Audio Engineering Society

Convention Paper
Presented at the 122nd Convention 2007 May 58 Vienna, Austria
The papers at this Convention have been selected on the basis of a submitted abstract and extended precis that have been peer reviewed by at least two qualied anonymous reviewers. This convention paper has been reproduced from the authors advance manuscript, without editing, corrections, or consideration by the Review Board. The AES takes no responsibility for the contents. Additional papers may be obtained by sending request and remittance to Audio Engineering Society, 60 East 42nd Street, New York, New York 10165-2520, USA; also see www.aes.org. All rights reserved. Reproduction of this paper, or any portion thereof, is not permitted without direct permission from the Journal of the Audio Engineering Society.

The Flexible Bass Absorber

Acoustics, Scion-DTU, Diplomvej, Bygn. 376, DK-2800 Kgs. Lyngby, Denmark Technology, rstedDTU, Technical University of Denmark, Building 352, rsteds Plads, 2800 Kgs. Lyngby, Denmark

2 Acoustic

Correspondence should be addressed to Niels W. Adelman-Larsen (Author) (nwal@flexac.com) ABSTRACT Multi-purpose concert halls face a dilemma. They host dierent performance types that require signicantly dierent acoustic conditions in order to provide the best sound quality to both the performers, sound engineers and the audience. Pop and rock music often contain high levels of bass sound energy but still require high denition for good sound quality. The mid- and high-frequency absorption is easily regulated, but adjusting the low-frequency absorption has typically been too expensive or requires too much space to be practical for multi-purpose halls. A practical solution to this dilemma has been developed. Measurements were made on a variable and mobile low-frequency absorber. The paper presents the results of prototype sound absorption measurements as well as elements of the design.

1. BACKGROUND The design of concert halls for classical music benets from centuries of acoustical research and experience. This is demonstrated by the quantity of books and papers describing the acoustics of these halls and providing guidelines for designing new halls. The amount of literature available on the recommended room acoustics for popular music is much smaller, and it seems fair to bring more attention to this subject. In the context of this paper, the term popular music is used to describe rhythmic genres, such as rock, pop, and jazz. Most concert halls that are used for rock and pop concerts are built to serve multiple purposes, including classical music con-

certs and speech performances. The acoustics of a venue can make or break the enjoyment of a musical performance. Performers and audience members may choose to not return to a venue because of a previous experience with unsatisfactory acoustical conditions that adversely affected the perceived quality of the performance or even affected the performance itself. In a recent study [1] among professional rock and pop artists, it was found that at least 1/3 of the artists surveyed avoid venues with unsatisfactory acoustics when planning their tours and all, including sound engineers, found the room acoustics to be very important to their job. Evidently, the audience is affected both by the qual-

Adelman-Larsen et al.

The Flexible Bass Absorber

ity of the performance on the stage and by the actual sound quality in the hall. The rst author of this paper is also a professional musician, educated at the Berklee College of Music, and has performed in well over 1200 concerts within the genres of popular music. This experience provided the background and personal motivation for more in depth investigations into recommendations for room acoustics for popular music. This research eventually led to the conclusion that additional bass absorption is a necessity, and particularly a variable bass absorber. Popular music performances are almost always amplied by powerful PA systems. The sound pressure level at rock concerts is often well above 100 dB in the center of the hall, several meters from the stage. On stage, the sound sources (i.e. speakers, monitors, instruments) are also radiating sound levels of this magnitude in a frequency range from 40 Hz and up toward the performers and the audience. A typical four-string bass guitar has a low E string tuned to E1 or about 41 Hz, and the bass drum and oor tom of a typical rock drum set vibrate in a similar frequency range. So the acoustic design and measurement range for a popular music venue should extend down to include at least the 63 Hz frequency band, and not begin at 125 Hz as is typical for the design of classical music halls. In [1], 20 Danish popular music halls were measured acoustically, and a survey of 25 rock musicians and 8 sound engineers that perform in these halls was made. In the survey, the respondents were asked their opinion on the halls acoustics. The survey showed that the most important subjective parameter for a hall to score a high rating was clarity. In fact bass clarity correlated with a coefcient of 0.74 to overall impression and mid-treble clarity correlated with a coefcient of 0.70 to overall impression. Clarity is generally inversely proportional with the reverberation time and, therefore, is proportional to the amount of absorption in the room. In the mid-treble frequency range, sound absorption, and thereby clarity, is usually obtained through the presence of the audience and can be easily regulated with curtains. However, in the low-frequency range, clarity is not so easily obtained. Bass sound radiates from speaker cabinets in all directions. Since bass frequencies have a very long
Absorption Coefficient ()

