`Buerli
`
`[11] Patent Number:
`[45] Date of Patent:
`
`5,066,118
`Nov. 19, 1991
`
`[75]
`
`[54] OPTICAL FAULT FINDER USING
`MATCHED AND CLIPPING FILTERS
`Inventor: Richard Buerli, Thousand Oaks,
`Calif.
`[73] Assignee: Minnesota Mining and
`Manufacturing Company, St. Paul,
`Minn.
`[21] Appl. No.: 508,834
`[22] Filed:
`Apr. 12, 1990
`Int. CI.s ............................................. GOIN 21/88
`[51]
`[52] U.S. CI •............................... 356173.1; 250/227.15
`[58] Field of Search ................... 356173.1; 250/227.15
`
`[56]
`
`References Cited
`U.S. PATENT DOCUMENTS
`3,981,592 9/1976 Williams .............................. 356/237
`4,070,118 1/1978 Maslowski et al. ................. 356/237
`4,197,007 4/1980 Costa et al. ........................ 356/73.1
`4,212,537 7/1980 Golob et aI. ....................... 356/73.1
`4,289,398 9/1981 Robichaud ......................... 356/73.1
`4,397,551 3/1983 Bage et al. ......................... 356/73.1
`4,497,575 2/1985 Philipp ............................... 356/73.1
`4,674,872 6/1987 Wright ............................... 356173.1
`4,685,799 8/1987 Brininstoo1 ......................... 356/73.1
`4,708,471 11/1987 Beckmann et al. ................ 336/73.1
`4,732,469 3/1988 Souma ................................ 356/73.1
`4,743,753 5/1988 Cheng et al. ........................ 250/227
`4,838,690 6/1989 Buckland et al. .................. 356/73.1
`4,870,269 9/1989 Jeunhomme et al. ............... 250/227
`4,963,020 10/1990 Luthra et al. ...................... 356/73.1
`
`FOREIGN PATENT DOCUMENTS
`2456293 6/1976 Fed. Rep. of Germany.
`
`1560124 1/1980 United Kingdom.
`2182222 5/1987 United Kingdom.
`
`OTHER PUBLICATIONS
`Three Brochures Depicting Commercially Available
`OTDR's-Model 5400XQ from 3M Photodyne (Copy(cid:173)
`right Notice Dated 1989-Model MW90lOA from An(cid:173)
`ritsu (Printing Date 1989)-Model TFS2020 from Tek(cid:173)
`tronix (Printing Date 10/89).
`Article Entitled "Optical Time-Domain Reflectometer
`Specifications and Performance Testing" by B. Daniel(cid:173)
`son (From Applied Optics, vol. 24, pp. 2313-2321, Aug.
`1, 1985).
`Primary Examiner-Vincent P. McGraw
`Attorney, Agent, or Firm-Gary L. Griswold; Walter N.
`Kim; Jack V. Musgrove
`[57]
`ABSTRACT
`An optical fault finder employing a novel processing
`technique to achieve greater sensitivity to loss detection
`and iocation of faults. The technique includes the use of
`a matched filter which sequentially operates on a set of
`datapoints in the trace signal. The resulting matched
`filter function generates peaks at those locations corre(cid:173)
`sponding to discrete losses in the trace signal. In order
`to optimize response of the matched filter with respect
`to reflective faults, a clipping filter is applied to the
`trace signal, prior to the matched filter, to remove re(cid:173)
`flective signals. Means are also provided for determin(cid:173)
`ing the value of the loss, and for optimizing the pulse
`width of the test signal launched into the fiber under
`test.
`
`19 Claims, 5 Drawing Sheets
`
`68
`
`----4-.:..... _____ 1
`B
`
`Page 1
`
`
`
`u.s. Patent
`
`Nov. 19, 1991
`
`Sheet 1 of 5
`
`5,066,118
`
`10"
`
`50
`
`VARIABLE PULSE
`WIDTH LASER
`
`POWER
`SUPPLY
`
`.--_-'--_ 14
`
`HIGH SPEED
`MEMORY
`
`Page 2
`
`
`
`u.s. Patent
`
`Nov. 19, 1991
`
`Sheet 2 of 5
`
`5,066,118
`
`68
`
`1ij
`c:
`C)
`Ci5
`'0
`Q)
`0-
`.9-
`(3
`
`c
`
`B
`
`Page 3
`
`
`
`u.s. Patent
`
`Nov. 19, 1991
`
`Sheet 3 of 5
`
`5,066,118
`
`0
`
`74
`
`76
`78
`80
`
`94--....
