EP0350652B1 - Electrodynamic loudspeaker - Google Patents

Abstract

In such a loudspeaker having several loudspeaker units (13, 14, 15) which are connected to a frequency filter (3) via one power amplifier each (10, 11, 12), power amplifiers having a negative source impedance (-Ri) are proposed which reduce the linear and non-linear distortions to such an extent that the application of modern filters becomes possible and further measures for reducing non-linear distortions become applicable. <IMAGE>

Description

The invention relates to an electrodynamic loudspeaker.

From DE-A-27 13 023 an electrodynamic loudspeaker with an amplifier for feeding a voice coil is known. The loudspeaker is designed as a woofer. The amplifier's effective output impedance is equivalent to a negative resistance in series with a parallel resonant circuit. The negative resistance has practically the same value as the resistance of the speaker's voice coil. By operating the loudspeaker with such an amplifier, the change in the bass characteristics of the loudspeaker can be achieved, which is equivalent to changing the mechanical parameters of the loudspeaker element and its movement mass, damping and modulation. In other words, the natural resonance frequency of the loudspeaker is to be fought and at the same time a different resonance frequency is to be enforced, which is better matched to the loudspeaker housing.

The disadvantage of this solution is that although it has advantages for low-frequency loudspeakers or for the reproduction of deep tones through other loudspeakers, it cannot be used to improve medium and high-frequency loudspeakers or to improve the reproduction of medium and high-frequency tones because these problems with the natural resonance frequency do not occur at all or only in a very weakened form. In addition, this solution does not result in a predictable, clear trend-indicating frequency response, as is usually required for other links (eg amplifiers, recording devices, etc.) of an electroacoustic chain. The loudspeaker thus remains the worst link in such a chain, which begins with the microphone and ends with the loudspeaker. The loudspeakers are linear and nonlinear distortions much worse than with the other links of such an electroacoustic chain such as microphone, amplifier, mixer, memory, etc. The improvements achieved according to the above patent application and other known measures are still so small in effect that they often reduce the effort not worth it.

From audio engineering, August 1951, W. Clements: "A new approach to loudspeaker damping" an amplifier circuit with negative impedance is known which is designed in such a way that the impedance of the voice coil of a downstream loudspeaker almost disappears. The advantage of this is that a very high attenuation of the loudspeaker is achieved.

It is therefore clear that such an amplifier circuit, as described there, is used in particular for damping resonances of the loudspeaker at low frequencies. In the case of reusable loudspeakers, the resonance frequency lies in the transmission range only with the woofer. With the mid-range and high-frequency tones, this resonance frequency lies outside the transmission range. In reusable systems, such an amplifier circuit is therefore used exclusively as a driver for woofers with success. In the case of mid-range tweeters and high-frequency tones, the frequency response is not improved with such an amplifier circuit, because other means are used to achieve this.

The invention as characterized in the claims solves the problem of creating a loudspeaker with a plurality of electrodynamic loudspeaker units in which the signal distortions are significantly reduced over the respective frequency range.

The advantages achieved by the invention are essentially to be seen in the fact that the loudspeaker now has a straight frequency response which rises continuously over its entire frequency range. This creates a mathematically ideal situation which allows a straight horizontal frequency response to be generated with the aid of a compensation network containing an integrator which is connected upstream of the power amplifier. Another advantage is the fact that the frequency response obtained in this way is phase-linear, ie has a linear relationship between phase and frequency that is easily predictable across all frequencies. Although according to the aforementioned DE-A-27 13 023 power amplifiers with negative impedance for woofers are known, a surprising and previously unused effect occurs when such power amplifiers are combined several times in a reusable loudspeaker together with a crossover. This effect consists in the above-mentioned predictable and constant properties of frequency and phase response over the entire transmission range. Once, according to the invention, it has been possible to make the frequency response straight and phase-linear over the entire frequency range, further possibilities open up with which reusable systems can be significantly improved. However, these improvements only have a full effect if the frequency and phase response are sufficiently well and reliably known. Such improvements can be achieved in the magnet systems of the loudspeaker units and in the crossovers. For example, further proposed measures for linearizing the driving force of the voice coil in the magnetic field of the magnet system only come into play when the nonlinearities of the restoring forces have already been combated in the manner according to the invention. Although certain of the improvements mentioned are already known per se, their effect is surprisingly increased if they are carried out on loudspeakers which are driven by power amplifiers with a negative source impedance.

