EP0145997B1 - Reproduction errors compensation device for electroacoustic transducers - Google Patents

Description

The invention relates to a device according to the preamble of claim 1.

All electroacoustic transducers are mechanical vibration systems, which are characterized by eigenvalues such as spring constant, mass and damping. Speakers, i.e. Transducers that receive electrical signals and emit acoustic signals are e.g. with the help of a voice coil excited and damped vibrations. Conversely, microphones are converters that convert acoustic signals into electrical signals. With electrodynamic microphones, this is also done with the help of a voice coil attached to a membrane. Electrodynamic pickup systems also pick up mechanical vibrations and generate electrical signals using voice coils. For this reason, there are no fundamental differences between electrodynamic microphones and electrodynamic pickup systems.

The design and functional principle of coupling the various influences, which also influence each other again, results in two major main errors that apply equally to electrodynamic loudspeakers and microphones.

1.) Error in the amplitude frequency response

Due to the eigenvalues of the vibration system, a characteristic transfer function results over a larger frequency range. Typical for the so-called amplitude frequency response is e.g. the non-linear course with resonance points and the low efficiency at the upper and lower end of the transmission range. As an example of this, a conventional, softly suspended bass loudspeaker of approx. 30 cm 0 at 20 Hz built into a closed housing shows only a low sound pressure effect with amplitude values that are too small, but achieves over-loudness with its resonance frequency in the range of approx. 40 - 80 Hz too high amplitude values and decreases again against high frequencies effectiveness in the sound transmission by too small amplitude values. The amplitude ratio is clearly shown by the frequency ratio related to the resonance frequency in FIG. 1 with a different damping factor a. This representation is known prior art and is not further explained here.

2.) Error in the phase frequency response

Due to the mass and damping of the vibration system, the swing-in and swing-out processes are clearly distorted in the case of pulsating vibrations of any frequency. This is because such vibrating systems, excited below and above the resonance frequency, have different phase positions with respect to the exciting signal. The phase angle curve over the frequency ratio related to the resonance frequency is illustrated in FIG. 2 for various damping factors a. This representation is also known prior art and is not further explained here.

The membrane starts to move in the same direction with vibration pulses above and below the resonance frequency, but only reaches low amplitude values in the case of pulses near and below the resonance frequency, especially during the first half of the oscillation period, since the phase shift occurs during the transient process . It is only when the phase shift corresponding to the frequency has taken place that the amplitude values corresponding to the exciting signal are reached, albeit out of phase.

Pulse-like vibrations, such as striking a guitar string, striking a note on the piano or hitting a drum, show the maximum amplitude at the first strike and then vibrate in the torn tone frequency. A loudspeaker system or microphone, which is operated in the range of its resonance frequency, only has to settle slowly with such pulses until it has the phase position corresponding to the frequency, and only then, depending on the quality, usually reaches the maximum amplitude after one or two full oscillation periods. In the event of sudden damping, as a result of the vibrating guitar or piano string or the drum head being suddenly stopped, the transducer oscillates at least for a period of time predetermined by the phase shift. In the subsequent decay process, the more or less well damped natural or resonant frequency of the converter can be seen.

The human ear only evaluates the volume according to the amplitude of pure sine tones. Mixtures of notes that music always consists of are evaluated using the envelope.

While the sound distortions of the converter system due to errors in the amplitude frequency response, which are perceived as too loud or too quiet pitches, rarely interfere with the music transmission, since anyway you can never judge exactly whether the musician himself played this pitch louder or quieter, errors in the swing-in and swing-out processes are perceived as sound discoloration, especially with pulsating music. The envelope curve is changed by the transient and decay errors. In addition, phase errors affect the location of the sound sources and thus create an artificial spatiality. Above all, the errors in the swing-in and swing-out processes as well as an impaired location of the sound sources let the listener recognize that the music is not live.

Methods are known which use equalizers to enable different volume influences in the different frequency ranges and thus improve the first error, that is to say only the amplitude frequency response. The disadvantage here is that the errors in the phase frequency response and thus the transient and decay processes as well as the positionability are not improved but rather worsened.

From DE-C-31 30 353 a method is already known which alone improves the swing-in and swing-out errors. The disadvantage here is that if there are no pulses in the sound material, the error in the amplitude frequency response is not improved.

Attempts were also made to compensate for the principle-related errors of the dynamic transducers when converting from electrical to acoustic vibration by means of feedback. Fig. 3 shows the known arrangement of a loudspeaker with a sensor for the membrane movement.

For this purpose, the movement of the membrane is scanned capacitively, inductively, piezoelectrically or optically and the electrical actual value signals generated in this way are compared with the setpoint signals. The readjustment is carried out via a differential amplifier. The capacitive motion sensors record not only the entire diaphragm movement but also all partial vibrations of the diaphragm, the inductive sensors move in the strongly changing magnetic field, which is influenced by the excitation winding through which current flows. They therefore only allow a rough detection of errors. The piezo transducers are relatively heavy and, due to their own weight, increase the error to be corrected. They cannot be used for the mid and high range. The optical pickups with their own control electronics are uneconomically expensive.

Because of the phase-changing properties of the loudspeaker and the transducer, the control loop would start to oscillate if the loop gain was high. To prevent this, the loop gain must be set to small values, e.g. 20, can be reduced, which greatly affects the effectiveness of the feedback.

By readjustment, moreover, only the amplitude error that occurs is always recognizable, recordable and controllable.

If errors occur in the phase position of pulses, they are expressed, for example in too small amplitudes. However, a pure amplitude adjustment requires excessive correction current pulses in the case of the phase-in transient oscillation, which the amplifier cannot usually deliver by providing its power for the music pulse. In addition, such readjustments of the membrane can only take effect with a certain delay after the error has occurred, and thus, in particular if the phase position is incorrect, the errors can never be prevented in principle.

With high amplitude changes, as they often occur in modern entertainment and dance music, the high readjustment correction signals can lead to short-term overloads of the power amplifier and thus to high distortion factors.

