New Measurement Methods for Anechoic Chamber Characterization

Transcription

1 New Measurement Methods for Anechoc Chamber Characterzaton Gomez-Alfageme J.J. AES Member, Sanchez-Bote J.L, Blanco-Martfn E. AES Member Grupo de Electroacustca y Tratamento de Senales, Unversdad Poltecnca de Madrd,, Madrd, Span alfaqeme@dac.upm.es, bote@dac.upm.es, eblanco@dac.upm.es ABSTRACT As a contnuaton of the work presented n nd AES Conventon, ths paper tres to study n depth the anechoc chambers qualfcaton. The purpose of ths paper s to fnd parameters that allow the characterzaton of ths type of enclosures. The proposal that becomes n ths work s tryng to obtan data of the anechoc chambers absorpton by means of the transfer functons between pars of mcrophones, or by means of the mpulse response between pars of mcrophones. Based on the results of the transfer functons between pars of mcrophones can be checked easly agreement of the nverse squared law, allowng to determne the chamber cut-off frequency. Makng a band-pass flterng t could be confrmed the anechoc chambers qualfcaton.. NTRODUCTON As a contnuaton of the work presented n nd AES Conventon [], ths paper tres to study n depth the anechoc chambers qualfcaton. The purpose of ths paper s to fnd, through dfferent measurement methods, parameters that allow the characterzaton of ths type of enclosures. One of the man problems found n the characterzaton of anechoc chambers s the determnaton of the absorpton coeffcent, or the reverberaton tme, of the same ones. The tradtonal methods only try to verfy the wellness of chamber as far as the agreement of the nverse squared law of wthn ths type of enclosures. These methods are only based on the measurement of the sound pressure level nsde the enclosure. The proposal that becomes n ths work s tryng to obtan data of the anechoc chambers absorpton by

2 means of the transfer functons of between pars of mcrophones, or by means of the analyss of the mpulse response between pars of mcrophones. Based on the results of the transfer functons between pars of mcrophones can be checked easly agreement of the nverse squared law, allowng to determne the chamber cut-off frequency. Makng a band flterng t could be confrmed the anechoc chambers qualfcaton. Based on the results of mpulse response measurements between pars of mcrophones may be obtaned EDT nformaton and through Schroeder reverse ntegraton t could be obtaned nformaton of the anechoc chambers reverberaton tme. Also, by flterng technques of the mpulse response, t s possble to obtan nformaton of the absorpton coeffcent of the chamber. The measurement methods proposed are based on the accomplshment of these measures n the zones where the agreement of the nverse squared law nsde the chamber takes place, usng dfferent sgnal types (mpulsve, broadband nose, MLS, etc.). Also a method sets out to obtan the absorpton coeffcent makng measurements between pars of mcrophones n the proxmtes of the walls of the anechoc chamber, where the absorbent s placed. n ths method, and through the assumpton of the anechoc chamber behavor an LT system can be found expressons for the absorpton coeffcent, f we consder that n the zone where the nverse squared law s agreed, the role of transfer s equvalent to an attenuaton and a fxed delay, dependng on the separaton between mcrophones. nformaton that s not obtaned by means of the use of tradtonal technques of measurement untl now n the qualfcaton of anechoc chambers. Let us thnk that the contrbutons of ths work are of suffcent nterest lke beng able to present conclusons that untl the moment are not by the tradtonal methods.. ANECHOC CHAMBER DESCRPTON Once placed the wedges, the nner dmensons of the chamber between the wedges are mmxmmx mm, what s equvalent to a free volume between wedges of 9.9m. f we compare t wth the dmensons of the prevous camera, of m, we see that the ncrement of volume s of %. The nternal area of the surfaces lmts of the chamber s.m. As a datum, the dagonal of the chamber s mm. The materal of the new wedges was decded to be rock wool wth a densty of kg/m wth a trangular wedge shape of mm of heght and wth a square base of mmxmm and wth a heght of mm. The total the heght of the wedge s mm. The base surface of each wedge s of.m and the volume of the wedge s of.m. To avod the deteroraton of the wedges and to mnmze as possble that was powder n suspenson of the rock wool nsde the chamber due to the degradaton process t was opted to manufacture a case of porous cloth for each wedge. The color of the cloth s whte n order of ncreasng the brghtness nsde the chamber. Both proposed methods are based on consderng that, wthn certan margns, the anechoc chamber behaves lke a LT system, and therefore the expressons that allow that are tred to obtan the parameters are relatvely smple. mm. mm The measurement system that s gong away to use s based on the Brel&Kjaer PULSE platform and the use of three mcrophones n smultaneous way. Ths platform allows the upset the data to Matlab, software whch wll make all the post-processng technques. The supposton of whch the anechoc chamber behaves as a LT system and the smultaneous use of multmcrophone measures allows to obtan a great amount of Fgure Wedge shape and dmensons Fnally t was decded that the materal was the rock wool of kg/m densty and the trangular shaped wedge that t was commented prevously. The advantage ths wedge wth a square base was that t could get placed n dfferent geometrcal shapes. The decson of the fnal wedges dsposton was assumed accordng to ncrease the effectve surface of absorpton