1.2 1 0.8 0.6 0.4 0.2 0 63 125

Standing, 2.7 pers/m2 Seated, 2 pers/m [2] 2 Seated, 1 pers/m [2] 250 500 1000 Frequency (Hz) 2000 4000
2

Fig. 1: Absorption coefcients for an audience. Standing audience with a density of about 2.7 people per m2 (circles, from [1]) and an audience seated on wooden chairs with a density of 2 people per m2 (diamonds) and 1 person per m2 (squares). Data for the seated audience are from [2], adapted by [3].

wavelength (3-10 meters), the sound radiates from sub-bass cabinets almost omnidirectionally. Therefore, bass sound can not be aimed at the audience in the same way as mid- and high-frequency sound can be. This sound will radiate in all directions and will be reected from most surfaces in the room. Sound is reected by hard surfaces of the hall. Most building materials or construction methods do not absorb bass sound energy. Therefore, most halls have a long reverberation time in the low-frequency range. On account of this long reverberation time and the omnidirectionality of the subwoofers, the distance from the subwoofer at which the reverberant sound is as loud as the direct sound (often referred to as the critical distance) becomes very short - often just a few meters. Further away than this distance the bass sound is often described as muddy, boomy and undened. Bass sound is not absorbed by the audience. The audience absorbs a lot of sound, but mainly midand high-frequencies. The sound absorption provided by a tightly packed standing rock audience (23 people per m2 ) is quite small in the low frequencies, relative to the high frequency absorption. According to [1], a standing audience absorbs less than

AES 122nd Convention, Vienna, Austria, 2007 May 58 Page 2 of 8

Adelman-Larsen et al.

The Flexible Bass Absorber

a quarter of the sound energy in the low-frequencies as it does in mid-to-high-frequencies (see Figure 1). Very high levels of bass sound. In rock and pop music, the bass frequencies from 40-125 Hz get amplied to very loud levels. Easily half of the electrical power of the PA and monitor system goes to these 1.5 octaves. Why? A) Because it sounds right. B) Because it feels good. High levels of bass have been shown to stimulate the part of the brain that controls such basic instincts as sexual desire and hunger [4]. Low-frequency sounds at these high levels will also mask mid- and high-frequency sounds, thereby affecting the perceived clarity in the higher frequency ranges.

2.1. Theory of membrane absorbers A membrane absorber consists of a light plate in front of a closed cavity, which is often lled with a porous material that provides damping for the system. When deriving the theoretical characteristic equations for a membrane absorber, the walls and back of the cavity are assumed to be rigid and the bending stiffness in the plate is assumed to be negligible compared to the stiffness of the air column in the cavity. In addition, the system is assumed to be locally reacting and the sound is assumed to reach the membrane with normal incidence. The system is characterized by the mass per unit area of the plate, M, the depth of the cavity, d, and the internal losses of the system, ri , e.g., losses due to the ow resistance of the porous material, internal losses in the plate, and losses in the joints along the edges of the plate [3]. The acoustic impedance of the system, Z, is given as Z = ri + j 2 f M 0 c2 2 f d , (1)

Articial reverberation is usually added to recordings and also live music performances for the mid- and highfrequency ranges. However, informal discussions with audio engineers and bass players give the perspective that reverberation is rarely, if ever, added to bass-frequencies. This supports the idea that a hall should be as dry as possible in the low-frequencies, in exactly the frequency range where the audience provides little sound absorption. This is why additional bass absorption is usually needed in popular music concerts. 2. MEMBRANE ABSORBERS Membrane absorbers are often used as efcient lowfrequency absorbers and therefore were selected as a starting concept in the attempt to create a variable lowfrequency absorber. An idea arose that an air mattress type construction could be used as a variable membrane absorber because it consists of an inatable cavity with a thin membrane. By inating and deating the mattresses, the sound absorption of the air mattress could be switched on and off. Commercially available air mattresses were chosen for initial proof-of-concept tests. The rst goal of this project was to investigate the sound absorption properties of air mattresses. Later, traditional soft layer membrane absorbers (specially constructed boxes with a membrane of a soft material hereafter referred to as test boxes) were used for nding a suitable material and to develop a knowledge base. Finally, these results were used to design a variable sound absorber with signicant peak absorption, a resonance frequency in the range of 63 to 125 Hz and a reasonable half-power bandwidth.

