only a subset of the RTCM/SC-104 messages, and some follow the older Version 2.0 of Reference T10 rather than Version 2.1, upon which this guidance was developed. Some manufacturers have also implemented proprietary DGPS messages which frequently bear a close resemblance to the RTCM/SC-104 messages.
For applicants implementing a real-time DGPS data link, it is preferred that RTCM/SC-104 messages be employed for this purpose. Type-1 or Type-9 messages, each of which contains the actual DGPS corrections, should be selected and transmitted at a rate of 0.5 Hz or higher. Other message types (e.g., Type-3 ground-station location and Type-5 satellite health) may be used but should be sent at a rate of once per minute or slower. There is no recommended requirement for storing real-time DGPS correction data. The da ta status or quality flag (see 302(e)(2)(C)(i)) should however provide an indication that the correction data have been properly received and processed by the aircraft.
302(e)(2)(C)(iv) Messages for multi-path testing [ETM 3.2.5.2.3.4] Applicant-designed systems using code-based DGPS processi ng should collect and save data from dedicated multipath tests to be conducted prior to aircraft noise testin g (see 302(e)(2)(E)). Data co llected during multi-path tests should include individual satellite pseudo-ranges and signal-to -noise ratios. These parameters are provided only by receiver manufacturers’ proprietary messages. It is not necessary for applicants to conduct a dedicated test for systems using carrier-based DGPS processing.
302(e)(2)(D) System accuracy and sources of DGPS error [ETM 3.2.5.2.4] If only divergence (spherical spreading) of the noise is considered, and atmospheri c absorption mechanisms are ignored, then a 0.1 dB change in the noise level corresponds to a change of approximately 1.1 percent of the distance between the aircraft noise source and th e measurement microphone. Thus, for an aircraft altitude of 400 ft (122 m), the approximate minimum altitude during noise certification tests, a position error of 4.3 ft (1.3 m) along the line-ofsight vector connecting the microphone and aircraft can be expected to introduce a 0.1 dB error in the processed noise data. A position error of 10 ft (3.0 m) along the line-of-sight vector can be expected to introduce a 0.23 dB error in the processed noise data.
For most DGPS systems, the most important error sources are, in decreasing order of importance, multi-path, correction latency and tropospheric delay. When these error sources are properly controlled, DGPS systems can provide accuracies between a few centimeters and approximately 15.1 ft (4.6 m) for an aircraft in low-dynamics flight regimes. Even the poorest of these accuracies is superior to that achieved by other conventional TSPI systems used for aircraft noise tests, including micr owave and photo-scaling. The best accuraci es are superior to those of a laser tracker.
DGPS systems suitable for consideration by noise certification applicants can be expected to achieve an accuracy of a few inches to 5 ft (1.5 m). The highest accuracy is achieved using carrier-based techniques and post-flight processing of data collected from both the aircraft and ground-sta tion computers. Code-based solutions which use carrier smoothing (e.g., Novatel RT-20) achieve accuracies of 3.0 ft to 4.9 ft (0.9 m to 1.5 m), provided that the error sources discussed in this section are addressed properly. Consequently it is expected that the DGPS systems used for noise certification tests will introduce less than a 0.2 dB error into the noise data in a worst-case scenario (i.e., a noise certification approach measurement). Typical errors will be less than 0.1 dB for noise certification flyover and lateral measurements.
In addition to the three error sources cited above, increases in DGPS position errors can also occur when the manufacturer and model of the ground station and aircraft GPS receivers are not the same, or when the ground station and aircraft receivers use different satellite ephemeris/clock data to specify the satellite orbital parameters.
Sections 302(e)(2)(E)and 302(e)(2)(F) address all of the above errors and include methods for minimizing these errors or eliminating them entirely.
302(e)(2)(E) Multi-path errors [ETM 3.2.5.2.5] 302(e)(2)(E)(i) Characteristics [ETM 3.2.5.2.5.1] Multi-path refers to signals from GPS satellites which are reflected from objects (e.g., the ground, buildings and aircraft structural elements) before reaching the GPS antenna. Multi-path signals add algebraically to the desired lineof-sight signal and thereby decrease the accuracy of meas urements made with the latter. Multi-path conditions can occur independently at the aircraft and ground station antennas. Thus the differential correction data from the ground are not directly useful for correction for multi-path errors at the aircraft antenna. Rather, the broadcast corrections can contain ground-station multi-path errors which, in a statistical sense, add to those in the aircraft.
