The loudspeaker system and the listening environment conditions in many ways the quality of domestic musical reproduction. The present Hi-fi components, even the best, cannot inspire the same feelings as a live listening, because of the great differences in the acoustic field. For a better quality of domestic reproduction the first step to take is therefore to take into account these differences and then find the way to reduce them.
DIFFERENCES BETWEEN LIVE MUSIC AND REPRODUCED MUSIC
If we want to have at home the same feelings experienced in live listening, it is necessary first of all to define the factors responsible for these feelings. We consider the following three essential factors: dynamics, imaging, and transparency. They represent the principal difference between live music and music reproduced in domestic environments.
DYNAMICS: the experiments done in typical conditions of live listening point out that a single acoustic instrument could cause a peak level between 110 and 130 dB; the human voice could reach levels above 110 dB. An orchestra causes a total level equal to the sum of those generated by each of the employed instruments; in a concert hall the level obviously depends on the listening distance and could reach a 130 dB peak during the transitory with maximum intensity in the first raws. Such sound levels (130 dB) are at least 10 times greater than those attainable from the hi-fi components. On the other hand, the noise in concert rooms is generally much lower than that in the domestic environments. If we consider a difference of only 10 dB in floor noise, the ensuing difference in dynamics between Iive listening and home listening is around 30 dB at best. In most practical cases this difference is greater (typical noise of 50 dB with maximum levels of little more than 100 dB) and the dynamics of a reproduced musical event is also 1000 times lower than that of the same event played live. This conditions in a heavy way the listening quality: better dynamics surely involves a better quality. We know that dynamics is the interval between the maximum and the minimum level; since the minimum level is limited from the environmental noise, the dynamic quality of a loudspeaker depends on the maximum reproducible level (in conditions of low distortion). Due to the known limitations in domestic environments, if we want a realistic listening it is necessary to use systems that can produce peak levels of 130 dB. Even though this value can seem exaggerated to many people, such is not the case, as shown from the following examples: during a battery solo (live execution) we have measured a peak level of 128 dB, with a RMS level of 90 dB; during the reproduction of a CD (one of those in which the dynamics of the support are exploited, with the classical WARNING! writing in cover) we have recorded a 121 dB peak at 5 meters from a couple of bookshelf loudspeakers, with a RMS level of 86 dB. As we can see, even in presence of a not so high mean level (RMS power under 1 Watt), in both cases the peak level is greater than 120 dB (in the case of the reproduction of a CD the peak level is limited from the power of the amplifier used and from the use of bookshelf loudspeakers; moreover, the environment has a notable absorption: T60 between 0.7 sec. at low frequencies and 0.3 sec. at mid-high). Therefore, to experience the same feelings during a concert, it is necessary to have at home a comparable dynamics, i.e. substantially similar maximum peak level. In music these peaks have an elevated value, but their duration is generally very brief, so the relative energy is limited: a peak level of 130 dB is very different from a mean level of 130 dB (we have seen that to a 130 dB peak could correspond a 90 dB of mean level, that is a value of acoustic pressure 100 times lower). If we want a realistic and exciting listening, the reproduction chain must be able to reproduce these peaks to the corresponding live level.
