EP0133948B1 - Method and apparatus for negative feedback of movement induced voltage in a loudspeaker - Google Patents

Abstract

1. Method for negative feedback of movement induced voltage of a moving-coil loudspeaker with respect to the signal voltage with the aid of a network, in which the movement induced voltage is obtained by subtraction between the loudspeaker terminal voltage and an offset voltage, characterized in that the electric signal to be reproduced is supplied to the loudspeaker (3) as an impressed current, that the input of a network (4, A'F , A'M ) comprising capacitors (CK1 etc.) resistors (R02 etc.) and operators and simulating at its output the voltage drop at the impedance of the firmly braked loudspeaker (3) is supplied with a voltage proportional to the impressed current and the output voltage of the network (4, A'F , A'M ) is used as the offset voltage for subtraction purposes.

Description

The invention relates to a method for negative feedback of the dynamic voltage of a dynamic loudspeaker with respect to the signal voltage, with the aid of a network in which the dynamic voltage is obtained by forming the difference between the loudspeaker terminal voltage and a compensation voltage, and a device for carrying out the method.

Particularly in the area of the basic resonance of the dynamic loudspeaker and low cabinet resonances, the loudspeaker diaphragm does not exactly follow the applied signal voltage, because the mechanical oscillation circuits, which are more or less firmly coupled to the diaphragm, tend to vibrate naturally when triggered by impulses, which are still disruptive even if they are caused by damping are significantly smaller than the actual signal. In addition, the mass-affected speaker membrane has difficulties, e.g. B. to follow pulse-shaped signal voltages. However, the transparency of the reproduction depends at least in the mid-range (200 Hz to 1 kHz) on the correct phase behavior, i.e. that is, a loudspeaker should be able to reproduce rectangular sound pressure profiles to a certain extent correctly in this frequency range. This means that the loudspeakers of the low and mid-range parts must be installed closely adjacent to avoid differences in delay, which generally leads to small loudspeaker housings. Because of the volume required of the woofer, the control of the diaphragm movement becomes even more urgent.

It is known - (Philips technology practice; technology of the loudspeaker box 545, Studio MFB-Philips contacts, issue 39/1976) - to use a special transducer, such as an accelerometer, to sense the membrane movement. This has the disadvantage that the loudspeaker is initially deteriorated by an additional transducer mass, and a standard loudspeaker system cannot be used. An attempt has already been made to use the movement voltage of the loudspeaker itself for feedback. This means that the loudspeaker terminal voltage has to be freed from the parts resulting from the signal voltage. It has been proposed to form a generic type (DE-A-719 499; DE-A-2 922 112) to use a bridge circuit for decoupling the kinetic voltage, which essentially consists of the impedances of the loudspeaker which is moved and braked. This circuit has - as impressively simple as it looks at first glance - the disadvantage that its power dissipation is relatively high due to the low impedance, and therefore the accuracy of the simulation deteriorates under high load because the braked system behaves differently thermally. On the other hand, the correct phase position of the feedback voltage is a prerequisite for stable behavior with the required degrees of negative feedback. The bridge circuit presents difficulties here, which is why it has so far hardly been used, although it has been repeatedly proposed theoretically.

In US-A-4393353, a voltage amplifier is connected upstream of the loudspeaker, a negative current feedback circuit and a negative voltage feedback circuit also being matched to one another in such a way that the weakening of the voltage circuit essentially corresponds to the amplification of the circuit. Two networks are therefore provided, but none of them is a network that simulates the voltage drop at the impedance of the loudspeaker that is locked.

With the AU-B-512 725, a contact of the loudspeaker is potentially controlled so that the internal resistance of the voltage source connected upstream is compensated for. The circuit thus behaves like an amplifier with a negative output resistance (steel method). This is usually implemented by positive feedback. In the publication, however, the special type of feed-forward prevents a possible instability of the circuit at the expense of a second power amplifier. The voltage controlled current source only serves to convert the loudspeaker current into a proportional voltage (i.e. to measure the loudspeaker current) which controls the mentioned (lower) contact of the load via amplifier 16. In the simplest case, it can be replaced by an ohmic resistor. Since in this document the voltage-controlled current source is included in the negative feedback of the amplifier, the load is also operated on a voltage source with a very low internal resistance. Furthermore, the braked impedance is approximated by an ohmic resistance, so that the imaginary part, which is particularly important at frequencies above the resonance frequency, is disregarded. It is not a negative feedback system, in which a correction signal would be generated by means of a comparable controlled variable with an input signal. The system behaves in such a way that the control and input variables are approximately proportional, which is not desirable over the entire frequency range.