Measurements have consistently shown that the presence of multi-path conditions at the ground station is significantly more deleterious than at the aircraft. This is because ground-station multi-path conditions vary slowly, acting like a bias over a test run of a few minutes, whereas the more dyna mic motion of the aircraft causes the effects of airborne multi-path conditions to behave like noise, which can be reduced somewhat by processing techniques such as filtering and averaging.
For code-based processing, ground station multi-path error is typically between 1.0 ft and 10 ft (0.3 m and 3 m). Under very adverse conditions (e.g., GPS antenna near the side of and well below the top of a large building) multi-path errors can be several hundred meters. Multi-path errors associated with carrier-based processing techniques are significantly less than those for code-based methods and are usually of the order of centimeters.
The extent of the multi-path error primarily depends on two factors: the capability of the ground station antenna and the location of the ground station antenna relative to reflec ting objects such as paved runways, buildings and parked aircraft. Receiver processing (e.g., the us e of narrow correlators available in most Novatel receivers) and/or carrier smoothing, available from several manufacturers, can reduce multi-path errors.
302(e)(2)(E)(ii) Code-based system ground station [ETM 3.2.5.2.5.2] To mitigate the effects of multi-path conditions on DGPS-based TSPI performance, the applicant’s ground station installation should meet the following requirements:
a) the ground station should employ a multi-path-limiting antenna, such as one with a choke ring or an absorbing ground plane; and
b) the ground station antenna should be mounted on a pole or tower, with unobstructed visibility of the sky.
A minimum height of 10 ft (3 m) above ground level is recommended for the ground-station antenna.
Additionally, to ensure that significant undetected multi-path errors do not corrupt the TSPI data collected during aircraft noise testing, the applicant’s ground-station inst allation should be tested for adequate multi-path condition performance prior to commencing the flight test. This can be done by collecting GPS receiver data during the same hours of the day that the system will be used for noise tests, with additional one-hour buffers on either side of this period. The data are then examined on a per-satellite basis, rather than navigation solution basis, for multi-path signatures. This examination should include at least pseudo-ranges and signal-to-noise ratios.
If multiple periods of significant multi-path errors (i.e., several feet) are found, then a new location for the groundstation antenna should be selected and te sted. If only one or two isolated, brief multi-path incidents are found, then the antenna location can be retained but aircraft testing should not be conducted during these periods.
Note.— The satellite-user geometry repeats over a cycle of approximately 23 hours 56 minutes. Thus if a ground Page 74
station multi-path incident is observed one day, it is expected that a similar incident will occur 4 minutes earlier the following day.
These procedures are similar to those utilized by the United States Coast Guard in checking out a marine DGPS station installation (References T10 and T19).8 After establishing a ground-station antenna site/configuration that satisfies the multi-path conditions criterion, the ground-station antenna should not be moved without performing another multipath test. The ground-station GPS receiver and any computer used in conjunction with the receiver may be removed and reinstalled without repeating the multi-path test. The multi-path verification test data should be saved as part of the permanent test-series data archive and should be made available for inspection by the FAA.
302(e)(2)(E)(iii) Carrier-bas ed system ground station [ETM 3.2.5.2.5.3] To mitigate the effects of multi-path conditions on DGPS-b ased TSPI performance, the applicant’s ground-station installation should meet the following recommended specifications:
a) the ground station should employ a multi-path-limiting antenna, such as one with a choke ring or an absorbing ground plane; and
b) the ground station antenna should be mounted on a pole or tower, with unobstructed visibility of the sky. A minimum height of (9.8 ft (3.0 m) above ground level is recommended for the ground-station antenna.
There is no recommended requirement for collecting data to assess multi-path errors when carrier-based processing is employed.
302(e)(2)(E)(iv) Ai rcraft installation [ETM 3.2.5.2.5.4] It is expected that aircraft manufacturers will select a location on each aircraft model that minimizes multi-path effects.