• Figur 1 eine schematische Darstellung eines erfindungsgemaessen elektrodynamischen Lautsprechers,
• Figur 2, 3 und 4 je eine Ausfuehrungsform einer im Lautsprecher verwendeten Frequenzweiche in schematischer Darstellung,
• Figur 5 einen Schnitt durch eine Lautsprechereinheit,
• Figur 6 ein elektrisches Ersatzschaltbild,
• Figur 7 eine Schaltung zur Erzeugung einer negativen Quellenimpedanz,
• Figur 8 eine Darstellung verschiedener Charakteristiken,
• Figur 9 eine Schaltung fuer einen Teil des Lautsprechers und
• Figur 10 eine Filtercharakteristik.

In the following, the invention will be explained in more detail with the aid of drawings showing only one embodiment. It shows

• FIG. 1 shows a schematic illustration of an electrodynamic loudspeaker according to the invention,
• 2, 3 and 4 each show an embodiment of a crossover used in the loudspeaker in a schematic representation,
• FIG. 5 shows a section through a loudspeaker unit,
• FIG. 6 shows an electrical equivalent circuit diagram,
• FIG. 7 shows a circuit for generating a negative source impedance,
• FIG. 8 shows various characteristics,
• Figure 9 shows a circuit for part of the speaker and
• Figure 10 shows a filter characteristic.

Figure 1 shows an electrodynamic speaker 1 in a schematic representation with an input 2 for an electrical signal, with a crossover 3 with three outputs 4, 5 and 6, each via an integrator 7, 8 and 9, each with a power amplifier 10, 11 and 12 are connected, to each of which a loudspeaker unit 13, 14 and 15 is connected. The integrators are to be considered as equalization networks that contain an integrator, since their characteristics should not correspond to that of an integrator over the entire frequency range, as is already known per se. The power amplifiers 10, 11 and 12 consist, for example, of an operational amplifier 16 and a circuit for generating a negative source impedance -R _i as they do is known, for example, from U.S. Patent No. 4,720,665. Such a power amplifier can also have a different structure, as long as it has a negative source impedance. The crossover 3 can be designed in a manner known per se. For example, such is described in JEI Journal of the Electronics Industry, Vol 32, September 1985 pages 38 to 44 and shown in Figures 3 and 4 of this article. Such a crossover with one input and two outputs then consists at least of an addition circuit with a positive counting and a negative counting input and a filter. The input of the filter is possibly connected via a time delay unit to the positive counting input and the output of the filter to the negative counting input of the addition circuit. If more than two outputs on the crossover are required, several such crossovers can be connected in series for two outputs. Usual dynamic speakers are provided as speaker units.

Figure 2 shows a crossover with an input 17 and three outputs 18, 19 and 20. This has an addition unit 21 with a positive counting input 22 and a negative counting input 23 and a filter 24 with an input 25 and an output 26. The output 26 is connected via a line 27 to the negative counting input 23 of the addition circuit 21 and to the output 20. The input 25 of the filter 24 is connected via a line 28 and via a time delay unit 29 to the positive counting input 22 of the addition circuit 21. The addition circuit 21 also has an output 30. The elements 17 and 21 to 30 thus form a first crossover with two outputs 20 and 30. A second crossover is connected in series to the output 30 and has the same elements. These are the addition circuit 31 with inputs 32 and 33, a filter 34 with an input 35 and an output 36 and lines 37 and 38. A time delay circuit 39 is turned on in line 37.

FIG. 3 shows a crossover which largely has the same elements as the crossover according to FIG. 2. It also contains a phase correction circuit 40 which is connected upstream of the filter 34. This phase correction circuit 40 is designed as an all-pass filter.