IMAGING: the imaging of a reproduction system is tightly linked to the type of radiation employed in the source, while the acoustic general result depends obviously also on the microphone recording technique. The majority of the loudspeakers use a mono-pole radiation type (fig. 1a): this is what we get, in fact, when mounting a loudspeaker (that is for its nature a dipole, being constituted from a vibrating membrane whose two surfaces radiate in phase opposition) in a box that absorbs the back radiation and theoretically cancels it. In the ideal case the only front radiation of the membrane would reach the listener with a direct and an indirect component (due to the reflections on the walls of the listening environment). With this type of radiation, the delay between a direct wave and a reflected wave is, in a normal domestic environment, very short. As the information relative to the imaging and to the environment depends on the phase shift between direct and reverberated components, with a monopole radiation loudspeaker the imaging perceived is limited. A notable advantage in terms of extension of the acoustic scene can instead be achieved (fig. 1b) by using a dipole system; in this case to the direct radiation is added the reflected component (due mainly to the back surface of the vibrating membrane, that radiates in opposite direction and with a l80° phase shift) that arrives to the listener attenuated and with notable delay: a situation similar to that in a live listening. A further step ahead in the reconstruction of the acoustic scene could be made using a multipole (fig. 1c) instead of a simple dipole so as to have full control on the irradiation diagram; in theory, by varying the number and the relative position of the employed transducers, we could carry out any type of diagram. In the example shown in figure 1c the principal lobes are more crushed respect to a simple dipole, indicating greater directivity; besides, there are secondary lobes of small entity. What we have in practice is more directivity for the principal radiation and an increased number of secondary components addressed along opportune directions, experimentally assessed. This situation is even more similar to what happens in live executions, allowing an elevated feeling of likelihood of the acoustic scene. TRANSPARENCY: in domestic listening it is usually impossible to hear the details and shades that are perceived in live events. This is due to the fact that a good part of the information contained in the original event is lost: a small part gets lost during the record production, while a significant quantity is lost during the reproduction process. The loss of information during the reproduction is due to the low resolution of the hi-fi systems, to the bad acoustic characteristics of the domestic places and to a high floor noise. The last limit, obviously, the least audible level, since our ear could not interpret correctly sounds whose level is not remarkable higher than floor noise. The main limit of the hi-fi systems lies in electroacoustic transducers.
First, low quality loudspeakers cannot carefully reproduce the musical signal: because of their inertia, the loudspeakers incline rather to store and release the energy with time constants very greater than those of the musical signal. The ensuing effect is that all the transitory signals with short rising times and little amplitude are not reproduced from the loudspeaker. The situation is even worse, since it is necessary to consider also the effects introduced by the acoustic box and by the crossover, whose contribution to the worsening of the transparency is generally greater than that of the same loudspeaker. All this results in the loss of information concerning low level signals, with a substantial reduction of the feeling of transparency.
Second, it is necessary to take into account the non-linearity in the transfer of acoustic energy: perhaps a few people know that only up to a certain value of pressure - around 100dB - there is direct proportionality between the vibration velocity of the particles of the elastic transmission medium (the air) and the acoustic intensity (that is the pressure by surface unit). In simpler terms, we could say that the air, that is the half within which the transmission of acoustic energy happens, is linear only for limited intensity signals. Since pressure is the product of the intensity for the emitting surface, we can deduce that it is better to have low intensity and big radiant surface rather than high intensity with small radiant surfaces (in this case the linearity limits of the air are overcome). Now, the substantial difference from this point of view between live and reproduction is that in the first case the acoustic energy density is, for the same radiated power, much lower respect to domestic listening, because the emitting surface is greater. For instance, if we consider an orchestra that irradiates a power of 10 W (acoustic, corresponding to around 120 dB), we have an intensity near the source of some cent of watt per square meter; if the same power is delivered by a loudspeaker with an emitting surface of 0,013 m2 (a 6" mid-woofer, that is 16 cm.) the acoustic intensity would be instead of around 750 W/ m2, that is around 10,000 (!) times greater, with consequent non-linearity due to the overcoming of the maximum values allowed. In order to avoid that, it is advisable to have the bigger radiant surface possible to minimize the acoustic intensity for a given pressure. On the other hand, a big emitting surface allows an elevated output and also good acoustic impedance; this leads to a good interface with the amplifier and the environment. In case this does not happen, most of the energy sent from the amplifier to the transducers is reflected towards the same amplifier and not converted into pressure, with consequent loss of transparency due to the reflections in the connection line. Consider, for example, the characteristics common to all the planar systems, renowned for their imaging and transparency: the vibrating membranes are very wide, allowing a low density of energy and an excellent acoustic impedance; they are very light and, usually, they have no acoustic boxes and crossover. This ensure a significant transparency, despite the fact that the output of these systems is always very low; the imaging is clearly due to the type of radiation, that has also a back radiation.