In this regard no recommended specifications have been developed. For most smaller aircraft (e.g., 10 seats or fewer), it has been found that the roof area directly behind the windshield is most advantageous. Manufacturers of larger aircraft have found forward positions on the roof to be desirable, although some have mounted the GPS antenna on the tail structure. Selecting a location for the GPS antenna on a helicopter may be more challenging since the main rotor will momentarily obscure most areas on the airframe.
302(e)(2)(F) Other sources of DGPS error [ETM 3.2.5.2.6] 302(e)(2)(F)(i) Correction latency [ETM 3.2.5.2.6.1] Correction latency, also called staleness, refers to the delay between the time of validity of a differential correction at the ground station and the time that the correction is applied in the aircraft. Delays in pro cessing at both ends of the ground-to-air data link can cause stale corrections to introduce unacceptably large errors.
A second form of latency, solution latency, refers to the delay between the time at which a GPS receiver’s measurement is valid and the time when it is available at the output of the receiver. Solution delays are inherently smaller than correction
8. Coast Guard DGPS ground stations employ two GPS receiver/antenna pairs. The “additiona l” receiver/antenna pair (termed the i ntegrity monitor) provide a real-time continuous ch eck on the validity of the differential corrections generated by the “basic” receiver /antenna pair (termed the reference station). DGPS ground-station architectures be ing investigated for the FAA LAAS program employ between 2 and 4 receiver/antenna pairs to verify the corrections sent to the aircraft. No requirement for redundant ground-station equipment is recommended for DGPS-based TSPI systems used in noise certification tests.
delays and, in this context, are of concern only for aircraft guidance.
For a system with a real-time data link which employs code-based DGPS solutions, it is strongly recommended that ground-to-aircraft messages conform to the RTCM/SC-104 standards used by the Coast Guard DGPS system.9 These messages contain pseudo-range rates-of-change, as well as th e correction at an identified time, to allow the user to correct for most of the latency-induced er ror. It is also preferred that the corr ections be computed and transmitted at least at a 0.5 Hz rate.
302(e)(2)(F)(ii) Tropospheric delay [ETM 3.2.5.2.6.2] The troposphere is that portion of the atmosphere between the earth’s surface and an altitude of approximately 20 miles (32 km). Differences in meteorological conditions between the ground station and the aircraft can cause dissimilar changes in the propagation times of signals from a satellite to these two locations. The effect is most pronounced for low-elevation-angle satellites. Since these changes are no t common to the two locations, they are not removed by differential corrections. Such tropospheric effects can contribute up to 66 ft (20 m) of ranging error on GPS signals, which can translate into as much as 33 ft to 39 ft (10 m to 12 m) of positioning error if not modeled and corrected. In differential mode this positioning error is typically less than 6.6 ft (2 m). Approximately 90 percent of these tropospheric propagation-related errors are due to the hydrostatic, or dry, component of tropospheric delay.
Experiments performed for the FAA LAAS program have found tropospheric differences to introduce DGPS errors between 1 ft and 3 ft (0.3 m and 0.9 m) when the aircraft was at 3000 ft (914 m) altitude, but only a few centimeters when the receiver antennas were at the same altitude. To reduce the effects of tropos pheric errors on DGPS-based TSPI systems used in noise certification tests, it is recommended that use of these systems be limited to the aircraft being within a lateral distance of 20 nm (37 km) and a height of 5000 ft (1524 m) relative to the ground station.
If desired, the hydrostatic component of the tropospheric delay can be effectively removed with the application of the tropospheric delay model (Reference T9) developed by the Radio Technical Commission for Aeronautics (RTCA) as per ICAO Annex 10 Standards and Recommended Practices, al ong with local meteorological measurements at the ground station. The relevant portion of this model is driven by local barometric pressure and satellite geometry (i.e., elevation angle). Reference T18 provides a functional overview of the RTCA model, as well as comparisons with other tropospheric propagation delay models.
302(e)(2)(F)(iii) Mismatched GPS receivers [ETM 3.2.5.2.6.3] Experiments have shown that DGPS errors are increased when the GPS recei vers at the ground station and in the aircraft are not “matched” in terms of manufacturer and model. With mismatched receivers, errors are increased moderately (e.g., 1.5 to 3 times) compared to those when the receivers are matched and when the satellites are operating normally. When a rare, soft satellite-failure, or signa l degradation, occurs, errors of several thousand feet have been observed.10 It is required that applicants’ systems have the same manufacturer/model GPS receiver on the ground and in the aircraft.