FIG. 4 shows a further embodiment of a crossover with loudspeaker units 41, 42 and 43 connected to it, all of which end in a common baffle 44. This means that their magnet systems 45, 46 and 47 have different distances from this baffle 44. If these distances are measured in terms of the running times of the sound waves, the voice coil in the magnet system 45 of the loudspeaker unit 41 has such a distance Δt 1 + Δt 2, the voice coil in the magnet system 46 of the loudspeaker unit 42 has a distance Δt 2 and the voice coil in the magnet system 47 of the loudspeaker unit 43 has a negligible distance. The power amplifiers and integrators, not shown here, as are known from FIG. 1, are not essential for the purposes of this illustration, but are in fact present. The crossover again has the same elements as are already known from FIGS. 2 and 3. These elements are therefore given the same reference numerals. However, new elements have been added, namely a time delay circuit 48, which is connected to the output 30 of the addition circuit 21 and generates a time delay that depends on the distance Δt 1, and a time delay circuit 50 connected to an output 49 of the addition circuit 31. This generates a time delay, which depends on the distance Δt₂.

Figure 5 shows two symmetrical magnet systems. The magnet system 52 on the left of a center line 51 and the magnet system 53 on the right. The structure of both magnet systems 52, 53 is partly the same. Common elements can therefore be given the same reference numerals. These include a pole piece 54, a pole plate 55 and a magnet 56. The magnet system 52 has a voice coil 57 which is connected and supplied in a manner known per se. A short-circuit ring 59 made of copper is inserted between the pole plate 55 and the pole piece 54 or more precisely an associated base plate 58 in a manner known per se. In the area of the voice coil 57, a short-circuit ring 93 made of copper is also used on the pole piece 54.

A voice coil 61 is arranged in the magnet system 53, which also has an annular air gap 60. A further coil 62 is arranged on the pole piece 54 coaxially to the voice coil 61. There are various ways of connecting the voice coil 61 and the further coil 62 to a power amplifier. Therefore only two connections 63 and 64 are shown. One possibility is to connect one end 65 of the voice coil 61 to the other end 66 of the further coil 62, as is indicated by the line 67. However, the further coil 62 is preferably connected to the same power amplifier as the voice coil 61, both coils 61 and 62 have the same number of turns and are wound in the opposite direction of rotation. The aim of the further coil 62 is to generate a magnetic field which at all times counteracts that magnetic field that is generated by the current in the voice coil 61 and compensates for it, so that the sum of the two magnetic fields is zero and consequently only in the air gap 60 the uniform magnetic field occurs, which is generated by the magnet 56.

FIG. 6 shows a simplified electrical equivalent circuit diagram of a loudspeaker with its supply. This contains a voltage source 68, a negative resistor 69, a resistor 70, which represents the ohmic resistance of the voice coil, an inductance 71, which represents the inductance of the voice coil, an inductance 72, the mechanical restoring forces as they produce the suspension of the membrane and the air cushion in the housing of the loudspeaker, and a capacitance 73, which represents the masses of the membrane, etc. All of these elements are connected to one another via lines 74, 91 and connected in series, with the exception of the inductance 72, which is connected in parallel to the capacitor 73 and is connected to the line 74, 91 via a line 75. The elements 68 and 69 together form the power amplifier and the other elements together form a loudspeaker unit.

• Die Impedanz Z_s soll viel kleiner sein als die Widerstaende R₂ und R₄
• Die Impedanz ${\text{Z}}_{\text{q}}{\text{= -Z}}_{\text{s}}\text{(1 + V₁ )/(V₂ - V₁)}$
  
  mit
  $\text{V₁ = R₂/R₁}$
  
  und $\text{V₂ = R₄/R₃}$
  
  .