While the readjustment in practice with the amplitude errors in the transfer function of the loudspeaker e.g. at its resonance frequency can have a balancing effect over several oscillation periods, it shows little effect in the phase-dependent improvement of the transient and decay processes in the event of sudden changes in amplitude in the decisive first half oscillation period. Readjustments of the type described cannot, of course, be applied to microphones and pickups.

In order to avoid the problems with the sensors on the loudspeaker diaphragm, attempts have already been made to work with the aid of an electrical simulation of the loudspeaker as an equivalent circuit according to FIG. 4.

In this electrical equivalent circuit made up of passive components, with coils, capacitors and resistors, the various characteristic values and parameters depend on each other and act in the same way as with the electrodynamic loudspeaker. Does the electrodynamic converter change e.g. its vibrating mass, this results in changes in its phase errors in the low and high frequencies as well as changes in the entire transmission range of the sound pressure curve. Any change in a value of the passive components of the electrical equivalent circuit thus affects the entire transmission range of the equivalent circuit and is not limited to a limited frequency section of the transmission range of the circuit, as is the case, for example, with adjustment measures with the aid of equalizers.

The electrical values as an example for a bass, middle and tweeter according to FIG. 4 are in compiled in the following table and show great differences.

Another bass speaker with a resonance frequency of 37 Hz can already have values of C = 300 u.F, L = 60 mH and R = 50 9. Passive components in this order of magnitude that can be matched to different loudspeakers are only feasible with a disproportionately large, uneconomical effort.

An equivalent circuit for the actual loudspeaker was used to try to get a better correction signal. The equivalent circuit is used for this purpose in a feedback circuit according to FIG. 5a. The disadvantage of this circuit is that replacement circuits constructed discretely with coils, capacitors and resistors, as well as the electrodynamic converters themselves, already show considerable differences in the assembled end product, even with small component and manufacturing tolerances. Such an equivalent circuit constructed with discrete components is therefore difficult to match to the actual speaker conditions, cannot be tuned and is expensive. 4 can also be arranged inversely in series with the loudspeaker (FIG. 5b), which is known from US Pat. No. 3,988,541. The current was also injected in order to neglect the components of the voice coil impedance and voice coil inductance in the equivalent circuit. However, the disadvantages of the large component tolerances of the loudspeaker and equivalent circuit and the almost impossible comparability for a specific loudspeaker remain here, which make this method not applicable in practice.

It is also not possible to avoid the disadvantages described above of the electrical equivalent circuit of a loudspeaker constructed with discrete components by using its more easily tunable electrical equivalent circuit as an analog computing circuit with active components according to FIG. 6. Since the exact electrical replica for a loudspeaker system as an analog arithmetic circuit already consists of several feedback loops and changes the eigenvalues through a further feedback loop, it cannot be switched into a feedback branch like a loudspeaker replacement circuit constructed with passive components according to FIG. 5a. This also makes the circuit unstable and starts to oscillate.

To operate the analog arithmetic circuit for the loudspeaker in accordance with FIG. 5b inversely in series with the loudspeaker is also not possible, because, like all electronic circuits with operational amplifiers, it only works in one direction and the inputs and outputs cannot be interchanged to reverse the effect.

From US-A-4 052 560 a compensation device of the type mentioned for playback errors of an electroacoustic transducer is known, which is used to straighten the sound pressure curve at the upper and lower end of the transmission range of electrodynamic loudspeakers. The circuit arrangement consists of active components. The individual circuit sections or filters are effective for the separate compensation of errors in different frequency ranges, i.e. Adjustment measures in a circuit section only affect the corresponding frequency range and do not affect the entire transmission range of the converter. The known circuit arrangement cannot fully take into account or compensate for the complex transmission behavior of the converter and its transmission errors. An approximate linearization of the frequency response or sound pressure curve is provided. After the compensation has been carried out, there are still significant phase errors.

According to a proposal by RA Greiner and M. Schoessow, "Electronic Equalization of Closed-Box Loudspeakers", J. Audio Eng. Soc., 1983, pages 125-134 become the poles of the transducer transfer function to improve the frequency response and transient response of a speaker system compensated by zeroing in the transfer function of an equivalent electronic filter or equalizer. This proposal does not enable distortion-free reproduction of rectangular pulses.

From US-A-4 340 778 it is also already known to compensate individually for the influence of the voice coil, the acoustic efficiency, the mechanical suspension, the damping, etc., by means of a circuit. Several compensation circuits are arranged one after the other. However, since all influences of the mechanical vibration system of the electrodynamic loudspeaker are dependent on one another and also influence one another again, such compensation circuits which are only effective in limited frequency ranges cannot effectively prevent the errors, but rather create new, other errors which are also manifested as distortion factors or sound distortions .

Circuits to influence the sound pressure curve of loudspeakers in any way are equalizers. These filter circuits are suitable for linearizing the sound pressure curve or for straightening the amplitude frequency response. In the case of the so-called graphic equalizers, the transmission frequency range from 20 to 20,000 Hz is divided into predetermined, frequency-limited sub-ranges, in which corrections are carried out in each case. A setting for other correction areas is not possible. The so-called parametric equalizers, of which US Pat. No. 4,052,560 describes a circuit arrangement, also enable the subsequent adjustment of special partial frequency ranges and the filter characteristic, as a result of which adaptation to any correction points can take place.

With the graphic equalizers, the phase frequency response of the converter is generally not improved, but often the phase error of the equalizer circuits is superimposed on it. The parametric equalizers also allow for very limited improvements in the phase frequency response.

A major advantage of the equalizer is the limitation of the setting options of its active filters to relatively small sub-frequency ranges, which makes setting any, even linear, sound pressure curves relatively easy and clear.

The complex relationships of the individual parameters in the electrodynamic converter, which in their interdependency extend over the entire transmission range, are only insufficiently recorded in the equalizers. The phase frequency response is influenced by the approximately linearization of the sound pressure curve, but is not properly compensated. The correction in each case in partial frequency ranges does not take sufficient account of the complex overall relationship of the electrodynamic converter. Instead of compensating for the phase errors, these are at most brought to smaller values.