302(e)(2)(F)(iv) Mismatched sa tellite ephemeris/clock data [ETM 3.2.5.2.6.4]
9. The United States Coast Guard DGPS system’s broadcast messages (as well as marine systems of other nations) include the rate-of-change of each pseudo-range error, in addition to the pseudo-range error at a reference time. The user’s receiver is required to apply an adjusted correction consisting of the broadcast pseudo-range error, plus its rate-of-change multiplied by the time elapsed between the time the adjusted correction is applied, and the validity time for the pseudo-range correction.
10. Beginning on or before 21 October 1993, some differential users with mismatched ground and aircraft receivers experienced position errors of thousands of feet. The GPS Joint Program O ffice (JPO) of the United States Department of Defense attr ibuted the cause to a “def iciency” in Coarse/Acquisition code broadcast by satellite SVN19. It announced that the problem was corrected on 10 January 1994. Official statements are found in Notice Advisory to NAVSTAR User (NANU) 343-93294, 396-93337 and 006-94010.
GPS satellite broadcasts include a navigation message in the form of 50 bits per second modulation superimposed on the pseudo-random codes used for ranging. Within the navigation message are data sets that describe the satellite orbit
(i.e., ephemeris information) and clock. These data sets are transmitted every 30 seconds. The GPS Control Segment uploads multiple ephemeris and clock data sets to the satellites, typically once per day.
Note.— The Control Segment is the ground-based portion of the total GPS system. It includes the Operational Control facility in Colorado Springs, Colorado, United States, where the satellite ephemeris and clock data are calculated, five worldwide sites which collect satellite broadcast signals and provide data to the operational control facility, and three locations from which new ephemeris/clock data are uploaded to the satellites.
Satellites typically change their broadcast ephemeris and clock message every four hours. The ephemeris/clock data sets are used by a receiver to compute its own position and, in the case of a reference station, differential corrections for use by other receivers. For a DGPS system to achieve full accuracy, both the ground station and aircraft receiver must use the same ephemeris and clock data sets. Internal receiver logic ensures that the ephemeris and clock data sets used by a given receiver are consistent for each satellite. However, occasionally the ground and aircraft receivers may use different ephemeris/clock data sets unless measures are taken by the user to ensure that the sets match. Mismatched ephemeris/clock data sets can occur for several reasons, e.g., a receiver is too busy performing other tasks when the data sets change, or a receiver encounters an error while decoding new data and continues to use an old data set.
The RTCM/SC-104 messages used by the Coast Guard DGPS system guard against mismatched ephemeris/clock data sets by including the issue of data (I OD), an eight-bit data set label broa dcast by each satellite, in the broadcast messages (References T10 and T19). User receivers which conform to the RTCM/SC-104 standards will not apply differential corrections unless the IOD from the satellite and the DGPS correction message agree. The applicant should ensure that the ground station and aircra ft use the same ephemeris and clock data sets during testing. One way is to purchase GPS receivers and select DG PS messages which cause this check to be performed automatically. Another way to ensure agreement between the ground and aircraft ephemeris/clock data sets is to store in a permanent file for record keeping, at a rate of once each 30 seconds, the IOD used by each receiver and compare the IODs during posttest processing.
302(e)(3) System approval recommendations [ETM 3.2.5.3] This section summarizes approval recommendations for DGPS-based TSPI systems proposed for use during noise certification tests.
302(e)(3)(A) Design issues [ETM 3.2.5.3.1] Each applicant’s TSPI system design should address the issues identified in Table 4. The applicant’s documentation (see 302(e)(3)(C)) should address each item in the table.