Starting from the generic state of the art, the object of the invention is to develop a method and a device for negative feedback of the dynamic voltage of a dynamic loudspeaker in such a way that a higher fidelity and thus a higher transparency of the reproduction is achieved.

According to the invention, the above-mentioned object is achieved in a method of the type mentioned at the outset in that the electrical signal to be reproduced is supplied to the loudspeaker as an impressed current, that the input of a capacitor, resistors and operators, at its output the voltage drop at the impedance of the braked After the speaker a voltage proportional to the impressed current is fed to the network and the output voltage of the network is used as a compensation voltage for difference formation.

A device according to the invention provides that the loudspeaker is connected to a voltage-controlled current source with high internal resistance, that at the input of the network of capacitors, resistors and operators a current-tapped loudspeaker is present at the loudspeaker and is proportional to the current and that the output of the network is present at a subtractor amplifier .

In a preferred embodiment, it can be provided that the signal current impressed on the loudspeaker is sampled at a frequency above the hearing range, and in the switch-off phases the movement voltage at the loudspeaker terminals is sensed at suitable times and regenerated by a downstream low-pass filter, and that the impedance of the stalled loudspeaker caused interference voltage and the simulation network are eliminated.

The invention enables further development by means of a loudspeaker with small woofers and mid-range loudspeakers in order to obtain a small acoustic center, through which the radiation of rectangular sound pressure curves is possible on all sides up to frequencies of approx. 3 kHz and with a woofer arranged at the end for use the mirror sound source of the woofer when the housing is mounted on the wall.

• Figur 1 Ein erstes prinzipielles Schaltbild der Erfindung ;
• Figur 2 eine Darstellung von Ersatzschaltbildern ;
• Figur 3a die typische Ortskurve eines festgebremsten Lautsprechers ;
• Figur 4b ein Ersatzschaltbild zur Darstellung der Schwächung eines Originalfeldes durch Wirbelströme ;
• Figur 4 die Darstellung als grundsätzlich bei der Erfindung einzusetzenden Netzwerks ;
• Figur 5 Darstellungen zur Bestimmung der Dimensionierung der einzelnen Elemente des Neutzwerks ;
• Figur 6 ein konkretes Anwendungsbeispiel ;
• Figur 7 eine vereinfachte Ausführungsform der Erfindung ;
• Figur 8 ein konkretes Schaltbild einer Ausführungsform für eine Mitteltonkalotte ;
• Figur 9 ein Blockschaltbild zur Anwendung des Zeitmultiplexprinzips bei der Erfindung ;
• Figur 10 den Ablauf bei einer Schaltung nach Figur 9 ;
• Figur 11 ein ausführliches Blockschaltbild zur Durchführung des Zeitmultiplexvertahrens ;
• Figur 12 Ausgestaltung und Anordnung eines bei der Erfindung eingesetzten Lautsprechers.

The invention is explained below with reference to exemplary embodiments with reference to the drawing. It shows:

• Figure 1 is a first schematic diagram of the invention;
• Figure 2 is an illustration of equivalent circuit diagrams;
• Figure 3a shows the typical locus of a loudspeaker locked in place;
• FIG. 4b shows an equivalent circuit diagram to illustrate the weakening of an original field by eddy currents;
• FIG. 4 shows the representation as a network to be used fundamentally in the invention;
• Figure 5 representations for determining the dimensioning of the individual elements of the Neutzwerk;
• Figure 6 shows a specific application example;
• Figure 7 shows a simplified embodiment of the invention;
• FIG. 8 shows a concrete circuit diagram of an embodiment for a mid-tone dome;
• FIG. 9 shows a block diagram for the application of the time-division multiplex principle in the invention;
• Figure 10 shows the sequence for a circuit according to Figure 9;
• FIG. 11 shows a detailed block diagram for carrying out the time-division multiplexing;
• Figure 12 Design and arrangement of a speaker used in the invention.