This is stated in V. Staggs, "Transient-Response Equalization of Sealed-Box Loudspeakers", J. Audio Eng. Soc., 1982, pages 906 to 910, measures in the frequency range of the resonant frequency of the transducer and a small part of the bass range falling sharply below it being limited. Several filter characteristics are compared with respect to the resulting transient distortion. When using biquadratic filters with a lower order, this results in better transient processes than with simple high-pass filters of a higher order. In both cases, however, errors in the transient processes remain clearly demonstrable. This confirms that with all filters that are only used in limited frequency sections of the transmission range of an electrodynamic loudspeaker, the phase errors can only be compensated for inadequately.

Bass equalization in the lower frequency range is described in numerous references. The known circuits often do not record the entire mechanical vibration system of the electrodynamic loudspeaker in its full transmission range. In addition, these circuits are often not assigned to a single electrodynamic loudspeaker, but they are connected upstream of multipath loudspeakers which contain high, medium and bass loudspeakers and a crossover network. However, since crossovers themselves cause phase errors and the phase errors of two sub-frequency loudspeakers produce further errors when they are superimposed, such measures can actually only straighten the sound pressure curve; the phase errors cannot be compensated for.

The invention has for its object to provide a device for compensating playback errors of an electroacoustic, in particular working according to the electrodynamic principle, by which the signals occurring in the electrical section of the transmission path are changed so that the system-related errors are at least largely compensated. The compensation devices should consist of inexpensive active electronic components and setting elements and should be easily and individually adjustable in a wide range to different converter types.

This object is achieved by the features in the characterizing part of patent claim 1.

The fact that the different loudspeaker copies of the same type have large electrical differences even with small component and manufacturing tolerances means that they are light, individual Adjustability to the respective copy a considerable advantage.

The advantages of the compensation circuit are even clearer if one takes into account that the easy adjustability is just as easily possible not only for loudspeakers of the same series type, but even for loudspeakers that are as different as woofers, midrange or tweeters. Compared to the manufacturing costs of passive, i.e. with capacitors, coils and resistors, built-up equivalent circuits with large component values, the material cost of the active electronic components and the adjustability of the actuators result in a great cost advantage.

However, the fact that the compensation circuit can be used universally, i.e. for all electrodynamic loudspeaker systems and electrodynamic headphones as well as for all electrodynamic microphones and electrodynamic pickups, results in a large area of application with a repeated increasing cost advantage and manufacturing advantage due to mass or series production.

If, in the case of using the compensation circuit in all branches of a reusable loudspeaker box, the crossover is constructed in accordance with DE-C-33 04 402 and thus ensures correct settling and all phases in all frequency ranges, the transient response of the bass, mid-range and tweeters results in Sound bursts from sound mixtures, as they are common in music, for example when playing the piano, guitar and drum, there are no phase shifts and no sound changes over the entire reusable loudspeaker box due to overlapping frequency ranges that have experienced different phase shifts. The membranes of the tweeter, midrange and bass speaker remain in the same phase with all suggestions, whether through impulses or through long-lasting tones. This solves the problem of the crossover frequency between bass and midrange or midrange and tweeter for the first time in a practical and cost-effective way. In the previous practice, for the reasons listed, only the compromise was possible that either the respective membranes could move in phase with steady tones or with impulses, and deliver acoustically correct superimpositions.

Another advantage is that standard loudspeaker specimens can be used in loudspeaker construction. You don't need any special designs, e.g. with sensors for readjustment or expensive, narrow-tolerance components and special manufacturing processes to maintain certain parameters.

Another advantage is that the electrical characteristics of the compensation circuit do not change during operation, as is the case with coils and capacitors due to heating during operation. It is also advantageous that non-linearities due to components such as e.g. in the coil due to hysteresis, saturation and eddy current, do not occur in the adjustable compensation circuit with operational amplifiers.

The easy and universal tunability is also advantageous if a converter is destroyed and has to be replaced. Here the compensation circuit provides a high utility value for repairs.

But also the adjustability to speaker developments in the future, such as on new loudspeakers with magnetic liquid in the air gap of the magnet or loudspeakers with new flat membranes, increases the value in use.

A significant advantage of the compensation circuit is also to be emphasized that it can be realized extremely inexpensively by only a few active components.

Likewise, the small space requirement of the compensation circuit, which is easily conceivable in the size of an operational amplifier customary today, should be compared to the large passive components of a loudspeaker equivalent circuit, e.g. when used in the bass range.