Table 4. TPSI systems design development issues Number Issue Major considerations 1 Selection of processing method (real-time versus post-test) Need for aircraft guidance, ability to check testrun quality 2 Selection of solution method (carrier versus code) Accuracy (favors carrier), robustness (favors code), cost (favors code) 3 Use of geodetic or waypoint coordinates Waypoints can simplify post-processing but are not available for all receivers 4 Selection of GPS receiver and antenna Items 1, 2, 3 and others (antenna multi-path control, data messages, solution latency, matched air/ground receivers, and IOD capability) 5 Selection of data link equipment (if real-time system) Assigned frequency, data rate, error detection/ correction, flexible interface 302(e)(3)(B) Data storage (logging) during noise testing [ETM 3.2.5.3.2] 302(e)(3)(B)(i) For a system with real-time data link [ETM 3.2.5.3.2.1] For applicants employing a real-time data link, the ground-station GPS receiver should output RTCM/SC-104 Type- 1 messages at a rate of 0.5 Hz or great er, which should be transmitted to and us ed by the aircraft GPS receiver. The applicant’s aircraft computer should collect data from the aircraft GPS receiver and generate permanent data files containing:
a) the three-dimensional aircraft pos ition copied directly from the receiver’s data port (i.e., in raw/native form) and not processed;
b) if waypoint navigation is used, the waypoints (i.e., latitude, longitude and altitude) used to define the local coordinate frame;
c) the time (e.g., UTC or GPS time), with or without a local offset, associated with each sample of position data copied directly from the receiver’s data port; and
d) the data quality/validity indication associated with each sample of position data.
If waypoints are used they should be included in the header of each data file. New waypoints should not be able to overwrite existing waypoints. If new waypoints are defined then a new data file should be created.
For consistency with the noise data collected during a certif ication test, it is recommended that the data associated with a), c) and d) above be saved in th e GPS receiver’s raw/native format at a rate greater or equal to 2 Hz, the rate associated with the noise data. However, if hardware limitations preclude following this recommendation, a sampling rate of 0.5 Hz or greater is acceptable.
302(e)(3)(B)(ii) For a system not using real-time data link [ETM 3.2.5.3.2.2] Page 78
TSPI systems which do not use a real-time data link should save data from both the ground and aircraft GPS receivers in raw/native format in a permanent file for record keep ing. Manufacturers’ proprietary formats should be used.
NMEA standard messages do not support this application.
Stored data should include: time (e.g., UTC or GPS time) with or without a local offset, satellite ephemeris, pseudorange, signal-to-noise ratio, and carrier phase. If tropospheric delay is being modeled, as described in 302(e)(2)(f) then local meteorological conditions should be measured and stored as well. It is recommended that applicants using dualfrequency (L1/L2) receivers also save L2 carrier phase data. Typically, post-processing of the ground-based and airborne GPS data will be performed using manufacturer-supplied software. If this is not the case then any applicantdeveloped software should be approved by the FAA.
302(e)(3)(C) Documentation [ETM 3.2.5.3.3] The applicant should prepare and submit documentation which includes:
a) System description. Identifies, at a minimum, the issues in Table 4.
b) Hardware description. Model and version number of all system components, including DGPS receivers, antennas, transceivers and computer.
c) Software description. Software functionality and capabilities, da ta file formats, hardware required and operating system.
d) System setup and operation. Ground and aircraft installation of the system including antennas, operating procedures, site survey procedures, power requirements and system limitations.
e) Validating of the installation. A method often used is to park the aircraft at a known surveyed location and to read its position from the DGPS system. From a comparison of the DGPS and surveyed positions the installation can be verified. This can be performed either at the test site or at another location, such as the aircraft home base. As a minimum this process should be performed at the start and end of each measurement program and preferably at the beginning and end of each measurement day.
302(e)(3)(D) Accuracy verification test [ETM 3.2.5.3.4] The applicant should perform a one-time verification of the system accuracy, based on a minimum of six aircraft flight-test runs which encompass the conditions (i.e., speed, altitude, range and maneuvers) for which the system will be later used as a reference. The accuracy verification te st should involve a comparis on of the DGPS-based TSPI system’s position data with those from an accepted referen ce, such as a laser tracker or another approved DGPS system. This test should be performed on the complete DGPS-based TSPI system developed by the applicant. It is not adequate for an applicant seeking system approval to simply cite prior approval of another applicant’s system designed around the same GPS receiver.
302(e)(3)(E) Softwa re verification [ETM 3.2.5.3.5] Prior to using the system during a noise measurement program, any applicant-developed software for data logging and post-processing used to obtain data listed herein should be approved by the FAA. The approved software should be placed under version management.