The device according to FIG. 1 has an adder 1, a voltage-controlled current source 2 with output amplifier, a loudspeaker 3, a network 4 for simulating the voltage drop at the impedance of the stuck loudspeaker and a feedback amplifier 5. In deviation from the usual technology, the loudspeaker is operated with impressed current, i.e. on a power amplifier with a very high internal resistance, in order to keep the feedback of the voice coil distortion on the current low and on the other hand, in time-division multiplex operation - as explained later - the induced To be able to tap motion voltage unloaded, a voltage which is proportional to the current drawn by the loudspeaker (IL) is subtracted via the simulation network 4 (A ' _F ) and from the separately tapped loudspeaker terminal voltage UL (FIG. 1, FIG. 2a). The “inner” induction voltage Um of the moving membrane, which cannot be measured in any other way, is obtained in a very simple way. The structure and dimensioning of the simulation network result from the following considerations:

FIG. 3a shows the typical locus of a loudspeaker that is braked. Their characteristic course is essentially due to the existence of eddy currents in the conductive core and magnet material in the case of an AC coil. According to Lenz's rule, they weaken the original field increasingly with increasing frequency, whereby the electrical energy is converted into thermal energy together with the hysteresis losses in the conductive material. This fact can be represented in an equivalent circuit diagram according to FIG. 3b.

For low frequencies, R _K and L _{K are} frequency independent. At high frequencies, a different spatial field distribution will occur for each frequency f due to the field displacement (skin effect). The flux linkage with the original field and consequently L _K and R _K become frequency-dependent.

The impedance Z _K can be described with sufficient accuracy by a network according to FIG. 4. 3 to 5 chain links were sufficient for the speakers previously used. Of course, this equivalent circuit diagram does not take into account any non-linearities. Above all, this includes field distortions with large modulation through the hysteresis, the dependence on the respective position of the voice coil due to the inhomogeneous field profile and temperature influence.

The individual elements of the simulation can be determined with little mathematical effort and good accuracy by the system's response to a defined «test signal». For this, z. B. a current jump, at time t = 0, the falling voltage being measured in response. For metrological reasons, however, it is cheaper to use a sequence of legal signals instead of the one-time jump.

Half the period must be chosen so large that the response of the system decays to a negligible final value during this time (Fig. 5a, 5b, 5c). If the voltage response U _K (t) is now divided into n time ranges (n = number of the desired chain links) with 2 base values each, the impedance Z _K can be easily calculated using a permissible approximation method.

For reasons of fast practical feasibility with almost ideal frequency behavior and high accuracy, capacitors are used (FIG. 4). A system A ' _{f is} required which responds to the input voltage U ₁ = L _L - _R with the output voltage U' _f = L _L · Z ' _f . R is the resistance over which the loudspeaker current is measured.

Since the use of capacitors is desired, a duality relationship can be established starting from coils.

A chain connection according to FIG. 4 results for Z _Cf. The equivalent sizes are calculated as follows:

An application example for a 13 cm woofer is shown in Fig. 6. The movement voltage UM induced in the voice coil (FIG. 2) contains, in addition to the distortion-related deviations, above all the portion that is determined by the mechanical resonance system (membrane together with voice coil, suspension) and the coupled housing (FIG. 2b). This can be indicated by an electrical resonant circuit with a constant loss resistance, as long as the frequency-dependent losses due to radiation in the range of the fundamental resonance can be neglected. If one therefore only wants to correct the basic behavior of a loudspeaker in which the influence of the linear mass-spring system dominates by far over the non-linearities, it is sufficient to do this by a network A ' _M (Fig. 7a) of the same quality and resonance frequency (Fig 7b). If the radiation impedance Z _s plays a role, it can acc. Fig. 7c are taken into account. In addition, the electrical compensation of the resistance R required for the current injection brings about an additional improvement in the reproduction quality (FIG. 8). This method is particularly suitable for loudspeakers in the mid-range and high-frequency range, which in themselves show good distortion factor behavior and in which the system is isolated from the influences of the loudspeaker housing by encapsulation. An application example for a mid-tone dome is shown in FIG. 8.

A further development for gaining the movement tension is to operate the loudspeaker in time multiplex as both a transmitter and a receiver. The terminal voltage U _L is keyed to the loudspeaker. The key frequency must be outside the listening area, e.g. B. 38 kHz. The principle is first explained using a voltage source with a purely ohmic internal resistance (FIG. 9). The circuit has a linear gate circuit 11 and a power amplifier 12, a voltage source 13 with ohmic resistor 11 and a low-pass filter 14.

In the switch-off phase, the internal induction voltage can then be sampled in proportion to the current using a small measuring resistor (FIG. 10a). However, since the impedance of the braked loudspeaker itself contains inductances, the stored inductive energy results in a time-increasing decaying interference voltage after switching off the terminal voltage, the effect of which disturbs the measurement of the movement voltage (Fig. 10b).

