Round Robin for Optical Fiber Bragg Grating Metrology.NIST (National Institute of Standards & Technology, Washington, DC, www.nist.gov) The standardsdefining agency of the U.S. government, formerly the National Bureau of Standards. It is one of three agencies that fall under the Technology Administration (www.technology. has administered the first round robin of measurements for optical fiber Bragg gratings A short length of optical fiber that filters out a particular wavelength. Periodically spaced zones in the fiber core are altered to have different refractive indexes slightly higher than the core. . We compared the measurement of center wavelength, bandwidth, isolation, minimum relative transmittance, and relative group delay among several grating types in two industry groups, telecommunications Communicating information, including data, text, pictures, voice and video over long distance. See communications. and sensors
Science of measurement. Measuring a quantity means establishing its ratio to another fixed quantity of the same kind, known as the unit of that kind of quantity. needs improvement in most areas. Specifically, when tunable lasers A laser that can change its frequency over a given range. In time, tunable lasers are expected to be capable of switching frequencies on a packet by packet basis. are used a filter is needed to remove broadband emissions from the laser. The linear slope of relative group delay measurements is sensitive to drift and systematic bias in the rfmodulation technique. The center wavelength measurement had a range of about 27 pm in the sensors group and is not adequate to support longterm structural monitoring applications. Key words: dispersion dispersion, in chemistry dispersion, in chemistry, mixture in which fine particles of one substance are scattered throughout another substance. A dispersion is classed as a suspension, colloid, or solution. ; metrology; optical fiber Bragg grating; optical fiber sensors; reflectance re·flec·tance n. The ratio of the total amount of radiation, as of light, reflected by a surface to the total amount of radiation incident on the surface. Noun 1. ; relative group delay; transmittance; wavelength division multiplexing See WDM. (communications) wavelength division multiplexing  (WDM) Multiplexing several Optical Carrier n signals on a single optical fibre by using different wavelengths (colours) of laser light to carry different signals. . Accepted: August 22, 2000 Available online: http://www.nist.gov/jres 1. Introduction At the request of industry, NIST has administered the first round robin of measurements for optical fiber Bragg gratings (FBG FBG Fiber Bragg Gratings FBG Fasting Blood Glucose FBG Functional BrainGut Research Group FBG Florida Brewer's Guild FBG Fluidized Bed Generator FBG Flavor Blasted Goldfish (gaming) FBG Forum Battle Group ). In this paper we report on that work. Fiber Bragg gratings are wavelengthselective reflectors that can be written into the core of optical fibers. These relatively new components are extremely important for telecommunication telecommunication Communication between parties at a distance from one another. Modern telecommunication systems—capable of transmitting telephone, fax, data, radio, or television signals—can transmit large volumes of information over long distances. and sensor A device that measures or detects a realworld condition, such as motion, heat or light and converts the condition into an analog or digital representation. An optical sensor detects the intensity or brightness of light, or the intensity of red, green and blue for color systems. applications. In new wavelength division multiplexed (WDM (1) (Wavelength Division Multiplexing) A technology that uses multiple lasers and transmits several wavelengths of light (lambdas) simultaneously over a single optical fiber. ) optical fiber communication systems FBGs are used as wavelength filters and dispersion compensators A device that removes pulse distortion in an optical fiber network. It is usually placed near the optical amplifier spaced some 25 to 50 miles apart. See dispersion. (dispersion in optical fiber spreads optical pulses Optical pulses Bursts of electromagnetic radiation of finite duration. Optical pulses are used to transmit information or to record the chronology of physical events. The simplest example is the photographic flash. in time and limits the data rate of the fiber link.). Also, FBGs make excellent strain sensors that can be networked to obtain distributed strain measurements of large structures, such as bridges and ships. In spite of in opposition to all efforts of; in defiance or contempt of; notwithstanding. See also: Spite the numerous and growing commercial applications of fiber Bragg gratings, there are no standard measurement procedures for the gratings See diffraction grating and fiber Bragg grating. and a variety of definitions are being used for important measurement parameters. At an informal meeting during the Optical Fiber Communication Conference in February of 1999 (OFC'99), a quorum A majority of an entire body; e.g., a quorum of a legislative assembly. A quorum is the minimum number of people who must be present to pass a law, make a judgment, or conduct business. of industry representatives recommended measuring several specific spectral spectral /spec·tral/ (spek´tral) pertaining to a spectrum; performed by means of a spectrum. spec·tral adj. Of, relating to, or produced by a spectrum. and relative group delay (RGD RGD Rijksgebouwendienst RGD Rat Genome Database RGD Registered Graphic Designer (Canada) RGD ArginineGlycineAspartic Acid RGD Rapid Gas Decompression RGD Reacting Gas Dynamics RGD Range Gate Deception RGD Returned Goods Damaged ) properties of a chirped FBG. (The derivative of RGD versus wavelength gives a measure of dispersion in an optical component.) In June of 1999 NIST launched the round robin among two industry groups; one group named "Telecom" measured spectral properties and RGD for their gratings, the other industry group named "Sensors" measured only the spectral properties of their gratings. The roundrobin participants in the two groups were: ADC (1) See A/D converter. (2) (Apple Display Connector) A peripheral connector from Apple that combines digital video display, USB and power in one cable. , Agilent, Corning, Perkin Elmer, GNnet test, NPL 1. NPL  New Programming Language. IBM's original (temporary) name for PL/I, changed due to conflict with England's "National Physical Laboratory." MPL and MPPL were considered before settling on PL/I. Sammet 1969, p.542. 2. , and 3M in the telecom group, and Blue Road Research, CiDRA, EXFO, Micron Optics optics, scientific study of light. Physical optics is concerned with the genesis, nature, and properties of light; physiological optics with the part light plays in vision; and geometrical optics with the reflection and refraction of light as encountered in the study , and NRL Noun 1. NRL  the United States Navy's defense laboratory that conducts basic and applied research for the Navy in a variety of scientific and technical disciplines Naval Research Laboratory in the sensors group. Raw data from the participants were sent to NIST because no formal methods for analyzing the spectral or RGD data existed. The participants were each sent a set of instructions to follow, which are pr esented in Appendices ap·pen·di·ces n. A plural of appendix. A and B. The purpose of this round robin was to determine the state of industrial metrology concerning spectral and RGD measurements in FBGs. From the data and results collected we are making recommendations on the methods of measurement and analysis for FBG. The participants each received a box containing two gratings with about 10 m of Corning SMF28 fiber [1] spliced on each side of the grating. Figure 1 shows a sketch sketch, a rapidly executed kind of pictorial notetaking. The sketch is not usually intended as an autonomous work of art, although many have been considered masterpieces in their own right. of the box and gratings. The telecom group received an International Telecommunication Union International Telecommunication Union (ITU), specialized agency of the United Nations, with headquarters at Geneva. It was created in 1934 as a result of the merging of the International Telegraph Union (est. (ITU (International Telecommunication Union, Geneva, Switzerland, www.itu.ch) A telecommunications standards body that is under the auspices of the United Nations. Comprising more than 185 member countries, the ITU sets standards for global telecom networks. ) channel 0, 50 GHz bandwidth grating and a mildly chirped, [approximate] 16 nm