302(e)(3)(F) Ground-station multi-path mitigation and verification [ETM 3.2.5.3.6] Page 79
303 302(e)(3)(F)(i) All systems [ETM 3.2.5.6.1] The ground-station GPS receiver antenna should have a ch oke ring, absorbing ground plane, or other multi-path - reducing technique. The antenna should be positioned on a pole or tower at a minimum height of 10 feet above ground level.
302(e)(3)(F)(ii) Code-based systems [ETM 3.2.5.3.6.2] Prior to each measurement program, applicants using code-based DGPS systems should perform a multi-path investigation using the ground-station receiver and antenn a, as described in 302(e)(2 )(E)(ii). The results of the investigation should be saved as part of the permanent te st-series data archive and be made available for inspection by the FAA.
302(e)(3)(G) Airport survey [ETM 3.2.5.3.7] Additional information on survey requirements may be found in 302(e)(2)(B). Prior to and after completion of each measurement program, the applicant should use the DGPS-based TSPI system to survey the locations of:
a) if no other method of survey is used, all microphones and, if used, waypoints; or
b) if another method of survey is used, a recommended minimum of at least three points common to both methods.
Survey data should be stored as part of the measurement-program permanent archive. If two survey methods are used, the common points should be reconciled to an accuracy of 1 ft (0.3 m) and the adjustment procedure submitted to the FAA for approval.
ON-BOARD FLIGHT DATA ACQUISITION [ETM 3.3] 303(a) General [ETM 3.3.1] It is necessary to obtain the values of a variety of flight and engine parameters during the noise measurement period in order to:
a) determine the acceptability of noi se certification flight tests;
b) obtain data to adjust noise data; and
c) to synchronize flight, engine and noise data.
Typical parameters would include airsp eed, climb angle, height/altitude, gro ss weight, flap position, landing gear position, jet-engine thrust (power) setting parameters (e.g., compressor rotor speed, engine pressure ratio and exhaust gas temperature), helicopter rotor speed, engine torque and propeller rotational speed.
A number of methods for collecting this information have been employed:
a) manual recording;
b) magnetic tape recording;
c) digital recording;
d) automatic still photographic recording;
e) cine recording; and
f) video recording.
Clearly, when a large number of parameters have to be co llected at relatively short time intervals, it may not be practicable to manually record the data. Thus the use of one of the automatic systems listed in b) to f) becomes more appropriate. The choice of a particular system may be influenced by a number of factors such as the space available, cost and availability of equipment.
For systems which optically record the flight deck instruments, care must be taken to avoid strong lighting contrast such as would be caused by sunlight, deep shadow and reflections from the glass fronts of instruments, which would make data unreadable. To avoid this, it may be necessary to provide additiona l lighting to “fill in ” the deep shadow regions. To prevent reflections from the front of instruments, it is recommended that light-colored equipment or clothing on the flight deck be avoided. Flight crews should be required to wear black or dark-colored clothing and gloves.
Furthermore, for systems which record the readings of dials it is important that the recording device be as near as possible, directly in front of the instruments to avoid parallax errors.
303(b) Magnetic Tape Recording [ETM 3.3.2] Multi-channel instrumentation tape recorders designed fo r airborne environments are employed for continuous recording of flight and engine performance parameters. Typical recorders are compact intermediate/wide band and can take both one-half-inch and one-inch magnetic tapes with a 24 to 28 volt DC power requirement. Six tape speeds as well as direct and FM recording are available in a tape recorder weighing about 60 lb (27) kg.
303(c) Automatic Still Photographic Recording [ETM 3.3.3] Photographs of the flight deck instrument panel can be taken by using a hand-held 35 mm single-lens reflex (SLR) camera with an 85 mm lens and high-speed slide film. The indications on the instruments can be read by projecting the slides onto a screen.
303(d) Cine Recording [ETM 3.3.4] Cine cameras with a one frame per second exposure rate have been used to acquire flight deck data. Care must be taken in mounting the camera to ensure that all the instruments that have to be photographed are within the field of view. Typical film cassettes containing about 2000 frames have been used with a frame counter to allow film changes to be anticipated.
