FR2790353A1 - Loudspeaker system with several electrostatic transducers, e.g. for hi-fi system, provides direct DAC action by weighting of individual transducers and drive system - Google Patents

Abstract

A controller (8) provides a separate drive signal for each bit position, based on comparison of current and previous samples, which is fed to a corresponding transducer. The transducers comprise a number of metal plates across which a high voltage is applied. A magnetic field is established perpendicular to the electric field using Helmholtz coils (36, 37), creating a movement of ionized air according to Lorenz's law. The resultant pressure waves, being in proportion to bit weighting, effect an inherent digital-to-analogue conversion process in the surrounding air.

Description

The present invention relates to an audio device making it possible to restore

directly from recorded sounds

  in digital format where the amplitude is coded on N bits.

  Usually, we use to obtain this restitution

  tion a chain with four links, which are, in order: the digital to analog electronic converter, the power amplifier, the selective two or three-way filter which distributes the signal between the different speakers, these are loading the emission of sound. The device according to the invention proposes to eliminate all these intermediaries by directly operating the digital to analog conversion in the air. Its principle is as follows: With each new sample of the digital signal N bits supplied to the input arriving at frequency fixed, an electronic control system subtracts this new sample with the previous one kept in memory, and in fact determines the value in N bits of the variation of this signal, therefore of the pressure variation to be produced by displacement of air to operate the

sound reproduction.

  Each bit of this quantified variation in N bits is associated with a loudspeaker such as the volume of air displaced by this loudspeaker at each instant of its activation.

  either of the form of the product of an amplification coefficient

  cation K, identical for the N loudspeakers, and of a power of two whose exponent corresponds to the rank of the associated bit in the digit N bits of the variation, the loudspeaker being activated when switching to the high state of this bit; the direction of the air movement carried out being determined by the sign of the variation and the intensity of this air movement by an identical reference voltage for the N loudspeakers, which determines the amplification coefficient K. A potentiometer will modify this reference voltage and therefore

to adjust the volume of the whole.

  The sub-assemblies which make it possible to create the displacements-

  air elements must be able to restore an indefinite number of times the same variation immediately, an always probable case, which is impossible for a diaphragm loudspeaker which would then be at the end of its travel,

  can return to its rest position without re-emitting an inverse variation. To do this, each of these sub-assemblies, which will be called -5 by analogy loudspeaker, consists of two lon-

  gues parallel perforated metal plates, placed in a tube, brought to a sufficiently high potential difference to ionize the air circulating through the holes between the plates, then creating a weak electric current between these plates, perpendicular to the axis of the tube. The tube

  is also placed between two Helmholtz coils which allows

  are trying to create a magnetic field directed perpendicular to

  direction relative to both the axis of the tube and the direction of the current between the metal plates. The direction and intensity of this magnetic field depend on the control current flowing in the coils. As soon as this magnetic field is established , the ionized air is directed en masse along the axis of the tube, due to Lorenz's law; the direction of this movement and its intensity being directly linked to that of the previous magnetic field. outside, the rest of the tube being arranged

in a box.

  The reference voltage, or its opposite, is transmitted to the coils, when the loudspeaker must be activated in one direction or the other, the high electric voltage between the

  plates remaining fixed, obtained for example by a food

  switching mode, rectified high-voltage AC

  or by a montage derived from the doubler Schenkel.

  The N preceding speakers must therefore all have a different efficiency for the same control voltage,

  depending on their geometry and the value of the coils.

  According to the particular embodiments:

  -the front of the device, which carries the vents of the top-

  speakers, can be curved in an arc,

  center the emissions of the different speakers at one point.

  - The loudspeakers can be controlled, for each sample for which the loudspeaker is to be activated, by a period of a positive40 or negative triangular signal depending on the direction of the air movement to be carried out, 3- the signal being zero at the start and at the end of the command, of amplitude equal in absolute value to the reference voltage, of period twice that of reception of the samples of the digital input signal, two successive commands to the same loudspeaker, shifted in principle by a sampling period, therefore only by half a period of the control signal described above, added together, so as to eliminate the traces of

the original sample.