3 and that the coeffcent of absorpton was the hgher as possble, n functon of the measures of the same one that were carred out n the reverberant chamber of the Laboratoro of Sondo of the Unversty. Once resolved the fnal dsposton that they wll have those whose, rotated 9 one wth regardng the prevous one, t was approached n prevously commented problem of makng a case of porous cloth (transparent to the sound) to avod the deteroraton of the absorbent materal. To be able to move nsde the camera a square standng floor (tramex) looommxlooomm of metallc grll loommxloomm, t was desgned mm above the floor of the chamber and supported by heght adjustable stands and supported to the walls. To approach the topc of the nternal and external connectons of the chamber oblque holes were desgned n descent of mm of dameter each one. n the nteror part of the chamber were prepared two connecton boxes wth ndependent audo connectors (XLR, BNC and LEMO). sound source n each one of the four drectons mentoned prevously, data from whch the SPL dfference can settle down regardng the nverse squared law, that anyway wll be nsde the tolerances establshed down n the SO -9 norm [], and that they are presented n the followng table. Central frequency of / octave band (Hz) < Hz De a Hz Table > Hz Allowed devaton (db) ±. ±. ±. SO -9 allowed devatons as functon of frequency... One mcrophone SPL measurements wth real tme analyzer (RTA) Ths study conssts on the realzaton of numerous measures of SPL receved along dverse longtudnal and dagonal drectons defned n the chamber, n such a way that one can have a good dea of the acoustc chamber behavor n each one of these drectons. n each drecton measures were carred out n the two possble ways as well as to two source-mcrophone heghts, correspondng n half and to a thrd of the chamber heght [] - []. As exctaton sgnal a pnk nose / octave band-pass fltered s used. Fgure Actual chamber vew wth dfferent measure arrangements. CHAMBER QUALFCATON ACCORDNG TO SO -9 For the realzaton of the measurements a drectve sound source was used, placed V meter away wall for four dfferent drectons (length, wdth and two dagonals) and for two opposte ways n each drecton. The sound source was a coaxal loudspeaker Tannoy n order to avod the hgh frequency drectvty losses. The measurements have been carred out from dstances to mm up to mm/mm (f possble) from the Fgure Example of sound source/mcrophone dsposton Some of the results of these measurements can be observed n the next fgures.

