Ultrahigh Resolution Optical Time -Domain Reflectometey. B. K. Garside

Transcription

1 Ultrahigh Reolution Optical Time-Domain Reflectometry Ultrahigh Reolution Optical Time -Domain Reflectometey B. K. Garide B. K. Garide Opto-Electronic Inc. Unit 9, 2538 Speer Road, Oakville, Ontario, Canada L6L 5K9 Opto- Electronic Inc. Unit 9, 2538 Speer Road, Oakville, Ontario, Canada L6L 5K9 ABSTRACT ABSTRACT There i an increaing need for an OTDR ytem with a centimeter cale patial reolution capability (ie the capability of reolving ignal from two target feature with eparation ~1 cm) in hort haul fiber network and enor ytem. In order to achieve centimeter cale patial reolution along a fiber, it i neceary to employ optical ource and detector with repone on a loop timecale. In thi paper, we decribe the reult of an invetigation into the limiting capability in term of patial reolution and return ignal trength which can be achieved uing both direct and photon counting method of detecting the return ignal from the fiber ytem under tet. Appropriate model are preented for the two detection procee and the prediction of the model are compared with experimental reult in elected cae; particular attention i given to the poibility of detecting Rayleigh backcattering with centimeter reolution. There i an increaing need for an OTDR ytem with a centimeter cale patial reolution capability (ie the capability of reolving ignal from two target feature with eparation.1 cm) in hort haul fiber network and enor ytem. In order to achieve centimeter cale patial reolution along a fiber, it i neceary to employ optical ource and detector with repone on a 100p timecale. In thi paper, we decribe the reult of an invetigation into the limiting capability in term of patial reolution and return ignal trength which can be achieved uing both direct and photon counting method of detecting the return ignal from the fiber ytem under tet. Appropriate model are preented for the two detection procee and the prediction of the model are compared with experimental reult in elected cae; particular attention i given to the poibility of detecting Rayleigh backcattering with centimeter reolution. 1. INTRODUCTION 1. INTRODUCTION There are an increaing number of application in hort haul optical fiber communication and multi-enor network for which a very high patial reolution (on cm cale or better) Optical Time-Domain Reflectometer (OTDR) i needed. Example here are fiber enor and control network currently being planned and prototyped for intallation in both military and commercial aircraft, helicopter, hip and ubmarine. Such enor network may be formed with either dicrete element or ditributed enor and can be ued to monitor vehicle operating parameter. A network of thi type can form the bai for fiber optic "mart tructure" projected for ue in pace vehicle. There are an increaing number of application in hort haul optical fiber communication and multi- enor network for which a very high patial reolution (on cm cale or better) Optical Time -Domain Reflectometer (OTDR) i needed. Example here are fiber enor and control network currently being planned and prototyped for intallation in both military and commercial aircraft, helicopter, hip and ubmarine. Such enor network may be formed with either dicrete element or ditributed enor and can be ued to monitor vehicle operating parameter. A network of thi type can form the bai for fiber optic "mart tructure" projected for ue in pace vehicle. Other example of area where high patial reolution of back reflected (or cattered) light i important are for reolving reflection from dicrete component in the meaurement of fiber train, multiple reflection from an expanded beam connector and in lo meaurement and fault location in fiber optical gyro and integrated optical device. Other example of area where high patial reolution of back reflected (or cattered) light i important are for reolving reflection from dicrete component in the meaurement of fiber train, multiple reflection from an expanded beam connector and in lo meaurement and fault location in fiber optical gyro and integrated optical device. 