bandwidth, grating. The sensor group received a similar ITU channel 0, 50 GHz bandwidth grating and a narrowbandwidth, low reflectance, sensor grating. To stabilize stabilize See peg. the ITU and chirped gratings against temperature changes they were packaged in an a thermal package and placed on a thermoelectric ther·mo·e·lec·tric also ther·mo·e·lec·tri·cal adj. Characteristic of, resulting from, or using electrical phenomena occurring in conjunction with a flow of heat. cooler (TEC), controlled to within less than 0.1 [degrees]C. The sensor grating was packaged in a silicon rubber tube, strain free, and placed on the same TEC. Again, the TEC was controlled to within less than 0.1 [degrees]C. The raw data from the each of the participants are shown in Appendices C and D. The labels A, B, C, etc. on the raw data and analysis represent participants and were determined by a random number generator A program routine that produces a random number. Random numbers are created easily in a computer, since there are many random events that take place such as the duration between keystrokes. . During the OFC'99 meeting the participants agreed that NIST should determine an analysis method and apply it to the raw data to determine a grating parameter (1) Any value passed to a program by the user or by another program in order to customize the program for a particular purpose. A parameter may be anything; for example, a file name, a coordinate, a range of values, a money amount or a code of some kind. of interest. If formal methods were determined at a later time, the data could be reanalyzed with those methods and valuable insight could be gained without having to repeat the round robin. For transmittance measurements, most of the participants launched on port No. 1 or No. 4, of Fig. 1. Reflectance data were obtained with the aid of a coupler Refers to a myriad of different types of sockets for plugging in electric or electronic cables or devices. See network coupler. (participant supplied) and only one port; usually No. 1 or No. 4 in Fig. 1 was used. Relative values for transmittance and reflectance were used since only relative features were needed to determine many of the parameters. 2. Measurement Techniques Measurement techniques used by the various participants depended on the group. The telecom group used primarily a grantingtuned diode laser See laser diode. , power meter, and wavelength meter system for the spectral measurements. The sensors group used primarily a wide band source and an optical spectrum analyzer A hardware device or software used to examine the frequency and power components of a signal. It provides more information than an oscilloscope, because it can display the signals over a range of frequencies. for spectral measurements. In some cases a swept diode laser or wideband source and swept fiber FabryPerot filter were used. The NIST system employed for spectral measurements is shown in Fig. 2. A tunable fiber FabryPerot filter (FFP FFP  Formal FP. A language similar to FP, but with regular sugarless syntax, for machine execution. See also FL. ["Can Programming be Liberated From the von Neumann Style? A Functional Style and Its Algebra of Programs", John Backus, 1977 Turing Award Lecture, CACM ) was used to filter amplified spontaneous emissionLight (ASE (Adaptive Server Enterprise) A relational DBMS from Sybase that runs on Windows NT/2000, Linux and a variety of Unix platforms. ASE is a comprehensive and robust data management product with a long history dating back to the late 1980s. ) from a gratingtuned diode laser. The ASE was filtered, because the measurement system photodetectors respond to light over a wide bandwidth. The ASE power, though small over narrow bandwidths, becomes significant over the photodetector A device that senses light. It uses the principle of photoconductivity, which is exhibited in certain materials that change their electrical conductivity when exposed to light. See photoelectric, photocell and photodiode. bandwidth. ASE represents spectral noise on the narrow spectral laser output and will limit the dynamic range of the measurement system. A wavelength meter provided the wavelength scale. Uncertainty analysis for this system is presented in Appendix E. The effect of the FFP filter can be seen in Figs. 3 and 4. Figure 3 shows the unfiltered Please wikify (format) this article or section as suggested in the Guide to layout and the Manual of Style. Remove this template after wikifying. This article has been tagged since and filtered output of the tunable laser over the 1350 nm to 1650 nm band. These data were taken with a 1 nm resolution, optical spectrum analyzer. The laser peak power at 1553 nm was about 308 [micro]W and the integrated ASE across the measured spectra in Fig. 3 was about 1.7 [micro]W. The ratio of these powers is about 23 dB and is a good measure of the dynamic range of the system without the FFP. With the FFP filter the ASE is suppressed sup·press tr.v. sup·pressed, sup·press·ing, sup·press·es 1. To put an end to forcibly; subdue. 2. To curtail or prohibit the activities of. 3. , as shown in Fig. 3. With the FFP filter the laser power is at about 174 [micro]W, and the integrated ASE power is about 76 pW for a power ratio of about 64 dB. If a second FFP filter with a different free spectral range was used, further ASE suppression suppression /sup·pres·sion/ (supresh´un) 1. the act of holding back or checking. 2. sudden stoppage of a secretion, excretion, or normal discharge. 3. could be realized. Figure 4 shows the effect of the ASE on a measurement of a FBG's relative transmittance. Without the FEP See front end processor. filter the minimum relative transmittance is only about 25 dB. With the FFP filter the minimum relative transmittance is about 65 dB. Laser ASE will also affect the isolation or crosstalk (1) Electromagnetic interference that comes from an adjacent wire. "Alien" crosstalk is interference that comes from a wire in an adjacent cable, for example, when two or more twisted wire pair cables are bundled together. measurement of a FBG. The RGD of the chirped grating was determined by various phaseshift techniques. The NIST phaseshift system is shown in Fig. 5. A detailed description can be found in Ref. [1] and a summary of the uncertainty is presented in Appendix E. NIST also employed a new, lowcoherence interferometer interferometer: see interference under Interference as a Scientific Tool. See also virtual telescope. An instrument that measures the wavelengths of light and distances. method to determine the RGD of the chirped grating. That system is shown in Fig. 6 and details of this system are described in the literature [2, 3]. A full uncertainty analysis is in progress. 3. Analysis Methods At the time of this report, a fiberoptic test procedure (FOTP FOTP Friends of the Parks (Chicago, Illinois) FOTP Fiber Optic Test Procedure FOTP Father of the Pride FOTP Fear of the Park (UK's wintersports freestyle website) ) titled, "Amplitude amplitude (ăm`plĭtd'), in physics, maximum displacement from a zero value or rest position. response measurement of narrowband, passive fiber optic components," was being written by the Telecommunication Industry Association (TIA (1) (Telecommunications Industry Association, Arlington, VA, www.tiaonline.org) A membership organization founded in 1988 that sets telecommunications standards worldwide. It was originally an EIA working group that was spun off and merged with the U.S. ) FO6.3.5 Subcommittee sub·com·mit·tee n. A subordinate committee composed of members appointed from a main committee. subcommittee Noun on Fiber Optic Interconnecting Devices. NIST is working with this subcommittee on the FOTP and any appropriate results from this round robin will be included in the FOTP. Another FOTP on dispersion measurements is also being drafted by the same subcommittee. From the TIA sources and the meeting at OFC'99, several parameters of interest were identified that could be obtained from the raw data files sent to NIST. The spectral parameters of interest to TIA and the OFC'99 group were center wavelength, bandwidth, isolation, and transmittance. The RGD parameters of interest were the linear slope across the operating bandwidth of the grating and the ripple magnitude and period. To determine the center wavelength, bandwidth, and isolation of a grating from the reflectance data, NIST used the following methods (see also Appendices F and G). First, the maximum reflectance in the plateau plateau, elevated, level or nearly level portion of the earth's surface, larger in summit area than a mountain and bounded on at least one side by steep slopes, occurring on land or in oceans. region was determined. Then, wavelengths at reflectance values of 3 dB and 0.5 dB from the maximum plateau reflectance were found by interpolating between data. The center wavelength [[lambda].sub.c] is defined as ([[lambda].sub.