303(e) Video recording [ETM 3.3.5] Flight and engine performance parameters can be recorded with a video camera, although as with cine cameras, care must be taken to ensure that all the instruments that have to be photographed are within the field of view. The recorded information is played back using freeze-frame features to obtain individual instrument readings.
TIME SYNCHRONIZATION OF MEASURED DATA [ETM 3.4] 304(a) General [ETM 3.4.1] Sections A36.2.3.2 and H36.101(d)(3) of part 36 specifies that there be precise time synchronization between noise measurements and aircraft position. Several methods have been used, such as noting the synchronization time on a clock mounted on the instrument panel which itself is recorded by the data acquisition system. One such system uses a ground camera that operates a radio transmission which, when received by an aircraft, lights two high-intensity light emitting diodes (LEDs) mounted in an analog clock attached to the instrument panel. Other methods for acquiring and processing TSPI are described in subsequent sections.
A common time base should be used to synchronize noise, aircraft tracking and meteorological measurements. TSPI data should be determined at half-second intervals throughout the sound-measuring period (i.e., within 10 dB of maximum tone corrected perceived noise level (PNLTM)) by an approved method that is independent from systems installed aboard and normally used to control the aircraft. During processing, measured TSPI data should be interpolated, over time, to the time of sound emission of each half-second acoustic data record within the 10 dB-down period. Although the simplified procedure requires adjustment of only the PNLTM record to the reference flight path, sound emission coordinates should be determined for each half-second record for use in background noise adjustment procedures and for determination of incidence-dependent free-field microphone and windscreen corrections.
304(b) TSPI Equipment and Software Approval [ETM 3.4.2] Some off-the-shelf TSPI equipment may require software enhancement to accommodate the specific installation. All TSPI equipment and software should be demonstrated to, and approved by, the FAA to ensure the system’s operational accuracy.
304(c) Continuous Time-Dode Recording [ETM 3.4.3] This method uses a time-code signal, such as IRIG B, which is a modulated audio-frequency signal used for encoding time-base data developed by the Inter-Range Instrumentation Group (IRIG). In this method, the time-code signals from individual generators that have been synchronized to a common time-base are continuously recorded by both the noise data recorder(s) and by the TSPI system during measurement test runs. Synchronization of multiple generators can be performed either physically, by interconnecting via cable, or by means of radio transmission. The transmitted continuous time-code signal can be recorded directly, or used either continuously or in bursts to maintain synchronization of an independent time-code generator which is being recorded directly. This method allows for high-quality continuous timecode recording when there are intermittent reception problems.
Note 1.— Synchronization should be accomplished at the start of each measurement day and checked at the end of each measurement day to minimize the effects of generator time drift. Any such drift should be documented and accounted for in processing.
Note 2.— GPS-based measurement systems are often used for acquisition of TSPI data. GPS receivers are capable of providing the user with precise time-base information broadcast from the GPS satellite system, in some cases eliminating the need for a separate time-keeping device in the TSPI system.
Note 3.— For noise data recording or for non-GPS-based TSPI systems, dedicated IRIG B time-code generators are available that use the GPS signal to constantly update and maintain time synchronization. Use of such a universal broadcast time-base can greatly simplify the logistics of time synchronization between measurement systems.
Note 4.— There are two available time-bases for GPS-based systems: GPS Time and Coordinated Universal Time Page 82
(UTC), whose values differ by more than 10 seconds at any given instant. Although the GPS signal includes both timebases, not all GPS receivers give the user access to both. Therefore, the user should exercise caution in identifying which time-base is used by each instrument.
Note 5.— Many acoustical data recorders provide separate annotation channels in addition to the normal data channels. These channels are often not suitable for recording a modulated time-code signal because of limitations on dynamic range or bandwidth. In such cases a normal data channel of the recorder should be dedicated to recording the time-code signal.
Note 6.— When continuous time-code recording is used, a nalysis of the recorded acoustic data can be initiated by routing the time-code channel output into a time-code reader and triggering the analyzer based on readout time.