  -Each speaker can have an independent box

  as well as its own high voltage generator.

  -Each speaker can be replaced by an assembly of several speakers (4) controlled simultaneously by the same local system (6), the assembly having an identical performance to that of the speaker substituted for the same

control voltage.

  -Each speaker can be controlled by a voltage

  proportional to the reference voltage.

  - We can use less than N loudspeakers for the conversion of a digital signal into N bits. Indeed, the difference between two successive samples is unlikely to be worth more than two power Nl, we can therefore be satisfied with N -1 loudspeakers at the output of the N bit subtractor. On this principle, the number of loudspeakers can be reduced to P (P <N), by connecting only

  the P least significant outputs of the N bit subtractor.

  the accompanying drawings illustrate the invention and its constituents:

  Plate 1 in Figure 1 shows a side view of the device

  tif according to the invention, with a partial section showing

  the various internal components of this device.

  Plate 1 in Figure 2 shows a front view of this device. Plate 2 figure 3 shows the electronic diagram of the

main control system.

  Plate 2 in Figure 4 shows the different signals and

  supply wires which form the main bus.

  Plate 3 in Figure 5 shows the electronic diagram of one of the N local control systems, each system

  room controlling only one speaker.

  - 4 Plate 3 in figure 6 gives the electronic diagram of an example of high voltage supply for the device. Plate 4 in Figure 7 shows a side section of a loudspeaker. Plate 4 in Figure 8 shows a longitudinal section

a speaker.

  Plate 4 in Figure 9 shows a perspective view

a speaker.

  Plate 4 in Figure 10 shows a loudspeaker with elements

multiple seen from the front.

  Plate 5 Figure 11 shows an example of an 8-bit binary subtractor for 8-bit input binary numbers

unsigned.

  Plate 6 in Figure 12 shows a local control system intended to smooth the original sampled signal, using a triangular control with a period double that of the original sampling for the loudspeakers. Plate 7 in Figure 13 shows the electronic diagram of a triangular signal generator synchronized by an external control, necessary for the smoothing system of the

figure 12.

  Plate 7, Figure 14, gives a diagram of the signals associated with the triangular smoothing system.

  double the period of the original sample.

  With reference to these drawings, the device comprises, in the overall views of FIGS. 1 and 2; (certain references have only been mentioned for one of the N

  repetitive elements of the device so as not to over-

  load the drawings, but they are valid for all.): a box (1), the front face (2) of which is pierced by N vents (3) corresponding to the N speakers (4); by

  convenience, we took N = 8 for the drawing of Figures 1 and 2.

  Each loudspeaker has an independent box (5) in order to avoid possible interactions during their simultaneous operation; it also has a local control system (6), connected to the main data and power bus (7 ), which itself comes from the general electronic control system (8). The device also has a high-voltage generator (9), high -5- which is distributed to the speakers by another network of wires (10 ), as well as a low voltage power supply for the entire device (14). A rotary button (11) placed at the bottom of the front face (2) drives a potentiometer ((19) in FIG. 3) which provides the reference voltage (V1) and thus serves to control the

  volume of the sound emitted by the device.

  The device also includes a socket for the electrical supply of the assembly (12), as well as a

  input socket for digital signal (13).

  The general control system (8), described in FIG. 3, consists of the following elements: a clock (17), clocked at the sampling frequency of the signal supplied to the input (13) of the device, which can be synchronized by an external signal. We assume that the digital signal is transmitted on a parallel link

  N bits, otherwise, in the case of a serial link, we inter-

  will set a shift register with serial input and N-bit parallel output, coupled to a clock N times faster than the main clock (17). At each rising edge of the clock (17), the sample in N bits present at the input (13) is memorized by the buffer (15) (consisting for example of flip-flops D controlled by a rising edge.) While the value previously present at the output of this buffer (15) is memorized by the buffer

(16) identical to the previous one.