+]  [[lambda].sub.])/2, and the bandwidth is defined as ([[lambda].sub.+]  [[lambda].sub.]), where [[lambda].sub.+] and [[lambda].sub.] are the wavelengths at data values x dB (x = 0.5 or 3) on each side of the plateau region. A similar approach was used to find the isolation I of a grating. The isolation was determined by the maximum reflectance value M and the maximum outofband reflectance value O in an adjacent channel (I = M  O). The ITU channel center frequency spacing is defined as (193.1 + J X CS) THz where J is an integer integer: see number; number theory and CS is the channel spacing The amount of bandwidth allotted to each channel in a communications system that transmits multiple frequencies such as fiber optics. It is measured as the spacing between center frequencies (or wavelengths) of adjacent channels. See guard band. [4]. The integer J is either even or odd. An adjacent channel would be a channel where [J.sub.a] = J [+ or ] n and n is an integer. The minimum relative transmittance was determined from relative transmittance data by fitting a spline In computer graphics, a smooth curve that runs through a series of given points. The term is often used to refer to any curve, because long before computers, a spline was a flat, pliable strip of wood or metal that was bent into a desired shape for drawing curves on paper. See Bezier and Bspline. function to the data and locating the minimum. From the RGD data, the linear slope of the chirped grating was determined from data within the 3 dB reflectance bandwidth (determined from a robust method described in Appendix F) using a leastsquares fit. The residual RGD is found by subtracting the linear slope from the RGD reflectance data. The ripple magnitude and period varied across the bandwidth of the chirped grating. We constructed two simple methods for comparing the RGD ripple between the different participants. 4. Telecom Group 4.1 Grating History To assure that there was no bias to a grating measurement as the round robin progressed, periodically we returned the round robin boxes to NIST for measurement. In this section a history of the center wavelength of each of the telecom gratings and the linear slope of the RGD of the chirped grating is given as a function of time. At the end of the first 60 days of the telecom round robin, the box received a shock that caused a splice tray to come lose from its moorings and break some fiber leads inside the box. After repairs were made at NIST, the gratings in the telecom box showed a change in the center wavelength that exceeded the uncertainty of our system. Data from participants that subsequently measured the telecom gratings received an appropriate wavelength correction. Figure 7 shows the center wavelength versus time for the telecom ITU grating. The NIST measurements are solid dots and those from the roundrobin participants are open circles. The uncertainties on the NIST data represent the expanded uncertainty of our measurement system, U = 6 pm (two standard deviation In statistics, the average amount a number varies from the average number in a series of numbers. (statistics) standard deviation  (SD) A measure of the range of values in a set of numbers. estimate and hence a coverage factor of k = 2, taking into account all known components of uncertainty). This uncertainty was subsequently reduced to U = 0.2 pm. The uncertainties on the round robin participants' measurements represent only the fit uncertainty (two standard deviation estimate) and are shown as a dashed line. The center wavelength is determined from the mean of the 3 dB wavelength values. After the box was returned due to damage, a shift in the center wavelength of about 37 pm was recorded and the grating center wavelength was monitored periodically afterwards af·ter·ward also af·ter·wards adv. At a later time; subsequently. afterwards or afterward Adverb later [Old English æfterweard] Adv. 1. . To give a fair correction to the later round robin participants, a linear leastsquaresfit was applied to the NIST measurements th at just preceded and followed a participant or group of participants. The fit is shown as the solid line in Fig. 7. A correction factor was obtained that gives a center wavelength equal to the beginning NIST value. Figure 8 shows the  3 dB center wavelength versus time for the telecom chirped grating. The NIST measurements are solid dots and the round robin participants are open circles. The uncertainties shown were determined in the same way as in Fig. 7. A shift in the center wavelength of 75 pm was recorded after the box was repaired. A linear leastsquaresfit to the NIST measurements is shown as a solid line in Fig. 8 and a correction factor was determined for the comparison of subsequent participants. Figure 9 shows the RGD linear slope history of the telecom chirped grating versus time in days. The history was obtained from the lowcoherence interferometer system and shows no appreciable ap·pre·cia·ble adj. Possible to estimate, measure, or perceive: appreciable changes in temperature. See Synonyms at perceptible. change over 350 days. The uncertainties shown represent the repeatability of this system, 0.053 ps/nm (two standard deviation estimate; full uncertainty analysis is being conducted). Apparently the damage done to the system did not affect the RGD linear slope and no correction was added to the participants values. These histories show that the gratings remained stable enough for the telecom round robin to be useful. When they did drift, periodic measurements at NIST corrected the bias. 4.2 Summary of Telecom Group Results From the data received from each participant, the following parameters were determined: the center wavelength and bandwidth were found using the  3 dB and  0.5 dB points from the relative reflectance of the JTU JTU Jackson Turbidity Unit JTU Jyllands Tennis Union (Denmark) grating. Also, the isolation was determined for the ITU grating. From the relative transmittance data, a minimum relative transmittance was measured. From the relative reflectance of the chirped grating, the center wavelength and bandwidth were found at  3 dB, and the minimum relative transmittance was determined from the relative transmittance data. From the RGD data, the linear slope was determined over the  3 dB bandwidth (see Appendix F) by a leastsquaresfit method. Also, a comparison of the RGD ripple among the participants was made. 4.2.1 ITU Grating Several comments can be made about the relative reflectance data for the ITU grating (Appendix C, Fig. 29). Participants C, E, and F all had coarse data sets (large wavelength intervals between values), so that not enough data were obtained in the plateau region and band edges for reliable measurements to be made. In the following figures, these coarse data sets increase the fit uncertainty and bias the center wavelength, bandwidth, and isolation measurements. Participant D may not have normalized the reflectance to source power fluctuations, as the ITU grating reflectance in the plateau region has more noise than any other participant; see Fig. 29. The 0.5 dB bandwidth is very sensitive to the shape of the ITU grating reflectance. Figure 10 shows the results of the ITU center wavelength determined using the 3 dB and 0.5 dB points. The expanded uncertainties for the NIST measurement are 6 pm (k = 2) and take into account all known components of uncertainty. The uncertainties for the participants are just the fit uncertainty (two standard deviation estimate). For the 3 dB center wavelength, the range of values is 27.5 pm, the standard deviation s is 8 pm, and the mean is 1552.521 nm. For the 0.5 dB center wavelength, the range of values is 42 pm, the s = 12 pm, and the mean is 1552.515 nm. The difference in the 3 dB and 0.5 dB center wavelength values is about 6 pm. These mean values are only 5 pm (3 dB measurement) and 11 pm (0.5 dB measurement) from the ITU channel 0 wavelength of 1552.526 nm. For either criterion (0.5 or 3 dB), this grating would pass the Telcordia specification that the specified and actual center wavelengths differ by less than 20 % (10 GHz or 80 pm in this case) of the specified bandwidth [4]. Figure 11 shows the results of the ITU bandwidth determined at 3 dB and 0.5 dB. The uncertainties are the same as discussed for Fig. 10 converted to GHz (the NIST uncertainty is 748 MHz (MegaHertZ) One million cycles per second. It is used to measure the transmission speed of electronic