a band-by-band basis using separate gain stages for each one-third-octave-band filter) are not allowed. In addition, use of a pre-emphasis/de-emphasis system will require testing and documentation of all filters and gain stages involved to ensure that any errors are quantified and minimized, and that the system performs predictably and reliably.
(6) Attenuator Specification: This requirement allows for the use of switchable voltage input range settings (now commonplace on DAT recorders) as controllable attenuation steps for gain-setting purposes. In all cases, attenuators must have fixed repeatable steps. Any devices in the measurement system that use vernier or continuously–adjustable gain controls must also have some demonstrable means of being fixed, or locked at a specific setting to eliminate non-traceable gain errors.
164. Section A36.3.7 Analysis systems
165. Section A36.3.7.1
The analysis system must conform to the specifications in sections A36.3.7.2 to A36.3.7.7 for the frequency bandwidths, channel configurations and gain settings used for analysis.
166. Section A36.3.7.2
The output of the analysis system must consist of one-third octave band sound pressure levels as a function of time, obtained by processing the noise signals (preferably recorded) through an analysis system with the following characteristics:
a) A set of 24 one-third octave band filters, or their equivalent, having nominal midband frequencies from 50 Hz to 10 kHz inclusive;
b) Response and averaging properties in which, in principle, the output from any one-third octave filter band is squared, averaged and displayed or stored as time-averaged sound pressure levels;
c) The interval between successive sound pressure level samples must be 500 ms ±5 milliseconds (ms) for spectral analysis with or without slow time-weighting, as defined in section A36.3.7.4;
d) For those analysis systems that do not process the sound pressure signals during the period of time required for readout and/or resetting of the analyzer, the loss of data must not exceed a duration of 5 ms; and
e) The analysis system must operate in real time from 50 Hz through at least 12 kHz inclusive.
This requirement applies to all operating channels of a multi-channel spectral analysis system.
167. Section A36.3.7.3
The minimum standard for the one-third octave band analysis system is the class 2 electrical performance requirements of IEC 61260 as amended, over the range of one-third octave nominal midband frequencies from 50 Hz through 10 kHz inclusive.
Note: IEC 61260 specifies procedures for testing of one-third octave band analysis systems for relative attenuation, anti-aliasing filters, real time operation, level linearity, and filter integrated response (effective bandwidth).
168. Section A36.3.7.4
When slow time averaging is performed in the analyzer, the response of the one-third octave band analysis system to a sudden onset or interruption of a constant sinusoidal signal at the respective one-third octave nominal midband frequency, must be measured at sampling instants 0.5, 1, 1.5 and 2 seconds(s) after the onset and 0.5 and 1s after interruption. The rising response must be - 4 ± 1 dB at 0.5s, -1.75 ± 0.75 dB at 1s, -1 ± 0.5 dB at 1.5s and -0.5 ± 0.5 dB at 2s relative to the steady-
state level. The falling response must be such that the sum of the output signal levels, relative to the initial steady-state level, and the corresponding rising response reading is -6.5 ± 1 dB, at both 0.5 and 1s. At subsequent times the sum of the rising and falling responses must be -7.5 dB or less. This equates to an exponential averaging process (slow time-weighting) with a nominal 1s time constant
(i.e., 2s averaging time).
169. Section A36.3.7.5
When the one-third octave band sound pressure levels are determined from the output of the analyzer without slow time-weighting, slow time-weighting must be simulated in the subsequent processing. Simulated slow time-weighted sound pressure levels can be obtained using a continuous exponential averaging process by the following equation:
Ls(i,k) = 10 log [(0.60653)10 0.1Ls[i,(k-1)] + (0.39347) 100.1 L(i,k) ]
where Ls(i,k) is the simulated slow time-weighted sound pressure level and L(i,k) is the asmeasured 0.5 s time average sound pressure level determined from the output of the analyzer for the k-th instant of time and the i-th one-third octave band. For k =1, the slow time-weighted sound pressure Ls[i,(k-1=0)] on the right hand side should be set to 0 dB. An approximation of the continuous exponential averaging is represented by the following equation for a four sample averaging process for k ‡ 4:
Ls(i,k) = 10 log [(0.13)10 0.1L[i,(k -3)] + (0.21) 10 0.1L[i,(k -2)] +(0.27) 10 0.1L[i,(k -1)] + (0.39)10 0.1L[i,k] ]
where Ls(i,k) is the simulated slow time-weighted sound pressure level and L(i,k) is the as measured 0.5s time average sound pressure level determined from the output of the analyzer for the kth instant of time and the i-th one-third octave band.