FIG. 7 shows an example of a power amplifier circuit with a negative source impedance, as can be substituted for the voltage source 68 and the negative resistor 69 in FIG. It consists of an operational amplifier 76 with an inverting input 77, a non-inverting input 78 and an output 79. An impedance Z _{s is} connected to this. From the output 80 a feedback 81 with a resistor R₂ is fed back to the input 77. This also includes a resistor R₁, which is connected to the input 77. From the output 79 a further feedback 82 with a resistor R₄ is fed back to the input 78 and is also connected to the earth via a resistor R₃. A load 83 is connected to the output 80. This load 83 simply represents the resistances, capacitances and inductivities of the loudspeaker units. From FIG. 6, these are the elements 70, 71, 72, 73, 74, 75 and 91. The impedance Z _q is understood to be the impedance at the unloaded output 80 occurs. The following applies to this circuit according to FIG. 7:

• The impedance Z _s should be much smaller than the resistors R₂ and R₄
• The impedance ${\text{Z.}}_{\text{q}}{\text{= -Z}}_{\text{s}}\text{(1 + V₁) / (V₂ - V₁)}$
  
  With
  $\text{V₁ = R₂ / R₁}$
  
  and $\text{V₂ = R₄ / R₃}$
  
  .

FIG. 8 shows a representation of different frequency courses. The frequencies are plotted on the horizontal axis 84, and the acoustic output i₀, on which the sound pressure, output powers etc. can be understood, is plotted on the vertical axis 85. Line 86 shows the frequency response generated by the power amplifier 10, 11, 12 (FIG. 1) with the negative source impedance. A line 87 shows the frequency response which the integrator 7, 8, 9 (FIG. 1) generates. A line 88 shows the frequency response as it is generated by the series connection of the integrator with the power amplifier.

FIG. 9 shows a circuit as can be used, for example, for the magnet system 53 according to FIG. 5 together with the further coil 62. In this circuit, the coil 62 is represented by an ohmic resistor 62a and an inductance 62b. Via a connection 63, it is connected to the output 95 of an amplifier 94 with a negative source impedance. The input 96 of amplifier 94 is connected to ground, as is the other end 66 of coil 62.

Figure 10 shows a frequency response as it can be provided for the filter 24. The frequencies are plotted on the horizontal axis 100 and the acoustic output is plotted on the vertical axis 101. A curve 102 represents a filter characteristic known per se for a low-pass filter of at least 2nd order according to Bessel. Another curve 103 above a point 104 represents a filter characteristic also known per se for a 2nd order low-pass filter according to Butterworth (provided the good 1 / √ 2 corresponds exactly) or Tchebycheff (if the quality is greater than 1 / √2). According to the invention, this characteristic expires after the point 104 horizontally in an area 105 in which the damping is essentially constant. The damping, which corresponds to a distance 106, should only be so great that signal components that this filter passes through and that in the frequency range of the Loudspeaker unit 42 (mid-range) are located, the loudspeaker unit 41 (low tone) or its output signal is not noticeably disturbed.

The mode of operation of the loudspeaker according to the invention will be explained below. For this purpose it is advantageous to first deal with the mode of operation with the aid of the electrical equivalent circuit according to FIG. 6. There the inductance 72 represents the properties of the suspension of the membrane and the capacitance 73 the inertia of the membrane of a loudspeaker unit. A current flows in line 91 which corresponds to the acoustic output. This current should depend in a simple and precise manner on the voltage and frequency at the voltage source 68. Ideally, the voltage at node 89 should correspond to the voltage at voltage source 68. However, this does not seem to be the case since impedances 69, 70 and 71 are arranged in series between voltage source 68 and node 89 and a current flows through them. Voltage drops thus occur at these impedances, the sum of which indicate a voltage difference ΔU between the voltage source 68 and the node 89. This voltage difference ΔU can disappear if the voltage source 68 is connected directly to the node 89. This is true even if the sum of the impedances 69, 70 and 71 is zero. This is achieved in that the impedance or resistance 69 has a size which corresponds to the negative sum of the impedance or resistance 70 and the impedance or inductance 71. It is thus possible to completely eliminate the undesired influence of the resistor 70, the inductance 71 and also the inductance 72 on the current in the line 91. Their influence on the frequency response and the distortions is therefore eliminated.

Such an impedance or resistor 69 with negative impedance together with a power amplifier is represented more precisely by the circuit according to FIG. Resistor 83 represents the connected loudspeaker unit.