`Event fL o s s
`
`~ 8a8J3J3a8 km
`8 Event ! B.8 dB
`
`m
`
`ft
`mls
`
`92
`98~ Enter
`OFF
`SCAN
`100
`
`102
`
`96
`Option
`NEXT ~ 16
`
`Page 4
`
`
`
`u.s. Patent
`
`Nov. 19, 1991
`
`Sheet 4 of 5
`
`5,066,118
`
`104
`
`106
`
`110
`
`ILLUMINATE ·UNIT,·
`·THRESW AND ·,NDEX·
`ANNUNCIATORS
`
`108
`
`118
`
`BLINK ·UNIT" AND
`CURRENT UNITS
`
`"-----1 STORE CURRENT
`UNITS
`
`DEPRESS·NEXT"
`UNTIL DESIRED UNIT
`IS BLINKING
`
`122
`
`116
`
`BLINK ·'NDEX- AND
`CURRENT INDEX J4 - - t
`
`126
`
`DEPRESS-NEXT"
`UNTIL DESIRED INDEX
`APPEARS
`
`134
`
`128
`142
`BLINK -THRESH- AND
`CURRENT THRESHOLD~-I
`
`146
`
`END
`
`Page 5
`
`
`
`u.s. Patent
`
`Nov. 19, 1991
`
`Sheet 5 of 5
`
`5,066,118
`
`CALCULATE OPTIMAL
`PULSE WIDTH
`-------
`LAUNCH PULSE
`
`178 r--=:-:~~ ___ -_
`CLIPPING FILTER
`-------
`CALCULATE AND
`STORE DISTANCE
`TO REFLECTIONS
`
`180
`
`-------
`MATCHED FILTER
`CALCULATE AND
`TORE DISTANCE AND
`LOSS AT EVENT
`
`ILLUMINATE "TESr
`------
`DIAGNOSTICS AND
`CALIBRATION
`
`154
`
`158
`
`~~ DISPLAY
`-ERROR-
`
`160
`
`162
`
`LAUNCH
`STANDARD PULSE
`------
`RECORD
`ORIGINAL TRACE
`
`CALCULATE AND
`STORE LENGTH
`OF FIBER·
`
`166
`
`170
`
`YES
`
`ILLUMINATE
`-OUT OF RANGE-
`
`CALCULATE AND
`STORE NUMBER OF t.----J
`EVENTS AND
`172
`LOCATIONS
`
`Page 6
`
`
`
`1
`
`5,066,118
`
`OPTICAL FAULT FIrmER USING MATCHED AND
`CLIPPING FILTERS
`
`BACKGROUND OF THE INVENTION
`1. Field of the Invention
`The present invention generally relates to devices for
`testing the transmission quality of optical fibers, and
`more particularly to an optical time domain reflectome-
`ter having improved fault detection and location.
`2. Descriotion of the Prior Art
`In recent years, fiber optic cables have replaced tradi(cid:173)
`tional copper wire as the preferred medium for telecom(cid:173)
`munications Although optical fibers have certain ad(cid:173)
`vantages over copper wire, they are still subject to 15
`faults which may result during installation of the fibers
`or from environmental factors after installation. Also,
`the practical length of an optical fiber is limited by
`attenuation of the light signals travelling therein, since
`there can never be 100% transmission of light through 20
`these fibers.
`Accordingly, it is necessary to occasionally test the
`transmission quality of optical fibers. One device which
`has established itself as one of the more versatile instru(cid:173)
`ments for this purpose is the optical time domain reflec- 2S
`tometer, commonly referred to as an "OTDR." In its
`simplest construction, an OTDR includes a light source,
`such as a pulsed laser diode; an optical coupler, includ(cid:173)
`ing a beam-splitter, connecting the light source to the
`near end of the fiber under test (FUT); and a photode- 30
`tector positioned adjacent the beam splitter. When a test
`signal is sent down the FUT, back scattering and reflec(cid:173)
`tions within the fiber core return to the near end of the
`FUT and are sensed by the photodetector. The trace
`signal of the backscattering and reflections provides 3S
`clues as to faults in the FUT. Numerous of which are
`disclosed in the following patents and applications:
`
`2
`the fiber. The strength of the reflected signal is primar(cid:173)
`ily dependent upon the peak power of the test pulse.