304(d) Recording of Single Time Marker [ETM 3.4.4] This method involves transmittal and recording of a radio “hack”, or tone, usually used to indicate the “recorders on” or “overhead” time instant. This method typically requires a dedicated channel on both the noise and the TSPI recording systems. When such a system is used, analysis can be triggered manually by an operator listening for the hack, or by a detector circuit responding to the tone. When the operator wishes to start analysis at a time other than that of the time marker, a stopwatch or delay circuit can be used to delay triggering of the analyzer. When manual triggering is employed, the operator sh ould use extreme care to perform the tr iggering as accurately as possible.
Accuracy to within one-tenth of a second can be expected from a conscientious human operator.
304(e) Measurement of th e Interval Between Recorder Start and Overhead [ETM 3.4.5] This method of synchronization involves use of a stopwatch or elapsed-time indicator to measure the interval between start-up of the noise data recorder and the instant that the aircraft position is overhead the centerline noise measurement point. This method can be employed successfully as long as the operator exercises care in timing, the determination of the overhead instant is performed accurately, and the star t-up characteristics of the r ecorder (in both record and playback modes) are known and repeatab le. Some recorders have variable star t-up times that cannot be predicted.
Such recorders are not suitable for this method of synchronization.
304(f) Setting of Internal Time-Stamp Clock [ETM 3.4.6] Many digital recorders maintain a continuous internal time-of-day function by encoding time data in the recorded data stream. This method uses a digital recorder’s sub-code time, synchronized to the time-base used for the TSPI data. As with the continuous time-code recording method, synchronization by this method should be checked at the beginning and end of each measurement day and any drift accounted for in processing.
Unfortunately, the time-setting function on many recorders does not provide for the necessary precision. The “second” digits cannot be made to “tick” in synchrony with an external clock. Such recorders are unsuitable for this method of synchronization.
304(g) Additional Time-Synchronization Considerations [ETM 3.4.7] Regardless of the synchronization method used, all elements affecting time synchronization, such as analyzer start-up delay, head displacement between normal and annotation data channels on analog recorders, and delays in automated triggering circuits, should be identified, quantified and accounted for in analysis and processing. Whenever human response to a timing event is required, errors cannot be accurately predicted, and conscientious operation is required to minimize such errors. The use of automated methods is preferred. Other methods, or variants of the listed methods, may be appropriate, but the use of all methods and instrumentation is subject to prior approval by the FAA.
CALCULATION OF CONFIDENCE INTERVALS [ETM 3.5] 305(a) Introduction [ETM 3.5.1] The use of NPD curves requires that confidence intervals be determined by using a more general formulation than is used for a cluster of data points. For this more general case, confidence intervals may have to be calculated about a regression line for:
a) flight test data;
b) a combination of flight test and static test data;
c) analytical results; or
d) a combination thereof.
Items b) and c) are of particular significance for noise certification of an aircraft model range and require special care when pooling the different sources of sampling variability.
Sections 305(b) to 305(e) provide an insight into the theory of confidence interval evaluation. Application of this theory and some worked examples are provided in 305(f). A suggested bibliography is provided in Appendix 3 to this AC for those wishing to gain a greater understanding.
305(b) Confidence Int erval for the Mean of Flight Test Data [ETM 3.5.2] 305(b)(1) Confidence Interval fo r the Sample Estimate of the Mean of Clustered Measurements [ETM 3.5.2.1] If n measurements of EPNLs y1, y2,….,yn are obtained under approximately the same conditions and it can be assumed that they constitute a random sample from a normal population with true population mean, , and true standard deviation, σ, then the following statistics can be derived:
ӯ = estimate of the mean s = estimate of the standard deviation of the mean = From these and the Student’s t-distribution, the confidence interval, CI, for the estimate of the mean, ӯ, can be determined as:
௦ ܫܥ ൌ ӯ േ ݐሺଵିഀ,ఋሻ √,, మ Where ݐሺଵିഀ,ఋሻ denotes the ሺ1 െ ఈଶሻ percentile of the single-sided Student’s t-test with ߜ degrees of freedom (for a మ clustered data set ߜ ൌ ݊ െ 1 and where ߙis defined such that 100ሺ1 െ ߙሻ percent is the desired confidence level for the confidence interval. In other words it denotes the probability with which the interval will contain the unknown mean, . For noise certification purposes, 90 percent confidence intervals are generally desired and thus f ݐଽହ,ఋ is used (see Table 5 for a listing of values of ݐଽହ,ఋ for different values of ߜ .)