Development of a hi-fi loudspeaker enclosure by using modal analysis

f A modal analysis system, 1 based on t h e D u a l - C h a n n e l Signal Analyzer T y p e 2032 or 2034, can be u s e d for optimizing t h e design of s t r u c t u r a l dyn a m i c systems. By m e a s u r i n g t h e compliance a t a set of p o i n t s on a s t r u c t u r e , t h e m o d al p a r a m e t e r s can be found, a n d a c o m p l e t e d y n a m i c model of t h e system can be p r o d u c e d . T h i s allows discovery of p o t e n tial vibration p r o b l e m s a t an early stage in t h e design. A n d the effect of s u b s e q u e n t m a s s , stiffness, a n d d a m p i n g modifications, designed to solve a n y problems, can be investigated. T h i s application n o t e pre. sents a s t u d y of how t h e Bruel&Kjaar modal analysis system was used to solve a colouration problem existing in t h e p r o t o t y p e design of a high quality loudspeaker system.

Modal analysis of a hi-fi loudspeaker. The Dual-Channel Signal Analyzer Type 2032/2034 is used to measure the compliance between the driving point and a number of response points; then a computer aids the extraction of the mode shapes.

Introduction
The objective of a modern high-fidelity loudspeaker is to "reproduce" natural sound images over a spacially large area. Technically speaking, a speaker must exhibit a flat frequency response over a wide range, and exhibit an even and smooth directional pattern, for the audible frequencies, in a reasonably large listening room. As a single speaker can never fulfil this goal, the modern loudspeaker is a rather complex system consisting of a set of dynamic speakers (drive units), a crossover network, and an enclosure. Building a system from high-quality individual elements does not guarantee a high quality result. A good loudspeaker should extract the best performance from each component, while ensuring optimum system compatibilBO 0243-11

ity. In particular, interaction with the listening room should be of prime consideration. The hi-fi loudspeaker comprises three fundamental parts, each having different objectives: D r i v e units. These provide the acoustic energy. The required frequency range is shared between a set of dynamic speakers: large membranes for the low-frequency range, and small membranes for the high-frequencies. C r o s s - o v e r n e t w o r k . This directs the part of the signal, with energy in a certain frequency band, to the appropriate drive unit. E n c l o s u r e . This is to provide a dynamically rigid structural sup-

port for the drive units, and to provide the appropriate acoustic loading for the drive units to obtain a good frequency response. This application note discusses how to solve the enclosure vibration problems, by using the tools of structural dynamic analysis.

Enclosure problems
An ideal enclosure would be infinitely rigid. It would secure the drive units in a fixed position, and be acoustically transparent. However, real enclosures are elastic structures having specific dynamic characteristics. The reaction forces of the drive units, and the sound pressure, excite the enclosure, causing it to vibrate. The structural vibrations are transmitted to the

air, contributing to the total sound image. Panel vibration can produce a significant contribution to the sound, being of the same order of magnitude as the direct drive-unit sound, at panel resonance frequencies. T h e panel sound is generally uncontrolled, interfering destructively with the direct sound, causing magnitude and phase errors in the frequency response, and causing the radiation pattern to deteriorate. Counter-measures can be taken against cabinet vibration. These include increasing the mass and/or stiffness of the panels to move the resonance frequencies, decreasing t h e compliance, and damping the panels to reduce resonance amplification. The enclosed air volume exhibits resonances (standing wave modes), which are excited by the internal sound pressure, and radiated through the drive-unit membranes and t h e panels, unless counteracted by internal damping and shielding. Enclosure design is a compromise between quality, price, size, mass, materials, vibration treatment, and appearance. Fig. /. Modelling and testing methods used to investigate the loudspeaker's colouration problem.