In principle, the stored reactive energy reduces the crosstalk attenuation between the "transmit and receive channel". One possibility of improving the "crosstalk" is to select the sampling time during the switch-off phase in such a way that the lowest-interference value of the movement voltage is detected (FIG. 10b). This is then regenerated by a downstream low-pass filter. Another option is to take measures which, despite the inductances, cause the transmission current to drop as quickly as possible to "zero" as a result of "countermeasures". This can be done e.g. B. achieve by a differentiator, which is connected upstream of the power amplifier. However, it is better to impress the loudspeaker current directly using a voltage-controlled current source. This also ensures that the membrane can continue to move electrically undamped during the dimensional phase due to the high internal resistance of the power source. The time-division multiplex method can advantageously be combined with the method of simulating the braked impedance. A detailed block diagram, in which FIG. 2 was used, is shown in FIG. 11. There are: adding 1, a linear gate circuit 2, a voltage-controlled current source 3 with output amplifier, a loudspeaker 4, a simulation network 5, an ideal scanner 6, a low pass 7 and a feedback amplifier 8.

Experiments with motion-controlled loudspeaker systems, both according to the time-division multiplex method de as well as with the methods of Fig. 1 and Fig. 7a showed that the speaker systems were able to generate rectangular sound pressure curves with rectangular excitation. A negative feedback in the area of the basic resonance of approx. 14 dB turned out to be good. It should be mentioned that the methods according to FIGS. 1 and 7a enabled stable operation at degrees of negative feedback in the range of 40 dB. The method according to FIG. 1 showed that a woofer in a 51 housing is linearized excellently. Even at sound pressures of approx. 90 dB (1.2 m distance in anechoic chamber, f = 100 Hz), the signals were reproduced clearly and free of natural vibrations of the basic resonance. With the method according to FIG. 7a, the reproduction quality of a midrange dome could be considerably improved with regard to the transmission and dynamic behavior.

The favorable results with regard to the reproduction of rectangular sound pressure curves can of course only be achieved in loudspeakers with several frequency ranges if small loudspeaker systems are used and arranged closely adjacent to one another in the small housing in order to maintain a uniform acoustic center for all frequencies. With regard to the woofer, however, satisfactory behavior can only be achieved in this case by negative feedback. An exemplary embodiment is shown in FIG. 12a. It is a 51 housing, in which the bass system is arranged in the top of the housing, so that the mirror sound source is also used when it is mounted on the wall. This means that the effective sound pressure is doubled if the woofer is at a distance of 2d <Al4 from the wall. In the embodiment according to FIG. 12b with d = 6 cm and a max. Operating frequency of approx. 450 Hz this requirement is sufficiently fulfilled.

Claims (5)

1. Method for negative feedback of movement induced voltage of a moving-coil loudspeaker with respect to the signal voltage with the aid of a network, in which the movement induced voltage is obtained by subtraction between the loudspeaker terminal voltage and an offset voltage, characterized in that the electric signal to be reproduced is supplied to the loudspeaker (3) as an impressed current, that the input of a network (4, A'_F, A'_M) comprising capacitors (C_K1 etc.) resistors (R_o2 etc.) and operators and simulating at its output the voltage drop at the impedance of the firmly braked loudspeaker (3) is supplied with a voltage proportional to the impressed current and the output voltage of the network (4, A'_F, A'_M) is used as the offset voltage for subtraction purposes.

2. Method according to claim 1, characterized in that the signal current impressed on loudspeaker (3) is scanned with a frequency above the audible range and in the disconnection phases the movement induced voltage at the loudspeaker terminals is scanned at appropriate times and regenerated by a series- connected lowpass filter (7, 14) and that the interference voltage brought about by the impedance of the firmly braked loudspeaker (3) and the balancing network are eliminated.

3. Apparatus for negative feedback of the movement induced voltage of a moving-coil loudspeaker with respect to the signal voltage via a network associated with the loudspeaker, the movement induced voltage being obtained by subtraction between the loudspeaker terminal voltage and an offset voltage, particularly for performing the method according to claims 1 or 2, characterized in that the loudspeaker

(3) is connected to a voltage-controlled power supply (2) with a high internal resistance, that at the input of the network (4) comprising capacitors (C_K1 etc.), resistors (R_co etc.) and operators is applied a voltage tapped at the loudspeaker and proportional and to the impressed loudspeaker current and that the output of the network (4) is applied to a subtract amplifier.

4. Apparatus according to claim 3, characterized by a loudspeaker (3) with small low and medium frequency loudspeakers (HT, MT) in order to obtain a small acoustic centre, which permits the emission of rectangular sound pressure variations to all sides up to frequencies of approximately 3 kHz and with a frontally arranged low frequency loudspeaker (TT) for using the mirror sound source of the low frequency loudspeaker in the case of wall fitting of the casing.

Images