Jun 1, 1997 - Minho Song and Byoungho Lee. School of Electrical Engineering, Seoul National University, Shilim-Dong, Kwanak-Gu, Seoul 151-742, Korea.
790
OPTICS LETTERS / Vol. 22, No. 11 / June 1, 1997
Interferometric temperature-insensitive strain measurement with different-diameter fiber Bragg gratings Minho Song and Byoungho Lee School of Electrical Engineering, Seoul National University, Shilim-Dong, Kwanak-Gu, Seoul 151-742, Korea
Sang Bae Lee and Sang Sam Choi Applied Physics Group, Korea Institute of Science and Technology, P.O. Box 131, Cheongryang, Seoul 130-650, Korea Received January 2, 1997 An interferometric strain-sensing system that combines an unbalanced Mach – Zehnder wavelength discriminator and spliced different-diameter fiber Bragg gratings is presented that has the capability to distinguish strain from temperature effects. Because the gratings are written in 135- and 165-mm-diameter fibers drawn from the same preform, when they are spliced they show similar temperature sensitivities but different responses to applied strain. Thus the relative Bragg wavelength difference of the two gratings is affected only by strain, and one can measure it accurately by monitoring the amplitude variation of the interference signal rather than the conventional phase variation. 1997 Optical Society of America
Fiber sensors based on photogenerated Bragg gratings have been investigated intensively for the past few years because of their advantages over conventional fiber sensors, which include suitability for use in multiplexed sensor networks and smart structures. Because the sensing principle is based on wavelength encoding, various Bragg wavelength interrogation methods have been reported and demonstrated successfully to improve measurement accuracy to better than that of an optical spectrum analyzer.1 – 3 However, use of a Bragg grating sensor in a harsh environment is limited by the fact that the Bragg grating is sensitive to both temperature and strain, which complicates independent measurement of each quantity. To solve this problem, many authors have presented simultaneous-measurement techniques for measuring temperature and strain or temperaturechange-compensated strain.4 – 9 Many of the reported techniques use two gratings with different characteristics in a sensor head and calculate both quantities from the different wavelength shifts of the gratings. But in these two-grating geometries the conventional high-resolution interferometric method1 cannot be used because the light from the two gratings overlaps at the detector and produces complicated phase variations that cannot discriminate between applied strain and temperature-induced wavelength shifts. Recently, spliced different-diameter f iber gratings were used for simultaneous measurement of strain and temperature6 and for dual-wavelength laser tuning.7 In Ref. 7 we showed that spliced f iber gratings written in two different-diameter fibers of identical material characteristics have different wavelength shifts in response to induced strain but the same responses to temperature change. Therefore the relative wavelength difference of the two gratings is immune to temperature variation and is affected only by the applied strain. In this Letter we show that, by taking advantage of this immunity to temperature, one can use the sensitive interferometric method with an unbalanced Mach– Zehnder interferometer1 to 0146-9592/97/110790-03$10.00/0
measure the applied strain by simple amplitude monitoring rather than by the conventional phase-variation technique. The experimental schematic for the proposed sensor system is shown in Fig. 1. At the two gratings, the light that satisf ies the Bragg condition of the gratings is ref lected back and generates the interference signals at the photodetector. The two beams of light ref lected from the two gratings do not interfere with each other, but each beam suffers interference that is due to the unbalanced Mach –Zehnder interferometer. (The Mach –Zehnder interferometer can be moved to location A in Fig. 1.) From the basic interference theory, the interference signal at the detector can be expressed as Idet Idc 1 a cos f1 1 b cos f2 Idc 1 sa2 1 b2 1 2ab cos Dfd1/2 n o 3 cos f1 2 tan21 fb sin Dfysa 1 b cos Dfdg , (1)
Fig. 1. Schematic diagram of the experimental setup. The sensor head is composed of 135- and 165-mm-diameter fiber Bragg gratings: ISO’s, isolators; EDF, erbium-doped fiber; WDM, wavelength division multiplexing; PZT, piezoelectric transducer; PC, polarization controller; PD, photodetector. 1997 Optical Society of America
June 1, 1997 / Vol. 22, No. 11 / OPTICS LETTERS