1 LASER 1 [TRANSMITTER! TTER < RECEIVER TIME DELAY OENERATOR} SAMPLER Figure 1(a) The Electrical Sytem LASER TRANSMITTER RECEIVER TEST FIRER RECEIVER j SAMPLER TIME DELAY GENERATOR Figure l(a) The Electrical Sytem 1 LASER U-. TRANSMITTER! * RECEIVER 13 TEST FIBER REFERENCE FIBER OPTI A PROCESSOR DISPLAY -,,-I I P I PRINTER 3. RS232 OPIO NEU a W AuNIPERSj` PISTAILS JUMPERS/ FO CONNECTORS * FO. CONNECTORS REFERENCE FIBER (OPTIONAL) NAL Figure l(b) The Optical Sytem Figure 1(b) The Optical Sytem ' COUPLER 'ú I=. TEMPORARY QUICK SPLICE 13 T-> TEMPORARY 1. QUICK SPLICEU! The baic OTDR technique i illutrated in Figure 1. In thi technique, a hort light pule i launched into the fiber under tet and the light reflected or cattered back toward the ource i monitored. Thi returning light i meaured a a function of time after the pule tart to propagate down the fiber. Each value of thi time delay (for a ingle fiber link) i uniquely related to a pecific location along the fiber. Thu, in principle, backcattering and back reflection can be determined in magnitude and located in ditance down the link. The preciion with which a given feature can be located, and the eparation between two target feature which can be reolved depend on the width of the light pule and the bandwidth -of the detection ytem. In order to achieve centimeter cale patial reolution along the fiber, it i neceary to employ optical ource and detector with ubnanoecond repone time. In thi paper, we decribe the reult of an invetigation into the limiting capability in term of patial reolution and return ignal trength which can be achieved uing both direct and photon counting method of detecting the return ignal from the fiber ytem under tet. The baic OTDR technique i illutrated in Figure 1. In thi technique, a hort light pule i launched into the fiber under tet and the light reflected or cattered back toward the ource i monitored. Thi returning light i meaured a a function of time after the pule tart to propagate down the fiber. Each value of thi time delay (for a ingle fiber link) i uniquely related to a pecific location along the fiber. Thu, in principle, backcattering and back reflection can be determined in magnitude and located in ditance down the link. The preciion with which a given feature can be located, and the eparation between two target feature which can be reolved depend on the width of the light pule and the bandwidth.of the detection ytem. In order to achieve centimeter cale patial reolution along the fiber, it i neceary to employ optical ource and detector with ubnanoecond repone time. In thi paper, we decribe the reult of an invetigation into the limiting capability in term of patial reolution and return ignal trength which can be achieved uing both direct and photon counting method of detecting the return ignal from the fiber ytem under tet. 670 / SPIE Vol. 954 Optical Teting and Metrology Il (1988) 670 / SPIE Vol. 954 Optical Teting and Metrology II (1988)

2 2. OTDR SENSITIVITY AND SPATIAL RESOLUTION The received returning ignal trength (PD(R)) from a given location along the fiber i given by The received returning ignal trength given by PD(R) = Po - (R + C) - PR (1) where Po i the input power launched into the fiber, R i the "reflection coefficient" at the location in quetion, C repreent the loe at the receiver due to coupler, plice etc and PR i the lo incurred due to two -way propagation along the fiber link (range lo). In what follow, we conider the magnitude (db) of the variou term on the right hand ide of the above equation and thereby determine repreentative value of PD(R). Subequently, expreion are obtained for the minimum detectable power for both direct and photon counting detection. In thi way, the limiting capability of the two approache in term of patial reolution and enitivity can be etablihed. The value of R to be ued in equation (1) depend on the overall nature of the returned ignal from the elected location. Any location along the fiber will produce a mall amount of backcattering of the light pule due to mall cale denity fluctuation in the optical fiber (Rayleigh cattering). In addition, there may be reflection at pecific location due to the preence, for example, of connector or reflective crack or break in the fiber. Thu, -R may vary over a wide range, from -14db for a well cleaved fiber end (or an unmated connector end) down to the value appropriate to the backcattered ignal from the fiber itelf. The value of R for backcattering in a fiber i dependent on the fiber parameter, and i functionally different for ingle and multimode fiber. In the cae of a ingle mode fiber the reflection coefficient i given by:1 3.8x10-2 X2 RSM =-10 log 10 1a n2 w2 1 astvg 1iI J where X i the optical wavelength, nl i the refractive index of the core, w i the optical mode width, a i the lo due to Rayleigh cattering, T i the length of the probe pule and Vg i the pule velocity in the fiber. Note that thi "reflection" coefficient depend on the length of the probe pule. A imilar expreion for RSM in the ingle mode cae ha alo been given by Brinkmeyer.2 A typical value of a a a function of wavelength i 1 db /km.