devices, including channels, buses and the computer's internal clock. A onemegahertz clock (1 MHz) means some number of bits (16, 32, 64, etc. ). For the 3 dB bandwidth the range of values is 9.5 GHz, the standard deviations is 3 GHz, and the mean is 51.2 GHz. Inmost in·most adj. Farthest within; innermost. inmost Adjective same as innermost Adj. 1. cases the participants would pass this as a 50 GHz ITU grating using the 3 dB criterion. For the 0.5 dB bandwidth the range of values is 12 GHz, the standard deviation s is 4 GHz, and the mean is 39.7 GHz. In all the cases the roundrobin participants would reject this grating as a 50 GHz ITU grating at the 0.5 dB criterion. The difference in the mean bandwidth values from 3 dB to 0.5 dB is 11.5 GHz. Figure 12 shows the results of the isolation measurement of the ITU grating determined by finding the value of the plateau region of the reflectance data and the highest reflectance value in an adjacent channel. The uncertainties are the fit uncertainty only. Telcordia specifies that the isolation for a branching/filtering component should be at least 25 dB for data rates up to 10 Gb/s [4]. For most participants this grating would not pass this test. Figure 13 shows the values for the minimum relative transmittance of the ITU grating taken from the relative transmittance data shown in Appendix C, Fig. 30. For strong gratings the minimum relative transmittance measurement is quite sensitive to the spectral purity of the laser light source. NIST's fiber FabryPerot filtered laser shows the lowest minimum transmittance. The uncertainties shown are only the uncertainty of the fit (two standard deviation estimate). 4.2.2 Chirped Grating The data sets for the chirped grating are shown in Appendix C, Figs. 31, 32, and 33. These data sets were used to measure the center wavelength, bandwidth, minimum relative transmittance, and RGD linear slope and ripple. Participants C and F provided coarse data sets. Again, because the data interval is large, the uncertainties of the fits to their data are larger than those for the other participants, and in some cases the results are biased. Because of the coarseness Coarseness Billingsgate site of a London fishmarket, known for foul and abusive language. [Br. Hist.: Brewer Dictionary, 106] Branghtons, the Evelina’s rude, coarse cousins. [Br. Lit. of the C and F data sets, no ripple information could be determined from these data. Figure 14 shows the 3 dB center wavelength measurement for the chirped grating. The uncertainties shown for NIST represent the expanded uncertainty U = 6 pm (k = 2) of this measurement system. The uncertainties shown for the participants are due to the uncertainty of the fit only (two standard deviation estimate). The range of values is 165 pm with a standard deviation s of 57 pm. Figure 15 shows the 3 dB bandwidth of the chirped grating. The uncertainties shown are similar to those in Fig. 14 converted to frequency. The range is 21 GHz with a standard deviation s of 7 GHz. Figure 16 shows the minimum relative transmittance of the chirped grating determined from the data shown in Appendix C, Fig. 32. As stated earlier, the minimum transmittance measurement is sensitive to the spectral purity of the laser light source. NIST's filtered laser shows the lowest minimum transmittance. The uncertainties shown (for NIST and the round robin participants) are only the uncertainty of the fit (two standard deviation estimate). Figure 17 shows the RGD linear slope determined from the data in Appendix C, Fig. 33. The linear slope was found using a leastsquaresfit of a linear function to the 3 dB bandwidth of the data (see Appendix F). The uncertainties shown in Fig. 17 represent the repeatability (two standard deviation estimate) for NIST measurements using the phaseshift and lowcoherence systems. The uncertainties for the participants represent only the fit uncertainty (two standard deviation estimate). Figure 17 shows several values for the RGD linear slope due to a systematic bias that the phaseshift systems have. Because these systems are highly coherent they measure the RGD of the entire interferometer, not just the grating. Thus, the [approximate]10 m of fiber on each side of the FBG plus any fiber in the measurement system between the modulator Modulator Any device or circuit by means of which a desired signal is impressed upon a higherfrequency periodic wave known as a carrier. The process is called modulation. The modulator may vary the amplitude, frequency, or phase of the carrier. arid ar·id adj. 1. Lacking moisture, especially having insufficient rainfall to support trees or woody plants: an arid climate. 2. the detector detector: see particle detector. will add some RGD to the measurement of the grating. To remove the bias from the surrounding fiber, the participant should make RGD measurements from both directions of the grating and, assuming the grating's absolute RGD is is independent of direction, subtract A relational DBMS operation that generates a third file from all the records in one file that are not in a second file. the surrounding fiber RGD. NIST and participants A, B, and E provided data for both directions. Shown in Fig. 17 are both the RGD linear slope for one direction, input on No. 1 (see Fig. 1), and the mean of both directions. Also included is the RGD measured with NIST's lowcoherence interferometer system. The lowcoherence system has a better repeatability than the phaseshift system, possibly because the lowcoherence system has faster data acquisition and does not require temperaturesensitive components such as the MachZender modulator. The two NIST systems agree well, differing by about 0.1 ps/nm. The range for the value of the RGD linear slope, for one direction only, is 1.1 ps/nm with a standard deviation s of 0.4 ps/nm. The range for the mean of both directions and the lowcoherence values of the RGD linear slope is 0.1 ps/nm with a standard deviation s of 0.04 ps/nm. The mean value is6.81 ps/nm. Figure 18 shows the residuals of the linear fit, across the 3 dB bandwidth, for each participant. To compare the agreement between NIST and participants A, B, E, and G, we calculated the difference from the mean residual at each wavelength and the standard deviation s from the mean residual. To compare the finer data sets with the coarser data sets, we compressed the finer data by determining a mean residual value Residual value Usually refers to the value of a lessor's property at the time the lease expires. residual value The price at which a fixed asset is expected to be sold at the end of its useful life. using several data points over a small wavelength interval that matched the coarser data interval. Figure 19 shows the difference from the mean residual and Fig. 20 the standard deviation s of the mean residual. The agreement at each wavelength interval is quite good, on average about 1 ps for the difference from the mean residual. The standard deviation s at most wavelengths is less than 1 ps. Thus, the various phaseshift systems record the same RGD ripple value to within 1 ps for RGD values ranging from about 40 to 150 ps. The agreement with the lowcoherence system is still being improved. The current rms difference is 1.5 ps [2]. Figure 21 shows a portion of the residual RGD spectra taken with the NIST rf and lowcoherence systems and with the systems of participants A, B, E, and G. For most cases the ripple measurements agree, but wavelength accuracy, measurement uncertainty, and rf sideband averaging can lead to differences of several picoseconds [5]. The difference between the lowcoherence and rf phase shift systems is still being investigated, but no major differences have been observed [2]. The other round robin participant's RGD data could not be used to compare the ripple because of coarse wavelength steps. Figure 20 illustrates the need for precision RGD ripple measurements, because over a 0.5 nm wavelength span the RGD changes rapidly, i.e., from +5 to  4 ps. Chirped gratings with larger RGD linear slopes will have larger RGD ripple amplitudes, increasing the need for more precision in RGD ripple measurements for WDM systems. 