The sum of the weighting factors is 1.0 in the two equations. Sound pressure levels calculated by means of either equation are valid for the sixth and subsequent 0.5s data samples, or for times greater than 2.5 s after initiation of data analysis.
Note: The coefficients in the two equations were calculated for use in determining equivalent slow time-weighted sound pressure levels from samples of 0.5 s time average sound pressure levels.
The equations do not work with data samples where the averaging time differs from 0.5 s.
170. Section A36.3.7.6
The instant in time by which a slow time-weighted sound pressure level is characterized must be 0.75s earlier than the actual readout time.
Note: The definition of this instant in time is needed to correlate the recorded noise with the aircraft position when the noise was emitted and takes into account the averaging period of the slow time-weighting. For each 0.5 second data record this instant in time may also be identified as 1.25 seconds after the start of the associated 2 second averaging period.
171. Section A36.3.7.7
The resolution of the sound pressure levels, both displayed and stored, must be 0.1 dB or finer.
a. Explanation
This section specifies the required performance characteristics of an analysis system for use in airplane noise certification.
b. Supplemental Information
(1) Analyzer Specifications: IEC 61260 specifies the Class 2 electrical performance requirements of one-third-octave-band filters, including tolerances for the attenuation in the transition bands (“skirts”) adjacent to the one-third-octave pass-bands. Most digital one-third-octave-band analysis systems offer only hardwired filtering algorithms that emulate the response of a traditional third-order analysis filter having a maximally-flat pass-band. However, some analysis systems allow the selection of other filtering algorithms which might not provide equivalent performance. FAA policy requires an applicant to demonstrate the effects that alternate filter design response characteristics might have on noise certification EPNL values.
(2) Determination of Bandwidth Error Corrections: The manufacturer can establish the geometric center frequencies of one-third-octave-band filters using either Base 2 or Base 10 systems. While the use of either method results in frequencies close to the nominal center frequencies referred to in part 36, it is important to note which system is used so that the bandwidth error adjustment can be properly determined. Use of test frequencies calculated by a different base-number system than that for which the analyzer was designed can result in erroneous values for these adjustments.
(3) Externally Controlled Linear-Integrating Analyzers: In cases where a computer or other external device is used to control and/or communicate with an analyzer performing linear integration, extra care should be taken to ensure that the integration period requirements are met. Some analyzers from major manufacturers have required factory modification in order to provide an integration time within 5 milliseconds of the specified 500-millisecond integration period.
c. Procedures
None
172. Section A36.3.8 Calibration systems
173. Section A36.3.8.1
The acoustical sensitivity of the measurement system must be determined using a sound calibrator generating a known sound pressure level at a known frequency. The minimum standard for the sound calibrator is the class 1L requirements of IEC 60942 as amended.
174. Section A36.3.9 Calibration and checking of system
175. Section A36.3.9.1
Calibration and checking of the measurement system and its constituent components must be carried out to the satisfaction of the FAA by the methods specified in sections A36.3.9.2 through A36.3.9.10. The calibration adjustments, including those for environmental effects on sound calibrator output level, must be reported to the FAA and applied to the measured one-third-octave sound pressure levels determined from the output of the analyzer. Data collected during an overload indication are invalid and may not be used. If the overload condition occurred during recording, the associated test data are invalid, whereas if the overload occurred during analysis, the analysis must be repeated with reduced sensitivity to eliminate the overload.
176. Section A36.3.9.2
The free-field frequency response of the microphone system may be determined by use of an electrostatic actuator in combination with manufacturer’s data or by tests in an anechoic free-field facility. The correction for frequency response must be determined within 90 days of each test series.
The correction for non-uniform frequency response of the microphone system must be reported to the
FAA and applied to the measured one-third octave band sound pressure levels determined from the output of the analyzer.