4 o o o o o o o o o o - ^ f - ^ - ^. o" m" o" o" o" n o" o" o" o" - ^ to N m OCMCQOlf-OOCOO -' ---cvcvn^-ncdeo Fgure nverse squared law devaton for dstance couples from rnm-mm to mm-mm n longtudnal drecton o' o' o" o" LO o" o" o" o" - H <o OJ n *~. OOJ<DOlO-OOCOO -' f\co -T--WOJnTflO(DK ty O OJ <Q O OJ Frequency(Hz) Fgure nverse squared law devaton for dstance couples from mm-mm to mm-mm n longtudnal drecton Fgure Theoretcal and measured nverse squared law for khz / octave band!' B k. k k. k.k k k. k k k. k,k k, -, * * ' -. : '. -. -,? ~, - J. -. -, -,,,,9,,,,,,,9,9,,,,9,,9,,,,,,,,,,,,,, 9 ' L - - -,,,, A - -, ± * -, J. ± L L -, -, -, -, -,, 9,,,,,,,9,,,,,9,,9,,,,,,,9,,,,, ,, ",,.,, J. L , " J. Fgure Theoretcal and measured nverse squared law for Hz / octave band Table Allowed devatons accordng to SO - 9 as functon of frequency dstance.... Mult-mcrophone measurements wth fast Fourer transform analyzer (FFT) Wth ths method, the measurements realzed n secton before wll be done agan, n the same four drectons and two ways per each one, but now wth three mcrophones smultaneously and fxng couples of separaton between sound source and mcrophones logarthmcally spaced (always duplcatng the dstance between source and mcrophone and between source and mcrophone ). The arrangement could be observed n fgure. All these measurements were done wth a

5 four channel FFT Analyzer Brtel&Kjaer model PULSE and three condenser Brel&Kjaer mcrophones. As exctaton sgnal a broad band whte nose s used (wth dfferent bandwdth or frequency span). Fgure Example of sound source and three mcrophone (,, ) dsposton The use of a multchannel mcrophones measurement system s gong to allow not obtan only the sound pressure level at every mcrophone, but also to obtan relatons among the sgnals n pars of mcrophones. t wll be able to measure these relatons usng the transfer functons and the mpulse responses that there provdes the FFT analyzer. The frst mcrophone (mcl) wll be n use for determnng the anechoc response of the loudspeaker, wth the hgh frequency lmtatons by placng the mcrophone n the near-feld of the loudspeaker. Mcl s placed at two fxed dstances from the loudspeaker (mm and mm). For our loudspeaker, the hgh frequency lmtaton of appears over khz, when t s placed at mm away and khz when t s placed at mm away. The other two mcrophones wll arrange n a logarthmc form along each of the drectons of measurement exposed before, so that the thrd mcrophone (mc) always s to the double of dstance from the loudspeaker that the second mcrophone (mc). Wth ths dstances relaton t wll be able to obtan an dea of the accomplshment of the nverse squared law for dstances up to mm from the source. The dsposto of the mcrophones and loudspeaker nsde the anechoc chamber can be observed n the followng fgure. ^MW ft / ' Fgure 9 Loudspeaker and mcrophones dsposton nsde anechoc chamber Snce t has been commented prevously, not only there are obtaned sound pressure levels at every mcrophone, but also nformaton of the loudspeaker response, both n tme and n frequency domans. [db/p/v] tol -) -«- - FrequencyResponseHXPULSE Saldaamplf)-Mark (Magntude) Workng: BasePand_Om_m :nput: FFT Analyzer DO fc k k Ek k [Hz] Fgure Loudspeaker frequency response as a transfer functon between mcl and power amplfer

6 mpulse Res h(pulselsaldaamplf)-mark (Real Part) Workng:E L.m_m :nput :FFT Analyzer ResponseHXPULSE,PULSE)-Mark (Magntuc Workng:Baseband_Om_m :nput:fft Analyzer - J\ \ V V v^ J-., rft. n A. ^^s^k_ u u,m,m,m m Fgure Loudspeaker mpulse response as an nverse Fourer transform form transfer functon n fgure Also, t could be measured the transfer functons between mc and mc wth respect of the output of the power amplfer of that exctes the loudspeaker for dfferent source dstances. k k k k ResponseHXPULSE,PULSE ) -Mark (M agntude Workng :B as epand_m_tn : nput: FFT Analyzer ^^www^^^^ft' [db/l.dd P9/V] Frequency Response H (PULSE S.Salda amplf) - Mark (Magntude) k k k ResponseHXPULSE,PULSE ) -Mark (M agnltude Workng: BasePand_tn_m : nput: FFT Analyzer : ^jb^rttj k [Hz] k k k Fgure Loudspeaker frequency responses as transfer functons between mcs and power amplfer for dstances from mm up to m At fgure t s possble to observe the effect of the anechoc chamber on the low frequency response, due to the lack of accomplshment of the nverse squared law when the mcrophones are far away from the loudspeaker. The followng step conssts n calculatng the relatve transfer functons between mcrophone couples for dfferent dstances among these and the loudspeaker. Ths measurement allows to elmnate the loudspeaker response and therefore t wll not be nfluenced the measurement by the fact that the loudspeaker frequency response would be not flat. k k k k k ResponseHXPULSE,PULSE ) -Mark (M agnltude Wo rkng:basepand_m_m:lnp ut: FFT Analyzer fap \ \H^A Alt vr mrrr Mtwgn DO K k k Dk k [Hz] Fgure Transfer functon between mc and mc for couples of dstances from mm-mm to mm- mm from loudspeaker