Enclosure vibration analysis in terms of structural dynamics

From now onwards we'll disregard the direct radiated sound, and concentrate on the undesired sound caused by the enclosure vibration. There are three elements to consider with every dynamics problem: the source, t h e path, and the receiver. Either one can be the cause of the problem, or the key to the solution. See Figure la. A loudspeaker enclosure can be characterised in the way described above. In this case: The receiver is the listener. If significant enclosure vibration is present, a box-sound will be perceived, the sound will be dependant on the listening position, and the quality of the stereo image will be dramatically degraded. If this is the case, not much can be done. If the listener doesn't like the sound, then he/she won't buy the loudspeaker. The vibration source is, in our opinion, the internal sound pres-

sure inside the enclosure, plus to a lesser extent, the reaction forces of the drive units. This point is a little controversial, as it is a common opinion that the reaction forces are dominant. The position of the drive units greatly affects the influence of the reaction forces. The dynamic energy is transmitted from the source through the enclosure's structure to the receiver. The sensitivity of the cabinet to vibration is described by its compliance (response/force). This is affected by the cabinet's mass, stiffness and damping distribution.

tern uses a process called curve fitting to compare the measured frequency response function with a theoretical model. From this model, the modal parameters, which give a statistically best curve fit, are extracted.

Development 01 a protOtvpe enclosure for a ,. , . r. , , . quality hl-tl loudspeak^Yl AudionOrd Dali 18. T e s t object The Dali 18 is a new high quality loudspeaker. The design is a threeway system comprising two 8 in woofers, one 4,5 in mid-range, and one 1 in moving-coil direct-radiation tweeter. The cross-over network is a third-order linear-directivity filter with asymmetric slopes. The enclosure is a bass reflex design of 90 litre volume, and is constructed from heavy chipboard and fiberboard, with a number of rib stiffeners. The selling price is approximately $800 on the US market.

The most efficient way of modelling the structural dynamics of the problem is to use modal analysis. This involves exciting the test structure, with a measurable force, in the frequency range of interest. The resulting dynamic response is measured, and the frequency response is calculated by using a Dual-Channel Analyzer Type 2032 or 2034. The modal analysis sys-

Subjective evaluation Measurements on the prototype had shown that the design goals had apparently been fulfilled. The next step was to subjectively evaluate the speaker. The listening panel concluded that the speaker was coloured in the 600 Hz region. The designer realized that the problem was caused by an enclosure resonance, and that a trial-and-error approach to the solution would be too time consuming and would produce dubious results. Consequently, he decided to attack the roots of the problem from a knowledge of the dynamic behaviour of the loudspeaker, obtained by using modal analysis. The objective of the measurements was to determine the inherent dynamic properties of the enclosure alone. A number of frequency response measurements were made at randomly chosen positions to gain an initial impression. Figure lc shows the instrumentation used for the introductory measurements. For convenience, hammer excitation was used. The results indicated high compliance in the troublesome frequency range (Figure 4 shows a frequency response measurement before and after enclosure modifications). To obtain a more comprehensive knowledge of the dynamic behaviour of the enclosure, a full modal test was made.

Fig. 2. Examples of estimated mode shapes, model frequencies and dampings for the loudspeaker enclosure. Frequency response measurem e n t s : d e t e r m i n i n g t h e compliance over the structure. A force transducer, mounted at the reference position, measured the force. The vibration exciter was connected to the force transducer by a flexible rod (stinger) to prevent any structural constraints. The response signal was produced by an accelerometer, which was mounted on the panels by using bees-wax. For each measurement, the accelerometer was moved to a new DOE The force and acceleration signals were supplied to the dualchannel analyzer via charge amphfiers. A frequency response measurement H I , representing the compliance between the reference point and each measurement point, was produced by the Type 2032/2034 analyzer. Each frequency response measurement was transferred to magnetic disk by a computer running a modal analysis software-package WT9100. Parameter estimation: curvefitting the m e a s u r e d f r e q u e n c y response. After completing the frequency response measurements (84 m total), the modal parameters (that is the moda frequency, damping and associated mode shapes) are extracted by the modal software. This is achieved by curve-fitting the frequency response data. The curve-fitter chosen is a "global" curve-fitter, which finds the modes representing the complete data set, in a least squares sense. This curve-fitting method works well with structures such as enclosures because it emphasizes "global" modes, and suppresses local modes, which can sometimes obscure the global modes.