• Fig. 1 zeigt das Amplituden-Resonanz-Verhalten bekannter elektrodynamischer Wandler für verschiedene Dämpfungsfaktoren α,
• Fig. 2 zeigt das Phasen-Resonanz-Verhalten bekannter elektrodynamischer Wandler für verschiedene Dämpfungsfaktoren α,
• Fig. 3 zeigt das Schema bekannter Membranrückkopplungen bei Lautsprechern,
• Fig. 4 zeigt eine mit passiven Bauteilen aufgebaute elektrische Ersatzschaltung eines bekannten elektrodynamischen Lautsprechers,
• Fig. 5a zeigt das Schema einer Rückkopplung über eine den elektrodynamischen Lautsprecher simulierende, mit passiven Bauteilen aufgebaute, bekannte elektrische Ersatzschaltung,
• Fig. 5b zeigt eine zu der Schaltung gemäß Fig. 5a elektrisch gleichwertige Schaltung mit invers und in Reihe geschalteter, bekannter, mit passiven Bauteilen aufgebauter, elektrischer Lautsprecherersatzschaltung für den elektrodynamischen Lautsprecher,
• Fig. 6 zeigt eine bekannte elektrische Ersatzschaltung eines elektrodynamischen Lautsprechers in einem Aufbau als Analogrechenschaltung,
• Fig. 7 zeigt eine mit passiven Bauteilen aufgebaute, bekannte elektrische Lautsprecherersatzschaltung für den elektrodynamischen Lautsprecher mit anschließender Differenzierstufe,
• Fig. 8a zeigt den Dämpfungsverlauf, der sich aus dem Lautsprecher oder seiner Ersatzschaltung nach Fig. 7 für das Beispiel eines elektrodynamischen Baßlautsprechers ergibt,
• Fig. 8b zeigt den Phasenwinkelverlauf, der sich aus dem Lautsprecher oder seiner Ersatzschaltung nach Fig. 7 für das Beispiel eines elektrodynamischen Baßlautsprechers ergibt,
• Fig. 9a zeigt den Prinzipaufbau einer erfindungsgemäßen Kompensationsschaltung mit 3 Integratoren,
• Fig. 9b zeigt ein abgewandeltes Ausführungsbeispiel einer erfindungsgemäßen Kompensationsschaltung nach Fig. 9a,
• Fig. 9c zeigt ein abgewandeltes Ausführungsbeispiel einer erfindungsgemäßen Kompensationsschaltung mit 4 Integratoren,
• Fig. 9d zeigt ein abgewandeltes Ausführungsbeispiel einer erfindungsgemäßen Kompensationsschaltung nach Fig. 9c,
• Fig. 9e zeigt ein abgewandeltes Ausführungsbeispiel einer erfindungsgemäßen Kompensationsschaltung nach Fig. 9a,
• Fig. 10a zeigt den entsprechenden Verlauf der Dämpfungsfunktion der Kompensationsschaltung für das errechnete Beispiel des elektrodynamischen Baßlautsprechers,
• Fig. 10b zeigt den entsprechenden Verlauf des Phasenwinkels der Kompensationsschaltung für das errechnete Beispiel des elektrodynamischen Baßlautsprechers,
• Fig. 11a zeigt den Verlauf des Dämpfungsfehlers gegenüber der idealen Übertragungsfunktion in einem Diagramm,
• Fig. 11 b zeigt den Verlauf des Phasenfehlers gegenüber dem idealen Phasenverlauf in einem Diagramm,
• Fig. 12 zeigt ein Blockschaltbild der erfindungsgemäßen Einrichtung unter Verwendung einer digitalen Rechenschaltung,
• Fig. 13 zeigt eine Einrichtung mit Aufteilung des gesamten Frequenzbereichs des Eingangssignals in drei Teilfrequenzbereiche und
• Fig. 14 zeigt eine Variante der Einrichtung nach Fig. 13.

The invention is described in more detail below with the aid of schematic drawings, formulas and a specific application example for a bass loudspeaker. In the following list, the drawings Fig. 1 to 6 already dealt with are included.

• 1 shows the amplitude-resonance behavior of known electrodynamic transducers for various damping factors α,
• 2 shows the phase-resonance behavior of known electrodynamic transducers for various damping factors α,
• 3 shows the schematic of known membrane feedback in loudspeakers,
• 4 shows an electrical equivalent circuit of a known electrodynamic loudspeaker constructed with passive components,
• 5a shows the diagram of a feedback via a known electrical equivalent circuit which simulates the electrodynamic loudspeaker and is constructed with passive components,
• FIG. 5b shows a circuit that is electrically equivalent to the circuit according to FIG. 5a, with an inverse and series-connected, known electrical loudspeaker constructed with passive components equivalent circuit for the electrodynamic loudspeaker,
• 6 shows a known electrical equivalent circuit of an electrodynamic loudspeaker in a structure as an analog computing circuit,
• 7 shows a known electrical loudspeaker replacement circuit for the electrodynamic loudspeaker with a subsequent differentiating stage, constructed with passive components,
• 8a shows the attenuation curve which results from the loudspeaker or its equivalent circuit according to FIG. 7 for the example of an electrodynamic bass loudspeaker,
• 8b shows the phase angle curve which results from the loudspeaker or its equivalent circuit according to FIG. 7 for the example of an electrodynamic bass loudspeaker,
• 9a shows the basic structure of a compensation circuit according to the invention with 3 integrators,
• 9b shows a modified embodiment of a compensation circuit according to the invention according to FIG. 9a,
• 9c shows a modified embodiment of a compensation circuit according to the invention with 4 integrators,
• 9d shows a modified embodiment of a compensation circuit according to the invention according to FIG. 9c,
• 9e shows a modified embodiment of a compensation circuit according to the invention according to FIG. 9a,
• 10a shows the corresponding course of the damping function of the compensation circuit for the calculated example of the electrodynamic bass loudspeaker,
• 10b shows the corresponding course of the phase angle of the compensation circuit for the calculated example of the electrodynamic bass speaker,
• 11a shows the course of the damping error versus the ideal transfer function in a diagram,
• 11 b shows the course of the phase error versus the ideal phase course in a diagram,
• 12 shows a block diagram of the device according to the invention using a digital arithmetic circuit,
• 13 shows a device with division of the entire frequency range of the input signal into three sub-frequency ranges and
• FIG. 14 shows a variant of the device according to FIG. 13.

7 shows a known loudspeaker equivalent circuit diagram with a differentiating stage connected downstream. The values for the example with the bass speaker are determined dynamically on the bass, i.e. measured the complex input impedance at different frequencies and used it to mathematically calculate the component values for the known equivalent circuit. The behavior of the equivalent circuit corresponds exactly to that of the loudspeaker itself.

The voltage U _{1 is} applied to the input terminals of the loudspeaker or its electrical exact replica by the equivalent circuit, the voltage U _{2 can be} tapped at the output terminals.

The damping function results from the ratio U1 / U2, and the phase angle curve results from the phase shift from U ₁ to U ₂ . The general mathematical damping function for the example above is:

(Equation 1)

The component values are standardized to simplify the calculation. The freely selectable reference values (index B) are chosen in such a way that the relationships are as simple as possible.

The normalized values are used in equation (1) and give the dimensionless coefficients of equation (2).

(Equation 2)

or again in a different spelling than

(Equation 3)

• the coefficients result
• q ₁ = 0.494082
• q ₂ = 2.439917
• q ₃ = 12.54577
• _Tn ⁼ _T / Ts
• C _o ⁼ 0.031609

This damping function to be compensated for by the compensation circuit as a function of the frequency is shown in FIG. 8a for the example of the bass loudspeaker, but runs in basically the same way for all electrodynamic converters. Likewise, the phase angle curve to be compensated for by the compensation circuit as a function of the frequency was recorded in FIG. 8b for the example of the bass loudspeaker, but this curve also runs schematically for all electrodynamic converters (see also FIG. 2). Simply reversing equation (3) in order to get the entire loudspeaker behavior in inverse form does not provide a solution, since this function is not stable in terms of circuitry and vibrates in itself.