(pm)4 ie 2db /km at 0.8 pm. Putting in typical parameter for optical fiber at thi wavelength, auming a pule length of 100 p, give a value for RSM in the region of -78db. For multimode fiber, Peronick3 give R a: RMM = -10 log10 (NAZ2 L 4n1 atvg] where NA i the numerical aperture of the fiber. For the ame 100 p long input pule, and for a numerical aperture of 0.2 at X =0.8 pm, we obtain: = -68 db A reaonable value for the receiver lo (C), including a contribution for a launching lo in the plitter, i 10db. For operation at 0.8 pm, a practical value for the input pule peak power i 100 mw (+ 20dbm). Thu, the ignal level we need to detect i: PD(R) = _ -(70 + PR)dbm; inglemode fiber -(60 + PR)dbm; multimode fiber In the next ub -ection, conideration i given to the minimum detectable power for direct and for photon counting detection. Minimum Detectable Power The minimum detectable power (Pmin) i determined by the requirement that the ignal -to -noie ratio at the receiver be 1. Of coure, thi i not ufficient for a practical OTDR, but it erve a ueful mean of evaluating different detection method. For direct detection, it can be hown(1) that Pmin i given by: Pmin = (NEP)a (B)2 (2) where the bandwidth of the detection ytem (B) i = ZT for a probe pule of length T (ie a matched filter i ued at the detector) and: 2. OTDR SENSITIVITY AND SPATIAL RESOLUTION (P (R) ) from a given location along the fiber i Pn (R) = - (R + C) - P (1) u where PQ i the input power launched into the fiber, R i the "reflection coefficient" at the location in quetion, C repreent the loe at the receiver due to coupler, plice etc and PR i the lo incurred due to two-way propagation along the fiber link (range lo) In what follow, we conider the magnitude (db) of the variou term on the right hand ide of the above equation and thereby determine repreentative value of P D (R). Subequently, expreion are obtained for the minimum detectable power for both direct and photon counting detection. In thi way, the limiting capability of the two approache in term of patial reolution and enitivity can be etablihed. The value of R to be ued in equation (1) depend on the overall nature of the returned ignal from the elected location. Any location along the fiber will produce a mall amount of backcattering of the light pule due to mall cale denity fluctuation in the optical fiber (Rayleigh cattering). In addition, there may be reflection at pecific location due to the preence, for example, of connector or reflective crack or break in the fiber. Thu, -R may vary over a wide range, from -14db for a well cleaved fiber end (or an unmated connector end) down to the value appropriate to the backcattered ignal from the fiber itelf. The value of R for backcattering in a fiber i dependent on the fiber parameter, and i functionally different for ingle and multimode fiber. In the cae of a ingle mode fiber the reflection coefficient i given by: where A i the optical wavelength, n- _ i the refractive index of the core, w i the optical mode width, a i the lo due to Rayleigh cattering, T i the length of the probe pule and Vg i the pule velocity in the fiber. Note that thi "reflection" coefficient depend on the length of the probe pule. A imilar expreion for RSM in the ingle mode cae ha alo been given by Brinkmeyer. 2 A typical value of a a a function of wavelength i 1 db/km.(ym) 4 ie 2db/km at 0.8 ym. Putting in typical parameter for optical fiber at thi wavelength, auming a pule length of 100 p, give a value for RSM in the region of -78db. For multimode fiber, Peronick give R a: where NA i the numerical aperture of the fiber. For the ame 100 p long input pule, and for a numerical aperture of 0.2 at A=0.8 ym, we obtain: -"-MM = ~ 68 db A reaonable value for the receiver lo (C), including a contribution for a launching lo in the plitter, i lodb. For operation at 0. 8 ym, a practical value for the input pule peak power i 100 mw (+ 20dbm). Thu, the ignal level we need to detect i: P D (R) = -(70 + P R )dbm; inglemode fiber -(60 + P~)dbm; t multimode fiber In the next ub-ection, conideration i given to the minimum detectable power for direct and for photon counting detection. Minimum Detectable Power The minimum detectable power (Pmin) i determined by the requirement that the ignal-to-noie ratio at the receiver be 1. Of coure, thi i not ufficient for a practical OTDR, but it erve a ueful mean of evaluating different detection method. For direct detection, it can be hown (!) that Pmi n i given by: Pmin = (NEP) a (B) (2) where the bandwidth of the detection ytem (B) i * i- for a probe pule of length T (ie a matched filter i ued at the detector) and: SPIE Vol. 954 Optical Teting and Metrology II (1988) / 671 SPIE Vol. 954 Optical Teting and Metrology 11 (1988) / 671