5. Sensors Group 5.1 Grating History In this section a history of the  3 dB center wavelength of each of the sensor gratings is given. Figure 22 shows the  3 dB center wavelength versus time for the Sensors group ITU grating. The NIST measurements are indicated by solid dots and the round robin participants by open circles. From the time of the construction of the roundrobin box until the beginning of the round robin, the center wavelength changed by about 12 pm. After the round robin was completed the center wavelength showed an almost insignificant, change of about 5 pm, which is within the expanded uncertainty (coverage factor k = 2). The uncertainties on the NIST data represent the uncertainty of our measurement system, 6 pm (two standard deviation estimate). The uncertainties on the roundrobin participants' measurements represent only the fit uncertainty (two standard deviation estimate). The center wavelength is determined from the mean of the  3 dB wavelength values. Figure 23 shows the  3 dB center wavelength versus time for the Sensors Group sensor grating. The NIST measurements are indicated by solid dots and those of the round robin participants by open circles. From the construction of the grating until the end of the round robin, the center wavelength changed by about 2 pm, which is insignificant compared to the 4 pm expanded uncertainty (k = 2) shown in the figure. The grating showed further drift over 188 days of about 4 pm. The smaller uncertainty for this grating is due to the narrow bandwidth and shape of the reflectance profile. Because the center wavelength change was so small, no correction was added to the values of the participants. These histories show that the gratings remained stable enough for the Sensors Group round robin to be useful. 5.2 Summary of Sensors Group Results From the data received from each roundrobin participant, the following parameters were determined: the center wavelengths and bandwidths were found at  3 dB from the relative reflectance of the ITU and sensor gratings. From the relative transmittance data the minimum relative transmittance was determined. Some of the participants in the Sensors Group gave two sets of data taken with different measurement systems. For the participants that gave two data sets, we labeled these with a numeral numeral, symbol denoting anumber. The symbol is a member of a family of marks, such as letters, figures, or words, which alone or in a group represent the members of a numeration system. 0 or 1 following the letter used to designate des·ig·nate tr.v. des·ig·nat·ed, des·ig·nat·ing, des·ig·nates 1. To indicate or specify; point out. 2. To give a name or title to; characterize. 3. the participant. 5.2.1 ITU Grating Some of the participants' measurement systems did not have a large dynamic range and outofband features are missing (See Appendix D, Fig. 34). If we were to measure the isolation of this grating with these systems, we would obtain false values. However, because the Sensors Group was concerned primarily with wavelength accuracy, the isolation was not measured for these data. Also, from the relative transmittance data presented in Appendix D, Fig. 35 the dynamic range of a participant's measurement system will affect the measurement of the minimum relative transmittance. The following figures show the center wavelength, bandwidth, and minimum transmittance. Figure 24 shows the results of the center wavelength determined at  3 dB. The expanded uncertainty for the NIST measurement is U = 6 pm (k = 2). The uncertainties for the participants are just the fit uncertainty (two standard deviation estimate). The  3 dB center wavelength range of values is 124 pm, the standard deviation s is 42 pm, and the mean is 1552.530 nm. If the value of participant B0 is removed, the range of values is 35 pm, the standard deviation s is 12 pm, and the mean is 1552.5 16 nm. These mean values are only +4 pm and 10 pm (excluding B0) from the ITU channel 0 specified wavelength of 1552.526 nm. For most participants, this grating would pass the Telcordia specification, which requires that the specified and actual wavelength differ by less than 20 % of the bandwidth (80 pm) [14]. Figure 25 shows the results of the bandwidth determined at  3 dB. The uncertainties are the same as discussed for Fig. 24 converted to GHz (the NIST uncertainty is 748 MHz, expanded uncertainty k = 2). For the 3 dB bandwidth the range of values is 1 GHz, the standard deviation s is 0.3 GHz, and the mean is 51.2 GHz. Figure 26 shows the values for the minimum transmittance taken from the relative transmittance data shown in Appendix D, Fig. 35. The minimum transmittance measurement is sensitive to the spectral purity of the laser light source or filter bandwidth of the detector. For some participants, the minimum relative transmittance was as much as 9 dB lower than others. The uncertainties shown are only the uncertainty of the fit (two standard deviation estimate). 5.2.2 Sensor Grating In Appendix D, Figs. 36 and 37, the round robin data for the sensor grating shows that most participants measured with a fine enough wavelength interval, but with some participants the dynamic range of the system was low, rendering the measurement insensitive in·sen·si·tive adj. 1. Not physically sensitive; numb. 2. a. Lacking in sensitivity to the feelings or circumstances of others; unfeeling. b. to detailed features. However, because this grating has a low relative reflectance, [less than] 40 %, the minimum relative transmittance is not as sensitive to the dynamic range of the system or spectral purity of the source. Figure 27 shows the results of the center wavelength determined at  3 dB. The expanded uncertainty for the NIST measurement is U = 4 pm (k = 2). The uncertainties for the participants are just the fit uncertainty (two standard deviation estimate). The  3 dB center wavelength range of values is 128 pm; the standard deviation s is 43 pm, and the mean is 1552.125 nm. If the value of participant B0 is removed, the range of values is 27 pm, the standard deviation s is 11 pm, and the mean is 1552.111 nm. If this were a typical sensor grating in a strain or temperature sensing application, the range of values for the center wavelength, 27 pm, converts to about a strain of 25 X [10.sup.6] or about a 3 [degrees]C temperature variation. In some applications this variation would be acceptable and in others the variation would be too high. Figure 28 shows the values for the minimum transmittance taken from the relative transmittance data shown in Appendix D, Fig. 37. The uncertainties shown for the participants are only the uncertainty of the fit (two standard deviation estimate). 6. Conclusions Metrology for WDM components, such as FBGs, must improve to meet the demands of current and future WDM networks. From the roundrobin results we can draw the following conclusions. The state of FBG metrology appears inadequate for measurements of isolation, minimum relative transmittance, bandwidth, and RGD linear slope. The source spectral purity is critical; ASE from diode sources must be substantially reduced, and more care is needed among the two groups concerning ASE. The spectral purity or detector filter bandwidth is important when measuring transmittance. The bandwidth of the optical spectrum analyzer (OSA 1. OSA  Open Scripting Architecture. 2. OSA  Open System Architecture. ) must be carefully considered when this instrument is used. Wavelength uncertainty of [less than]1 pm and step sizes [less than]10 pm are necessary for bandwidth and RGD ripple measurements. Removal of fluctuations in the source spectral power is necessary to measure bandwidth. The criteria for determining the center wavelength and bandwidth appear to be important for ITU gratings. The use of the average of the 3 dB points may be standard practice but is not as practical for the system designer as using the 0.5 dB points. If 0.5 dB is accepted by industry as the value used to determine the center wavelength and bandwidth, many gratings being produced and in use will not pass the bandwidth specification. The center wavelength value did not change significantly for either the 3 dB or 0.5 dB criterion. Comparing the two groups for the measurement of the center wavelength of the ITU gratings, we note that the sensors group had a 35 pm range and s = 12 pm, while the telecom group had a 28 pm range and s = 8 pm. The telecom group primarily relied on wavelength meters to set the wavelength scale and the sensors group relied on OSAs. Surprisingly, this comparison would indicate that there is little difference between the two systems when measuring the ITU grating. Sensor applications need better absolute calibration calibration /cal·i·bra·tion/ (kal?ibra´shun) determination of the accuracy of an instrument, usually by measurement of its variation from a standard, to ascertain necessary correction factors. on the wavelength scale. The spread in values for the center wavelength of the sensor grating will not meet the needs of longterm structural monitoring. The measured RGD linear slope of the chirped grating had a range of 16 % when only one direction of the grating was measured. When using rf phaseshift