The path to a compensation circuit is shown below which, like the equivalent circuit of the loudspeaker as an analog computer, has similarly complex cross-connections to one another, but only represents the inverse function in a sufficiently good approximation in the loudspeaker transmission range.

Outside the transmission range, e.g. for a bass loudspeaker in the mid-range or for a mid-range speaker in the bass and tweeter range or for a tweeter in the bass and mid-range, an error that can be determined arbitrarily by the setting of the circuit is permitted. However, because the loudspeakers are operated via a crossover which strongly attenuates the frequency range outside the transmission range, this permitted error does not appear in practice at all. It is therefore expedient in this exemplary embodiment to arrange the compensation circuit after the crossover and in front of the respective sub-frequency loudspeaker.

In the method according to the invention, the inverse function H (p) is generally applied as a polynomial in such a way that the numerator from equation (3) with the coefficients that were determined on the loudspeaker comes into the denominator of equation (4) and the new counter in equation (4) is generally applied. The mathematical stability criterion here requires that the degree of the numerator of the polynomial is equal to or greater than the degree of the denominator.

(Equation 4)

A general approach would also be possible, in which the coefficients of the denominator are also calculated, or another approach with the numerator according to the fourth order or even higher. But if all coefficients e.g. the numerator and denominator are freely selectable, the computing effort to achieve a good approximation is greater. Even if the order of the denominator is set higher than necessary, on the one hand there is more computing effort, and on the other hand there are many integration levels in the circuit arrangement corresponding to the degree of the order, which then, the more complicated it becomes, can itself have errors in signal processing. In practice, the last integration stages due to the weakening of the signal have only a slight influence on the compensation curve by adjusting the potentiometers. A circuit of fourth, fifth and higher order with 4, 5 or more integration levels is therefore no better than the exactly matched compensation circuit with 3 integration levels.

• 1. Die Einstellung und Verbesserung der frei wählbaren und zu bestimmenden Koeffizienten muß immer am Gesamtübertragungssystem erfolgen, da nur so die komplexen Rückwirkungen der Einstellung eines Koeffizienten auf die anderen gewährleistet sind.
• 2. Die Annäherung der Übertragungsfunktion erfolgt nur im gewählten Übertragungsbereich an die inverse Dämpfungsfunktion nach Gleichung (3).

It is important to determine the coefficients for equation (4) or another higher degree equation from several points of view in an iterative solution procedure to the desired accuracy. These are:

• 1. The setting and improvement of the freely selectable and to be determined coefficients must always be carried out on the overall transmission system, since this is the only way to ensure the complex repercussions of the setting of one coefficient on the other.
• 2. The transfer function is only approximated in the selected transfer range to the inverse damping function according to equation (3).

• 3. Die Form der Annäherung der Übertragungsfunktion im gewählten Übertragungsbereich an die inverse Dämpfungsfunktion nach Gleichung (3) soll vorzugsweise in monotoner Form erfolgen. Wenn sich die angenäherte Kurvenform des Dämpfungsverlaufs nicht monoton an die gegebene Kurvenform annähert, sondern sich z.B. um die gegebene Kurvenform mit positiven und negativen Abweichungen herumwindet, ergibt sich keine gute Übereinstimmung in der Annäherung des Phasenwinkelverlaufs. Die monotone Annäherung der Dämpfungsfunktion kann gut in der Darstellung des Dämpfungsfehlers gegenüber der idealen Übertragungsfunktion nach Fig. 11a beurteilt werden.
• 4. Die Form der Annäherung des ermittelten Phasenwinkelverlaufs im gewählten Übertragungsbereich an den inversen Phasenwinkelverlauf soll optimal sein.

Such a curve is shown for the example of the bass speaker in Fig. 10a.

• 3. The form of the approximation of the transfer function in the selected transfer range to the inverse damping function according to equation (3) should preferably be in monotonous form. If the approximated curve shape of the damping curve does not approximate monotonously to the given curve shape, but instead winds around the given curve shape with positive and negative deviations, there is no good agreement in the approximation of the phase angle curve. The monotonous approximation of the damping function can be judged well in the representation of the damping error compared to the ideal transfer function according to FIG. 11a.
• 4. The shape of the approximation of the determined phase angle curve in the selected transmission range to the inverse phase angle curve should be optimal.

• 5. Es soll eine Fehlerabschätzung der Annäherung für Dämpfungsfunktion nach Fig. 11a und des Phasenwinkelverlaufs nach Fig. 11b b in dem gewünschten Übertragungsbereich, am Rand des gewünschten Übertragungsbereichs, außerhalb des gewünschten Übertragungsbereichs erfolgen.

Such a curve is shown for the example of the bass loudspeaker in FIG. 10b.

• 5. An error estimate of the approximation for damping function according to FIG. 11a and the phase angle curve according to FIG. 11b b is to take place in the desired transmission area, at the edge of the desired transmission area, outside the desired transmission area.

• bis das gewünschte Ergebnis erzielt ist. Die Koeffizientenverbesserung erfolgt immer schrittweise und im Gesamtübertragungssystem. Die einzelnen Rechenschritte können numerisch,
• mit Hilfe von Rechencomputern,
• mit Grafikcomputern erfolgen.
• Hierbei kann die Koeffizientenveränderung direkt in der Wirkung auf die Kurvenveränderung beurteilt und dadurch das Verfahren beschleunigt werden.

The approximation process itself is carried out by a suitable choice of the coefficients, which are improved as long as

• until the desired result is achieved. The coefficient improvement is always gradual and in the overall transmission system. The individual calculation steps can be numerical,
• with the help of computing computers,
• done with graphics computers.
• Here, the change in the coefficient can be assessed directly in terms of the effect on the change in the curve and the process can thereby be accelerated.