The output 80 and the node 92 are indicated in both circuits (FIGS. 6 and 7) for better orientation. The operation of the circuit according to Figure 7 can be specified as follows: The resistors R₂ and R₁ form the negative feedback 81 with the tendency to reduce the output signal at the output 79. The resistors R₃ and R₄ form the positive feedback 82 with the tendency to increase the output signal at the output 79. If there is no current at outputs 79 and 80, both outputs 79 and 80 have the same voltage. This depends on the size of the resistors R₁ to R₄. If the load 83 is switched on, which means that the loudspeaker unit is in operation, a current flows in the impedance Z _s . The voltage at output 80 is thus somewhat lower than at output 79. The positive feedback 82 is thus taken from a greater voltage than the negative feedback 81. The effect of the positive versus the negative feedback increases and the signal at the outputs 79 and 80 is getting bigger. A larger load at output 80 results in an increase in signal. This corresponds to a negative source impedance.

In the electrodynamic loudspeaker according to the invention, as shown in FIG. 1, an AC signal is applied to input 2. This is divided in the crossover so that, for example, signal parts with low frequencies pass through the output 6 into the integrator 9, signal parts with medium frequencies via the output 5 into the integrator 8 and signal parts with high frequencies via the output 4 into the integrator 7. These signal parts are treated in a manner known per se in these integrators in such a way that their amplitudes decrease with increasing frequency, as indicated by line 87 in FIG. After these integrators 7, 8 or 9, these signal parts reach the power amplifiers 10, 11 and 12, where these signal parts receive an amplitude / frequency characteristic according to line 86 from FIG. Together this results in a characteristic according to line 88 or just an even frequency response. The integrators and the power amplifiers with the negative source impedance are to be designed so that the slopes of lines 86 and 87 are exactly the same in opposite directions. The slope is preferably 6dß / octave. It would be conceivable to design the power amplifier 10 for the high-frequency loudspeaker unit 13 without a negative source impedance, but then the straight frequency response can no longer be generated into the high-frequency range. However, this compromise could at best be accepted and the disadvantages neglected.

If an electrical signal is applied to the input 17 of the first crossover 21 to 30 according to FIG. 2, it reaches the input 25 of the filter 24 via the line 28, which is designed as a low-pass filter. The unfiltered signal is also applied via line 28 to the positive counting input 22 of the addition circuit 21. The low frequencies are subtracted from this signal in the addition circuit 21 and only the high frequencies appear at the output 30. The low frequencies then reach the output 20. The elements 31 and 34 of the second crossover 31 to 39 work exactly accordingly, so that the high frequencies occur at the output 18, the medium frequencies at the output 19 and the low frequencies at the output 20. This applies if the filter 34 is also designed as a low-pass filter. Irrespective of whether time delay circuits 29 and 39 are provided or not, it can be achieved with such a crossover that the signals from the outputs 18, 19 and 20, which via the power amplifiers 10, 11 and 12 to the loudspeaker units 13, 14 and 15 are emitted, together emit an acoustic output signal which has a linear phase / frequency characteristic.

A significant improvement can be achieved if the time delay circuits 29 and 39 are designed to accommodate the time delays of the signal in the filters 24 and 34 exactly balance and if the filters 24 and 34 are designed as a low-pass filter of at least fourth order according to Bessel. This results in the outputs 18, 19 and 20 signals which let the loudspeaker units 13, 14 and 14, 15 work in phase at the relevant takeover frequency. This means that the diaphragms of the loudspeaker units 14 and 15 operate synchronously and without a phase difference at any frequency of signals in the range of the upper limit frequency of the filter 24 designed as a low-pass filter. The radiation characteristics of these two loudspeaker units are thus stable. The same effect can be achieved for the crossover frequency of the loudspeaker units 13 and 14.