systems, the best way to remove the system bias is to take the mean of both directions on the grating. When both directions were used the RGD linear slope range was 1.5 %. If the same relative range of RGD linear slope (16 %) were applied to a 100 km dispersion compensating grating, the effects on a telecommunication link would be dramatic; the uncertainty in the dispersion compensating grating would be around 100 ps/nm. The actual industry measurement range for RGD on a 100 km dispersion compensating grating is probably less than this. A round robin for this type of grating would provide a useful check. RGD ripple values have [+ or ]2 ps differences among participants. Stabilizing stabilizing, v to hold a limb motionless in order to ground its energy; a standard isometric resistance technique, it releases tension and lengthens muscle fibers. rf phaseshift measurement systems and working at lower rf frequencies so the wavelength resolution is [less than]10 pm is necessary for RGD ripple measurements. A standard way of discussing or presenting the RGD ripple is needed. Also, RGD resolution [less than]1 ps will be required for most WDM components as data rates increase. RGD measurements with the lowcoherence system compare well with phaseshift systems and may be preferred for rapid evaluation of components. This round robin with FBGs showed that with proper packaging and monitoring, reasonable results for a comparison could be achieved. The gratings used in this round robin, when handled well by the shipper SHIPPER. One who ships or puts goods on board of a vessel, to be carried to another place during her voyage. In general, the shipper is bound to pay for the hire of the vessel, or the freight of the goods. 1 Bouv. Inst. n. 1030. , showed little change in central wavelength. Those gratings that endured a large shock/acceleration showed moderate drift but, with interval monitoring at NIST, corrections could be made. About the authors: A. H. Rose is a physicist in the Optoelectronics See optoelectronic. Division of the NIST Electronics and Electrical Engineering electrical engineering: see engineering. electrical engineering Branch of engineering concerned with the practical applications of electricity in all its forms, including those of electronics. Laboratory. C.M. Wang is a mathematical statistician Noun 1. mathematical statistician  a mathematician who specializes in statistics statistician mathematician  a person skilled in mathematics in the Statistical Engineering Division of the NIST Information Technology Laboratory. S. D. Dyer is an electrical engineer in the Optoelectronics Division of the NIST Electronics and Electrical Engineering Laboratory. The National Institute of Standards and Technology National Institute of Standards and Technology, governmental agency within the U.S. Dept. of Commerce with the mission of "working with industry to develop and apply technology, measurements, and standards" in the national interest. is an agency of the Technology Administration, U.S. Department of Commerce. (1.) Certain commercial equipment, instruments, or materials are identified in this paper to foster understanding. Such identification does not imply recommendation or endorsement by the National Institute of Standards and Technology, nor does it imply that the materials or equipment identified are necessarily the best available for the purpose. 14. References (1.) S. E. Mechels, J. B. Schlager, and D. L. Franzen, Accurate measurements of the zerodispersion wavelength In a singlemode optical fiber, the zerodispersion wavelength is the wavelength or wavelengths at which material dispersion and waveguide dispersion cancel one another. In all silicabased optical fibers, minimum material dispersion occurs naturally at a wavelength of in optical fibers, J. Res. Natl. Inst. Stand. Technol. 102 (3), 333347 (1997). (2.) S. D. Dyer, K. B. Rochford, and A. H. Rose, Fast and accurate lowcoherence interferometric measurements of fiber Bragg grating dispersion and reflectance, Optics Express Optics Express is a peerreviewed rapidpublication scientific journal published by the Optical Society of America. It reports on all fields of optical science and technology. The journal is free. The costs are paid by the authors of the published articles. 5 (11), 262266 (1999). (3.) S. D. Dyer and K. B. Rochford, Lowcoherence interferometric measurements of fiber Bragg grating dispersion, Electron. Lett. 35, 14851486 (1999). (4.) Generic requirements for fiber optic branching components. GR1209CORE, Issue 2, Feb. 1998, Telcordia. (5.) R. M. Fortenberry, Enhanced wavelength resolution chromatic dispersion See dispersion. measurements using fixed sideband sideband, any frequency component of a modulated carrier wave other than the frequency of the carrier wave itself, i.e., any frequency added to the carrier as a result of modulation; sidebands carry the actual information while the carrier contributes none at all. technique, Optical Fiber Communication Conference OSA Technical Digest Digest: see Corpus Juris Civilis. (1) A compilation of all the traffic on a news group or mailing list. Digests can be daily or weekly. (2) Any compilation or summary. , Optical Society of America The Optical Society of America (OSA) is a scientific society dedicated to advancing the study of light—optics and photonics—in theory and application, by means of worldwide research, scientific publishing, conferences and exhibitions, partnership with industry, and the , Washington DC, 2000, TuG8, pp. 107109. (6.) A. M. Mood, F. A. Graybill, and D. C. Boes, Introduction to the Theory of Statistics, McGrawHill, New York New York, state, United States New York, Middle Atlantic state of the United States. It is bordered by Vermont, Massachusetts, Connecticut, and the Atlantic Ocean (E), New Jersey and Pennsylvania (S), Lakes Erie and Ontario and the Canadian province of (1974), p. 252. (7.) M. Abramowitz and I. A. Stegun, Handbook of Mathematical Functions In mathematics, several functions or groups of functions are important enough to deserve their own names. This is a listing of pointers to those articles which explain these functions in more detail. with Formulas, Graphs, and Mathematical Tables Before calculators were cheap and plentiful, people would use mathematical tables —lists of numbers showing the results of calculation with varying arguments— to simplify and drastically speed up computation. , National Bureau of Standards National Bureau of Standards: see National Institute of Standards and Technology. National Bureau of Standards  National Institute of Standards and Technology (1964), p. XII. 7. Appendix A. Telecom Group, Bragg Grating See fiber Bragg grating. Round Robin Procedure [1] Once you receive the package, open it up and splice the gratings into your measurement system. The Chirped grating has fiber numbers 1 and 2, the ITU grating has fiber numbers 3 and 4. For all measurements please record the input fiber number. If you make only one dispersion measurement of each grating, launch into fiber No. 1 for the chirped and fiber No. 4 for the ITU. [2] Apply power to the unit, enable the thermoelectric cooler (TEC) by pressing the green button, and allow about a 30 mm warmup period. The thermometer thermometer, instrument for measuring temperature. Galileo and Sanctorius devised thermometers consisting essentially of a bulb with a tubular projection, the open end of which was immersed in a liquid. should read 25.0[degrees]C. (If not call me at 3034975599.) [3] Begin your measurements: * ITU grating * 4 nm wide, high resolution ([less than]0.1 nm steps), scan over the reflectance/transmittance window; suggested range is 1550.5 nm to 1554.5 nm. * 40 nm wide, low resolution (0.1 nm to 0.2 nm steps), reflectance/transmittance scan; suggested range 1532.5 nm to 1572.5 nm. Chirped grating * 40 nm wide, low resolution, reflectance/transmittance scan, suggested range 1532.5 nm to 1572.5 nm. * Relative dispersion measurement over a 25 am wide scan. [4] Send your results by email, preferably pref·er·a·ble adj. More desirable or worthy than another; preferred: Coffee is preferable to tea, I think. pref as an attached file (my address is arose@boulder Boulder, city, United States Boulder, city (1990 pop. 83,312), seat of Boulder co., N central Colo.; inc. 1871. A Rocky Mountain resort and a suburb of Denver, it is the seat of the Univ. of Colorado (1876). .nist.gov). The data for each scan should be in ASCII ASCII or American Standard Code for Information Interchange, a set of codes used to represent letters, numbers, a few symbols, and control characters. Originally designed for teletype operations, it has found wide