With already known coefficients e.g. for loudspeakers of the same series type, the fine adjustment can be carried out with the oscilloscope by correctly setting the phase angle curve. For this purpose, the compensation circuit is connected in series with the electrodynamic loudspeaker system and the entire transmission system comprising the compensation circuit and electrodynamic converter or its exact equivalent circuit is supplied with square-wave signals of different frequencies. The variation of the coefficients corresponds to the adjustment of the adjustable potentiometers of the compensation circuit. The aim of the optimization is to reproduce the square-wave signal form, which is as error-free as possible, from the converter or its equivalent circuit and thus the transient and decay processes. This can be done optically very well on the oscilloscope compared to the input signal.

• q, = 0.494082
• q₂ = 2.439917
• q₃ = 12.54577
• nach mehreren Annäherungsrechenschritten folgende Koeffizienten
• C = 4.839
• ^w _" ⁼ 0.25
• Q = 0.707
• q = 50
• oder für die umgeformte Gleichung 5a

In the example for the bass speaker described so far, the equation (4) and the values for gave

• q, = 0.494082
• q ₂ = 2.439917
• q ₃ = 12.54577
• after several approximation steps, the following coefficients
• C = 4,839
• ^w _" ⁼ 0.25
• Q = 0.707
• q = 50
• or for the transformed equation 5a

(Equation 5a)

the coefficients

Reference frequency f _B = 65.05284 Hz

These are the coefficients which, in the circuit arrangement according to FIG. 9a, only have to be carried out as settings on the potentiometers P ₁ to P ₇ . Any fine adjustment on the electrodynamic loudspeakers that may be required by the circuit components is carried out, as described above, with the aid of an oscilloscope.

How exactly the compensation circuit can equalize the existing loudspeaker intrinsic values can be shown using the example of the bass loudspeaker in the error curves in FIGS. 11 a and 11 b.

The error in the area of the sound pressure transmission curve is 40 - 500 Hz less than 0.1 dB. The error in the phase angle curve is less than ± 10 ° in the range from 80 - 800 Hz.

The circuit arrangement according to the invention according to FIG. 9a is described in more detail below.

The circuit arrangement according to the invention according to FIG. 9a has three positive integrators B ₁ , B ₂ and B _{3 connected} in series according to the degree of derivations according to equation (5a). At the input, the input signal U _{1 is introduced} into a summer S ₁ . The returns Ro, R ₁ and R ₂ from the circuit are also introduced into this summer S ₁ , which have the adjustable potentiometers P ₇ , P ₆ and P ₅ arranged in their return branch. The returned signals are taken from the outputs of the integrators Bi, B ₂ , B ₃ and inverted using the inverters lo, I ₁ and 1 ₂ . The four decouplings Ao, A ₁ , A ₂ and A ₃ , which have the adjustable potentiometers P ₄ , Ps, P ₂ and P ₁ in their branches and in the summer S, take place from the circuit arranged in series of the input summer and the three integrators _{2 can be} initiated. The output voltage U _{2 can be} tapped at the output of the summer S ₂ . Integrators are available as integrated circuit modules (e.g. TL 071 CP or TL 074 from Texas Instruments).

The circuit arrangements according to FIG. 9b, 9c, 9d and 9e are modified exemplary embodiments of the circuit arrangement according to the invention according to FIG. 9a, which can be derived analogously from the circuit arrangement according to the invention according to FIG. 9a and the mathematical approach. S are summers, B integrators, R feedbacks, A decouplings, P potentiometers that can be set to coefficients and I inverters.

In the modified circuit arrangement according to the invention according to FIG. 9b, the three integrators do not follow one after the other, but only two. A third integrator is switched separately.

The mathematical approach for this is:

(Equation 5b)

The modified circuit arrangement according to the invention according to FIG. 9c was realized from the mathematical approach to a fourth-order equation with four integrators arranged one behind the other.

The mathematical approach for this is:

(Equation 5c)

The modified circuit arrangement according to the invention according to FIG. 9d was not implemented with four integrators in series compared to the circuit arrangement according to the invention from FIG. 9c, but rather with two times two integrators arranged one behind the other.

The mathematical approach for this is:

(Equation 5d)

The modified circuit arrangement according to FIG. 9e shows that an embodiment is also possible in which the integrators are not connected in series directly as in FIG. 9a, but each integrator is visible for itself in a circuit diagram closed by feedback and decoupling, and these circuit arrangements are then simply strung together.

The mathematical approach for this is:

(Equation 5e)

In the known circuit according to FIG. 7, the signal tapped off from the known equivalent circuit of the electrodynamic converter according to FIG. 4 is differentiated once. This gives the transfer function for damping or acceleration. The method and the circuit arrangement for compensating for the malfunction of electrodynamic transducers have been described in detail so far on the basis of this transfer function, which is proportional to acceleration or damping.

The predistorted signal, which is proportional to acceleration or damping, is suitable for being sent directly to the power amplifier for the electrodynamic converter in order to compensate for its own behavior. However, it is also possible to tap the signal from FIG. 4 directly without arranging a differentiating stage as in FIG. 7. In this way, the speed-proportional transfer function of the electrodynamic equivalent circuit or the converter is obtained.

Here too, a similar mathematical approach and an iterative solution of the inverse speed-proportional transfer function is possible with the same compensation circuit arrangement. You only get other coefficients. In order to be able to forward this predistorted, speed-proportional signal to the power amplifier for the electrodynamic converter, it has to be differentiated in order to obtain the acceleration-proportional, predistorted voltage function.

However, it is also possible to tap the signal from FIG. 4 and, instead of differentiating it once as in FIG. 7, integrating it once. In this way, the deflection-proportional transfer function of the electrodynamic converter or its equivalent circuit is obtained. Here too, a similar mathematical approach and an iterative solution of the transfer function that is proportional to the deflection are possible with the same circuit arrangement. But you get different coefficients here too. In order to be able to forward the predistorted signal proportional to the deflection to the power amplifier for the electrodynamic converter, however, it has to be differentiated twice in order to accelerate the predistorted one voltage function proportional to the supply.

According to US-A-3 988 541 it is also known to connect the inverse loudspeaker equivalent circuit in series with the loudspeaker without voice coil influence, that is to say without voice coil resistance and voice coil inductance. With this circuit arrangement, however, the current must be impressed on the loudspeaker, otherwise the voice coil influences would not be negligible.

