Tapped serial interferometric fiber sensor array with time division ...

wDD4-1 Reprinted with permission from Optical Fiber Sensors, 1988 Technical Digest Series, Vol. 2, Optical Society of America [Washington, DC., 1988). 0 1988 OSA.

Tapped Serial Interferometric Fiber Sensor Array with Time Division Multiplexing A. D. Kersey?and A. Dandridge Optical TechniquesBranch Code 6570 Naval ResearchLaboratory Washington, DC 203755000 Abstract: We report a new architecture for multiplexed interferometric fiber-optic sensors basedon time-division addressing.Intrinsic crosstalk of the system is analysed and compared to experimental results obtained using a three element network. Introduction: There have been a number of different approaches to the multiplexing of interferometric fiber-optic sensors reforted very recently 1. These include schemes based on conventional formats of frequency v3and time4-’ division multiplexing, and other, more optically complex and specialized schemes such as coherence multiplexing8*9. The primary objective of this research interest in multiplexing is the development of techniques which allow compact arrays, or other networks of interferometric sensors, to be efficiently addressed and demodulated using a minimal number of connecting fibers. This paper reports a new architecture of interferometric sensors in which a serial array of sensor elements is tapped between each element by a single output fiber bus. The array can be viewed as the transmissiveanalogueof the low-reflectance Fabry-Perot array, and is capable of supporting a large number of sensors. Principle: The architecture of the proposed array is schematically shown in Figure 1. Pulsed light from a laser source is launched into a long fiber line which forms the series of sensor elements each of length L. A small fraction ( t$ ) of the optical power in this fiber is tapped-off to an output fiber bus at points between each sensor element, and at points before the first and after the last (N th) elements. If the optical propagation delay in each sensor element is greater than the width of the input pulse, the output obtained from the array consist of a series of N+l pulses which are separated in the time domain. Apart from crosstalk effects, which will be discussed later, these pulses carry no direct interferometric information. Applying this output pulse train to a compensatinginterferometer of optical path imbalance L coherently mixes pulses obtained from consecutive tap points. A series of N+2 output pulses is thus generated by the compensating interferometer, of which only the first and last carry no interferometric signals, with the central N pulses carrying information generatedby the N sensorelements.Time division demultiplexing of this output pulse train can then be used to allow individual sensors to be addressed. As with other multiplexing topologies which utilize a compensatinginterferometer6S7, the maximum output duty-cycle which can be used is N/(N+l). Demodulation of the time demultiplexed sensoroutputs can be achievedusing either ‘y&e generatedcarrier’homodyne or synthetic-heterodynetechniquesapplied to the compensator The coupling ratios of the tapping couplers required to equalise the returned power from each tap can be derived by simply equating the power from the first tap to that from the n* ( 1~ n = s/ (I- G>, i.e., equal splitting

(K)

at each tap point; the power in each pulse at the array output is thus 80

(1)

WDD4-2 P,, = T’j.P, = K.(l - K)N.p, ,

(3)

wherePOis the peak power in the input pulse. This neglectsmultiple crosscoupling of the pulses in the array, and excess loss in the system. The intrinsic crosstalk between sensorscan be shown to be directly related to the power tapping ratio K, and can be assessedby consideringthe number of interfering pulses generatedin the output of the array. Taking into accountfirst order crosstalk effects only ( i.e pulses which cross-coupleback from the output fiber to the sensor chain and back again ) it can be shown that the number of crosstalk pulses received at the array output in the time slot associatedwith the arrival of the primary pulse from the n+l* tap (i.e. t = nz) is given by M = Nn - n2.

(4)

Each of these pulses is a factor ~~ weaker than the primary tapped pulses, but mix interferometrically at the compensatoroutput with primary pulses dervied from the nth, n+l* and n+2* tap points to produce crosstalk. This leads to a worse case time averagedcrosstalk (sensor to sensor) for the centrally located sensor of