application in computers. format, two columns [first column wavelength (nm), second column voltage or optical power (dB) or relative dispersion (ps), optional third column temperature ([degrees]C)]. In your email message describe your measurement methods, such as stepped tunable laser and spectrum analyzer, etc. If the temperature is not listed in the third column of your data set, mention it in the text of your message. [5] Lock the box up and send it on to the next participant by Friday or the date indicated on the schedule. We have included an envelope with a Federal Express shipping label for you to affix affix v. 1) to attach something to real estate in a permanent way, including planting trees and shrubs, constructing a building, or adding to existing improvements. to the outside of the box. Please make sure you are sending it to the next participant. Please put the envelope back in the shipping box for the next participant to use. Thanks to all. 8. Appendix B. Sensors Group, Bragg Grating Round Robin Procedure [1] Once you receive the package, open it up and splice the gratings into your measurement system. The Sensor grating has fiber numbers 1 and 2, the ITU grating has fiber numbers 3 and 4. For all measurements please record the input fber number. [2] Apply power to the unit, enable the thermoelectric cooler (TEC) by pressing the green button, and allow about a 30 min warmup period. The thermometer should read 25.0 [degrees]C. (If not call me at 3034975599.) [3] Begin your measurements: ITU grating * 4 nm wide, high resolution ([less than]0.1 nm steps), scan over the reflectance/transmittance window; suggested range 1550.5 nm to 1554.5 nm. * 40 nm wide, low resolution (0.1 nm to 0.2 nm steps), reflectance/transmittance scan; suggested range 1532.5 nm to 1572.5 nm. Sensor grating * 4 nm wide, high resolution, reflectance/transmittance scan, suggested range 1550 nm to 1554 nm. * 40 nm wide, low resolution (0.1 nm to 0.2 nm steps), reflectance/transmittance scan; suggested range 1532 nm to 1572 nm. [4] Send your results by email, preferably as an attached file (my address is arose@boulder.nist.gov). The data for each scan should be in ASCII format, two columns [first column wavelength (nm), second column voltage or optical power (dB), optional third column temperature ([degrees]C)]. In your email message describe your measurement methods, such as stepped tunable laser and spectrum analyzer, etc. If the temperature is not listed in the third column of your data set, mention it in the text of your message. In your email message, please give the center wavelength, reflectance, and bandwidth values for the ITU and Sensor gratings. [5] Lock the box up and send it on to the next participant by Friday or the date indicated on the schedule. We have included an envelope with a Federal Express shipping label for you to affix to the outside of the box. Please make sure you are sending it to the next participant. Please put the envelope back in the shipping box for the next participant to use. Thanks to all. 11. Appendix E. NIST Uncertainty The uncertainty of various measurements will be discussed in this section. Because only raw data were given to NIST, only a fit uncertainty can be attributed to various parameters determined from a participant's data. However, NIST has determined the complete uncertainty of the various measurement systems it uses and those uncertainties are included in the uncertainties associated with parameters determined from NIST data. The NIST system used to measure the spectral properties of a grating, shown in Fig. 2, has the following attributes. The laser has a stability of about 80 MHz over a 10 s period, with a line width less than 300 kHz, and a 3 dB tuning range of 1525 nm to 1585 nm. The tunable fiber FabryPerot filter has a 3 dB bandwidth of about 1.5 GHz. The tunable filter attenuates the amplified spontaneous emission Amplified spontaneous emission (ASE) or superluminescence is light, produced by spontaneous emission, that has been optically amplified by the process of stimulated emission in a gain medium. from the diode laser as shown in Fig. 3. The peak transmittance of the filter is dithered in wavelength across the peak output of the laser, which modulates the laser's transmitted intensity. This modulated mod·u·late v. mod·u·lat·ed, mod·u·lat·ing, mod·u·lates v.tr. 1. To adjust or adapt to a certain proportion; regulate or temper. 2. laser intensity used with a lockin amplifier A lockin amplifier (also known as a phase sensitive detector) is a type of amplifier that can extract a signal with a known carrier wave from a noisy environment. It is essentially a homodyne with an extremely low pass filter (making it very narrow band). allows a dynamic range of about 68 dB. The wavelength meter in this system has a wavelength resolution of [+ or ]0.1 pm and an expanded uncertainty of 0.2 pm (k = 2). The wavelength meter uncertainty is periodically checked by comparison with a rubidium rubidium (rbĭd`ēəm), metallic chemical element; symbol Rb; at. no. 37; at. wt. 85.4678; m.p. 38.89°C;; b.p. 686°C;; sp. gr. 1.53 at 20°C;; valence +1. absorption cell. The NIST repeatability of determining the center wavelength and bandwidth of the round robin ITU grating, including the wavelength meter and fit uncertainty, was U = 6 pm (k = 2). We find that for a narrowband sensortype grating the repeatability is [+ or ]4 pm (two standard deviation estimate). Included in the fit uncertainty are the effects of the coupler transmittance versus wavelength on the reflectance of the grating. After further investigation, we found the method we used to step the laser wavelength caused the laser wavelength to drift during a measurement and the laser did not step evenly in wavelength. A new method was used and the repeatability of our measurements is now limited by the wavelength meter's uncertainty. The gratings are in an athermal package that has a 0.5 pm/[degrees]C temperature dependence, and the cooler has a temperature stability of about 0.01 [degrees]C, which should provide an estimated wavelength stability of 0.01 pm. The uncertainty of the absolute transmittance data is dominated by the uncertainty of the linearity and responsivity of the detectors used, which is about 1 % (two standard deviation estimate). We measured the linearity of the detectors over a 40 dB range of output voltages and found the linearity to vary less than [+ or ] 0.1 % (two standard deviation estimate). We measured minimum relative transmittance to a level of less than 60 dB. Assuming a 1 % uncertainty in the linearity of the detectorlockinamplifier system, we have an uncertainty at 60 dB of about 2.5 dB (two standard deviation estimate). The wideband fiber couplers used have a small wavelength dependence. Because a reference measurement is taken from the unused port of the 2 X 2 coupler, much of the laser's, and detector's wavelength dependence can be removed from the transmittance and reflectance signals. The uncertainty of the reflectance is dominated by the uncertainty of the coupling ratios and the detector's response. In this report only relative values are used because of differences in measurements systems. The coupler splitting ratio has a weak wavelength dependence of about 0.01 dB/nm over the 1540 nm to 1560 nm range. The NIST system used to measure the RGD of the gratings, shown in Fig. 5, has an uncertainty of about 0.3 ps (two standard deviation estimate). However, for longterm measurements, the system can have a phase drift due to temperature changes of the modulator. Care must be taken to hold the temperature constant during a measurement or an additional bias is added to the RGD values. With the 1.92 GHz modulation modulation, in communications modulation, in communications, process in which some characteristic of a wave (the carrier wave) is made to vary in accordance with an informationbearing signal wave (the modulating wave); demodulation is the process by which used, the wavelength uncertainty is about 30 pm (two standard deviaton estimate). A full uncertainty analysis of the whitelight interferometer shown in Fig. 6 is still in progress. 