where c is the speed of sound in air (approx. 343 m/s) and 0 is the density of air (approx. 1.2 kg/m3 at standard ambient temperature and pressure). The resonance frequency of the system, f0 , is found when {Z} = 0, or when c 0 1/2 60 kg m f0 = . (2) 2 Md s Md This shows that the resonance frequency, where the absorption should be highest, is inversely proportional to the square root of both the mass of the membrane, M, and the depth of the cavity, d. According to this theory, in order to have a maximum absorption at around 63 Hz, with d = 0.2 m, the membrane must have a mass of about M = 5 kg/m2 . The impedance of the absorber can be tuned in order to maximize the absorption at the resonance frequency and the usable bandwidth of the absorber (half-power bandwidth, Br ). If the impedance is too high, relative to the radiation resistance of the membrane, rs = 0 c, then the incident sound eld will reect off of the membrane and not be absorbed. If the impedance is too low, then the internal losses will be too small and not enough sound energy will be absorbed. The impedance ratio of the internal losses and the external radiation resistance can be expressed as ri (3) = , rs

AES 122nd Convention, Vienna, Austria, 2007 May 58 Page 3 of 8

Adelman-Larsen et al.

The Flexible Bass Absorber

then the maximum absorption coefcient, max , and bandwidth, Br , can be written as max = 4 (1 + )2 0 d . M (4)

Br = (1 + ) f0

(5)

With these theoretical equations in mind, some parameters of the test samples were varied to achieve the project goals. It should be noted that many of the assumptions in deriving the theoretical equations describing membrane absorbers do not apply to an air mattress, including that the membranes are not simply supported around the edges, the walls are not rigid, the membrane is not locally reacting, and the sound eld is diffuse, not normally incident. In addition, the air pressure inside the mattress is above atmospheric pressure, so the density of the air inside the system is higher than 1.2 kg/m3 . Therefore, some deviations from the theory of equations 2 and 5 are expected. 3. MEASUREMENTS The measurements of the absorption coefcients were carried out according to ISO 354:1985 [5], with some deviations. The DIRAC software was used to generate test signals and measure the impulse responses of the reverberation room with and without test samples. A linear sweep of either 11 or 21 s duration was used for the signal and the results from six measurement positions were averaged. A dodecahedral speaker array with an additional subwoofer was used as the sound source and the room response was recorded by a microphone connected to the computer running the DIRAC software. 3.1. Air Mattresses Measurements were performed on four sets of off-theshelf air mattresses with different dimensions in order to determine the effect of the dimensions on the resonance frequency and absorption characteristics of the mattresses. The four mattress types will be referred to as FirstEver, Single, Double and DoubleThick. The relevant properties of the mattresses are shown in Table 1. The Single and Double mattresses were manufactured by the same company and were constructed of the same material, so only differed in mattress size. The other two mattresses had different manufacturers and slightly different materials, although all mattresses were made of PVC based membranes. The main body of each mattress did not have a uniform thickness, but had a regular

Table 1: Critical dimensions of the samples of the four air mattress types used in the experiments. Mattress Thickness Mass Sample Area Type d M A (cm) (kg/m2 ) (m2 ) FirstEver 20 0.5 9.9 Single 12 0.5 12.1 Double 12 0.4 11 DoubleThick 40 0.5 13.7