where fi vt 1 2pndyli 2 s2pndyli 2 dDli 1 f0 and Df f1 2 f2 . Here a and b are the coeff icients that denote the combined power level and the coherence degree of the ref lected light, v is the interference signal frequency determined by the piezoelectric ramp signal, d s,2 mmd is the fiber path imbalance of the interferometer, f0 is the thermal phase drift, li si 1, 2d is the Bragg wavelength of grating i, and Dli is the induced wavelength shift of each grating by the measurement quantities applied. As shown in Eq. (1), the phase of the overall interference signal contains mixed phase information from each grating, so strain and temperature effects cannot be separated with this phasevariation measurement. But the amplitude of the signal is related only to the relative phase difference Df of the two signals, and because it does not depend on temperature variation in the scheme of differentdiameter gratings that have the same temperature sensitivities, we can obtain temperature-insensitive relative phase information from this amplitude variation. If we assume that the two Bragg wavelengths are nearly the same, the relative phase difference Df can be expressed as Df > s2pndyl2 dsDl1 2 Dl2 d s2pndyl2 dfske De1 2 ke De2 d 1 skT 2 kT dDT g s2pndyl2 dke Desl1 1 l2 d 3 sr2 2 2 r1 2 dysr1 2 l2 1 r2 2 l1 d ,
(2)
791
wavelengths when the fibers are strained, the thicker grating was chosen to yield the shorter wavelength and was strained in the fabrication process to make its Bragg wavelength shorter than that determined by the phase mask (,1550 nm). After the two gratings were fabricated they were spliced by a fusion splicer with ,3-cm spacing. To test the assumed characteristics of temperature and strain, we carefully heated and strained several different diameter gratings, using a controlled heater plate and micrometer-driven translation stages. The shifts of the Bragg wavelengths were monitored with an optical spectrum analyzer with 0.1-nm resolution and are plotted in Fig. 2. In Fig. 2(a) the gratings show the same temperature sensitivities of 10.65 6 0.005 pmy±C. Different wavelength shifts of the spliced sensor head (135and 165-mm fibers combined) are plotted in Fig. 2(b) and correspond to the total strain. As a broadband light input to the interferometer we used superf luorescent light (, 250-mW ) from an erbium-doped fiber to obtain enough optical power to generate an interference signal at the photodetector. At the end of the erbium-doped fiber, a wavelengthdivision-multiplexing coupler was inserted to extract unabsorbed 980-nm pump laser light, which can be ref lected at the fiber end surface and can generate its own interference signal. Figure 3 shows oscilloscope traces of the maximum and the minimum interference signals and the piezoelectric modulation ramp signal (250 Hz), Fig. 3(a) for the 200-mstrain-
where Dei Dli yli and De sDl1 1 Dl2 dysl1 1 l2 d. In relation (2), Dei si 1, 2d is the strain of each grating, De is the total strain given to the sensor head, and l1 and l2 s20 cm each in the experiment) are the lengths of fiber between the splicing point and the anchoring point at each end. It is assumed that Dei of each grating is inversely proportional to the grating’s cross-sectional area with De1 yDe2 sr2 yr1 d2 , where ri si 1, 2d is the cladding diameter of each grating. Because both gratings have the same material characteristics, the strain and temperature coefficients ke and kT , respectively, of the two gratings are the same. To calculate the applied strain from the measured signal amplitude, we must determine the coefficients a and b in Eq. (1). By prestraining the sensor head, we can see the maximum and the minimum amplitudes, and from the observed maximum and minimum signal amplitudes Imax and Imin, respectively, the coeff icients a and b can be obtained as follows: Imax I sDf 2npd a 1 b, Imin I sDf 2np 1 pd a 2 b, a sImax 1 Imin dy2,
b sImax 2 Imin dy2 .
(3)
Finally, with the coeff icients a and b determined by Eqs. (3), we can calculate the applied strain by combining Eq. (1) and relation (2). The nominal Bragg wavelengths of the two gratings G1 and G2 written in 135- and 165-mm-diameter fibers are 1549.3 and 1546.4 nm (at 25 ±C), respectively. They were written with a KrF excimer laser (248 nm) and a phase mask. To avoid overlap of two Bragg
Fig. 2. Temperature and strain responses of differentdiameter Bragg gratings. All fibers were drawn from the same preform (KIST-1104, 18 mol. %, Ge doped). (a) Temperature sensitivities of (100-, 135-, 154-, and 165-mm-diameter) gratings. (b) Strain responses of spliced gratings.