177. Section A36.3.9.3
When the angles of incidence of sound emitted from the aircraft are within –30° of grazing incidence at the microphone (see Figure A36-1), a single set of free-field corrections based on grazing incidence is considered sufficient for correction of directional response effects. For other cases, the angle of incidence for each 0.5 second sample must be determined and applied for the correction of incidence effects.
178. Section A36.3.9.4
For analog magnetic tape recorders, each reel of magnetic tape must carry at least 30 seconds of pink random or pseudo-random noise at its beginning and end. Data obtained from analog taperecorded signals will be accepted as reliable only if level differences in the 10 kHz one-third-octaveband are not more than 0.75 dB for the signals recorded at the beginning and end.
179. Section A36.3.9.5
The frequency response of the entire measurement system while deployed in the field during the test series, exclusive of the microphone, must be determined at a level within 5 dB of the level corresponding to the calibration sound pressure level on the level range used during the tests for each one-third octave nominal midband frequency from 50 Hz to 10 kHz inclusive, utilizing pink random or pseudo-random noise. Within six months of each test series the output of the noise generator must be determined by a method traceable to the U.S. National Institute of Standards and Technology or to an equivalent national standards laboratory as determined by the FAA. Changes in the relative output from the previous calibration at each one-third octave band may not exceed 0.2 dB. The correction for frequency response must be reported to the FAA and applied to the measured one-third octave sound pressure levels determined from the output of the analyzer.
180. Section A36.3.9.6
The performance of switched attenuators in the equipment used during noise certification measurements and calibration must be checked within six months of each test series to ensure that the maximum error does not exceed 0.1 dB.
181. Section A36.3.9.7
The sound pressure level produced in the cavity of the coupler of the sound calibrator must be calculated for the test environmental conditions using the manufacturer’s supplied information on the influence of atmospheric air pressure and temperature. This sound pressure level is used to establish the acoustical sensitivity of the measurement system. Within six months of each test series the output of the sound calibrator must be determined by a method traceable to the U.S. National Institute of Standards and Technology or to an equivalent national standards laboratory as determined by the FAA. Changes in output from the previous calibration must not exceed 0.2 dB.
182. Section A36.3.9.8
Sufficient sound pressure level calibrations must be made during each test day to ensure that the acoustical sensitivity of the measurement system is known at the prevailing environmental conditions corresponding with each test series. The difference between the acoustical sensitivity levels recorded immediately before and immediately after each test series on each day may not exceed
0.5 dB. The 0.5 dB limit applies after any atmospheric pressure corrections have been determined for the calibrator output level. The arithmetic mean of the before and after measurements must be used to
represent the acoustical sensitivity level of the measurement system for that test series. The calibration corrections must be reported to the FAA and applied to the measured one-third octave band sound pressure levels determined from the output of the analyzer.
183. Section A36.3.9.9
Each recording medium, such as a reel, cartridge, cassette, or diskette, must carry a sound pressure level calibration of at least 10 seconds duration at its beginning and end.
184. Section A36.3.9.10
The free-field insertion loss of the windscreen for each one-third octave nominal midband frequency from 50 Hz to 10 kHz inclusive must be determined with sinusoidal sound signals at the incidence angles determined to be applicable for correction of directional response effects per section A36.3.9.3. The interval between angles tested must not exceed 30 degrees. For a windscreen that is undamaged and uncontaminated, the insertion loss may be taken from manufacturer’s data.
Alternatively, within six months of each test series the insertion loss of the windscreen may be determined by a method traceable to the U.S. National Institute of Standards and Technology or an equivalent national standards laboratory as determined by the FAA. Changes in the insertion loss from the previous calibration at each one-third-octave frequency band must not exceed 0.4 dB. The correction for the free-field insertion loss of the windscreen must be reported to the FAA and applied to the measured one-third octave sound pressure levels determined from the output of the analyser.
a. Explanation
Sections A36.3.9.1 through A36.3.9.10 specify the performance requirements for calibration systems and the procedures for calibration of the noise measurement system used for airplane noise certification.
b. Supplemental Information
(1) Pink Noise: Pink noise contains equal energy in each octave band or fractional octave band
(e.g., the octave from 100 Hz to 200 Hz contains the same amount of energy as the octave from 1 kHz to 2 kHz, although for the lower-frequency octave, it is distributed over a frequency range 10 times narrower).