This type of equivalent circuit constructed with passive components can also be approximated by a compensation circuit according to the invention. Because there is no voice coil influence, you only get a second order approach. The coefficients are determined using the same iteration method. The disadvantages of this circuit arrangement are that current-impressed amplifiers are not common because they are very difficult to dimension correctly and can easily become unstable. Damping the membrane movement by the current of the amplifier is not possible with current-impressed amplifiers, but it is possible with voltage-impressed amplifiers.

The measures described using a bass loudspeaker to compensate for system-related playback errors can in principle also be used unchanged in connection with inexpensive electrodynamic microphones or electrodynamic pickup systems, since the latter have the same vibration behavior as loudspeakers. The only natural difference is that the change in the electrical signals in terms of the signal flow in loudspeakers takes place in front of them and therefore represents a pre-distortion, while the signal change in microphones or pickups takes place behind them and is therefore an equalization.

Instead of designing the computing circuit to compensate for playback errors as a purely analog circuit, a digital computing circuit can also be used. This possibility, which is particularly advantageous when the electrical signals already exist as digital signals when converting electrical into acoustic signals, is described below.

FIG. 12 shows the block diagram of a corresponding device which is used to generate a predistorted control signal for the electroacoustic transducer derived from the original input signal. The predistortion must be dependent on the instantaneous course of the input signal and must be such that the inadequacies of the real electrodynamic converter, including the surrounding medium, are compensated for as far as possible.

According to FIG. 12, the original input signal U is converted into a sequence of digital signals DS1 by an analog / digital converter A / D. The digital signals DS1, which are output with a high repetition frequency (sampling frequency) of, for example, 100 kHz compared to the highest frequency of the input signal, represent the binary coding in each case one of e.g. 128 different amplitude values. Each e.g. The 7-bit data thus represents the (instantaneous) amplitude values at the time of sampling in the time course of the input signal U.

The sequence of digital signals DS1 is fed to the data inputs of a microcomputer R, which essentially consists of a microprocessor MP, at least one programmable read-only memory PROM and a read / write memory RAM as working memory, and a number of auxiliary devices, which are not dealt with in more detail is known.

The PROM read memory stores all the characteristic values which are relevant for the reproduction quality of the electroacoustic transducer, that is to say, for example, an electrodynamic loudspeaker built into a housing with an upstream power amplifier or a microphone. These parameters relate to variables such as slip, mass inertia of the sound-emitting membrane and the upstream air volume, clamping and restoring forces, damping, resonance frequencies, etc., as well as frequency response and internal resistance of the power amplifier, if applicable.

With the aid of a program likewise stored in the programmable read-only memory already mentioned or in a second, separately addressable memory of the same type, the digital signals DS1 entered into the computer, which are now referred to as primary digital signals, are converted into secondary digital signals DS2 in accordance with the characteristic converter characteristic values.

However, in order to be able to make determinations about the course of the input signal U _i , the computer R requires at least three successive samples of the curve course of the input signal. He can then see both the slope and the curvature of the curve. The changes in the curve of the input signal U, which are of particular interest for the present purpose, can be determined by comparing them with earlier samples.

The implementation of the conversions, which amounts to the solution of differential equations of the forced vibration (see Istvan Szabo, Introduction to Technical Mechanics, Springer-Verlag 1963, pages 348, 349), is no longer discussed here.

Since any necessary correction of the secondary digital signals DS2 should take place as early as possible, the input must be used to convert the digital signal assigned to the first of three samples wait for the next two digital signals. This results in a delay that must be taken into account in addition to the pure computing time.

According to FIG. 12, the sequence of the secondary digital signals DS2 is converted by an digital / analog converter D / A connected to the data output of the microcomputer R into an analog control signal U ₂ which is used to control the electroacoustic converter W. In general, however, the electroacoustic converter W is preceded by a power amplifier EV, which only amplifies the analog control signal U ₂ . Since the characteristics of the power amplifier EV, in particular its frequency response and internal resistance, are also included in the transmission chain from the original input signal U ₁ to the acoustic oscillation, as already mentioned, these variables together with the characteristic converter characteristic values must also be taken into account when calculating the secondary digital signals DS2 will.

Digital recording of music has become increasingly important in recent years. Devices for reading such recordings are able to immediately output a sequence of digital signals corresponding to the recorded information. It goes without saying that in such cases the provision of an analog / digital converter is not necessary.

Are electroacoustic transducers, e.g. Speakers, preferably used for playing music, then the entire frequency range of the input signal is usually divided into, for example, three sub-frequency ranges. A specially designed loudspeaker is provided for each partial frequency range. The frequency range is divided by crossovers, which can be designed as LC elements, as filters with operational amplifiers or as digital filters. The latter is particularly useful in connection with digital recording.

Correction of the input signal in the highest partial frequency range, the high frequency range, is often not necessary. This case is shown in Figure 13. The original input signal U ₁ is divided by crossovers FW1 to FW3, the crossover FW1 for the deepest and the crossover FW3 for the highest partial frequency range.

To compensate for the signal delay caused by correction units K1 and K2 consisting of an analog / digital converter, computer and digital / analog converter, a delay element DEL is provided in the highest partial frequency range. The electroacoustic transducers and the upstream power amplifiers are designated W1 to W3 and EV1 to EV3.

Instead of a passive delay element, a clock-controlled shift register arrangement can also be provided, which, however, must be preceded by an analog / digital converter and followed by a digital / analog converter. In the meantime, the analog / digital converter in connection with digital recording can be omitted again. Furthermore, the shift register arrangement can be replaced by a further microcomputer, the sole task of which is now to delay the signal.

By delaying the sampling clocks in the analog / digital converters A / D1 or A / D2 for the low-frequency and mid-range, preferably by half a clock period, it is possible, according to FIG. 14, for the primary digital signals DS11 and DS12 of both partial frequency ranges to be the data inputs of a common microcomputer Rg alternately fed and also processed alternately. A prerequisite for this is a sufficiently high processing speed of the microcomputer Rg and, of course, adapted programming.