3 (NEP) a = + ( NEPPD -Q) /Q W/ (3) (NEP)a = (NEP)PD C(ieq ) 2 + (NEPPD.Q)23 #/Q W Hz (3) i eq i the amplifier noie equivalent input current jav/hz, JQ = <M>, <M> i the average gain and n the quantum efficiency of the avalanche photodiode and: ieg i the amplifier noie equivalent input current C A iz,iq = <M> ñ-ey, <M> i the average gain and 11 the quantum efficiency of the avalanche photodiode and: C2e(IDS + M >2 +X IDM)/Q (4) < (NEP) pd = C2e(I DS + <M>2+X I DM )lvq (4) where IDSDM) are t ie unmultiplied (multiplied) component of the photodiode dark current and Q i the reponivity in amp/watt. The amplifier noie equivalent input current i given by: where IDS(IDM) are the unmultiplied (multiplied) component of the photodiode dark current and Q i the reponivity in amp /watt. The amplifier noie equivalent input current i given by: eq = (4kT)1/2 RL where RL i the amplifier input load reitor. For a typical avalanche ilicon detector, with a gain of 60, a Q of 10, and a dark current of 0.5 ya, the firt and econd term in (NEP) are equal if the value of R i 20 kft. Conequently, the thermal noie i dominant for an R- value of 50ft, which i typically employed. where RL i the amplifier input load reitor. For a typical avalanche ilicon detector, with a gain of 60, a Q of 10, and a dark current of 0.5 pa, the firt and econd term in (NEP) are equal if the value of RL i 20 k52. Conequently, the thermal noie i dominant for an RL value of 500, which i typically employed. Thu, for room temperature, the minimum detectable power i: Thu, for room temperature, the minimum detectable power i: 18.2 x 10" 13 ()2 for N = x (2Ót)ß for N = 1 which, for a 100 p pule, take the value of 128 nw before averaging. Thi minimum detectable power will not be achieved if there i inufficient gain between the detector and the ampling ytem. Thi arie due to the noie introduced by the ampler. For a typical ytem, thi noie current i 100 ya and the photodiode output current at the minimum detectable power (Q=10) i 1.2 ya. Thu, a gain of 44f0 Z 80 mut be provided. which, for a 100 p pule, take the value of 128 nw before averaging. Thi minimum detectable power will not be achieved if there i inufficient gain between the detector and the ampling ytem. Thi arie due to the noie introduced by the ampler. For a typical ytem, thi noie current i 100 pa and the photodiode output current at the minimum detectable power (Q =10) i 1.2 pa. Thu, a gain of mut be provided. The minimum detectable power of 128nW correpond to -42dbm, which i to be compared with a maximum backcatter return ignal ( p R= 0) of -70(60)dbm for ingle (multi-) mode fiber uing a pulelength of 100 p. A 20db improvement in ignal-to-noie ratio can be obtained by ignal averaging (10 average) thereby reducing the minimum detectable ignal to -62dbm. Clearly, thi i not adequate for detecting backcatter ignal atifactorily for either ingle or multimode fiber. The minimum detectable power of 128nW correpond to -42dbm, which i to be compared with a maximum backcatter return ignal (PR =0) of -70(60)dbm for ingle(multi -) mode fiber uing a pulelength of 100 p. A 20db improvement in ignal -to -noie ratio can be obtained by ignal averaging (104 average) thereby reducing the minimum detectable ignal to -62dbm. Clearly, thi i not adequate for detecting backcatter ignal atifactorily for either ingle or multimode fiber. Both the minimum detectable power and the backcattered power are dependent on the length of the pule ued in the OTDR. Figure 2 how the variation of the maximum backcatter return ignal, (PoViax ' an< p f r a inglemode fiber at 0.8 ym, auming that 10 4 average are ued and that the in load reitor RL i 50 ft for all pulelength. The crohatched region delineate the regime in which the maximum backcattered ignal exceed the minimum detectable power. Figure 3 how a imilar plot for the cae of a multimode fiber at 0.8 ym. In both cae, the figure