(2) Note on Pink Noise Usage: Because of the dynamic nature of the pink noise signal, longer samples produce statistically better measurements. At a minimum, recorded pink noise signals must have durations of 30 seconds.
(3) Acoustic Calibrator Output Adjustments: Acoustic calibrator outputs may require adjustment for ambient conditions (such as temperature and atmospheric pressure), coupler volume, etc. Section A36.3.9.8 requires that all such adjustments be applied in the data processing stage rather than by using an adjusted calibration value in the analyzer. In this way, a traceable record of the adjustments can be maintained.
(4) Calibration Traceability: All performance calibration analyses of calibration equipment must be traceable to the U.S. National Institute of Standards and Technology or an equivalent national standards laboratory as determined by the FAA.
c. Procedures
(1) Measurement System Field Calibration (All components of the measurement system except microphones): All components of the measurement system, except microphones, must be tested while deployed in the field using pink noise at a level within 5 dB of the calibration level (as specified in A36.3.9.5). The signal must be recorded for a duration of at least 30 seconds so that one-third octave-band system frequency response adjustments can be determined and applied during analysis. The pink noise generator must be calibrated within 6
months of the measurement, and is acceptable for certification use only if its output in each one-third octave band does not change by more than 0.2 dB between calibrations.
(2) Field Acoustical Calibrations: At the start and end of each measurement day, at the beginning of each physical recording (each tape reel, cartridge, cassette, disk, etc.), and at the end of the last physical recording, an acoustic calibration signal of known amplitude and frequency must be fed through the entire measurement system (including microphone) as deployed in the field, and recorded. All components of the system (excluding the windscreen) must be in place at this time, including cables, attenuators, gain and signalconditioning amplifiers, filters (including preemphasis) and power supplies. During calibration, attenuators and gain stages must be set to prevent overload, and to maintain the calibration signal level on the reference level range within the limits specified in section A36.3.6.6. If any switchable filters that could affect the calibration signal are utilized during measurements, then calibrations must be performed both with and without these filters enabled. Components of the electrical system must not be added, removed, or replaced without re-calibrating the entire system immediately before and after each change.
The 0.5 dB limit stated in section A36.3.9.8 requires that sufficient determinations be made of the entire system’s acoustical sensitivity during each measurement day to ensure that the response of the equipment is known for each test. The equipment shall be considered satisfactory if (after the required calibrator corrections are applied) the recorded variation over the period immediately before and immediately after each test series, within a given day, is not more than 0.5 dB.
(3) Measurement Program Equipment Calibration Schedules:
(i) Microphones and preamplifiers (also windscreens, if included in free-field calibration):
Pressure-response or free-field calibration within the 90 days of the measurement program.
(ii) One-Third-Octave-Band Analyzers: Within 6 months of the measurement program.
(iii) Recorders, Amplifiers, Filters, etc.: Within 6 months of the measurement program.
(iv) Calibration Equipment: Within 6 months of the measurement program, traceable to NIST or an equivalent national standards laboratory as determined by the FAA.
(4) Applications of Adjustments for Incidence: When using microphones whose frequency response is nearly flat at grazing incidence, and when the angles of incidence of sound emitted from the aircraft are within ± 30° of grazing incidence, a single set of data adjustments for free-field response (and windscreen insertion loss), based on grazing incidence, is considered sufficient to account for incidence effects. When it is impractical to orient the microphone properly to maintain grazing incidence, provided that a continuous record of TPSI is available, free-field (and windscreen insertion-loss) incidence data adjustments can be applied to the noise data on a spectral-record-by-spectral-record basis. These adjustments are obtained by calculating the angle of incidence for each record, using the point of time which characterizes the 2-second averaging period (per section A36.3.7.6), and determining the airplane's emission coordinates (and angle of incidence) for the sound measured at that time.
(5) Determination of Windscreen Data Adjustments: The physical condition of a windscreen can significantly affect its performance, and manufacturer-provided windscreen data adjustments for insertion loss are only valid for new, or clean, dry windscreens. For these adjustments, a single set of values based upon wind screen insertion loss tests at grazing incidence may be used when the angles of incidence of sound emitted from an aircraft are within + 30 degrees of grazing incidence. For other cases, the windscreen insertion loss adjustments must be determined and applied on the basis of intervals between angles tested not exceeding 30 degrees.