792
OPTICS LETTERS / Vol. 22, No. 11 / June 1, 1997
Fig. 3. Interference signal traces for (a) maximum- and (b) minimum-amplitude cases. The lower traces are modulation signals applied to a piezoelectric transducer. The increase in wavelength difference between these two cases was 0.6 nm.
Fig. 4. Normalized amplitude variation with the total strain. The curve was calculated from Eq. (1). Inset, wider-range output variation.
induced case and Fig. 3(b) for the 1500-mstraininduced one. From these maximum and minimum values we determined the coeff icients a and b in Eqs. (3). The change in relative wavelength difference between these two cases is 0.6 nm. The detailed amplitude variation at each strain was carefully measured, and the results are shown in Fig. 4 along with the applied strain. We compared these results with those calculated from Eqs. (1) and (3) and found them to be in good agreement in the measurement range. Because of the 2p ambiguity in measurement range, one should set the range and resolution of measurements by changing the ratio of the f ibers’ diameter or the path imbalance of the interferometer, in quasi-static strain measurement. The wavelength-difference resolution of the electronics used in the experiment was ,1.2 pm s,2.6 med, nearly 80 times better than that of the conventional optical spectrum analyzer. It could be further improved by use of a more sensitive analog-to-digital converter
without altering the measurement range. The possible sources of errors that could lead to meaningless variation in this measurement of signal amplitude are power f luctuation of the light source and other losses in the optical system. In the experiment described here, there were no noticeable errors originating from power f luctuation. A self-referencing technique for establishing long-term stability is nonetheless under construction in our laboratory. The nonuniform gain prof ile of the erbium-doped f iber broadband source could also be an error source. At the first trial we used gratings with , 1535-nm Bragg wavelengths, but because of the steep gain curve variation at the wavelength range the result deviated slightly from the expectation. We solved this problem by using the f lat gain region of the broadband spectrum. A gainf lattened broadband spectrum would be needed for a wide usable wavelength range. In conclusion, we have constructed an effective temperature-insensitive strain sensor that consists of spliced different-diameter f iber gratings and uses an interferometric wavelength-difference measuring technique. The output signal measured is not phase variation but the amplitude change of the interference signal at the detector. Because the Bragg wavelength difference of the sensor head shows immunity to temperature variation, the measured variation in signal amplitude showed no thermal drift and had much higher resolution than that of an optical spectrum analzyer. Furthermore, because the phase term of the resultant interference signal contains mixed information on strain and temperature, temperature could also be measured simultaneously if we subtracted the strain effect from the mixed phase information. References 1. A. D. Kersey, T. A. Berkoff, and W. W. Morey, Electron. Lett. 28, 236 (1992). 2. A. D. Kersey, T. A. Berkoff, and W. W. Morey, Opt. Lett. 18, 1370 (1993). 3. M. G. Xu, H. Geiger, and J. P. Dakin, J. Lightwave Technol. 14, 391 (1996). 4. M. G. Xu, J.-L. Archambault, L. Reekie, and J. P. Dakin, Electron. Lett. 30, 1085 (1994). 5. S. E. Kanellopoulos, V. A. Handerek, and A. J. Rogers, Opt. Lett. 20, 333 (1995). 6. S. W. James, M. L. Dockney, and R. P. Tatam, Electron. Lett. 32, 1133 (1996). 7. S. B. Lee, L. Yu, and S. S. Choi, presented at the 1996 Optoelectronics and Communications Conference, Tokyo, July 16 – 19, 1996. 8. H. J. Patrick, G. M. Williams, A. D. Kersey, J. R. Pedrazzani, and A. M. Vengsarkar, IEEE Photon. Technol. Lett. 8, 1223 (1996). 9. G. W. Yoffe, P. A. Krug, F. Ouellette, and D. A. Thorncraft, Appl. Opt. 34, 6859 (1995).