in parenthee given after the value of pule length indicate the change in range along the fiber correponding to each of the time. Figure 2 and 3 demontrate very clearly that backcatter ignal from either ingle or multimode fiber cannot be detected atifactorily uing light pule hort enough to give a two-point reolution of 1 cm or le. A a reult, an OTDR with centimeter cale reolution baed on a direct detection approach i limited to the meaurement of reflective feature along an optical fiber link or network. Some operating characteritic of uch a ytem are decribed in the ubequent ection of thi paper. Both the minimum detectable power and the backcattered power are dependent on the length of the pule ued in the OTDR. Figure 2 how the variation of the maximum backcatter return ignal, (PD)max, and P for a inglemode fiber at 0.8 pm, auming that 10 average are ued and that théln load reitor RL i 50 Q for all pulelength. The crohatched region delineate the regime in which the maximum backcattered ignal exceed the minimum detectable power. Figure 3 how a imilar plot for the cae of a multimode fiber at 0.8 pm. In both cae, the figure in parenthee given after the value of pule length indicate the change in range along the fiber correponding to each of the time. Figure 2 and 3 demontrate very clearly that backcatter ignal from either ingle or multimode fiber cannot be detected atifactorily uing light pule hort enough to give a two -point reolution of 1 cm or le. A a reult, an OTDR with centimeter cale reolution baed on a direct detection approach i limited to the meaurement of reflective feature along an optical fiber link or network. Some operating characteritic of uch a ytem are decribed in the ubequent ection of thi paper. The photon counting technique i one method whereby hot-noie limited receiver enitivity can be attained in principle. Baically, the OTDR ytem operate in the ame way a for direct detection except that the avalanche photodiode i operated in a photon counting mode by applying a bia voltage in exce of the breakdown voltage. In thi way, an incoming photon trigger an avalanche in the detector, and produce an output pule. The detector i then followed by a pule height dicriminator and a pule counter, and pule are accumulated in a multichannel analyer in which the different channel correpond to different range tep. For long optical pule, the minimum detectable power for the photon counting mode i baically the ame a for direct detection uing a high impedance amplifier. The advantage of thi technique occur in high reolution meaurement where, a we have een, thermal noie dominate for the direct detection approach. Photon counting, through the ue of a pule height dicriminator, eliminate the thermal noie component of the ignal and thereby permit a coniderable reduction of the minimum detectable power at high reolution. Of coure, high patial reolution till demand the ue of very hort input pule and detector with commenurate reolution in the photon counting mode. The minimum detectable optical power in the photon counting cae i given by (D : The photon counting technique i one method whereby hot -noie limited receiver enitivity can be attained in principle. Baically, the OTDR ytem operate in the ame way a for direct detection except that the avalanche photodiode i operated in a photon counting mode by applying a bia voltage in exce of the breakdown voltage. In thi way, an incoming photon trigger an avalanche in the detector, and produce an output pule. The detector i then followed by a pule height dicriminator and a pule counter, and pule are accumulated in a multichannel analyer in which the different channel correpond to different range tep. For long optical pule, the minimum detectable power for the photon counting mode i baically the ame a for direct detection uing a high impedance amplifier. The advantage of thi technique occur in high reolution meaurement where, a we have een, thermal noie dominate for the direct detection approach. Photon counting, through the ue of a pule height dicriminator, eliminate the thermal noie component of the ignal and thereby permit a coniderable reduction of the minimum detectable power at high reolution. Of coure, high patial reolution till demand the ue of very hort input pule and detector with commenurate reolution in the photon counting mode. The minimum detectable optical power in the photon counting cae i given by (1): hñ watt.. (ÑAt)11 watt. ( - N2NAt) ) where R i the dark count rate and N i the number of average (laer pule). Thu, from the detector we wih to maximie the quantum efficiency and minimie the dark count rate where R i the dark count rate and N i the number of average (laer pule). Thu, from the detector we wih to maximie the quantum efficiency and minimie the dark count rate 672 / SPIE Vol. 954 Optical Teting and Metrology II (1988) 672 / SPIE Vol. 954 Optical Teting and Metrology 11 (1988) _L Zo (5)