When the windscreen data adjustments provided by the manufacturer are presented in the form of curves, care must be taken to include the insertion loss throughout each one-third-octave band, rather than just at the nominal midband frequency. Windscreen insertion loss can vary substantially within the frequency range of a
single band and must be averaged or faired to more accurately correct one-third-octave-band data for the presence of the windscreen. Windscreen data adjustments may also be obtained by free-field calibration in an anechoic chamber.
185. Section A36.3.10 Adjustments for ambient noise
186. Section A36.3.10.1
Ambient noise, including both acoustical background and electrical noise of the measurement system, must be recorded for at least 10 seconds at the measurement points with the system gain set at the levels used for the aircraft noise measurements. Ambient noise must be representative of the acoustical background that exists during the flyover test run. The recorded aircraft noise data is acceptable only if the ambient noise levels, when analyzed in the same way, and quoted in PNL (see A36.4.1.3(a)), are at least 20 dB below the maximum PNL of the aircraft.
a. Explanation
This section specifies procedures for ambient noise measurements, and the acceptability limit for comparison of aircraft PNL against ambient PNL.
b. Supplemental Information
(1) Ambient Noise: The term “ambient noise” as specified in section A36.3.10.1 includes the acoustical background noise present at a microphone site, generated by sources other than the subject aircraft and the electrical background noise from the measurement system. This specification ensures that any signals that can affect the measured levels of aircraft noise can be properly quantified.
(2) Background Noise: The term “background noise” as used in Appendix 3 of the appended TM and Appendix 3 of this AC describes all noise from sources other than the subject aircraft which can influence or obscure the noise levels being measured including ambient noise, and analytical noise from the measurement system (e.g., the baseline of an analysis window or the amplitude non-linearity characteristics of measurement system components). The elements of background noise can be characterized as one of two types:
Pre-detection noise, which contributes energy to the signal level being measured; and Post-detection noise, which does not contribute energy to the measured signal level. Appendix 3 of this AC provides detailed information on the characteristics of background noise, and detailed procedures for adjusting measured aircraft noise levels for the effects of ambient and other types of background noise.
c. Procedures
(1) Measurement System Noise: Since measurement system noise can add energy to measured aircraft noise levels, the ambient noise measurement described in section A36.3.10.1 must be made with all gain stages and attenuators set as they would be used during the aircraft noise certification measurements. If it is expected that multiple settings will be required during the measurements, ambient noise data must be collected at each of these settings. Whenever possible, ambient noise recordings should be made for at least 30 seconds each, and care should be taken to ensure that the ambient noise is truly representative of that present during the aircraft noise certification tests.
(2) Mean Ambient Noise Assessments: At least 10 seconds, but preferably 30 seconds, of ambient noise data must be time-averaged to determine the mean level for each one-third octave band. The PNL value for this averaged spectrum must then be calculated using the procedures defined in section A36.4.2. The aircraft noise level data must also be analyzed, and PNL values calculated for each spectral record. The maximum aircraft PNL value must be at least 20 dB above the PNL of the averaged ambient noise spectrum for the data to be considered acceptable.
187. Section A36.3.10.2
Aircraft sound pressure levels within the 10 dB-down points (see A36.4.5.1) must exceed the mean ambient noise levels determined in section A36.3.10.1 by at least 3 dB in each one-third octave band, or must be adjusted using a method approved by the FAA; one method is described in the current advisory circular for this part.
a. Explanation
This section quantifies the amount by which aircraft sound pressure levels must exceed the mean ambient noise levels for noise certification testing, and specifies the requirement for data adjustment in cases where this exceedance amount is not met.