`Reflective signals may be used to determine the overall
`length of the fiber line, and to detect breaks in the fiber,
`5 reflective connectors, and splices of fibers having differ(cid:173)
`ent indices of refraction. Reflective signals also cause
`"deadzones," as explained more fully below.
`Although the trace signal is a function of time (i.e.,
`the amount of time passing from the initial test pulse
`until the return signal is detected), it can be directly
`correlated to positions along the FUT by the equation
`x =ct/2n, where x is the distance along the fiber, c is the
`speed of light in a vacuum, t is the elapsed time, and n
`is the index of refraction of the fiber material. Thus, the
`approximate location of a fault or splice may be deter(cid:173)
`mined.
`A difficulty arises in locating faults, however, due to
`the deadzone created by Fresnel reflections. If two
`faults are in close proximity, their reflections andlor
`losses will overlap and may appear in the trace signal as
`a single fault. The theoretical length 1 of the dead zone
`is 1=ctp",/2n, where tp'" is the duration of the pulse
`width. For example, an OTDR emitting a 500 nanosec(cid:173)
`ond pulse into an optical fiber having an index of refrac(cid:173)
`tion of 1.5 will result in a dead zone of about 50 meters,
`which is quite significant. Of course, other factors can
`exacerbate this effect, such as the response time of the
`photodetector, and the strength ofany reflected signals.
`In order to minimize the dead zone and thereby in-
`crease the effective resolution, a small pulse width may
`be selected. Prior art OTDR's provide for manual selec(cid:173)
`tion of pulse width from a set of a few discrete values.
`Some OTDR's provide a pulse width as small as one
`nanosecond. In minimizing the deadzone, however,
`other performance parameters of the OTDR are ad(cid:173)
`versely affected. As noted above, micro-bends and
`splice losses are detected by means of Rayleigh scatter(cid:173)
`ing which is dependent on the pulse width. Hence, if
`__ P_._te_n_t/....:ap""p_li_ca_ti_on __ ___ A_p;.;p_li_ca_n_t __ ____ 40 relatively small pulse widths are employed, low loss
`microbends and splices may ·go undetected, although
`u.s. Pat. No. 3,981,592
`D. Williams
`u.s. Pat. No. 4,070,118
`they would be distinguishable if the launch signal were
`Maslowski et aI.
`longer. Attenuation in the fiber may make it difficult to
`u.s. Pat. No. 4,197,007
`CoGOsltoabeettaall·.
`u.s. Pat. No. 4,212,537
`d
`d'
`.,
`I
`h
`d'
`I
`1
`f
`etect
`Istant ,au ts, urt er man atmg a onger pu se
`u.s. Pat. No. 4,289,398
`R. Robichaud
`u.s. Pat. No. 4,397,551
`4S width. More broadly stated, a single trace may provide
`Bage et al.
`u.s. Pat. No. 4,497,575
`an optimal pulse width for one section of the fiber path,
`H. Philipp
`~:~: ~::: ~~: ::~~~:~~~
`~.~~~~~stOOI
`but the pulse width will not be optimal for the majority
`u.S. Pat. No. 4,708,471
`of the path. This presents a clear dilemma which prior
`Beckmann et aI.
`u.s. Pat. No. 4,732,469
`art OTDR's have not adequately addressed.
`M. Souma
`u.s. Pat. No. 4,743,753
`The above problem relates only to the resolution of
`Cheng et a!.
`r. 1
`U.s. Pat. No. 4,838,690
`h OTDR fi
`f d
`.
`h
`h
`Buckland et aI.
`t e
`or purposes 0
`etectmg t e au t. Anot er
`u.s. Pat. No. 4,870,269
`leunhomme et aI.
`problem occurs with respect to the precision of the
`Brit. Pat. No. 1,560,124
`'Standard Tel. & Cables
`OTDR in determining the location of any given fault
`Brit. Pat. Appn. 2,182,222
`STC pIc.
`along the fiber path. Early OTDR's merely provided a
`graphic display of the return trace signal from which
`only the crudest estimates could be made. Instruments
`have since been devised which can automatically detect
`and toggle through the approximate locations of losses,
`but they still require heavy user interpretation with
`respect to the specific location of any given fault.