Following the above considerations, we have designed loudspeaker systems to improve the reproduction in terms of dynamics, imaging and transparency. The first prototype has required a search and development time of around 10 years, while the definitive product has seen the light only after an ulterior refinement period that has required, among other demands, the complete redesign of the section appointed to the low frequencies reproduction, surely the more critical point in any loudspeaker. In fact, if it is true that the acoustic box itself is responsible for poor musical performances, it is also true that without it, it is impossible to obtain satisfying dynamics in the low frequencies area: to overcome this drawback, we have developed a new and original system of low frequence reproduction, denominated "DuaI Balanced Line" (DBL). The DBL system employs two transmission lines, each of which is designed to compensate the irregularities of the other. A transmission line presents in fact, in its frequency response, a series of faults the amplitude and Q-factor of which depend on the line geometric characteristics (namely on its length), as well as on the material crossed by the pressure wave.
In substance a transmission line is nothing else than a duct into which the wave of pressure coming from the loudspeaker propagates (generally the back radiation of the membrane is used). When the pressure wave reaches the end of the duct it is reflected toward the loudspeaker. This happens independently from the fact that the duct is closed or open, considering the significant variation of acoustic impedance at the extremity of the duct. In both cases, the pressure waves from the membrane travel in a direction, while the waves reflected from the ducts extremity travel in opposite direction. Depending on the frequency, these waves could add in phase, causing an exaltation in the response, or out of phase, with consequent cancellation and attenuation. The frequencies at which these anomalies are verified depend mainly on the ducts length, while their amplitude depends on the material crossed by the waves inside the duct. Generally, the ducts are filled with fibrous material that increases the apparent length of the transmission line, improving low frequency response (the sound velocity in these materials is lower respect to the air) reducing at the same time the entity of the irregularities in the response. Unfortunately, this trick also greatly affects the efficiency.
Fig. 2 Simulated response of the closed line
The principle on which relies the DBL system is instead very simple and elegant: it is a matter of coupling two transmission lines so that the exaltations of the one correspond to the attenuations of the other and vice versa; in this way we achieve a regular response without losing the intrinsic efficiency of the system.
Fig. 3 Simulated response of the open line
Fig. 4 Simulated response of the DBL system. The red curve is relative to a typical position in environment (system at 2 meters from the back wall and 1 meter from the side one)
In the DBL system this result has been achieved by relying on the fact that a closed duct and an open duct resonate at different frequencies. The DBL system is therefore completely different from the Acoustic Wave proposed by BOSE that employs two open transmission lines of different length that act on the two faces of the membrane of a same loudspeaker; the DBL system is based on the employment of two separate loudspeakers the back radiations of which excite in a case a closed duct, (fig. 2), in the other case an open duct (fig. 3), while the front radiations are direct. The complete frequency response, i.e. the sum of the responses of the two lines, is shown in figure 4 where we can see the regularity and the extension at low frequencies, mainly in a listening environment. To reach the desired objective all the transducers have been developed and built specifically for this application. The total emitting surface is around 2000 cm2, to keep the acoustic intensity low, while, in order to increase dynamics, the linear travel of the moving part is almost 2 centimeters. The transducers produce a multipolar irradiation system to improve imaging. The drawback is the extreme design and tuning difficulty, due to the presence of lobes of secondary irradiation that impose a long series of trials in environments with different characteristics of absorption. The aim is to define the optimal number of transducers and of the relative mutual position. Because of the employment of the DBL system, there is no acoustic box properly said, while the custom realization of loudspeakers that acoustically implement the desired transfer functions has allowed eliminating the crossover. This first multipolar loudspeaker prototype has led to the HORNBLOWER model.
HORNBLOWER, PRAXIS, EXCALIBUR
EXCALIBUR HORNBLOWER PRAXIS
So far, we have designed three different multipolar loudspeaker models; the first represents the reference, uses the DBL system (Double Balanced Line) and is called HORNBLOWER for this reason; the second loudspeaker instead implements a multipolar radiation of the simpler type possible and has therefore been called PRAXIS. The EXCALIBUR model bears from the demand for a better impact in low frequency and for a free positioning in environment; the system is made of a pyramidal base (the rock) on which a long and narrow panel is inserted (the sword), hence its name. The characteristics of the three systems are presented, considering separately the radiation and the energetic aspects.