7 From these measures, whch are obtaned easly by the FFT analyzer, t could be verfed the accomplshment of the nverse squared law, calculatng the average level n bands of / of octave (theoretcally, between mc and mc always there would be a level dfference of db, assumng free feld condton nsde the chamber). The results can observe at fgures and. supposes the method s the velocty of makng the measurement for every mcrophone poston. The FFT analyzer B&K allows to connect up to mcrophones smultaneously, wth what t s possble to obtan an analyss of the anechoc chamber n a short tme, always mnor that the used n the tradtonal methods These measurements have been realzed for dfferent analyss frequency span of the FFT analyzer, n order to obtan more detaled nformaton of dfferent parts of the spectrum. n all the cases there has been confgured a ponts FFT analyss and Hannng wndowng. t Two frequency span have been used:.khz and khz. n the same way, t s possble to work wth the temporal nformaton of the mcrophones obtaned from nverse transforms of the prevous transfer functons of, so from mc/ampl as from couples of mcrophones. [Pa/V/s] mpulse Response h(pulse,salda amplf) - U ark (Real P art) Fgure nverse squared law devaton for dstance couples from mm-mm to mm-mm n longtudnal drecton. Mcl s placed mm away from loudspeaker l ' \ Vll WVl/V - W V^Vv*~T* p ~-* Ou u,m,m m,m,m m,m m,m,m [s] Fgure Loudspeaker mpulse response from mcl (mm), mc (mm) and mc (mm) Fgure nverse squared law devaton for dstance couples from mm-mm to mm-mm n longtudnal drecton. Mcl s placed mm away from loudspeaker t s possble to observe that the curve that represents the average value of the devaton supports nsde the lmts allowed by SO -9, from the Hz / octave band. Ths result agrees totally wth the obtaned one for the methods descrbed n []. The advantage that At fgure t s possble to observe the attenuaton that suffers the loudspeaker mpulse response dependng on the dstance nsde the anechoc chamber. Fgure shows the loudspeaker mpulse responses for the mcrophones placed at mm, mm, mm and mm. n the last case, the effect of the opposte wall reflecton s observed (sgnal that comes wth ms delayed wth respect of the loudspeaker drect sgnal). n ths response the characterstcs of the loudspeaker (the frst part of the mpulse response) are mxed wth the characterstcs of the room. Therefore, to try obtanng nformaton from the anechoc chamber s necessary to elmnate the nformaton from the loudspeaker.

8 lmpulseresponsel(pulse,saldaamplf]-mark (Real Part] Workng :Baseband_.m_m: nput :FFT Analyzer lmpnfeersponseh(pul E.= LLE = ;-\' ak {FfedPjt} Wbrfcng Basefamd SSw m : nput :FFT - = /.=' ft**' f r*fjr* mpulser esponse h(pulse,saldaamplf]-mark (Real Part] Workng :Baseband_.m_tn: lput :FFTAnE >, lmpulsefleporjl-tpul E.PUl E)-M*k{FfedPj:} Vc-'< r ] : = = == ;:= :_l! r '_" : f"s.: FFT Ard^r d ff'"" r F :v lmpulseresponsel(pulse,saldaamplt]-mark (Real Part] Workng : Baseband_tn_m : nput: FFT Analyzer mpulser esponse h(pulse,pulse ] - M ark (Real P art] Workng :Baseband_tn_m : nput: FFT Analyzer Tr MMf M E ^»«*»- f o" mpulser esponse h(pulse,saldaamplt]-mark (Real Part] Workg: Baseband_m_m : nput: FFT Analyzer mpulser espouse h(pulse,pulse ] - M ark (Real P art] \fttorkng: BasePand_m_m : nput :FFTAnatyzer LLuj p"*f * * D Fgure Loudspeaker mpulse responses from mcs at mm, mm, mm and mm To do ths, t wll be calculated the mpulse response between the mcrophones mc and mc for the same pars of dstances that n case of the transfer functons. m m m m m m m m 9m Cm [s] Fgure Anechoc chamber mpulse responses couples from mm-mm to mm-mm The nformaton so obtaned corresponds only to the anechoc chamber characterstcs.