b. Supplemental Information
(1) Background Noise Adjustments: Adjustments must be applied for the effects of ambient and other types of background noise on measured aircraft sound pressure levels. Determination of masking must be accomplished before any frequency-dependent adjustments are applied (such as for system frequency response or microphone free-field response). Appendix 3 of the appended ICAO TM presents a background noise adjustment method that represents the international recommended practice. FAA has developed procedures contained in Appendix 3 of this AC to provide the applicant with a more definitive methodology for accounting for the effects of background noise, including some procedural elements not contained in the TM Appendix 3. The Technical Issues Task Group of Working Group 1 of the ICAO Committee on Aviation Environmental Protection is considering possible revision to the ICAO TM Appendix 3 background noise adjustment method based upon review of the procedures contained in Appendix 3 of this AC.
c. Procedures
None
188. Section A36.4 Calculation of Effective Perceived Noise Level From Measured Data
189. Section A36.4.1 General
190. Section A36.4.1.1
The basic element for noise certification criteria is the noise evaluation measure known as effective perceived noise level, EPNL, in units of EPNdB, which is a single number evaluator of the subjective effects of airplane noise on human beings. EPNL consists of instantaneous perceived noise level, PNL, corrected for spectral irregularities, and for duration. The spectral irregularity correction, called "tone correction factor", is made at each time increment for only the maximum tone.
191. Section A36.4.1.2
Three basic physical properties of sound pressure must be measured: level, frequency distribution, and time variation. To determine EPNL, the instantaneous sound pressure level in each of the 24 one-third octave bands is required for each 0.5 second increment of time during the airplane noise measurement.
192. Section A36.4.1.3
The calculation procedure that uses physical measurements of noise to derive the EPNL evaluation measure of subjective response consists of the following five steps:
(a) The 24 one-third octave bands of sound pressure level are converted to perceived noisiness (noy) using the method described in section A36.4.2.1(a). The noy values are combined and then converted to instantaneous perceived noise levels, PNL(k).
(b) A tone correction factor C(k) is calculated for each spectrum to account for the subjective response to the presence of spectral irregularities.
(c) The tone correction factor is added to the perceived noise level to obtain tone-corrected perceived noise levels PNLT(k), at each one-half second increment:
PNLT(k) = PNL(k) + C(k)
The instantaneous values of tone-corrected perceived noise level are derived and the maximum value, PNLTM, is determined.
(d) A duration correction factor, D, is computed by integration under the curve of tonecorrected perceived noise level versus time.
(e) Effective perceived noise level, EPNL, is determined by the algebraic sum of the maximum tone-corrected perceived noise level and the duration correction factor:
EPNL = PNLTM + D
a. Explanation
This section specifies that the noise rating for an individual airplane event for purposes of noise certification is Effective Perceived Noise Level (EPNL in units of EPNdB). It is based on the results of extensive psycho-acoustical research conducted primarily in the 1950’s and early 1960’s concerning the subjective effects of airplane noise on human beings. EPNL is a single number measure of the noise of an individual airplane flyover that approximates laboratory annoyance responses. It is derived from calculations of perceived noise level (PNL, in units of PNdB), adjusted for the presence of audible pure tones or discrete frequencies (PNLT, in units of TPNdB), and for duration of the airplane flyover. The PNL metric accounts for variations in hearing sensitivity as a function of frequency. The PNLT metric additionally accounts for the effects of tones, which are often present in complex sounds such as airplane noise. The EPNL metric further accounts for the effect of the duration of the noise. The EPNL computation process effectively yields a time integrated annoyance level.
b. Supplemental Information
None
c. Procedures
None
193. Section A36.4.2 Perceived noise level
194. Section A36.4.2.1
Instantaneous perceived noise levels, PNL(k), must be calculated from instantaneous one-third octave band sound pressure levels, SPL(i,k) as follows:
(a) Step 1: For each one-third octave band from 50 through 10,000 Hz, convert SPL(i,k) to perceived noisiness n(i,k), by using the mathematical formulation of the noy table given in section A36.4.7.
(b) Step 2: Combine the perceived noisiness values, n(i,k), determined in step 1 by using the following formula:
( ) ( ) ( ) ( )N k n k n i k n k i = + − =∑015 124. , ( ) ( )= + = ∑0 85 015 1 24 . . ,n k n i k i
where n(k) is the largest of the 24 values of n(i,k) and N(k) is the total perceived noisiness.