4 ao á 40 g a0 :) P.r KXtamO 102(10.) I0t0c.) PULSE LENGTH IN p 1040m) lorn) 100m.) KHbnm) K)*(l«ii) 102(1..) I0100am) PULSE LENGTH IN p pt Kftlm) I0tI.) Figure 2 The maximum backcatter ignal and Figure 3 The maximum backcatter ignal and minimum detectable power for inglemode fiber the minimum detectable power for a multimode a a function of ource pulelength. fiber a a function of ource pulelength. (one way i to cool the detector). The above expreion how, a expected, that the mini- mini mum detectable power reduce a (N)2. (N). However, for a given laer pule repetition rate, we can accumulate many more pule in the photon counting mode than for direct detection, which i limited by drift in the analog circuitry to total accumulation time of of 015 "* min. Furthermore, the NEP (Noie Equivalent Power) will be reduced for ubnanoecond repone ytem ince the detection ytem i no longer limited by thermal noie generated in the load reitor. In the next ection, we decribe reult obtained uing both direct and photon counting detection. 3. HIGH RESOLUTION OTDR MEASUREMENTS In what follow, we decribe the reult obtained uing a direct detection ytem to meaure reflective feature in a fiber network with ubcentimeter reolution and the meaurement of fiber backcattering with a patial reolution of a few centimeter. Direct Detection The baic electrical and optical configuration of the OTDR ytem were illutrated in Figure 1. The Proceor (TDR10) control the ampling time relative to a delayed trigger pule upplied by the delay generator. Thi trigger delay i neceary to offet the delay of the optical pule in traniting the fiber. For thi purpoe, Opto- Electronic ha developed a new time-delay plug-in (PDG20) for the modular Picoecond Fiber Optic Sytem (PFOS) with an exceptionally low trigger jitter. The proceor enable a electable number of weep through a return feature of interet to be accumulated and ignal averaged. In thi way, the combination of the TDR10 and an appropriate detector module allow detection of return ignal over a wide dynamic range. The patial preciion which can be achieved uing thi technique depend on the length of the detected return pule and the ignal-to-noie -noie ratio (SNR) of thi ignal. Figure 4 how the variation in the accuracy with which a return pule can be located in time or pace a a function of the SNR for a number of received pule width (FWHM). Clearly, for detected pule with width in the region of 100 p, the location of a given reflection can eaily be fixed to a pre- preciion well below 1 millimeter. Thi capability can be ued for the very accurate location of reflective feature, uch a break, along the fiber a well a to determine fiber train, mode velocity etc. to unprecedented accuracy. Another important capability of the ytem i the patial reolution (which repreent the eparation between two target reflective feature which can be reolved) which can be achieved. The upper curve in Figure 5 how the return ignal from two feature, uch a might be obtained from a connector and an adjacent break in the fiber. The reflection from the two feature, eparated by 1.44 cm, are clearly reolved. The lower trace how the eparate reflection from the two feature, which i calculated by the proceor unit. Appropriate choie of the laer ource and detector unit allow a patial reolution of everal millimeter to be achieved. SPIE /E Vol. 954 Optical Teting and Metrology 11 II (1988) / 673

5 i ] t y»- IOJ r; B r11inininr11rm1ir p 01M uminmmiiomoru OVUMENS MMIMIENOaEMMaa MIME0ß momo11 1I11K pm111 0it0iiC000 ioccc v x t, E,,1, 111 N! ÌC I,111 ' _ ram.ow rrm = =31wvecrr11M Nur nmcuou 111M= M!! 11 r i N muiími:ri,mi,-imiii: 1 f' ` C! 11 '11"1ii ML"iéii: i.gl.ì ig,ili..1 -o11111vo` V< rv S S, y SS. S X- y !'ÌNO._.» A > r V VV v ' " >. I -H - t". S,! l_ 4;S *-. V X, f *"* K> P l h **« If I ITS «0 4 o MR 10 Ul OTv F? * * * HO TIME,EC NODE, PULSE DIFFERENCE Figure 4 Time and ditance reolution a a function of SNR for variou pulelength. Figure 4 Time and ditance reolution a a function of SNR for variou pulelength. -LASER TRANSMITTER Figure 5 Two point meaurement in a fiber link howing two reflective feature eparated by 1.44 cm. The upper figure i the meaured ignal, the lower i proceed to give the eparate reflection. Figure 5 Two point meaurement in a fiber link howing two reflective feature eparated by 1.44 cm. The upper figure i the meaured ignal, the lower i proceed to give the eparate reflection.. Photon Counting Detection Figure 6 how the baic element of a photon counting OTDR ytem. Short pule from a laer ource are launched into the fiber under tet and the backcattered ignal in a elected time window are accumulated. The detector i an avalanche detector operated in the "geiger" mode in conjunction with a pule height dicriminator. The time-delay generator elect the tart of the time-interval to be invetigated and initiate the operation of the detector for the elected interval. If a photon count i generated during thi interval, it i accumulated in the appropriate range bin of the proceor. The time delay i tepped equentially over a preelected range and count are accumulated until an Figure 6 Schematic diagram of the Photon adequate ignal-to-noie ratio i attained. Counting OTDR Sytem. An example of what can be achieved i illutrated in Figure 7. The three peak correpond to very mall reflection at location A, B and C a indicated on the Figure. The patial eparation between A and B i 24 cm and between B and C i 30 cm. The flat region to the left of each location correpond to the backcattered ignal from the fiber, which i a tandard 50/125 ym multimode type. The reflection at C correpond to the end of the fiber. At location A and B, the fiber wa looped everal time around a 3 mm diameter rod, which i ufficient to introduce a ignificant lo a can be een from the change in the backcattered ignal before and after both A and B. In fact, ince the zero ignal level i etablihed by the region to the right of location C, it i a traightforward matter to determine the loe induced at A and B which are 0.7db and 2.3db repectively for the one way lo. It i clear that the eparate reflection are wellreolved; the reolution limit i determined principally (the laer pulelength i loop) by the time the avalanche diode i puled on. The reolution achieved in thi cae, a can be een from the width of the pule reflected from the fiber end, i approximately 3 cm. The ue of a technique frequently ued in photon counting pectrocopy(4) f in which a photon count i ued to provide a "top" pule for a time-to-pule-height converter initiated by the detector bia, hould permit the patial reolution to be reduced to below 1 cm. Photon Counting Detection Figure 6 how the TIME DELAY TO baic element of a photon counting OTDR 6ENERAIOR 2-TEST ytem. Short pule from a laer ource are RAM launched into the fiber under tet and the Bu CONTROL PRAM C backcattered ignal in a elected time counmit co(pler window are accumulated. The detector i an avalanche detector operated in the "geiger" PROCE4oa mode in conjunction with a pule height 1 dicriminator. The time -delay generator I PULSE MOAT elect the tart of the time - interval to be DISCRIMINARA invetigated and initiate the operation of the detector for the elected interval. If a photon count i generated during thi interval, it i accumulated in the appropriate range bin of the proceor. The time delay i tepped equentially over a preelected range and count are accumulated until an Figure 6 Schematic diagram of the Photon adequate ignal -to -noie ratio i attained. Counting OTDR Sytem. An example of what can be achieved i illutrated in Figure 7. The three peak correpond to very mall reflection at location A, B and C a indicated on the Figure. The patial eparation between A and B i 24 cm and between B and C i 30 cm. The flat region to the left of each location correpond to the backcattered ignal from the fiber, which i a tandard 50/125 pm multimode type. The reflection at C correpond to the end of the fiber. At location A and B, the fiber wa looped everal time around a 3 mm diameter rod, which i ufficient to introduce a ignificant lo a can be een from the change in the backcattered ignal before and after both A and B. In fact, ince the zero ignal level i etablihed by the region to the right of location C, it i a traightforward matter to determine the loe induced at A and B which are 0.7db and 2.3db repectively for the one way lo. It i clear that the eparate reflection are well - reolved; the reolution limit i determined principally (the laer pulelength i loop) by the time the avalanche diode i puled on. The reolution achieved in thi cae, a can be een from the width of the pule reflected from the fiber end, i approximately 3 cm. The ue of a technique frequently ued in photon counting pectrocopy(4), in which a photon count i ued to provide a "top" pule for a time -to- pule- height converter initiated by the detector bia, hould permit the patial reolution to be reduced to below 1 cm. 674 / SPIE Vol. 954 Optical Teting and Metrology II (1988) 674 / SPIE Vol. 954 Optical Teting and Metrology II (1988)