This is the result which would be expected intuitivitely, and again neglects excesslosses and polarizationeffects. Experimental and results: Our experimental system comprised three sensorelements each with an imbalance between the sensorcoil and link between the power taps of 23.30 f 0.05 m, and a compensatorwith the sameimbalancebetweenits arms. The array was constructedusing four 2x2 fiber directional couplers with power splitting ratios (tap ratios) of 8 to 10 %. The compensator was constructed using nominally 1:l couplers. Light from a 830 nm single frequency diode laser ( Hitachi HLP 1400) was passedthrough an isolator and acousto-optic modulator (AOM) and launchedinto the fiber input to the array. The AOM was driven from a pulsed RF source and effectively pulsed the optical input to the array. Figure 2.a. shows the pulse train detectedusing a Si APD at the output of the compensator.Here, a piezoelectricphase shifter in one arm of the compensatorwas modulatedby 27~rads.p-p at w 1kHz to highlight the amplitude of the ‘interferometricsignal’carriedby eachof the central threepulses. In order to time demultiplex the output pulse train, a secondAOM was employed to gate the received signal. This modulator was synchronisedwith the input modulator, with a variable delay z between the input pulse and output gate. The interferometric signal obtained was demodu4 ated using synthetic-heterodyneprocessing12. The detection sensitivity for the second sensorin the chain can be assessedat 10 urads/dHz (signal frequencyof 1 kHz) with referenceto Figure 2.b. which shows the spectrumof the output with an applied signal of 20 mrads at 1 kHz. Figure 2.c shows an example of the crosstalk observedwith the array. Here, phaseshift signals of 200 mrads rms at frequenciesof 800 Hz and 500 Hz were applied to the sensorcoils S, and S, repectively. The photograph shows the frequency spectrum of the synthesisedheterodyne carrier output ( f 1 kHz about the carrier) generatedwith the output AOM set to select the pulses correspondingto the first sensorelement. The crosstalk sidebandgeneratedby S2 are clearly observedat m -30 dB down from the principle sidebandsat f 800 Hz. The use of couplerswith a lower tapping ratio ( * 1% ) will allow the operationof this multiplexing schemewith crosstalk betweenelementsw -40 dB. Conclusions: We have describeda new interferometric sensorarray architecturewhich can be multiplexed using time-division addressing.The configuration is based on a serial network of sensor elements which is tapped between each element by a single output fiber bus. Experimental results obtained for a three sensor array gave phase-detectionsensitivities w 10 PadsdHz at 1 kHz. Crosstalk levels betweenelementsof u -30 dB were also observed. 81

wDD4-3

This work was supportedby the Office of Naval Technologyprogramon Electro-Optics. References: 1. A. Dandridge and A.D. Kersey, “Signal Processing for Optical Fiber Sensor”, SPIE Conference‘Fiber SensorsIT’, The Hauge, Netherlands, 1987.2. I.P. Giles, D. Uttam, B. Culshaw and D.E.N. Davies, “Coherent Optical-Fiber Sensors with Modulated Laser Sources”, Electron. Lett., 19, p. 14, 1983. 3. A. Dandridge,A.B. Tveten, A.D. Kersey and A.M. Yurek, “Multiplexing of Interferometric Sensors using Phase Generated Carrier Techniques”, IEEE J. Lightwave Technology, LT-5, July 1987. 4. J.P. Dakin, C.A. Wade andM.L. Henning, “Novel Optical Fibre Hydrophone Array Using a Single Laser Sourceand Detector”, Electron. Lett., 20, p. 53, 1984. 5. J.L. Brooks, M. Tur, B.Y. Kim, K.A. Fesler, and H.J. Shaw, “Fiber-Optic Interferometric SensorArrays with FreedomFrom Source-Phase-Induced Noise”, Opt. Lett., pp. 473-475, 1986. 6. J.L. Brooks, B.Y. Kim, M. Tur and H.J. Shaw, “Sensitive Fiber-Optic Interferometric SensorArrays”, SPIE 86 Conf., 86 Cambridge,MA, Sept. 1986. 7. A.D. Kersey, A. Da&ridge and A.B. Tveten,“Multiplexing of InterferometricFiber Sensors Using Time Division Addressingand PhaseGeneratedCarrierDemodulation”, Acceptedfor publication in Optics Letters, Oct. 1987. 8. A.D. Kersey and A. Dandridge, “Phase Noise Reduction in Coherence Multiplexed Interferometric Fibre Sensors”, Electron. Lett,, Vol. 22, No. 11, pp. 616-618, 1986. 9. A. Dandridge, A.B. Tveten and T.G. Giallorenzi, “Homodyne Demodulation Schemefor Fiber-Optic SensorUsing Phase GeneratedCarrier”, IEEE J. Quantum Electron., 18, p. 1647, 1982. 10. D.A. Jackson, A.D. Kersey, M. Corke and J.D.C. Jones, “Pseudo-heterodyneDetection Schemefor Optical Interfermeters”, Electron. Lett., 18, p. 1081, 1982. 11. A.D. Kersey, A.C. Lewin, and D.A. Jackson, “Pseudo-Heterodyne Detection for the Fibre Gyroscope”, Electron. Lett., Vol. 20, pp. 368 , 1984. t - SachslFreeman Associates, Lmdover, MD 20785.

input

sensor array

demodulation

interferometer

Figure 1. Schematic of the new interferometric sensor array architecture using time division multiplexing. 82

0a

0C

Figure 2 (a) Pulses received at the compensator output. Note: compensator driven by 27~rads at - 1 kHz to highlight the interferometric signal carried by each of the three central pulses. Horizontal scale : 100 nsec/div (b) Phase detection sensitivity: 20 mrad. signal at 1 kHz applied to sensor 2, output S/N - 66 dB/dHz. (c) Crosstalk between sensors 1 and 2.(see text) 83