The HORNBLOWER system uses two 10" transducers for the reproduction of low frequencies, each of which is loaded by a transmission line, as requested by the DBL system; the distance between the centers of the two woofers is around 2 meters, wavelength which corresponds a frequency of around 170 Hz. Therefore, the radiation of the system could be considered omnidirectional only at very low frequencies, while at some hundreds of Hz the irradiation is not of a spherical type anymore; this is all the more true if we consider the fact that to the two 10" woofers are added two other 8" woofers that increase the system directivity factor, and also the number of secondary lobes. The distance between the two 8" transducers is about 130 cm. and they represent the heart of the system; each of them is situated on a panel not much wider than the transducer itself, so that the panels do not operate any type of acoustic loading; this involves that each transducer in its range start from a 4p str. operation and passes to 2p str. as the frequency grows, with an increase in directivity. Finally, there are 4 tweeters, aligned along a vertical plane; each of them has an irradiation of dipolar type, achieved through a particular conformation of the central pole and of the back plate in the magnetic circuit. The two radiations are not energetically similar as the back radiation is diminished by means of the employment of absorbent material (that also aims at linearizing the tweeter behaviour in the resonance area). The employment of 4 tweeters allows a notable increase in the directivity on the vertical plane, reducing the reflections from floor and ceiling.
Summarizing the directivity characteristics of the system, we could say that at low frequencies we are substantially in presence of a mono-pole, as the wavelength is much larger than the physical dimensions of the loudspeaker; at frequencies a bit higher we have a tri-pole, since to the anterior radiations of the two 10" woofers should also be added the one deriving from the back extremity of the open transmission line, delayed compared to the other two; at the same time also begin the contributions of the two 8" woofers, that added to the previous, form a 7-pole; the fall in the response of the two 10" woofers makes the middle range operation a quadri-pole type one (double dipole or 4-pole), while to higher frequencies (that is, those of pertinence of the tweeters) we have an 8-pole (however of asymmetrical type as the back radiation doesn't have the same amplitude of the front one).
From the brief, strictly qualitative description of the type of radiation, it is easy to understand that it is very difficult to obtain a linear frequency response, as there are a lot of sources, each of them varying its behaviour in frequency. Without electric filtering, the planning of a system of this kind becomes a notable exercise of synthesis, as it is necessary, from the start, to clearly focus on the characteristics of each loudspeaker in function of the global result, that depends strongly from the interaction between the single sources and between these and the environment. Each transducer must therefore be "imagined" and designed not only thinking at the characteristics that it should have in its own operation range, but also taking into account its behaviour in the transition band (that in multipolar systems is particularly wide) and its position respect to environment and the other sources: in fact, a few centimeter shift in the relative position of a transducer as regards the others can cause significant alterations in frequency response, due to destructive interference. In practice, the planning of a system of this type consists in building a gigantic mechanic-electric-acoustic puzzle in which all the three aspects must be manipulated at the same time; each single transducer constitutes a mosaic card in which only one element not perfectly integrated with the others can affect each effort. Since the behaviour of a transducer depends strongly from its position, it should be clear that the aesthetical appearance of the product is not a design exercise, but has a functional connotation: in other terms, a change in the form would affect the result too.
As the HORNBLOWER loudspeaker is concerned, it could be noticed that under the mechanical appearance is a three-way system, using three different types of transducers. From an acoustic point of view, considering the number of sources, with direct and back radiations, any attempt of classification under the acoustic profile would cause a headache. Electrically speaking the system is substantially not filtered, since the only passive component (a capacitor in series to the group of tweeters) does not have a filtering function but must rather adapt the impedance and the electric and acoustic phase correction.