pattern of indentations; either long columns or discrete points, from the internal webbing that helps the mattress maintain a generally at shape when inated. Therefore, the mattress thicknesses listed in Table 1 are an approximate average over the mattress area. 3.2. Test boxes Four wooden test boxes were constructed with a plywood bottom and 22 mm thick walls that were 10 cm tall and gave the box a total outer dimension of 240100 cm2 . The wood was sealed on the inside with a layer of liquid sealant and on top with a window seal lining. A soft plastic membrane was stretched across the top of the box and clamped around the edges with a 2222 mm2 wooden bar, screwed down around the top of the frame. A hole was drilled in a side of the box, to which an air pump could be tted, and was blocked with a cork during measurements. The membrane used for these initial tests was a 2 mm extruded soft PVC. This material was chosen because it had an appropriate mass (3 kg/m2 ) and was locally available. The soft material was chosen so that various depths could be tested through ination and deation without having to modify the test box, and to have a high internal damping so that there would not be as much need for extra damping material inside the box. It was expected that the high elasticity module of the membrane would provide a high degree of internal damping. For the tests, two Rockwool Flexi A-batts (306051000 mm3 , air ow resistance 8.3 kPas/m2 ) were used inside each test box. The absorption of the test boxes was measured with the membrane approximately level with the top of the frame (i.e., 10 cm depth, referred to as inated). In addition, measurements were made with two Rockwool batts in the box with the membrane inated to a depth of about

AES 122nd Convention, Vienna, Austria, 2007 May 58 Page 4 of 8

Adelman-Larsen et al.

The Flexible Bass Absorber

Absorption Coefficient,

0.6

Absorption Coefficient,

0.8

FirstEver Single Double DoubleThick

0.8

Hyperinflated Inflated Vacuum deflated

0.6

0.4

0.4

0.2

0.2

63

125

250 500 Frequency (Hz)

1000

2000

63

125

250 500 Frequency (Hz)

1000

2000

Fig. 2: Measured sound absorption coefcients for the four types of air mattresses: FirstEver (diamonds), Single (triangles), Double (squares), and DoubleThick (circles).

Fig. 3: Absorption coefcients measured on the test boxes with a 2 mm membrane. The three conditions represent three levels of ination: hyper-ination to d = 25 cm, normal ination to d = 12 cm, and vacuum deated with d = 2.5 cm.

25 cm (hyper inated) and vacuum deated so that as much air as possible was removed from the box. Since the Rockwool batts were in the box, the deated depth was still around 2.5 cm. In order to be installed in a concert hall, a sound absorber must conform with re safety codes. Therefore, a material with good predicted acoustical properties as well as a good re safety rating (UL 94 V-0) was identied. The density of the material is 1.15 g/cm3 . The sound absorbing properties of this membrane was also measured on the test boxes. The tested membrane thicknesses were 2.0 mm and 2.4 mm (2.3 kg/m2 and 2.8 kg/m2 , respectively) and the membranes were inated to depths of about 15 and 20 cm. According to equation 2, these congurations should yield maximum sound absorption at resonance frequencies around 80-100 Hz. These measurements were performed with 0, 2 and 4 Rockwool batts in the test boxes to see the effect of increasing the internal damping of the system. 3.3. Prototype tests Since a high degree of absorption was found to remain in the deated system, a design was developed that included a mechanism for stowing the absorber when not in use. A rolling mechanism was designed that removes the exible absorber (and its sound absorption) while providing easy installation and easy adjustment of the reverberation

in the multipurpose hall. The rolling mechanism only takes up about 5% of the total area of the absorber. A typical absorber unit may provide 30 m2 of absorbing area with about 1.4 m2 of rolling mechanism. Four prototypes, each with dimensions 175275 cm2 , were constructed for the measurements of the sound absorption of the system. Two prototypes included a porous sheet stretched inside the cavity and two had no additional damping. All prototypes had an internal webbing with a spacing of 80 cm to maintain an overall at shape when inated. 4. RESULTS 4.1. Air mattresses The rst measurements were made on the FirstEver mattresses as a proof-of-concept measurement, to verify that an air mattress could in fact be used as a sound absorber. The results of this measurement were very promising (see Fig. 2). The absorption curves show a signicant peak around 315-400 Hz with a maximum value of about 0.8 and a half-power bandwidth of about three octaves. This demonstrated that an air mattress can in fact be used as an effective mid-range sound absorber. However, the resonance frequency is too high for the de-

AES 122nd Convention, Vienna, Austria, 2007 May 58 Page 5 of 8

Adelman-Larsen et al.