A simplification can be achieved if the filter 24 is designed as a low-pass filter with two poles and with a good that is greater than or equal to 1 / √2, and if the falling frequency response of the filter is designed so that it falls into a range with essentially passes constant attenuation, as shown in Figure 10. This is in contrast to a normal frequency response that drops to infinite attenuation. As a result, the group delay in the filter 24 is significantly shortened, which significantly reduces the circuit complexity in the time delay circuit 29. Under certain circumstances, the time delay circuit 29 can then be omitted entirely. This simplification has the consequence that the loudspeaker units 14 and 15 operating in the range of the cut-off frequency of the filter 24 no longer work in exactly the same phase.

As is known from FIG. 3, the filter 34 can be preceded by a phase correction circuit 40. This results in an improvement and a phase-linear and in-phase acoustic output signal. The filter 34 designed as a low-pass filter is preferably designed as a second-order Butterworth filter. For this purpose, the phase correction circuit 40 must be designed in such a way that the phase up to sufficiently high frequencies. These sufficiently high frequencies include, in particular, frequencies in the region of the falling frequency response of the filter 34. The addition circuit 31 thus forms a high-pass filter with a steepness of a third-order filter.

All elements previously proposed for the crossover are combined in the illustration according to FIG. 4. But there are also the two time delay circuits 48 and 50. The time delay circuit 48 delays the signal from the output 30 by an amount that is dimensioned such that time delays of all elements connected in series, such as time delay circuit 48, phase correction circuit 40 and filter 34, add up the Time difference Δt₁ correspond. Imagine that the signal that is fed to the input 2, 17 is composed of two partial signals and that one partial signal is emitted via the loudspeaker unit 41 and the other partial signal is emitted via the loudspeaker unit 42. Then it is ensured by the time delay circuit 48 that both sub-signals, when they pass the baffle and are converted into sound waves, leave the baffle 44 exactly when and thus as they were composed in the signal at the input 2, 17. The time delay circuit 50 compensates for exactly that time difference Δt₂ which corresponds to the difference in transit time of the sound waves in the loudspeaker units 42 and 43 from the membrane to the baffle 44. In order to achieve the same purpose, such time delay circuits could also be provided at other locations in the crossover network.

Through improvements to the magnet systems (see FIG. 5) of the individual loudspeaker units 13, 14 and 15, the loudspeaker according to the invention can be further improved overall.

It is known that there are measures on a magnet system which aim to improve the distribution of the magnetic field in the air gap. This improvement is supposed to cause the magnetic field strength to decrease steadily and symmetrically at both ends of the air gap. This is achieved, for example, by the pole piece 54 projecting beyond the pole plate 55. This reduces nonlinear distortion when moving the voice coil. This reduction is not noticeable in many cases because the nonlinear distortions caused by the nonlinear restoring forces of the membrane suspension predominate. This means that such improvements to the magnet system only make sense if they are used together with the power supply via power amplifiers with negative source impedance.

Furthermore, it is known that the part 98 of the voice coil 57 protruding into a cavity 97 (FIG. 5) causes an additional excitation in the magnetic circuit as soon as the current flows through it. The force that the voice coil experiences is given by the vector product of the magnetic field and current. However, since, as mentioned, the magnetic field is again a function of the current through the voice coil, the vector product is not linear. It is therefore important to counteract the change in the magnetic field. The short-circuit ring 59 offers a possibility for this. In it, currents always form in such a way that the change in the magnetic flux is counteracted. There is also the possibility of providing a coil instead of the short-circuit ring 59. It is excited by the current in the voice coil and thus compensates for the unwanted excitation by the lower part 98 of the voice coil 57. This coil can also be short-circuited. The effect then corresponds to that of the short-circuit ring 59. If the ohmic resistance of this coil is disturbed, this can be compensated for by the short-circuit being effected via a negative resistor. This means that the coil is connected to an amplifier with a negative source impedance. A circuit according to FIG. 9 then also results for this coil, which takes the place of the short-circuit ring 59. The amplifier 94 does not receive any signal at the input 96. The amplifier 94 compensates the ohmic resistance 62a of the coil 62b. Thus the coil 62b no longer has any resistance and currents of any size can flow therein, that is to say it is short-circuited. This improvement is only really noticeable once the distortions caused by the nonlinear restoring forces have been eliminated.