In practice, to get an energetically correct operation in frequency domain, it is necessary to start from the DBL in the HORNBLOWER system: remembering that this requires the compensation of the irregularities of two transmission lines, it must be considered that the open line has a higher efficiency than the closed one; to get a good balancing, it is advisable to place the woofer loaded from the closed line near the floor. This solution has proved to be efficient in solving the problem; remember that the transmission lines are completely empty and that, therefore, the amplitude of the anomalies is significant (figg. 2,3,4). If we request a good response at lower frequencies, the two lines must at least be 1.5 meter long; if one of the two woofer is positioned on the floor, there is no other alternative than superimposing the two transmission lines. As it is necessary to keep a vertical alignment of all the transducers, and since these have a back radiation, putting the lines immediately behind the "midrange" and the tweeter must be avoided otherwise the sound has an annoying coloration due to early reflections; therefore, the column that contains both the lines must be moved sideways as regards the principal axle that contains all the transducers (in this way we can also partially control the back irradiation). The DBL system is therefore fitted in a 2.3 meter high structure; the higher woofer (the one loaded from the open line) is around 70-80 cm. from the ceiling, so avoiding the reinforcement due to the excessive proximity of the wall. We should remember that in an open transmission line the back door radiates with some milliseconds of delay as regards the front radiation, so it is also important to consider the behaviour in the time domain and not only that in the frequency domain; in practice there is some reinforcement that, from a certain point of view, could be similar to that resulting from placement near to one or more walls. The floor placement of the closed line driver and the contemporary positioning at around 2 meters of height of the open line driver accounts for an almost ideal synergy.
It is particularly important to make sure that the transition between a transducer and the other ensures the highest coherence possible in the mechanic-acoustic parameters: it is not possible interfacing a woofer with acceleration of 20g/A and moving mass of some ten grams with a tweeter with a 100 times higher acceleration and a 100 times lower moving mass (in example, consider very bad result obtained by the systems with electrostatic or isodynamic panels coupled with an electrodynamic woofer). The use of two 8" woofers reduces excursion (emitting surface is around 1000 cm2 instead of 80 cm2 of common 5" midrange), with great effects on transient velocity, distortion (with great excursions non-linearity are unavoidable) and air saturation. Taking into account these mechanical considerations, and those relative to the impedance of radiation (which varies with a certain continuity inside the operating range of a loudspeaker, while presenting a great discontinuity in the passage from a transducer to another of different type) it is advisable to ensure that the transition between transducers operating in adjacent bands takes place in the most gradual way: this means very little attenuation slopes and wide overlapping ranges.
The PRAXIS system instead employs two 8" transducers coupled to the same tweeters of the greater model; the height is of around 160 cm and the encumbrance very limited as regards to the other model. The two 8" woofers are situated at a distance of around 130 cm; while the higher is mounted on a panel analogous to that used in the HORNBLOWER system, the lower is near the floor and has employed in a bass-reflex configuration, to get a sufficient extension at the low frequencies. Also in this case the type of radiation broadly varies in frequency, with increase of the directivity as the frequency increases. At this point, however, it should be evident that the directivity control and the presence of secondary lobes added to the principal lobe interests only the vertical plane and not the horizontal one, unlike from what happens in all wide diaphragms (with consequent loss of the horizontal spatial information).
The EXCALIBUR model, differently from the other two, makes uses of a diagram of radiation asymmetrical on the vertical plane. In this way we can obtain a substantial invariance of the response for heights between 90 and 180 centimeters; in practice the response doesn't vary for sitting or standing listening positions. Mechanically it is a three ways system, with a 10" woofer bass-reflex box, a 8" woofer mounted on panel and six soft dome 1" tweeters.
Since the multipolar systems have been designed for music reproduction in semi-reverberant environments, many of the conventional measures on the loudspeakers have no meaning; however, we report two frequency responses performed in a 28 m2 (4 x 7) room, that is fairly small with respect to the dimensions of the bigger model. The diagram in fig. 5 is relative to the HORNBLOWER system situated near to the side wall; despite the proximity of the wall the regularity in middle and high range is impressive; remember that such result has been achieved with the employment of an elevated number of sources and practically without electric filtering, while the behaviour in low range is surely limited by the insufficient dimensions of the hall (to a frequency of 20 Hz corresponds a wavelength of around 17 meters).
Fig. 5 Response in environment of the HORNBLOWER system
In fig. 6 is instead reported the response of the PRAXIS system placed at 1 meter from the nearer wall; also in this case the energetic distribution in frequency is practically perfect in middle and high range. Respect to the bigger model there is a predictable smaller extension to the low frequencies; a better behaviour could be achieved placing the loudspeaker near the side wall (position that supply a reinforcement of the low range), with the evident advantage of having a smaller room occupation. The sensibility of the systems varies from the 100 dB of the big model to 95 dB for the smaller.
Fig. 6 Response in environment of the PRAXIS system
The systems also exist in XL version (that is with extended linearity transducers: please see the relative documentation); in this case the sensibility is around 6 dB higher. This version has maximum results when current driven, since the synergy between the two technologies is practically perfect (to examine the advantages obtainable with the current driving you are invited to read the appropriate documentation).
As concern the optimal room positioning of the multipolar systems, experiments have demonstrated that better performances are achieved distancing the loudspeakers from the back wall. The space of the acoustic scene increases as increase the distance between loudspeakers and back wall: statistically good results are obtained with distances around equal to the height of the loudspeakers, while if the distance is smaller than a meter generally we have an excessive coloration of the reproduction. Like it can be deduced from the two reported measures, the distance from the sides walls is instead less important, as it influences only the middle-low range, that often acquires a better roundness. The loudspeakers therefore can be placed near the sides walls also, chiefly if they are oriented toward the center; this is opportune to guarantee the maximum separation between direct and back radiation (remember that both are of multipolar type and therefore characterized by the presence of secondary lobes). Best results are achieved when the back radiation arrives with delays of 15 - 20 ms respect to the direct one; that means a path difference of 5 - 6 meters between front and back radiation. In these conditions the illusion of presence to the original event is practically perfect; the acoustic scene is very wide and each instrument has believable dimensions and is perfectly focused. This is obtainable practically always with live recordings, unfortunately not so easily with hardly manipulated recordings (each intervention modify original phase relationships, with consequent negative results).
The position that generally allows the betters results it is the following: loudspeakers nearer to the side walls that to the back one (so that the back radiation has the first reflection on the side wall), with a little inclination toward the center (more or less pronounced according to the characteristics of the listening room). The listening position also influence the result, as the proximity to the loudspeakers carries to a predominance of the direct component of the radiation, with a sound clear and detailed, while as distance increases becomes more important the contribution of the reflected component, with a better feeling of space, but with smaller perception of the details. In practice quite the same that happens in a concert hall.
Thanks to these characteristics, it is possible for each person to choose the type of preferred listening sensation, simply balancing the distance of the loudspeakers from the back wall with the distance of the listening point; it is opportune that the head of the listener is to at least a meter of distance from the nearer wall, so that avoid the consequences of a comb filtering. Over ten years of experiments in different acoustical conditions have shown that a particularly spectacular configuration exists, obtainable placing the loudspeakers at two vertexes of a rhombus, with the others two vertexes occupied from the listener and from the back wall: in practice, its a matter of doubling the known "listening triangle" flipping it respect the line joining the two loudspeakers.
Summarizing, the employment of these systems is recommended in rooms with length not smaller than 5 meters, since at least a meter must exist between loudspeakers and back wall, a meter should be let free behind the head of the listener, which in turn should be placed not less than three meters from the loudspeakers; if these conditions are not satisfied, it is not worth it of employing a multipolar system. If you instead have a listening room of opportune dimensions, we strongly recommend that the loudspeakers are placed 2 - 3 meters from the back wall (remember that its possible and often advisable approach them to the side walls); in these conditions it is possible to enjoy results unachievable with other systems (with the right software, of course). Because the efficiency and the transparency, notably higher than usual, all the components of the hi-fi chain must be adequate; at these qualitative levels a wrong cable or a not suitable foot could result in a unacceptable listening. A further trick is to check the absolute phase of the system as well. Finally, the most important consideration: generally, the results get better with the increase of the dimensions of the listening room; but, and this is valid in each case, if the room doesn't have adequate characteristics for music reproduction (i.e., poor acoustics), also the best hi-fi system in the world can obtain very small performance. So, we would give you a suggestion: if possible, it is better to search for an improvement of the acoustic characteristics of the listening room; often the benefits are greater that not replacing the whole hi-fi system.
Vicolo T. Aspetti, 18
35100 Padova (ITALY)
Tel./Fax: +39 049 8644085
© Copyright - Ultrasound - All rights reserved - 1999-2000