The Flexible Bass Absorber

membrane thickness: 2.4 mm

20cm, 2rw 15cm, 2rw 15cm, 0rw Absorption Coefficient ()

0.8

20cm, 2rw 15cm, 2rw 20cm, 4rw 15cm, 4rw

membrane thickness: 2.0 mm

0.6

0.6

0.4

0.4

0.2

0.2

40

50

63

80 100 Frequency (Hz)

125

160

200

40

50

63

80 100 Frequency (Hz)

125

160

200

Fig. 4: Absorption coefcients measured on the test boxes with the 2.4 mm re retardant membrane. Curves are shown for d = 20 cm with two Rockwool (rw) batts (squares), and for d = 15 cm with two rw batts (circles) and with no rw batts (diamonds).

Fig. 5: Absorption coefcients measured on the test boxes with the 2.0 mm re retardant membrane. Curves are shown for d = 20 cm (squares), and for d = 15 cm (circles) with two rw batts (open symbols) and with four rw batts (shaded symbols).

sign criteria, so the challenge remained to shift the frequency of maximum absorption toward lower frequencies without signicantly reducing the maximum absorption coefcient, , or the bandwidth. The resonance frequencies of the air mattresses do not follow the simple model suggested by membrane absorber theory (see section 2.1), and the four absorption curves are surprisingly similar, despite the different dimensions (see Table 1). All four membranes have a membrane mass of about 0.5 kg/m2 . With a depth of about 0.2 m, the resonance frequency of a membrane absorber should be about 190 Hz. The higher resonance frequency in the air mattresses could be due to the higher air pressure and membrane tension in the inated mattress. As the mattress is inated, the pressure and membrane tension increase, increasing the stiffness of the system. Increasing the stiffness increases the resonance frequency of the mattress. The internal structure of the mattress limits the motion of the membrane, provides additional moving mass, and subdivides the membrane into small panels. By subdividing the membrane into small panels, the small areas may not be large enough to vibrate freely at the low frequencies. If it is assumed that these factors only change the proportionality constant in equation 2, then the product of the mass and depth must increase by a factor of sixteen in order to reduce the measured resonance frequency by a factor of four to be

within the target range. 4.2. Test boxes, PVC membrane The absorption of the test boxes with the 2 mm membrane (M = 3 kg/m2 ) and two Rockwool batts in each test box was measured with normal ination (d = 12 cm), hyper-inated (d = 25 cm) and vacuum deated (see Fig. 3). In the hyper-inated position, max 0.8 is seen in the 63 Hz frequency band, as predicted by equation 2, which is within the design frequency range. The bandwidth of the absorption curve is over 4/3-octaves, so the three main sound absorption criteria are fullled. However, when the test box is switched off, in the vacuum deated state, there is still a signicant peak seen in the absorption curve at around 200 Hz. With two Rockwool batts in the test box, the membrane still had a residual depth of about 2.5 cm in the vacuum deated state. Without Rockwool and with the membrane completely ush with the bottom of the box, this peak should disappear, but then the damping of the system and the inated-state absorption curves would be changed. 4.3. Test boxes, re retardant membrane The same sound absorption measurements were made with the re retardant membranes on the test boxes. The measured absorption coefcients for a 2.4-mm-thick membrane and 2.0-mm-thick membrane are shown in Figs. 4 and 5, respectively. Both membranes were mea-

AES 122nd Convention, Vienna, Austria, 2007 May 58 Page 6 of 8

Adelman-Larsen et al.

The Flexible Bass Absorber

Absorption Coefficient ()

Absorption Coefficient ()

0.8 samples on floor 0.6

2.0mm, with damping 2.4mm, with damping 2.0mm, no damping 2.4mm, no damping

0.8 hanging samples 0.6

2.0mm, with damping 2.4mm, with damping 2.0mm, no damping 2.4mm, no damping

0.4

0.4

0.2

0.2

40

50

63

80 100 Frequency (Hz)

125

160

200

40

50

63

80 100 Frequency (Hz)

125

160

200

Fig. 6: Measured absorption coefcients for prototypes lying on the reverberation room oor. Prototypes had either a 2.0 mm (open symbols) or 2.4 mm (shaded symbols) membrane, and either had an internal damping layer (circles) or had no extra internal damping (squares).