6 ABC I J I Figure 7 A high reolution OTDR trace from a multimode fiber. The ditance between the loy feature at A and C i 54 cm; the limiting two-point reolution i approximately 3 cm. Figure 7 A high reolution OTDR trace from a multimode fiber. The ditance between the loy feature at A and C i 54 cm; the limiting two -point reolution i approximately 3 cm. 4. CONCLUSION 4. CONCLUSION Two approache to the development of centimeter cale reolution in an OTDR ytem have been invetigated. It ha been hown that direct detection i appropriate for the meaurement of reflective feature in a fiber link or network, but ha inufficient enitivity to meaure Rayleigh backcattering from the optical fiber. A photon counting detection technique uing an avalanche photodiode operating in the "geiger" mode ha alo been explored. It i hown that thi technique i not ubject to the thermal noie limitation, of the direct detection cheme, and i ufficiently enitive to direct Rayleigh backcatter with centimeter cale reolution. OTDR ytem baed on both the direct and the photon counting detection technique are decribed. It i hown that reflective feature in a fiber link or network can be reolved with a patial reolution of 1 cm or le uing the direct detection approach. A prototype OTDR ytem, baed on the photon counting approach i alo decribed which i ufficiently enitive to meaure the fiber backcattering. It i demontrated that a patial reolution of approximately 3 cm can be achieved with thi ytem. It i anticipated that the patial reolution can be reduced to 1 cm uing a more precie timing technique for the arrival of a photon counting pule. Two approache to the development of centimeter cale reolution in an OTDR ytem have been invetigated. It ha been hown that direct detection i appropriate for the meaurement of reflective feature in a fiber link or network, but ha inufficient enitivity to meaure Rayleigh backcattering from the optical fiber. A photon counting detection technique uing an avalanche photodiode operating in the "geiger" mode ha alo been explored. It i hown that thi technique i not ubject to the thermal noie limitation, of the direct detection cheme, and i ufficiently enitive to direct Rayleigh backcatter with centimeter cale reolution. OTDR ytem baed on both the direct and the photon counting detection technique are decribed. It i hown that reflective feature in a fiber link or network can be reolved with a patial reolution of 1 cm or le uing the direct detection approach. A prototype OTDR ytem, baed on the photon counting approach i alo decribed which i ufficiently enitive to meaure the fiber backcattering. It i demontrated that a patial reolution of approximately 3 cm can be achieved with thi ytem. It i anticipated that the patial reolution can be reduced to 1 cm uing a more precie timing technique for the arrival of a photon counting pule. 5. REFERENCES 5. REFERENCES 1. P. Healey, Optical and Quantum Electronic 16, 267 (1984). 2. E. Brinkmeyer, J. Opt. Soc. America 70, 1010 (1977). 3. S.D. Peronick, Bell Syt. Tech. Journal 56, 355 (1977). 4. See for example: S. Cova, M. Bertolaccini and C. Buolati, Phy. Stat. Sol. 18, 11 (1973). S. Cova, A. Longoni, A. Andreoni and R. Cubeddu, IEEE J. Quantum Electronic, QE-19, 630 (1983). C.G. Bethea, B.F. Levine, L. Marchut, V.D. Mattera and L.J. Peticola, Electronic Letter, 22, 302 (1986). 1. P. Healey, Optical and Quantum Electronic 16, 267 (1984). 2. E. Brinkmeyer, J. Opt. Soc. America 70, 1010 (1977). 3. S.D. Peronick, Bell Syt. Tech. Journal 56, 355 (1977). 4. See for example: S. Cova, M. Bertolaccini and C. Buolati, Phy. Stat. Sol. 18, 11 (1973). S. Cova, A. Longoni, A. Andreoni and R. Cubeddu, IEEE J. Quantum Electronic, QE -19, 630 (1983). C.G. Bethea, B.F. Levine, L. Marchut, V.D. Mattera and L.J. Peticola, Electronic Letter, 22, 302 (1986). SPIE Vol. 954 Optical Teting and Metrology 11 (1988) / 675 SPIE Vol. 954 Optical Teting and Metrology II (1988) / 675