(c) Step 3: Convert the total perceived noisiness, N(k), determined in Step 2 into perceived noise level, PNL(k), using the following formula:
( ) ( )PNL k N k= +40 0 102. log log
Note: PNL(k) is plotted in the current advisory circular for this part.
a. Explanation
This section specifies the method(s) used to calculate perceived noise level (PNL, in units of PNdB) for each one-third-octave-band spectral record, i.e., each one-half second spectrum. A plot of PNL(k) versus total noisiness (N) is presented in Figure 4-1 of Appendix 4 of the appended TM.
b. Supplemental Information
(1) “Instantaneous” Sound Pressure Levels: For the purposes of this procedure, “instantaneous” Sound Pressure levels are considered to be one-third-octave-band sound pressure levels for each half-second record obtained using a continuous exponential averaging process as described in section A36.3.7.4.
c. Procedures
None
195. Section A36.4.3 Correction for spectral irregularities
196. Section A36.4.3.1
Noise having pronounced spectral irregularities (for example, the maximum discrete frequency components or tones) must be adjusted by the correction factor C(k) calculated as follows:
a. Explanation
This section specifies a 10-step procedure for identifying the tonal content of each one-third-octaveband spectrum and for determining correction factors for converting PNL values to values designated as PNLT.
b. Supplemental Information
None
c. Procedures
(1) Data Precision for Tone Correction Computation: Prior to Step 1, all one-third-octave-band sound pressure levels should be temporarily rounded to 0.1 dB resolution. The tone-correction procedure presented here includes several steps that utilize decibel level criteria to characterize the significance of tonal
content. These criteria can become artificially sensitive to small variations in level if resolution finer than 0.1 dB is used in the computations.
(a) Step 1: After applying the corrections specified under section A36.3.9, start with the sound pressure level in the 80 Hz one-third octave band (band number 3), calculate the changes in sound pressure level (or "slopes") in the remainder of the one-third octave bands as follows:
( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )k,23SPLk,24SPLk,24s k,1iSPLk,iSPLk,is k,3SPLk,4SPLk4,s value no k,3s −=
•
• −−=
•
• −= =
(b) Step 2: Encircle the value of the slope, s(i, k), where the absolute value of the change in slope is greater than five; that is where:
( ) ( ) ( )∆s i k s i k s i k, , ,= − − >1 5
(c) Step 3:
(1) If the encircled value of the slope s(i, k) is positive and algebraically greater than the slope s(i-1, k) encircle SPL(i, k).
(2) If the encircled value of the slope s(i, k) is zero or negative and the slope s(i-1,k) is positive, encircle SPL(i-1,k).
(3) For all other cases, no sound pressure level value is to be encircled.
a. Explanation
Steps 1-3 specify the method used to identify significant tones in spectral data.
b. Supplemental Information
None
c. Procedures
None
(d) Step 4: Compute new adjusted sound pressure levels SPL'(i, k) as follows:
(1) For non-encircled sound pressure levels, set the new sound pressure levels equal to the original sound pressure levels, SPL'(i, k) = SPL(i, k).
(2) For encircled sound pressure levels in bands 1 through 23 inclusive, set the new sound pressure level equal to the arithmetic average of the preceding and following sound pressure levels as shown below:
( ) ( ) ( )[ ]SPL SPL SPL′ = − + +i k i k i k, , ,12 1 1
(3) If the sound pressure level in the highest frequency band (i =24) is encircled, set the new sound pressure level in that band equal to:
( ) ( ) ( )k,23sk,23SPLk,24LSP +=′
(e) Step 5: Recompute new slope s'(i,k), including one for an imaginary 25th band, as follows:
( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )k,24sk,25s k,23LSPk,24LSPk,24s k,1iLSPk,iLSPk,is k,3LSPk,4LSPk,4s k,4sk,3s ¢=¢ ¢-¢=¢
•
• -¢-¢=¢
•
• ¢-¢=¢ ¢=¢
(f) Step 6: For i, from 3 through 23, compute the arithmetic average of the three adjacent slopes as follows:
( ) ( ) ( ) ( )[ ]k,2isk,1isk,is31k,is +¢++¢+¢=
(g) Step 7: Compute final one-third octave-band sound pressure levels, SPL"(i,k), by beginning with band number 3 and proceeding to band number 24 as follows: