GB2196815A - Motional feedback system for loudspeakers - Google Patents

Abstract

The system compensates for non-linearity of the BI (gap flux densityxlength of wire within gap) factor in a moving coil motor system of a loudspeaker Memory (PROM) 62 stores digital words, each of which represents an appropriate correction coefficient to be applied to the speech-coil back-EMF at a specific displacement, to provide an accurate linear velocity signal. The PROM is addressed by a digitised displacement-proportional signal 61, and feeds the digital inputs 65 of a multiplying digital-to-analogue converter (MDAC) 66. The output of the MDAC is the product of the PROM correction coefficients and the scaled back-EMF 14 applied to the Reference input 67 of the MDAC. Addition of the analogue signal 71 to the back-EMF in summing amplifier 76 provides a corrected velocity signal 13. Path 14 to 13 completes a negative-feedback loop enclosing the audio power amplifier and loudspeaker. <IMAGE>

Description

SPECIFICATION Motional feedback system for loudspeakers Introduction Motional feedback systems which have been proposed or implemented fall into two groups.

The first group encompases systems which employ a separate transducer to convert motional information about the dynamic behaviour of the loudspeaker diaphram into a corresponding analogue electrical signal. The second group recovers a motional analogue electrical signal from a moving-coil loudspeaker directly, and without recourse to a separate feedback transducer, by means of a bridge circuit which is balanced in respect of the driving programme signal, such that a signal related to the loudspeaker speech coil velocity is recovered at the output terminals of the bridge.

As is well known, the gain of any negative feedback system is described by the formula A G= 1 +awl where G is the closed loop gain, A is the open loop gain, and ss describes the fraction of an output quantity fedback. In the limit condition of infinite open loop gain, G tends towards a value fi The term ss describes the ultimate reduction of distortion which may be achieved. In a motional feedback system ss is determined by the transfer characteristic of the feedback transducer.It is therefore evident that the feedback transducer transfer characteristic must be at least an order of magnitude more linear than the forward transfer characteristic of the loudspeaker if its motor system is to be effectively linearised and if a substantial reduction of acoustically radiated harmonic and intermodulation distortion is to be achieved. This poses problems with both the aforementioned groups of feedback system. In particular those systems of the said second group suffer the non-linearity of the loudspeaker motor system twice over. First in respect of non-linearity of the driving force applying to the loudspeaker diaphram, secondly in respect of the recovered motion-responsive signal.Apart from that disadvantage, systems of the second group have a very considerable advantage that in principle they can be applied to any loudspeaker, and the performance of the loudspeaker is not in any way compromised by appendages.

It is an object of the present invention to substantially remove said primary disadvantage motional-feedback systems-of the second group which derive the motion-responsive analogue signal to be fedback from the loudspeaker voice-coil; and to disclose a motional feedback system which is capable of yielding very substantial improvements in the levels of distortion products associated with the motor system of the loudspeaker. Said reduction of harmonic distortion is obtained in addition to, and compliments known advantages of motional feedback systems in respect of frequency response and transient or step function performance.

In order to carry the invention into effect, it is necessary to characterise the loudspeaker motor system and to quantify how the factor B1 varies with diaphram displacement. (B1 is the product gap flux density length of wire within the gap field). Simple and inexpensive means for characterising a loudspeaker and accurately determining the product B1 as a function of displacement and the correction required, will be described as an example of how the invention may be carried into effect.

Description Embodiments of the present invention will now be disclosed by way of example, with reference to the accompanying drawings in which: Figure 1 is a schematic representation of a conventional motional feedback system.

Figure 2 is an equivalent circuit of the loudspeaker at the low frequencies at which motional feedback is effective.

Figure 3 is a drawing of an assembly whereby the performance of the loudspeaker system may be characterised.

Figure 4 is a block diagram of an ultrasonic distance measuring system used to characterise the motor system of the loudspeaker.

Figure 5 is an illustrative example representative of a typical graph of relative incremental displacement versus mean diaphram displacement relating to the loudspeaker.

Figure 6 is a block diagram of an embodiment of the invention, whereby linear velocity feedback controls the loudspeaker diaphram motion, and wherein digital correction is applied to the velocity dependant back e.m.f, from the loudspeaker voice coil.

Fig. 1, is a diagram of a conventional motional feedback system of the "constant velocity" type, which maintains a loudspeaker diaphram velocity tending to be independant of frequency. Accordingly, terminal 1 is an input for a velocity signal. (Since sound pressure, desirably made independent of frequency is itself proportional to diaphram accelleration which is the rate of change of velocity; an integrator preceeds terminal 1 accordingly to known art, to maintain a flat frequency re response) .

The velocity signal at terminal 1 is con nected to one input of a power amplifier 2, said amplifier being preferably of the transconductance type, as shown. The current output of amplifier 2 is connected to one terminal 3 of the loudspeaker 4. The other terminal 5 of the loudspeaker is connected via a small resistor 6 (typically 0.5 ohms) to a ground reference point 7 common to the ground referenced input signal.

Fixed resistors 6,8,9,10,11 and the preset variable resistor 12 together with the operational amplifier 113 form a bridge circuit whose output is present at terminal 14. Proper balance of the bridge is achieved by adjustment of the variable resistor R12. Adjustment is conveniently accomplished in practice by temporarily applying a d.c. potential to terminal 1, which causes the loudspeaker diaphram to deflect from its first normal rest position to a second static position of stable equiiibrium between motive force and restoring force provided by the diaphram suspension.

This is a condition of zero velocity, and accordingly variable resistor 12 is adjusted until the bridge output at terminal 14 is zero in this condition of zero velocity.

Terminal 14 is connected via a link 15, to terminal 13 which forms the other input to power amplifier output 2. When link 15 is made, a feedback path is established between the output of the bridge (14) and the one input to the power amplifier. The feedback sense is negative if this is the non-inverting input of the power amplifier. Subsequently it will be shown that the elements of the present invention will be connected between terminals 13 and 14, and in place of link 15.

Because motional feedback systems are only effective over the lower frequency range where the diaphram acts as a rigid piston i.e.

where the diaphram diameter is small compared to a half-wavelength, a simplified equivalent circuit of the loudspeaker can be used for analysis purposes. This equivalent circuit is shown in Fig. 2. In this equivalent circuit the loudspeaker terminals 3, and 5 connect to a series network consisting of a resistor 16, which is the electrical resistance of the loudspeaker voice coil, and a generator 17. The generator e.m.f. has a value Blv, where B is the gap flux density, relating to the air gap in which the loudspeaker voice coil is positioned; 1 is the length of wire of the voice coil winding which interacts with the gap magnetic flux, and v is the velocity of the voice coil. It is seen therefore that the output of the bridge circit at terminal 14 of Fig. 1 is KBlv where K is a constant dependant on the values of resistors in the bridge circuit.

For a loudspeaker of nominal 8ohms impedence having a coil resistance of 6.5 ohms, the constant K has a value of the order 8 x 1 0-2 Unfortunately, due to the construction of a typical moving coil loudspeaker, and engineering compromises in design association with power handling capacity and transducer efficiency, the product B1 is not constant, but varies with diaphram displacement.

Accordingly the e.m.f. derived in Fig. 1 and used therein as feedback, is a function of diaphram velocity, but is not an accurate and true velocity signal. A first step in the implementation of the present invention is the characterisation of the variation of the product B1 as a function of diaphram displacement. Illustratively, means for obtaining this required information will now be described.

Since it is required to characterise the function B1 as a function of displacement of the diaphram, it is essential to have a means to accurately displace the diaphram by a known amount. The principle on which the following method is based is the constant velocity of sound in air at constant temperature and pressure. When therefore an ultrasonic transmission path is established between a first point on the diaphram and a second (moveable) point in space forward (or rearward) of the diaphram, the propagation delay between the two points is constant if a distance between the two points is constant.Following this principle, if a first transmitting ultrasonic transducer is (temporarily) attached at the first point on the loudspeaker diaphram, and a second receiving ultrasonic transducer is mounted e.g. on a rigid beam such that the axes of the transducers are aligned with the acoustic transmission path which is itself aligned with the axis of motion of the loudspeaker diaphram, then the time delay (or phase) between electrical signals associated with the transducers if the distance between them is constant and vice versa.

If there exists a phase detector having first input port connected to the first (transmitting) transducer, and a second input port connected to the second (receiver) transducer, then if the distance between the transducers is constant, the voltage output from the phase detector is a constant. The value of said constant may be zero, by choosing that the distance between the transducers shall be an exact multiple of a quarter-wavelength at the ultrasonic transmission frequency used.

If the volatge output and said phase detector is now fed to an integrator, the output the integrator is representative of the average or mean distance between the ultrasonic transducers. By interconnection of said integrator output voltage to a power amplifier which energises the loudspeaker voice coil, a servo system may be established. For convenience the power amplifier 2 of Fig. 1 may be used.

The servo system acts to maintain a constant distance between the transducers, by moving the loudspeaker diaphram, to which the first (transmitting) ultrasonic transducer is attached.

Details of construction of said mean-position servo system herein above described are given in Fig. 3 which is a cross section of drawing in which myay be identified on the left hand side the moving coil loudspeaker with magnet assembly Fig. 18, frame 19, voice coil 20, with its associated terminals 3 and 5 and the diaphram 21.

Attached to the frame of the loudspeaker (19) are two accurately threaded rods 22, 23, and these are each surrounded by a compression coil spring 24, and 25 respectively, which urge a rigid beam 26 away from the loudspeaker frame. The beam 26 is however restrained by wing-nuts 27, 28.

Be it noted that since threaded rods 22,23 are accurately machined, angular rotation of the wing-nuts translates to a known linear displacement of the beam 26.

Aforementioned first ultrasonic transducer is identified as 29 and has lead-out wires 30.

Aforementioned second ultrasonic transducer is identified as 31, and its associated lead-out wires are 32.

Between the two transducers is a volume of acoustically absorbent material 33 (such as cotton wool, or glass fibres) which shall provide at least 12dB attenuation to the ultrasonic wave so as to diminish the amplitude of standing waves which might otherwise be set up in the space between the transducers to an extent which might cause anomalous behaviour.

A block diagram of the mean position servo system is shown in Fig. 4 in which the components configured within the dotted area 34 are those already described with reference to Fig. 1. Item 35 constitutes summing means for adding together the velocity input signal at terminal 36 and the mean-position control voltage 37.

29 is the aforementioned transmitting ultrasonic transducer, and is for illustrative purposes considered to be of the piezo-electric type. Current sensing resistor 39, and maintaining amplifier 38 co-operate to sustain oscillations at the natural resonant frequency of the transucer 29 which frequency may typically be of the order 40 KHz.

Limiting amplifier 40 provides a first logic signal 41.

The receiving transducer 31 is connected to amplifier 42, the output of which feeds limiting amplifier 43 to provide a second logic signal 44.

Both logic signals are squarewaves. The first logic signal is fed to the reset port of set-reset flip-flop 45, and the second logic signal is fed to the set port (The ports are designated R and S respectively). The voltage waveform at the 0 output of the flip-flop 45 is a rectangular waveform whos duty-cycle and hence average value depends on the relative timing between logic signals 41 and 44.

Thus flip-flop 45 functions as the phase detector of the mean-position servo system.

Remaining elements of the mean position servo-system are an integrator composed of amplifier 46, resistor 47 and capacitor 48. 49 is a voltage reference which determines what the reference mean position shall be.

Amplifier 46 has an open loop gain of the order 100 dB hence the zero frequency loop gain of the mean-position servo system may be in excess of 100dB, depending on the gain amplifier 2 and the effective sensitivity of the phase detector 45.

Resistor 47 may have a value 1 megohm, and capacitor 48 may have a value 1 microfarad such that the unity gain frequency of the integretor is illustratively 0.159 Hz, but is essentially an appreciably lower frequency than the frequency of any velocity signal which it may be desired to apply whilst the mean-position servo system is operative.

The application of a pilot velocity input signal in conjunction with the velocity dependant feedback loop depicted in Fig. 1 (and as Box 34 in Fig. 4) and whilst the mean-position servo is operative, is the basis for the characterisation of B1 versus diaphram displacement.

A more detailed description of the process will now be given. An a.c. signal herein after referred to as the pilot tone, frequency preferably about twice the nominal resonant frequency is applied at terminal 36 in Fig. 4, with an amplitude such as to cause a small vibratory displacement of the loudspeaker diaphram e.g. +0.1 mm.

The preference for a frequency nominally about twice the loudspeaker resonant frequency is that the velocity dependant feedback I6op (operative via link 15 in Fig. 1) still has substantial gain. Also the loop gain is not very greatly influenced by variations of the actual resonant frequency of the loudspeaker.

(The actual resonant frequency of the loudspeaker varies considerably as the diaphram mean position is changed, because of an associated variation of the incremental compliance of the mechanical suspension applying to the diaphram).

The vibratory displacement amplitude is established by measurement of the pilot-tone amplitude at the Q output of the R-s flip-flop 45 in Fig. 4. Means to this end are d.c. blocking capacity 50, band-pass filter 51 tuned to pass the pilot tone frequency, and a.c. digital voltmeter 52.

The mechanical vibratory displacement is determined from a knowledge of the ultrasonic frequency radiated by transducer 29 and hence the acoustic wavelength in air of the ultrasonic emission; coupled with a knowledge of the d.c. supply voltage applied to the flipflap 45. In the preferred case where flip-flop 45 is a CMOS device, the peak-to-peak amplitude of the rectangular waveform of the Q output of the said device is substantially equal to the supply voltage. So, for example, if the ultrasonic emission has a wavelength of 8mm, and the supply voltage to flip-flop 45 is 12 volts, a vibratory displacement of the loudspeaker diapgram of +0.7mum (0.2mm peakto-peak) yields an a.c. voltage at the 0 output of flip-flop 45 of 12x0.2/8 volts=300 mV peak-to-peak. The digital voltmeter may be calibrated to read the true incremental displacement.

During initial setting up of the apparatus, it is convenient to establish an initial condition in which the loudspeaker diaphram is at its natural rest position (i.e. there is zero voltage across the voice coil) and in which the duty cycle of the rectangular waveform at the Q output of flip-flop 45 is approximtely 0.5 This initial condition is obtained by adjustment of the reference voltage 49 in Fig. 4 and by ajdustment of the spacing between the ultasonic transducers 29 and 31 in Fig. 3. Said spacing is readily adjusted by means of the wing-nuts 27, 28 in Fig. 3.

When the initial setting up has been done, the reading displayed by the digital voltmeter 52 in Fig. 4 is noted. The apparatus is now prepared to characterise the product B1 versus diaphram displacement, and the process of characterisation is as follows: Wing nuts 27 and 28 in Fig. 3 are simultaneously rotated in increments (e.g. 1/4 turn) and the digital voltmeter reading (52 in Fig. 4) corresponding to each increment noted. In this way all the required information is obtained.

The results are logged, and normalised relative to the natural rest position of the diaphram. A graphical plot may be assembled and a typical example is given in Fig. 5.

At this stage in the description of method a brief explanation of the operation may be helpful.

Rotation of the wing nuts (27,28) moves the beam in a controlled manner, and the mean position servo system causes the loudspeaker diaphram to track the beam movement, maintaining substantially constant distance between the beam and the loudspeaker diaphram by reason of its high loop gain in excess of 100 dB.

The velocity-dependant feedback loop (established via link 15) also is designed to have high loop gain and therefore maintains a substantially constant back-e.m.f.,(f)Blv at each increment of diaphram displacement. In the (unatainable) ideal design of loudspeaker wherein the product B1 would be a constant, the velocity-dependant feedback loop would succeed in maintaining constant velocity. The digital voltmeter reading would be independant of diaphram displacement, and the graph Fig.

5 would be a horizontal straight line indicative of constant, unity relative incremental displacement. In practice, the graph Fig. 5 gives directly the correction required to the back e.m.f. such as to obtain constant velocity, independent of diaphram displacement. (relative incremental displacement can be shown to be the reciprocal of relative B1 product).

Illustratively, the graph Fig. 5 contains arbitrary sample points marked X and labelled A, B, C, D, E, F respectively. It will be appreciated any number of sample points can be chosen to encompass the intended (safe) maximum excursion of the loudspeaker diaphram (horizontal axis) and that at each sample point the corresponding value of correction (vertical axis) can be represented by a digital word, wherein the number of bits corresponds to a desired accuracy or resolution.Neither the number of samples or the word length need to be very large in practice because first, that which is being digitised is only a correction component applied to the programme analogue signal, and secondly, at small signal levels (small displacements) where quantisation distortions or noise are likely to be subjectively most audible, the correction is zero or near zero, and hence quantisation noise is zero or near zero. At higher programme levels, quantisation effects are masked by the loudness of the programme.

Generally 256 samples each of 8-bit resolution are adequate, and this means that readily available electronic components of low cost can be used. (e.g. analogue-to-digital convertors/digital to analogue convertors).

It will be appreciated that the manually operated characterisation process could easily be automated.

However they are obtained, the multiplicity (e.g. 28=456) of digital correction words are written into a digital memory store e.g. programmable read-only memory (PROM). Preferably that PROM is henceforth paired with the particular loudspeaker which has been characterised, but alternatively, and especially if production variations are small, the PROM may be associated with loudspeakers of that specific type, or with specific production batch of loudspeakers of that type previously characterised.

Having described the object of the invention, and the preliminary process of characterising the loudspeaker (or loudspeaker type) which is to be used, it is now appropriate to describe the final implementation of the invention with reference to the block diagram Fig. 6. As has been peviously herein stated the principal components of the invention are inserted in an otherwise conventional feedback loop, and between terminals 13 and 14 in Fig. 1 (link 15 having been removed). The same terminals may be identified in Fig. 6, wherein at 14 the uncorrected velocity dependant feedback signal enters, and wherein a corrected and true velocity signal is fed out via terminal 13 to complete the velocity feedback loop.

A displacement proportional analogue voltage 53 is obtained either from a separate displacement transducer such as the ultrasonic mean-position servo herein above described at terminal 54 or from an integrator whos input is the true velocity signal; terminal 55. Link 56 is made between 53 and 54 or 53 and 55 as appropriate.

Displacement voltage 53 passes to a sample-and-hold circuit consisting of switch 57, hold capacity or 58, and buffer amplifier 59.

The buffered output of the sample and hold circuit feeds the analogue input terminal of an analogue-to-digital convertor 60. The digital output of the analogue-to-digital convertor (e.g. 8 parallel bits) feeds the address inputs 61, of the programmable read only memory 62.

A timing sequence pulse generator 63 provides repetitive timing pulses to the sample/ hold cricuit, analogue-to-digital convertor, and the write enable terminal 64 of the read only memory.

The digital outputs of the read only memory (e.g. 8 parallel bits) feed the digital inputs of the multiplying digital to analogue convertor 66. The reference input 67 of the digital-toanalogue convertor is the uncorrected velocity dependant signal entering at terminal 14, which is scaled in amplitude by amplifier 68, and associated gain determining resistors 69,70.

The analogue output 71 of the digitai-to- analogue convertor, which is the product of the correction data stored in the PROM 62 and the analogue signal 67, is fed as customary according to known art, to a de-glitcher configuration consisting of switch 72, capacitor 73, and buffer amplifier 74. A further feed from aforementioned sequence generator 63 controls switch 72.

The analogue correction signal present at the output of buffer amplifier 74 feeds the summing junction of amplifier 76 via resistor 75. Resistor 77 couples the uncorrected velocity dependant signal to the summing junction, and 78 is the feedback resistor. Resistors 75 and 77 have values chosen so as to add the analogue correction signal from buffer amplifier 74 to the uncorrected velocity-dependant signal from amplifier 68 in the correct proportion so as to negate the errors in the uncorrected velocity depenant signal to provide a true (corrected) velocity signal at terminal 13.

Since amplifiers 68 and 76 are both inverting amplifier, the path 14 to 13 is non-inverting, and hence the circuit described above, and shown in Fig. 6 may be directly inserted in the velocity feedback path shown in Fig. 1.

The velocity signal at 13 is fed to the integrator consisting of amplifier 79, resistor 81, and capacitor 80. The output of the integrator is the integral of velocity, namely displacement. Said displacement signal is fed optionally via d.c. restoring means 82 which included essentially a diode and resistor, and may then provide a displacementsignal at terminal 55, which is appropriate if no separate displacement transducer is used. The d.c. restorer 82 is desireable if there are substantial even-order non-linearities in the loudspeaker which promote a variation in mean diaphram position proportional to the displacement amplitude.A separate displacement transducer may provide the d.c. component associated with even-order distortion of mechanical origin, but the d.c. restorer 82 is required if the displacement signal is derived by integration of a velocity signal.

Particularly if the loudspeaker is mounted with the voice-coil axis vertical, it may be preferred to incorporate a mean position servo system as herein above described to provide electronic suspension of the diaphram, and prevent mechanical instability whereby the diaphram sags under its own weight over a period of time. (A situation which frequently occurs in modern loudspeakers of high compliance and low resonant frequency).

As refinement, the displacement signal is fed to a window discriminator 83, having reference voltages 84 and 85. The window discriminator has an output when the voltage analogue of displacement exceeds either of the reference voltages 84,85. The window discriminator output couples to muting switch field effect transistor 86 causing it to switch on, whereby the impedance from drain 87 to ground becomes low. Connection of the drain to a convenient point in the programme signal path facilitates muting, and thereby prevents physical damage to the loudspeaker as a result of excessive excursions of the diaphram.

It has been stated herein that the power amplifier 2 shall preferably be of the transconductance type i.e. shall have a high output resistance. This allows a higher velocity feedback loop gain to be achieved than would be the case with the hitherto more usual low output impedance amplifier. Higher loop gain promotes a correspondingly higher performance.

Claims (12)

1. A motional feedback system applied to an electro dynamic loudspeaker, providing a first uncorrected feedback signal, said feedback signal being a function of diaphram velocity. Means for characterising the first derived but uncorrected velocity signal as a function of diaphram displacement to provide characteristic data.

2. A emotional feedback system as in claim 1 wherein said characteristic data is represented by a multiplicity of digital words, said digital words being written into a digital memory store.

3. A motional feedback system as in claims 1 and 2 wherein an anlogue signal representative of the diaphram displacement is digitised, and wherein the digitised displacement signal is used to address the digital memory store of claim 2.

4. A motional feedback system according to the above claims 1 to 4 whereby the digital output read out from the digital memory store (which digital output contains pertinent correction data) is fed to the digital input of a multiplying digital to analogue convertor.

5. A motional feedback system as in claims 1 and 4 wherein the uncorrected feedback signal as in claim 1 forms the analogue reference input to a multiplying digital to analogue convertor, which d.a.c. is also fed at its digital input terminals with the outputs of the digital memory store.

6. A motional feedback system incorporating a digital memory store which is addressed by the digitised displacement signal, and which reads out into a multiplying d.a.c. as in claim 5, wherein the anague output of the m.d.a.c. is added to the uncorrected feedback signal in a proportion such as to provide true and substantially distortion fee velocity information.

7. A motional feedback system as in claim 3 wherein the analogue displacement signal is obtained by integration of the corrected and substantially distortion-free velocity signal.

8. A motional feedback system as in claim 7, containing additionally d.c. restoring means such that the derived displacement signal has a mean value representative of the mean value of the programme-dependant mean diaphram displacement relative to the natural rest position of the diaphram.

9. A motional feedback system as in the preceeding claims wherein the power amplifier driving the loudspeaker is configured as a transconductance amplifier.

10. A motional feedback system incorporating a displacement transducer, the output of which is integrated to provide mean diaphram position information, and wherein said mean position analogue information is compared to a reference voltage, and fed back to stabilise the mean position of the diaphram.

11. A system wherein a window-detector operative in response to the magnitude of a displacement-proportional signal in relation to prescribed voltage limits activates muting means if said voltage limits are reached, such as to protect the loudspeaker from damage.

12. Apparatus containing a motional feedback system as, or substantially as described herein.