It is also known that the voice coil as a whole generates a magnetic field which influences the magnetic field in the air gap 60 (FIG. 5) in such a way that, depending on the polarity of the fields, a field intensification occurs at one end of the air gap and a field weakening occurs at the other end. This does not seem to be bad at first, as the total flow in the air gap remains approximately constant. Only closer examinations show that this is not the case. The permeability of the iron in the pole plate 55 and in the pole piece 54 is dependent on the magnetic modulation. This leads to the fact that the total magnetic flux is again dependent on the current in the voice coil 61. A short circuit ring 93 in the area of the air gap can be used to combat the distortions that have arisen. Another possibility is the replacement of the short-circuit ring 93 by a stationary further coil 62, which is arranged in such a way that it exactly cancels the field generated by the voice coil 61. For this purpose, the voice coil current preferably also flows through it. It is connected in series or parallel to the voice coil. It is also possible to feed the coil 62 through a separate amplifier. It is also conceivable to connect the coil to an amplifier with a negative source impedance, as shown in FIG. 9. The improvements that can be achieved through these measures only really take effect when the dominant distortions caused by nonlinear restoring forces are eliminated.

Claims (10)

1. Electrodynamic loudspeaker (1) with at least two dynamically operating loudspeaker units (13,14,15) for at least one higher and one lower frequency range linked by means of at least one dividing network, characterized in that the loudspeaker unit (14,15) for the upper frequency range has a correcting network with an integrator (8,9) and a power amplifier (11,12) with a negative source impedance (-R_i), which compensates the voice coil impedance (70,71) to such an extent that the group delay in the loudspeaker unit (14,15) is at least approximately constant.
2. Electrodynamic loudspeaker according to claim 1, characterized in that the loudspeaker units (13,14,15) have a magnet system (53) with a voice coil (61), whose magnetic flux is substantially independent of a current flowing through the voice coil.
3. Electrodynamic loudspeaker according to claim 1, characterized in that the dividing network (3) is constructed in such a way that it emits signals, which allow the loudspeaker units (13,14,15) together to produce a phase-linear acoustic output signal.
4. Electrodynamic loudspeaker according to claim 1, characterized in that the dividing network (3) is constructed in such a way that it emits signals, which allow the loudspeaker units (13,14,15) to operate in in-phase manner at a take-over frequency.
5. Electrodynamic loudspeaker according to claim 2, characterized in that, in addition to the voice coil (61), the magnet system (53) has a further coil (62), which is arranged in in-phase, but fixed manner to the voice coil and which eliminates the magnetic field produced by the voice coil.
6. Electrodynamic loudspeaker according to claim 3, characterized in that the dividing network comprises a summer (21) with a positively counting and a negatively counting input (22,23) and a filter (24), which are connected in such a way that the input (25) of the filter (24) is connected to the positively counting input (22) and the output (26) of the filter (24) is connected to the negatively counting input (23) of the summer (21).
7. Electrodynamic loudspeaker according to claim 6, characterized in that the filter (24) is designed as a low-pass filter with two poles, a Q equal to or higher than 1/√2 and with a decreasing frequency response (103), which passes into a range (105) with substantially constant attenuation (106).
8. Electrodynamic loudspeaker according to claim 6, in which the dividing network comprises a first and a second series-connected dividing networks, characterized in that as filters (24,34) are provided low-pass filters, that to the filter (34) of the second dividing network is connected a phase correcting circuit (40) and that the filter (34) of the second dividing network is designed in such a way that the following summer (31) acts at least as a third order high-pass filter.
9. Electrodynamic loudspeaker according to claim 6, characterized in that the summer (21) is followed by a delay network (48), which at least partly compensates delay differences between the magnet systems (45,46) of the loudspeaker units (41,42) and the acoustic baffle (44) of the loudspeaker units.
10. Electrodynamic loudspeaker according to claim 1, characterized in that the negative source impedance compensates the ohmic resistance (10) and the inductance (71) of the voice coil of the connected loudspeaker units.

Images