12. Appendix F. Determination of Wavelength Increments This Appendix describes the procedures we used to determine the appropriate data spacing for a transmission/reflectance response curve. Let [y.sub.1], [y.sub.2], ... [y.sub.n] (in dB) be the measured response in the "top flat" area of the transmission/reflectance curve, then the  x dB transmission/reflectance response [y.sub.x] is obtained as [y.sub.x] = max([y.sub.1], [y.sub.2], ... [y.sub.n])  x. (1) If there are no outlying out·ly·ing adj. Relatively distant or remote from a center or middle: outlying regions. outlying Adjective far away from the main area Adj. 1. data points, max([y.sub.1], [y.sub.2], ... [y.sub.n]) is the estimate of the "plateau" level of the curve. For smooth data sets, where the source wavelength dependence has been removed and noise is low, this assumption will hold. We can determine the proper sample size, hence the proper wavelength increment To add a number to another number. Incrementing a counter means adding 1 to its current value. , based on the desired precision of this plateau estimate. If we assume the [y.sub.1] are independent and equally probable to lie anywhere within the interval a and b (i.e., the maximum possible measurement error is b  a), then it can be shown [6] that the standard deviation (s) of [y.sub.x] is given by s([y.sub.x]) = [squre root]n/(n+2)[(n+1).sup.2] (b  a) [approximate] b  a/n + 2. (2) We can then equate e·quate v. e·quat·ed, e·quat·ing, e·quates v.tr. 1. To make equal or equivalent. 2. To reduce to a standard or an average; equalize. 3. this standard deviation to a threshold value to obtain the sample size required. For example, if we want to have an estimate of x dB transmission/reflectance response with a standard deviation less than onetenth of the maximum possible error (in the top flat portion), we need to have at least eight measurements in that area. Once we have a "good" estimate of the x dB transmission/reflectance response, the lower and upper x dB wavelengths can be calculated. We consider only the lower x dB wavelength here. Let [y.sup.] and [y.sup.+] be the first two consecutive measured responses such that [y.sup.] [less than or equal to] [y.sub.x] [less than or equal to] [y.sup.+]. The corresponding wavelengths for [y.sup.] and [y.sup.+] are [[lambda].sub.1] and [[lambda].sub.1] + h (h [greater than] 0), respectively. The lower x dB wavelength based on linear interpolation Linear interpolation is a method of curve fitting using linear polynomials. It is heavily employed in mathematics (particularly numerical analysis), and numerous applications including computer graphics. It is a simple form of interpolation. is given by [[lambda].sub.L] = [[lambda].sub.1] + [y.sub.x]  [y.sup.]/[y.sup.+]  [y.sup.] h. (3) The maximum error of [[lambda].sub.L] can be estimated by [7]. [delta][[lambda].sub.L] [approximate] [delta]y/dy/d[[lambda].sub.L], (4) where [delta]y is the maximum possible error in the transmission/reflectance measurements. An approximate value for dy/d[[lambda].sub.L] is ([y.sup.+]  [y.sup.])/h, or [delta][[lambda].sub.L] [approximate] [delta]y/[y.sup.+]  [y.sup.] h. (5) The bandwidth ([[[lambda].sup.+].sub.x]  [[[lambda].sup.].sub.x]) determined from the transmittance/reflectance response will be effected by h, [delta]y, and [y.sup.+]  [y.sup.] as seen in Eq. (5). For a desired maximum error [epsilon] in [delta][[lambda].sub.L] (and the bandwidth), a wavelength step size can be found according to according to prep. 1. As stated or indicated by; on the authority of: according to historians. 2. In keeping with: according to instructions. 3. h [less than or equal to] [epsilon]([y.sup.+]  [y.sup.])/[delta]y. (6) The result in Eq. (6) indicates that when the response curve is slowly varying in regions where [y.sub.x] is located ([y.sup.+]  [y.sup.] is small), or [delta]y is large, we need a smaller increment. When the response curve is slowly varying around [y.sub.x] ([y.sup.+]  [y.sup.] is small), then according to Eq. (6), h will need to be smaller than if the response curve had a larger difference between [y.sup.+]  [y.sup.]. Also, if [delta]y is large around [y.sub.x], then h will need to be smaller. 13. Appendix G. Robust Estimation estimation In mathematics, use of a function or formula to derive a solution or make a prediction. Unlike approximation, it has precise connotations. In statistics, for example, it connotes the careful selection and testing of a function called an estimator. of the x dB Wavelengths This Appendix describes the robust statistical method we used to determine the lower and upper x dB wavelengths. When there are outlying measurements, the lower and upper x dB wavelengths based on the x dB transmission/reflectance response [y.sub.x] calculated as [y.sub.x] = max([y.sub.i], i = 1, 2, ...) x (7) may be misleading. For example, the dotted vertical lines in Fig. 38 are the lower and upper 0.5 dB wavelengths. Obviously, the results reflect only the presence of the hump hump (hump) a rounded eminence. dowager's hump popular name for dorsal kyphosis caused by multiple wedge fractures of the thoracic vertebrae seen in osteoporosis. at the right side. Thus, we need a robust estimate of [y.sub.x] representing the plateau level of the transmission/reflectance curve. Let [y.sub.1], [y.sub.2], ... [y.sub.n] be the measured responses in the upper region of the transmission/reflectance curve. This can be accomplished by accepting only the responses that are greater than a cutoff value. For the example in Fig. 38, we could use a cutoff value, say, 6 dB. A particular cutoff value is not critical; any reasonable value will yield almost identical results because of the robustness of the procedure. One might use the mean y = [[[sigma].sup.n].sub.i=1] [y.sub.i]/n to estimate the plateau level of the curve. However, the mean is sensitive to outliers. We propose two alternatives. The first is the median of [y.sub.i]. The second is a statistic statistic, n a value or number that describes a series of quantitative observations or measures; a value calculated from a sample. statistic a numerical value calculated from a number of observations in order to summarize them. , called shorth, which is similar to the median (in robustness) but has a convenient geometrical ge·o·met·ric also ge·o·met·ri·cal adj. 1. a. Of or relating to geometry and its methods and principles. b. Increasing or decreasing in a geometric progression. 2. interpretation. The shorth of [y.sub.i], i = 1, 2, ... n, is the midpoint mid·point n. 1. Mathematics The point of a line segment or curvilinear arc that divides it into two parts of the same length. 2. A position midway between two extremes. of the shortest interval that includes half of [y.sub.i]. This is done by finding the smallest of the values [[y.sup.*].sub.k]  [[y.sup.*].sub.1], [[y.sup.*].sub.k+1]  [[y.sup.*].sub.2],..., [[y.sup.*].sub.n]  [[y.sup.*].sub.nk+1], where k = [n/2] + 1, [p] is the integer part of p, and [[y.sup.*].sub.1] [less than or equal to] [[y.sup.*].sub.2] [less than or equal to] ... [less than or equal to] [[y.sup.*].sub.n] are the ordered measurements of [y.sub.i]. Then, the shorth simply equals the midpoint of this shortest interval. For example, let the ordered measurements of [y.sub.i], i = 1, 2, ..., 100, be 1 3 4 7 8 14 15 16 17 27 100. Then k = [11/2] + 1 = 6 and the six intervals that include half of the measurements are (1, 14), (3, 15), (4, 16), (7, 17), (8, 27), (14, 100). The shortest interval is (7, 17) and the shorth = (17 + 7)/2 = 12. Note that the median of the above eleven measurements is 14, while the mean is 19.3 (skewed skewed curve of a usually unimodal distribution with one tail drawn out more than the other and the median will lie above or below the mean. skewed Epidemiology adjective Referring to an asymmetrical distribution of a population or of data by a single measurement). If we fit a horizontal line (Descriptive Geometry & Drawing) a constructive line, either drawn or imagined, which passes through the point of sight, and is the chief line in the projection upon which all verticals are fixed, and upon which all vanishing points are found. See also: Horizontal to [y.sub.i], i = 1, 2, ..., n, the mean of [y.sub.i] is the line that minimizes the sum of the squared residuals (differences between the predicted and measured [y.sub.i]). The shorth of [y.sub.i] is the line that minimizes the median of the squared residuals. The median is not affected by the values of the outlying residuals and will not change unless more than half the residuals represent bad or spurious spu·ri·ous adj. Similar in appearance or symptoms but unrelated in morphology or pathology; false. spurious simulated; not genuine; false. measurements. In summary, the shorth is a robust estimate of the plateau level of the transmission/reflectance curve. Figure 39 displays the estimated plateau of the transmission/reflectance curve based on the mean (solid horizontal line) and the shorth (dashed horizontal line) of [y.sub.i]. It also shows the 0.5 dB wavelengths based on the shorth (dashed vertical lines) and the mean (solid vertical lines). [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] [Graph omitted] 

Reader Opinion