signal at a distance of 1 m from the PS-FBG senor can be detected precisely. Due to the high sensitivity, broadband property, simple structure, and low cost, this system presents the potential for many practical applications.
The authors would like to thank Mr. Koji Omichi and Mr. Ryujiro Nomura from the Fujikura Company for providing the PS-FBG and the BPD. The authors also would like to thank Dr. Takeda for lending us the TLS.
References and Links
1. C. U. Grosse and M. Ohtsu, Acoustic Emission Testing: Basics for Research—Applications in Civil Engineering, (Springer, 2008).
2. G. Wild and S. Hinckley, “Acousto-ultrasonic optical fiber sensors: overview and state-of-the-art,” IEEE Sens. J. 8(7), 1184–1193 (2008). [CrossRef]
3. G. Wild, S. Hinckley, and P. V. Jansz, “A transmit reflect detection system for fiber Bragg grating photonic sensors,” Proc. SPIE 6801, 68010N, 68010N-9 (2007). [CrossRef]
4. Y. Okabe, K. Fujibayashi, M. Shimazaki, H. Soejima, and T. Ogisu, “Delamination detection in composite laminates using dispersion change based on mode conversion of Lamb waves,” Smart Mater. Struct. 19(11), 115013 (2010). [CrossRef]
5. I. Perez, H. L. Cui, and E. Udd, “Acoustic emission detection using fiber Bragg gratings,” Proc. SPIE 4328, 209–215 (2001). [CrossRef]
6. Q. Wu and Y. Okabe, “Ultrasonic sensor employing two cascaded phase-shifted fiber Bragg gratings suitable for multiplexing,” Opt. Lett. 37(16), 3336–3338 (2012). [CrossRef]
7. H. Tsuda, K. Kumakura, and S. Ogihara, “Ultrasonic sensitivity of strain-insensitive fiber Bragg grating sensors and evaluation of ultrasound-induced strain,” Sensors (Basel) 10(12), 11248–11258 (2010). [CrossRef] [PubMed]
8. A. Rosenthal, D. Razansky, and V. Ntziachristos, “High-sensitivity compact ultrasonic detector based on a pi-phase-shifted fiber Bragg grating,” Opt. Lett. 36(10), 1833–1835 (2011). [CrossRef] [PubMed]
9. A. Minardo, A. Cusano, R. Bernini, L. Zeni, and M. Giordano, “Response of fiber Bragg gratings to longitudinal ultrasonic waves,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control 52(2), 304–312 (2005). [CrossRef] [PubMed]
12. D. Gatti, G. Galzerano, D. Janner, S. Longhi, and P. Laporta, “Fiber strain sensor based on a pi-phase-shifted Bragg grating and the Pound-Drever-Hall technique,” Opt. Express 16(3), 1945–1950 (2008). [CrossRef] [PubMed]
13. S. Avino, J. A. Barnes, G. Gagliardi, X. Gu, D. Gutstein, J. R. Mester, C. Nicholaou, and H. P. Loock, “Musical instrument pickup based on a laser locked to an optical fiber resonator,” Opt. Express 19(25), 25057–25065 (2011). [CrossRef] [PubMed]
15. A. Arie, B. Lissak, and M. Tur, “Static fiber-Bragg grating strain sensing using frequency-locked lasers,” J. Lightwave Technol. 17(10), 1849–1855 (1999). [CrossRef]
16. M. S. Islam, T. Chau, S. Mathai, T. Itoh, M. C. Wu, D. L. Sivco, and A. Y. Cho, “Distributed balanced photodetectors for broad-band noise suppression,” IEEE Trans. Microw. Theory 47(7), 1282–1288 (1999). [CrossRef]
17. A. Joshi, X. Wang, D. Mohr, D. Becker, and C. Wree, “Balanced photoreceivers for analog and digital fiber optic communications,” Proc. SPIE 5814, 39–50 (2005). [CrossRef]
18. A. Othonos, “Fiber Bragg gratings,” Rev. Sci. Instrum. 68(12), 4309–4341 (1997). [CrossRef]