Fig. 7: Measured absorption coefcients for prototypes hanging in the reverberation room. Prototypes had either a 2.0 mm (open symbols) or 2.4 mm (shaded symbols) membrane, and either had an internal damping layer (circles) or had no extra internal damping (squares).

sured with d = 20 cm and d = 15 cm, and with 0, 2, or 4 Rockwool (rw) batts in each test box. The measured resonance frequencies are between 80-100 Hz, as predicted by Equation 2. With the 2.4 mm membrane, f0 decreases with the increase from 15 to 20 cm (compare squares and circles in Fig. 4), and adding two Rockwool batts increases the absorption around the f0 . In Fig. 5, with a 2.0 mm membrane, there is a strong secondary peak in the absorption curve in the 50 Hz band with two Rockwool batts (open symbols). This peak is greatly reduced or eliminated by the addition of two more Rockwool batts in each test box (shaded symbols). With both membrane thicknesses, the maximum absorption coefcient, max , is around 0.7. 4.4. Prototype measurements Four prototype systems were made and tested for their sound absorption properties. Two prototypes were welded from 2.0 mm thick membranes and two with 2.4 mm membranes. One prototype of each membrane thickness had an internal damping layer, and the other had no additional internal damping. The measured absorption coefcients are shown in Fig. 6 with the prototypes against the oor. In all four cases, max is about 0.6 and is in the 80 Hz 1/3-octave band. There were only small differences between the four prototypes. The prototypes were also tested hanging freely in the reverberation room instead of against a wall or oor. When

hanging, the front and back of the absorber are acoustically active. This effectively halves the depth of the membrane absorber for calculation of the resonance frequency (equation 2), thereby increasing f0 by a factor of 2. The absorption coefcients shown in Figure 7 show that f0 is indeed increased to the 100 Hz 1/3-octave band, where max is still around 0.6. In all conditions, the bandwidth of the absorption is almost 2 octaves. 5. DISCUSSION AND DESIGN The concept of an inatable membrane sound absorber has been proven as an effective, exible low-frequency sound absorber. The low-frequency absorption can be switched on by inating the bladder so that the plastic membrane absorber is most effective around 80-100 Hz with a maximum absorption coefcient of 0.6 and a bandwidth of almost 2 octaves. In order to change the reverberation time in a room perceptibly, a large area of the walls and/or ceiling must be covered with the sound absorber. A exible sound absorber system can be installed with several absorbers which can be operated by the push of a single button. When the hall is to be set up for classical music, the absorbers can be rolled up, and when popular music is to be played, the absorbers can be rolled down and inated. A small demo system will be installed in April/May 2007 at the Technical University of Denmark.

AES 122nd Convention, Vienna, Austria, 2007 May 58 Page 7 of 8

Adelman-Larsen et al.

The Flexible Bass Absorber

The authors hope that this exible sound absorber system will help to enable multi-purpose concert halls to provide acoustics that allow the artists to deliver their best performance and the audience to enjoy the show. Patents are pending on the basic concept and several possible designs. 6. ACKNOWLEDGMENTS This project and future product development were and are supported by grants from DTU-Innovation/Seed Capital Denmark, including funding from the Danish government and a number of Danish companies, including Carlsberg. 7. REFERENCES [1] N. W. Adelman-Larsen, E. R. Thompson, and A. C. Gade, Acoustics in rock and pop music halls, in 122nd Audio Engineering Society Convention, (Vienna, Austria), May 5-8 2007. [2] E. Meyer, H. Kutruff, and D. Kunstmann, Uber einige Messungen zur Schallabsorption von Publikum [On some measurements of sound absorption of the audience], Acustica, vol. 14, no. 2, pp. 119 124, 1964. [3] H. Kuttruff, Room Acoustics. London: Spon Press, 4th ed., 2000. [4] N. P. M. Todd and F. W. Cody, Vestibular responses to loud dance music: A physiological basis of the rock and roll threshold?, J Acoust Soc Am, vol. 107, pp. 496500, Jan. 2000. [5] International Organization for Standardization, ISO 354:1985: Acoustics - Measurement of Sound Absorption in a Reverberation Room. International Organization for Standardization, 1985.

AES 122nd Convention, Vienna, Austria, 2007 May 58 Page 8 of 8