9 From the nformaton of the mpulse responses between pars of mcrophones, t s possble to try to get nformaton of the reverberaton tme n the anechoc chamber and later to obtan the nformaton of the absorpton coeffcent of the chamber, n order determne ts real cut-off frequency. The problem appears n the own nature of the anechoc chamber, whch produces mpulse responses of very short duraton and wth a quck energy decay of the. The method that one has proposed s a backward Schroeder ntegraton [] - [] realze from the mpulse response between pars of mcrophones separated always a double dstance. Ths method supposes a band-pass flterng of the mpulse response. Up to the moment the obtaned results are not too encouragng, though t wll contnue nvestgated n ths way.. CONCLUSONS AND FUTURE WORKS The conclusons that can be obtaned of the anechoc chamber characterzaton are the followng ones: The verfcaton of the nverse squared law (free feld condtons) completng s a complcated process by the great quantty of measurements that should be carred out and for the lack of exhaustve nformaton that the anechoc chambers owners provde. n employment of methods of havng multchannel measurements by means of FFT analyss and obtaned data post processng allow to become ndependent of the sound source characterstcs that we are usng (t usually presents radaton problems n low frequency and of drectvty leakage n hgh frequency). The faclty and rapdty of the nverse squared law (free feld condtons) checkng by means of the multchannel method, t does s preferable to the tradtonal methods. As for the future works: We wll contnue workng wth the transfer functons based on cross spectrum among mcrophones tryng to estmate the chamber absorpton coeffcent carryng out measurements nearest to the absorbent walls. Under the supposton of the behavor of the anechoc chamber lke a LT system when behaves as free feld, t wll allow to smulate the ntermedate dstances as an attenuaton and a delay, allowng to estmate the absorpton coeffcent of the chamber walls.. ACKNOWLEDGEMENTS We would lke to thank the work carred out by Oscar Vllarejo-Vllanueva and Teresa Haro-Marno durng ther Bachelor Degree Thess.. REFERENCES [] Gomez-Alfageme J.J., Sanchez-Bote J.L., Blanco- Martn E., "Desgn, Constructon, and Qualfcaton of the New Anechoc Chamber at Laboratoro de Sondo, Unversdad Poltecnca de Madrd", AES nd Conventon, Prepnt, Vena, May. [] SO :9, Determnaton of sound power levels of nose sources. Precson methods for anechoc and semanechoc rooms, SO Standards Handbook, Swtzerland, 99. [] Mlosevc M.A., Crc D.G., "Measurements of anechoc room characterstcs", J. Acoust. Soc. Am., (999) [] Cunefare K.A., Van Besel J., Holdhusen M, Tran J., Albanese A., "Anechoc chamber qualfcaton: Traverse method, nverse square law analyss method, and nature of test sgnal", J. Acoust. Soc. Am.,,. [] Wang J., Ca B., "Calculaton of free-feld devaton n an anechoc room", J. Acoust. Soc. Am., (99) [] Schroeder M.R. New Method of Measurng Reverberaton Tme J. Acoust. Soc. Am. 9 [] Schroeder M.R. ntegrated-mpulse method measurng sound decay wthout usng mpulses J. Acoust. Soc. Am. Vol. () 99 [] Karjalanen M, Antsalo P., Peltonen T., "Estmaton of Modal Decay Parameters from Nosy Response Measurements", J. Audo Eng. Soc, Vol., No., November.