`For example, some prior art OTDR's employ digital
`sampling and analysis of the trace signal, and use a
`moving least-squares fit of several datapoints to calcu-
`late an average slope function. Logic circuitry examines
`this function for deviations which are greater than a
`preset threshold value, and records the elapsed time
`(i.e., the distance along the fiber) to the datapoint corre(cid:173)
`sponding to the change in slope. The calculated dis-
`
`10
`
`50
`
`The backscattered signal (also known as Rayleigh SS
`scattering) is typically weak, and is due to refractive(cid:173)
`index fluctuations and inhomogeneities in the fiber core.
`The strength of the backscattered signal is primarily
`dependent on the peak power and width of the test
`pulse, i.e., a longer pulse width results in stronger back- 60
`scattering. The backscattered signal may be used to
`detect faults such as micro-bends or splice losses, and to
`measure overall attenuation. In fact, attenuation is pri(cid:173)
`marily due to back scattering, although it is also a func(cid:173)
`tion ofthe wavelength of the test pulse and any discrete 65
`losses along the fiber path.
`Reflective signals (also known as Fresnel reflections)
`are somewhat stronger, and are due to discontinuities in
`
`Page 7
`
`
`
`5,066,118
`
`3
`tance, however, is usually not the actual distance to the
`fault. In order to more accurately define the specific
`point at which the fault occurs, human interaction is
`necessary. These prior art OTDR's allow the user to
`graphically estimate the fault location by moving a 5
`cursor on the display to the point along the trace signal
`corresponding to the beginning of the fault. This is, of
`course, a very subjective step and requires experience
`and training for an reliable measurement. It is clear that 10
`a simpler and more accurate technique for fault location
`is long overdue.
`It would, therefore, be desirable and advantageous to
`devise an optical time domain reflectometer providing
`optimization of pulse width, and improved resolution in 15
`fault detection and location. It should also be capable of
`multi-fault operation, and should calculate the loss
`value at the fault. Finally, minimal operator training and
`interaction should be required.
`
`4
`DESCRIPTION OF THE PREFERRED
`EMBODIMENT
`With reference now to the figures, and in particular
`with reference to FIG. 1, there is depicted an optical
`time domain reflectometer (OTDR) 10 of the present
`invention. OTDR 10 is generally comprised of a hous(cid:173)
`ing 12, a display 14, a keyboard 16, a port 18 for receiv(cid:173)
`ing the optical. fiber to be tested, and associated elec-
`tronics mounted on internal circuit boards 20. Housing
`12 is constructed of any durable material such as poly-
`carbonite, and includes a lid 22 having a latch 24. Hous(cid:173)
`ing 12 forms a watertight container when lid 22 is
`closed. The inner surface oflid 22 may include a printed
`instruction summary 26, or have attached thereto one or
`more clips 28 for retaining small accessories such as an
`optical splice connector 30. Housing 12 may further
`include a recessed cavity 32 for storing other accesso(cid:173)
`ries, and be provided with a carrying strap 34. While
`20 OTDR 10 may be adapted for connection to an external
`power source, the disclosed embodiment includes a
`portable power supply, i.e., battery 36.
`Referring now to FIG. 2, a block diagram of the
`electronics of OTDR 10 is explained. The electronic
`system includes a microprocessor 40 connected to a
`programmable read-only memory unit (PROM) 42, a
`random-access memory unit (RAM) 44, and a timer 46.
`Microprocessor 40 is also connected to the display 14
`and keyboard 16, and to power supply 36 via a voltage
`control 48. In the disclosed embodiment, microproces(cid:173)
`sor 40 is an integrated circuit commonly known as a
`68000 processor, available from Motorola Corp. of Aus(cid:173)
`tin, Texas, Hitachi Inc. or Toshiba Corp of Japan.
`Power supply 36 comprises a set of six 1.2 volt batteries
`(nickel-cadmium rechargeable "D" cells). Voltage con(cid:173)
`trol 48 is a DC/DC converter, and provides a five volt
`output to microprocessor 40. As those skilled in the art
`wiIl appreciate, however, use of the specific compo(cid:173)
`nents described herein is not meant to be limiting of the
`present invention; rather, these components are merely
`deemed preferable in the use of OTDR 10.
`Microprocessor 40 is used to control timer 46 which
`in turn regulates a variable pulse width light source 50.
`In the preferred embodiment, timer 46 utilizes a 20
`45 megahertz clock, and light source 50 is a laser diode
`having an output wavelength of 1300 nanometers at a
`peak power of 10 milliwatts. Such a laser diode is avail(cid:173)
`able from STC pic. of England under model number
`LP3SA 10-18. Other wavelengths besides 1300 nm are
`acceptable, and OTDR 10 may optionally be provided
`with multiple light sources of different wavelengths to
`target the test results for specific operating conditions.
`Inasmuch as 1300 nm is outside the visible spectrum (as
`are most wavelengths used in optical fiber technology),
`it should be understood that the term "light" as used
`herein means a source of electrom;agnetic radiation of
`any wavelength which may be transmitted through a
`waveguide. Voltage control 48 also supplies 12 volt
`power to light source 50.
`Light source 50 is connected to the fiber under test
`(FUT) by means of a 3-way optical coupler 52 and port
`18. Actually, these two components are preferably com(cid:173)
`bined into one integral coupler/port. Combined cou(cid:173)
`pler/ports are available from Amphenol Corp. of Lisle,
`111., under model number 945J, and from Gould Elec(cid:173)
`tronics of Glen Burnie, Md. The laser diode comprising
`light source 50 is provided with a "pigtail" (a short
`section of optical fiber), and connected to the input of
`
`60
`
`SUMMARY OF THE INVENTION
`The foregoing objectives are achieved in an optical
`time domain reflectometer having a novel fault location
`method including use of a digital matched filter for 25
`detecting any discrete losses in the trace signal. A "clip(cid:173)
`ping" filter is also used to remove reflections from the
`trace signal. This technique improves accuracy in rec(cid:173)
`ognition of the fault, determination of the distance to
`the fault, and may further optionally be used to calcu- 30
`late the amount of any loss. After the device has re(cid:173)
`corded and calculated this information, a display con(cid:173)
`veys the essential data to the user in a sequential manner
`for each fault.
`The OTDR also employs means for optimizing the 35
`pulse width based on the backscattered signal level at
`each fault. An initial signal is launched revealing one or
`more faults (reflections or losses). These faults are re(cid:173)
`corded, and then a series of signals are sent, one for each 40
`fault. The pulse width of each such signal is optimized
`for the particular fault being analyzed. For added flexi(cid:173)
`bility, the device may be programmed by the user with
`such information as the index of refraction of the fiber,
`threshold loss levels, and output units.
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`The novel features and scope of the invention are set
`forth in the appended claims. The invention itself, how(cid:173)
`ever, will best be understood by reference to the accom- 50
`panying drawings, wherein:
`FIG. 1 is a perspective view of an optical fault finder
`construced in accordance with the present invention,
`with cutouts to illustrate interior elements;
`FIG. 2 is a block diagram of the electrical and optical 55
`subsystems of the present invention;
`FIG. 3 is a graph showing an original trace signal,
`and corresponding signals generated by the clipping
`and matched filters;
`FIG. 4 is a top plan view of the display used in one
`embodiment of the present invention;
`FIG. 5 is a flow chart depicting the programming
`steps for preparation of the OTDR; and
`FIG. 6 is a flow chart illustrating the steps performed 65
`by the OTDR in detection and location of faults along
`the fiber path, including pulse width optimization and
`use of a matched filter and clipping filter.
`
`Page 8
`
`
`
`5,066,118
`
`5
`coupler 52 by means of a FIBERLOK brand splice
`connector (FIBERLOK is a trademark of Minnesota
`Mining & Manufacturing Company, assignee of the
`present invention). The primary output of coupler 52 is
`connected to port 18, while the secondary output is s
`connected to a receiver 54.
`In the preferred embodiment, receiver 54 is an ava(cid:173)
`lanche photodiode manufactured by Fujitsu Corp. of
`Japan under model number FPDI3U512UX. It is, of
`course, imperative that the fiber from coupler 52 be 10
`properly aligned with the sensor of receiver 54 in order
`to maximize receiver sensitivity. The output of receiver
`54 is directed to an amplifier/filter 56 which provides
`signal conditioning. Conventional amplifiers and filters
`may be used to increase the gain of the return signal and 15
`filter out any unwanted signals. The inventor uses a
`three-stage amplifier. The first stage consists of a trans(cid:173)
`impedance amplifier used to capture the signal from the
`avalanche photodiode with maximum performance
`characteristics. This stage also provides clamping 20
`which is used to cut off excessively high signals (e.g.,
`reflective signals greater than 2 volts). The second and
`third stages consist of operational amplifiers and merely
`raise the gain of the return signal. The filter, which is
`built in to the amplifier, is simply a low-pass filter which 2S
`cancels high-frequency noise (e.g., greater than 16
`megahertz). Voltage control 48 supplies ISO volt power
`to receiver 54, and further provides ± 12 volts to ampli(cid:173)
`fier 56.
`The output of amplifier/filter 56 (in the range of zero 30
`to two volts) is sampled by an analog-to-digital con(cid:173)
`verter 5s (whose output is a digital value between zero
`and 255). The sampling rate may vary as a function of
`the desired resolution, the amount of available memory,
`and the switching speed of the electronics. It is antici- 35
`pated that a sampling rate of 100 nanoseconds will be
`sufficient for most applications; this is hardwired into
`timer 46 which controls operation of AID converter 58.
`Unfortunately, such a small increment makes it difficult
`for microprocessor 40 to adequately handle the data due 40
`to switching limitations. Therefore, a high-speed mem(cid:173)
`ory unit 60 (static RAM) is integrally provided with
`AID converter 58, having about 4096 bytes of memory
`(16 bit). An acceptable high-speed chip may be obtained
`from Performance, Inc. of Sunnyvale, Calif., under 45
`model number P4C1682. Assuming a sampling rate of
`100 ns and an index of refraction of 1.5 for the FUT, this
`amount of memory would be sufficient to store data for
`a fiber path of up to fifty kilometers. More memory may
`be provided if a shorter sampling rate, or a longer fiber 50
`path, were necessary. Of course, if a microprocessor
`having a faster switching speed were used, then RAM
`44 could store this information, and a separate high(cid:173)
`speed memory would be unnecessary.
`Microprocessor 40 is controlled by the program code. 55
`which is stored in PROM 42. This code may allow for
`user inputs (such as the value of the index of refraction
`for the FUT) as explained further below. After such
`information has been entered, microprocessor 40 initi(cid:173)
`ates the test by instructing timer 46 to launch a test 60
`signal, such as a 50 ns pulse, which is accordingly trans(cid:173)
`mitted by light source 50. This pulse is conveyed to the
`FUT by means of coupler 52 and port 18. Backscatter(cid:173)
`ing and reflections are returned down the FUT to port
`is, and conveyed thence from coupler 52 to receiver 54. 6S
`The return trace signal is processed by amplifier/filter
`56, and digital values are computed and stored by AID
`converter 58. Microprocessor 40 then uses these raw
`
`6
`datapoints to calculate backscattering and discrete
`losses due to faults, as discussed more fully below.
`In order to obtain more reliable data, the test signal
`may be launched several times and the datapoints aver(cid:173)
`aged. This reduces the effects of noise and statistical
`sampling, i.e., improves the effective signal-to-noise
`ratio (SNR). The test signal is preferably launched 256
`times, yielding an effective improvement in the SNR of
`about 12 decibels. As each successive trace signal is
`returned, the datapoints from that signal are respec(cid:173)
`tively added to the datapoints from the previous signals.
`In this regard, it is desirable to include a 16-bit adder
`(such as a 74F283 chip) with the AID converter and
`high-speed memory 58. The fmal sums of these values
`may be divided by 256 to yield average values but, since
`the results of the data analysis are given in a logarithmic
`(i.e., decibel) scale, there is actually no need to so divide
`the sums of the datapoints. Thus, the summed values are
`delivered to microprocessor 40 for further data analysis.
`While the foregoing construction has significant nov(cid:173)
`elty, the present invention actually lies in further en(cid:173)
`hancements to data acquisition and analysis. One en(cid:173)
`hancement relates to the threshold detection, and subse(cid:173)
`quent location, of any faults along the fiber path.
`is achieved by the use of a
`Greater sensitivity
`"matched" filter as described further below. Also, the
`preferred embodiment of the invention overcomes the
`limitations (discussed in the Description of the Prior
`Art) relating to the deadzone, by adaptively adjusting
`the pulse width based on the measured backscatter level
`at the point of interest.
`The pulse width optimization feature as described
`herein is independent of the novel use of the matched
`filter which forms the basis of this application; it is
`described herein, however, since pulse width optimiza(cid:173)
`tion is present in the preferred embodiment. It is accord(cid:173)
`ingly understood that certain features described herein
`(such as timer 46 used in conjunction with variable
`pulse width light source 50), are not necessary in the
`practice of the broadest scope of the present invention.
`The initial pulse launched by light source 50 is prefer(cid:173)
`ably short in duration, e.g., SO ns, to optimize accuracy
`with respect to close faults. The original return trace
`signal is analyzed (as described further below) to iden(cid:173)
`tify reflections and other losses. OTDR 10 then focuses
`on the first (closest) fault, determining the signal level at
`that point. If the measured signal level is too low (due to
`attenuation or other losses in the fiber path), then the
`test signal is launched again with a larger pulse width.
`On the other hand, if the measured signal level is too
`high, a new test signal is launched with a smaller pulse
`width. This comparison is performed by microproces(cid:173)
`sor 40.
`In this regard, the range of acceptable signal levels
`may be established in different ways. In the preferred
`embodiment, the acceptable minimum and maximum
`signal levels are based on the average background noise.
`Specifically, the minimum acceptable measured signal is
`about 3 dB above the background noise, and the maxi(cid:173)
`mum acceptable measured signal is about 9 dB above
`the background noise. In other words, if the measured
`signal in the vicinity of the fault under investigation is
`less than 3 dB above background noise, then another
`test signal will be launched with a longer pulse width; if
`the measured signal is greater than 9 dB above back(cid:173)
`ground noise, then the relaunched signal will have a
`shorter pulse width. This range is preferable although it
`may be narrowed or expanded.
`
`Page 9
`
`
`
`5,066,118
`
`7
`If the signal level is to be analyzed with respect to the
`background noise level, it is necessary to calculate the
`noise level. This may be computed using various meth(cid:173)
`ods. The inventor chooses to examine a portion of the
`trace signal beyond the end of the fiber (Le., that portion 5
`of the signal which takes more than about 400 microsec(cid:173)
`onds to return to OTDR 10), since portion of the signal
`represents pure noise. One hundred data points beyond
`this position are sampled and averaged to give the back(cid:173)
`ground noise value. This calculation is performed each 10
`time the pulse width is adjusted to correlate current
`measured signals to current noise values.
`Once it has been determined that the pulse width is
`too short or too long, a suitable adjustment to the pulse
`width must be made. This may be accomplished by 15
`simply increasing or decreasing the pulse width by a
`fixed increment, e.g., 50 ns. This technique, however,
`would not necessarily place the measured signal in the
`acceptable range, and reiterations might be necessary.
`Therefore, an alternative technique may be used which 20
`has been found to decrease optimization time. This
`technique requires that the pulse width be increased by
`an increment tine according to the empirical formula:
`
`tinc= [(PWold!2S0)+ II x 50
`
`8
`other laser sources may be, used to provide a pulse
`width of up to 20 Jls. Also, the minimum practical set(cid:173)
`ting is 50 ns, although smaller settings are conceivable.
`Once the optimum pulse width for the first fault is
`established, microprocessor 40 instructs light source SO
`(via timer 46) to launch a new test signal (actually, to
`launch a series of 256 pulses as explained above). Of
`course, the original 50 ns pulse may already be optimum
`for the first fault. If, however, a new pulse is launched,
`then the resulting trace reolaces the original trace. This
`replacement trace is processed in the same manner as
`the original trace (discussed more fully below) to rede(cid:173)
`fine the locations of any faults. The signal level at the
`first fault is then reexamined to confirm that the new
`pulse width is indeed optimal (Le., the measured signal
`level is within the acceptable range). If not, the optimi(cid:173)
`zation routine is repeated. Once optimization of the
`pulse width for the first fault is confirmed, the trace
`signal is analyzed further to determine its exact location
`and the associated signal loss.
`After analysis of the first fault is completed, attention
`shifts to the second fault. The same optimization routine
`is used to determine the best pulse width for the second
`2S fault; light source SO emits another series of pulses and
`the resulting trace again replaces the previous trace.
`The latest trace is similarly analyzed to redefine all fault
`locations, tQ confirm optimization for the second fault,
`and to calculate the loss at that fault. This process is
`repeated for as many faults as are detected. If no faults
`at all are initially found, the test signal may be succes(cid:173)
`sively relaunched at longer pulse widths, up to the max-
`imum of 6 Jls.
`The above procedure has clear advantages over prior
`art OTDR's. The optimization of pulse width provides
`better resolution of closely spaced faults without reduc(cid:173)
`ing sensitivity to losses. The dynamic range of OTDR
`10 is thus extended since dynamic range is a function of
`both peak power and pulse width, and this is accom-
`plished without excessive amplification of the trace
`signal which would result in a lower signal-to-noise
`ratio. Nevertheless, there is still room for improvement,
`namely with respect to threshold detection and location
`of losses. The present invention additionally provides
`means for improving the accuracy of fault location
`independent of pulse width optimization.
`This improvement in fault location, which forms the
`basis for this application, is achieved by the use of
`"matched" filter. A matched filter, which is a non-linear
`function, is known in digital processing, and sequen(cid:173)
`tially focuses on a given datapoint and a certain number
`of data points on either side thereof. In the preferred
`embodiment, the matched filter sequentially operates on
`a set of datapoints in the trace signal: the central or
`reference point, the two preceding datapoints, and the
`two succeeding data points. For each reference point, a
`matched filter function fm/is created according to the
`equation:
`
`where pWold is the duration of the old pulse width in
`nanoseconds, and the increment is given in nanosec(cid:173)
`onds. For simplicity, the division by 250 is an integer
`division, i.e., the quotient is rounded down to a whole
`number. This calculation is performed by microproces- 30
`sor 40. The increment used in decreasing pulse width is
`based on the same formula, with the minor change of
`dividing the old pulse width by 200 rather than by 250.
`For example, assume that the original launched pulse
`was 50 ns, but the measured signal at the first fault was 35
`below the minimum acceptable value. Using the above
`formula, an increment of 50 ns is calculated which,
`when added to the old pulse width of 50 ns, yields a new
`pulse width of 100 ns. As another example, assume a
`larger pulse width of 1 Jls had been transmitted, and that 40
`the measured signal at the fault under investigation was
`above the maximum acceptable value. The above for(cid:173)
`mula would indicate a decrease of 250 ns, yielding a
`new pulse width of 750 ns.
`As those skilled in the art will appreciate, if two faults 45
`are closely spaced together, an increase in the pulse
`width might "erase" the second fault since the deadzone
`would overlap both faults. In order to avoid this result,
`microprocessor 40 computes the estimated deadzone
`for the new pulse width prior to its being launched. If 50
`the new pulse width would so erase a fault, the optimi(cid:173)
`zation routine is aborted and'the last trace is used to
`analyze both faults. Alternatively, if enough memory
`(RAM 44) is available, the last trace may be stored for
`later analysis with respect to the erased fault. The opt i- 55
`mization routine could then be performed for all other
`faults.
`With appropriate control electronics,
Accessing this document will incur an additional charge of $.
After purchase, you can access this document again without charge.
Accept $ ChargeStill Working On It
This document is taking longer than usual to download. This can happen if we need to contact the court directly to obtain the document and their servers are running slowly.
Give it another minute or two to complete, and then try the refresh button.
A few More Minutes ... Still Working
It can take up to 5 minutes for us to download a document if the court servers are running slowly.
Thank you for your continued patience.
This document could not be displayed.
We could not find this document within its docket. Please go back to the docket page and check the link. If that does not work, go back to the docket and refresh it to pull the newest information.
Your account does not support viewing this document.
You need a Paid Account to view this document. Click here to change your account type.
Your account does not support viewing this document.
Set your membership
status to view this document.
With a Docket Alarm membership, you'll
get a whole lot more, including:
- Up-to-date information for this case.
- Email alerts whenever there is an update.
- Full text search for other cases.
- Get email alerts whenever a new case matches your search.
One Moment Please
The filing “” is large (MB) and is being downloaded.
Please refresh this page in a few minutes to see if the filing has been downloaded. The filing will also be emailed to you when the download completes.
Your document is on its way!
If you do not receive the document in five minutes, contact support at support@docketalarm.com.
Sealed Document
We are unable to display this document, it may be under a court ordered seal.
If you have proper credentials to access the file, you may proceed directly to the court's system using your government issued username and password.
Access Government Site