The secondary digital signals output by the microcomputer Rg must be distributed according to their affiliation to the two channels assigned to the low and mid range. This is done with the aid of a MUX multiplexer controlled by the microcomputer Rg. The multiplexer MUX can, however, be omitted if the subsequent digital / analog converters D / A1 and D / A2 are equipped for a clock-controlled transfer of the digital input information and the transfer clocks which are synchronous with the data output of the microcomputer Rg are mutually out of phase.

Claims (15)

1. Device for compensating reproduction errors in an electroacoustic transducer (W) within a given frequency range, the transmission path of the signals to be reproduced comprising an acoustic and an electrical section, there being provided in the electrical section of the transmission path a circuit arrangement that receives input signals (Ui) and emits altered output signals (U₂), characterised in that the circuit arrangement in the electrical section of the transmission path is a computer circuit the output signals (U₂) of which are derived according to amplitude and phase from the input signals (Ui) in accordance with a function H(p) that is the inverse to the complex damping function that applies over the whole of the given frequency range of the transducer (W), is described by characteristic values describing the inherent response characteristic of the transducer (W), such as slip, the inertia of a diaphragm that is caused to move and of the air volume, tensioning and restoring forces, damping, resonant frequencies and so on, and describes the complex inherent response of the transducer in relation to the amplitude response and phase-frequency response, the computer circuit having adjusting members (p) for matching the function H(p) to the inverse complex damping function.

2. Device according to claim 1, characterised in that the electroacoustic transducer (W) serves to convert electrical signals into acoustic signals and that the computer circuit is arranged upstream of the transducer (W) in the transmission direction.

3. Device according to claim 1, characterised in that the electroacoustic transducer (W) serves to convert acoustic signals into electrical signals and that the computer circuit is arranged downstream of the transducer (W) in the transmission direction.

4. Device according to claim 2, characterised in that as the computer circuit there is provided a digitally operating microcomputer (R) to which input signals (Ui) are supplied in the form of series of primary digital signals (DS1) and which emits series of secondary digital signals (DS2), that with the microcomputer (R) is associated a programmable read-only memory (PROM) in which the inherent properties of the transducer (W) and a programme for the conversion of the primary digital signals (DS1) into the secondary digital signals (DS2) are stored and that, furthermore, a digital/analogue converter (D/A) is provided for converting the series of secondary digital signals (DS2) into analogue output signals (U₂).

5. Device according to claim 4, characterised in that an analogue/digital converter (A/D) is provided for converting input signals (Ui) in the form of analogue signals into series of primary digital signals (DS1).

6. Device according to claim 5, characterised in that frequency dividers (FW1 to FW3) for dividing the frequency range of the input signals (U₁) into a plurality of predetermined partial frequency ranges are provided, that for each partial frequency range a final amplifier (EV1 to EV3) and an electroacoustic transducer (W1 to W3) are provided and that at least in the lowest partial frequency range a correction unit (K1, K2) comprising a microcomputer (R), a digital/analogue converter (D/A) and, optionally, an analogue/digital converter (D/A) is arranged, while in the remaining partial frequency ranges devices (DEL) for delaying the signal are provided.

7. Device according to claim 6, characterised in that the primary digital signals (DS11, DS12) of the lowest and at least the next-highest frequency range are multiplexed to the data inputs of a common microcomputer (Rg) and that connected to the data outputs of the microcomputer (Rg) is a multiplexer (MUX) controlled by the microcomputer (Rg), which multiplexer connects the secondary digital signals (DS21, DS22) associated with the lowest and at least the next-highest frequency range alternately through to the inputs of the corresponding digital/analogue converters (D/A1, D/A2).

8. Device according to claim 6, characterised in that the primary digital signals (DS11, DS12) of the lowest and at least the next-highest frequency range are multiplexed to the data inputs of a common microcomputer (Rg), that the inputs of the digital/analogue converters (D/A1, D/A2) for the lowest and at least the next-highest frequency range are connected in parallel and are connected to the data outputs of the microcomputer (Rg) and that the transmission of the secondary digital signals (DS21, DS22) into the digital/analogue converters (D/A1, D/A2) can be multiplexed by signals supplied by the microcomputer (Rg).

9. Device according to claim 1, characterised in that the computer circuit is designed as an analogue circuit comprising a plurality of integrators (B), adjusting members (P) and at least two summing circuits (S), that the input signals (U₁) are applied to one input of the first summing circuit (S) and the other inputs are connected via inverters (I) and adjusting members (P) to outputs of at least one integrator (B) connected downstream of the first summing circuit (S), that the output of the first summing circuit (S) and the outputs of the integrators (B) are connected by way of further adjusting members (P) to the inputs of the second summing circuit (S) at the output of which the output signal (U₂) can be taken and that the number of integrators (B) contained in the computer circuit is equivalent to the order of the function H(p) by means of which the complex inherent response of the transducer (W) is matched in inverse form in relation to the amplitude/frequency response and phase/frequency response.

10. Device according to claim 9, characterised in that the number of integrators (B) connected directly in series is in each case equal to the order of the factors of the mathematical function H(p) which is optionally reduced to factors, that with each group of integrators (B) connected directly in series is associated a first and a second summing circuit (S) and corresponding adjusting members (P) and inverters (I) and that the output of the second summing circuit (S) of a preceding group is connected to an input of the first summing circuit (S) of a subsequent group.

11. Device according to claim 10, characterised in that the construction of the computer circuit corresponds to a mathematical function which in inverse form matches the damping-proportional and/or acceleration- proportional transfer function of the transducer (W).

12. Device according to any one of claims 9 to 11, characterised in that the construction of the computer circuit corresponds to a third-order mathematical function.

13. Device according to any one of claims 10 to 12, characterised in that the mathematical function is divided into any desired number and mixture of factors of the first and higher orders.

14. Device according to any one of claims 1, 2 and 9 to 13, characterised in that it is combined with a known circuit arrangement for diaphragm readjustment.

15. Device according to any one of claims 1, 2 and 4 to 14, characterised in that the moving coil of a loudspeaker is current-driven.

Images