ReSllltS
TwQ their of the estimated frequencies modegj and plus damp_

modal

ings> a r e g h o w n i n p i g u r e 2 animation the mode

Computer

Modal test of the enclosure

Measurement positions: struct u r a l d e g r e e s of f r e e d o m . The modal test produces a dynamic description of a discrete set of defined points and directions, termed degreesof-freedom. The points on the enclosure panels were defined approximately equidistantly in a 0,1m mesh. The measuring direction, for each point, is perpendicular to the surface. D r i v i n g point: c h o i c e of e x c i t a tion and r e f e r e n c e point. To produce enough force at high frequencies, a vibration exciter was used in place of the impact hammer. The Dual-Channel Signal Analyzer Type 2032/2034 supplied the vibration exciter with a random signal having a flat spectrum across the selected frequency range (0 to 1600 Hz). The vibration exciter was attached to the loudspeaker enclosure at a reference point on the side panel, in this case point # 1 3 , as the pre-analysis indicated high compliance here.

(slow m o t i o n simuiation) ghapeg reyealg

^ _

of a
in _ the

m o d e s b e h a v e ag n o r m a l m o d e g p a r t of t h e s t r u c t u r e phage other or 18Qo out

each

moves either with

.of.phase
aU the

partg;

and

partg

pagg

t h r Q u g h zerQ

simultaneously constitute

The the

modal

parameters in the

parameters

mathematical Cm b e u g e d in

.modal
predkt

model; w h k h programs happeng t0

computer

enable us to
dynamic

what

to the

properties of a system (resonances, mode shapes, sensitivity etc.), when the system is physically modified. Enclosure modifications: Applyi n g a d d i t i o n a l r i b stiffeners.
A n y of t h e m o d a l p a r a m e t e r g ; stiffness> and damping, magg;

can be modif i e d to affect the structural dynamics. T h e m o d a l analygig g y g t e m c a n gimu_ late t h e effectg of p o t e n t i a l modifica_
tions> before t h e y a r e implemented

By studying the mode shapes, with the response model borne in mind, we can see that it is mode no. 3 that is the most likely cause of the colouration. 3

This is evident because the mode shape exhibits what is called an air mode, that is it breathes and changes volume. In this enclosure, only the bass sound is present, as the midrange and tweeter units are encased in their own enclosures. Consequently, only wavelengths which are long, relative to physical dimensions of the cabinet, are present. Assuming equal pressure fluctuations throughout the cabinet, only an air mode can be excited by the internal sound pressure fluctuations. Fortunately, air modes are easily modified by bracing the parts of the structure that move in opposite phase. From the appearance of the mode shape, it would seem reasonable to place a stiffener between the airmode's anti-phase points, illustrated in Figure 2. Reasoning along this path leads to a number of stiffness modifications, shown in Figure 3.

Fig. 3. Application

of rib-stiffeners to the enclosure.

F

Has the colouration problem been solved?

After the loudspeaker enclosure was modified, it was re-tested for verification purposes. Initially, frequency response measurements were made at suitable points to see if the resonant behaviour had decreased. The first measurements showed that further modifications were needed. In total, three modifications were completed. Figure 4 shows a driving point compliance measurement, before and after modification. The compliance in the critical frequency range has been attenuated by approximately 20 dB. Note that the modification effecting the compliance at mode 3 does not effect the compliance of the constant volume modes 1, 4 and 5. The modal model allows modifications to be tested analytically, before e they are implemented. However, a trial-and-error approach, based on a comprehensive knowledge of the enclosure dynamics found from a modal il test, was considered best. This was because the forcing function, being pressure distributed over all points, plus s drive-unit reactions, is complicated I and, due to the lack of knowledge of f the force distribution, is difficult to 3 model.

Fig. 4. Driving point compliance measurement, before (thick line) and after (thin line) structural modification.

Final results and conclusions

The last and most critical test - the subjective evaluation by a listening panel - was passed with flying colours. Science, as applied to arts such as hifi, is necessary, but certainly not sufficient. Mechanisms other than functions based on ideal linear models can behave in a unpredictable manner because they cannot be completely described by simplified models. However, a modal analysis system based around a Dual-Channel Analyzer Type 2032 or 2034 is ideal for investigating the measurable properties. In this application, modal analysis enabled the enclosure structure to be optimized dynamically. A modal analysis system aids comprehensive understanding of structures, and supports the designer in product and prototype development.