  These two successive stored values of the input signal are applied to the N bit synchronous subtractor (18), the type of subtractor will vary depending on whether the subtraction is done in signed or unsigned binary arithmetic,

  depending on the nature of the N bits samples supplied.

  An example of an 8-bit subtractor for unsigned binary digits can be seen in Figure 11, which uses the complement method to !; (the letters aO, al, a2, a3 and bO, bl, b2, b3 marked on the two adders (27) and

  (28) represent the entries for the numbers to be added

  ner, the result being presented on the outputs E0, El, E2, E3) To subtract the two numbers N bits from buffers (15) and (16), we take the complement to 1 of the number N bits present at the output of buffer (16), which is added - 6 - with the number N bits present at the output of the buffer (15), thanks to two adders 74LS83 (27) and (28) connected in cascade; if the carry-out output of the adder (28) is high (retained), the result is positive and to obtain the right value, we add 1 to the result of the addition (thanks to the carry-in entry of l 'adder (27)), on the other hand, if the carry-out output of the adder (28) is at 0, the result is negative and to obtain the correct

  value, we take the complement to 1 of the result of i'addi-

  tion, via the exclusive OR present at the output. The sign bit (21) is the complement of the carry-out output of the adder (28), so it will be worth 0 for a positive result and 1 for a result. negative of the

  subtraction performed between the two samples.

  (27) and (28). This device is made synchronous by the output buffer (29), controlled by a rising edge of the clock (17), the other circuits being asynchronous, which, given the working frequencies of the system for example

44 kHz is not a problem.

  On the next rising edge of the clock (17), the result of the subtraction is available, in the form of an absolute value in N bits on the output bus (20) and of a sign bit (21), the sign bit set to 0 for

  a positive variation and 1 for a negative variation.

  Each bit of the output bus (20) of the subtractor goes

  order, depending on its condition, the activation or not of the built-in

  speaker associated with it, the sign bit (21)

  determining the direction of the air movement carried out.

  To determine the amplitude of this movement, a third system is used: a rotary button (11) makes it possible to adjust a positive reference voltage (Vl) between the zero value and that of the low voltage supply,

thanks to a potentiometer (19).

  This reference voltage (Vi) associated with the bus N output bits (20) of the subtractor (18), the sign bit (21) and the three low voltage supply wires (positive pole + E, negative -E, and midpoint OV) intended for

  local control systems form the main bus (7).

This is illustrated in Figure 4.

  These signals are distributed to the communication centers.

  local signals (6) (figure 5) as follows: the reference voltage (V1), the sign bit (21) and the three low voltage supply wires are made available to all local centers (6) .However, each local center5 (6) will only be connected to a single bit of the N-bit bus (20) of the subtractor (18). Thus, the rank P bit of this bus will be associated with the local system which controls the loudspeaker (4) such that the volume of air displaced by the latter at each instant of its activation is in the form of the product of a

  amplification coefficient K, identical for N high-

  speakers, and with a power of two whose exponent corresponds to the rank of the bit in question, ie p,

p being between 0 and N-1.

  FIG. 5 shows the diagram of a local control system (6), it will be noted that the references of the resistors (R1, R2, R3) are given to show the relative values existing between all the resistors of the diagram. The operating mode of a local control system (6) is as follows: The operational amplifier (24) mounted as an inverter makes it possible to obtain the opposite (-Vi) of the reference voltage (Vl). voltages (V1) and (-Vi) arrive at the input of a reversing adder (26) built around an operational amplifier. The transistors (T1) and (T2) are used to ground when they are at the on state, the inputs which are not selected, on the other hand when (T1) or (T2) is in the blocked state, it allows the control voltage corresponding to it to pass to the inverting adder ((Vi) for T2 or (-Vi) for T1), and the assembly functions as an analog switch which transmits to the loudspeaker either (V1) to effect a movement of air in a positive direction (T1 blocked.), Or (Vi) for air displacement

  in a negative direction (T2 blocked.) and neither if the speaker (4) is not selected.

  The control of the two transistors (T1) and (T2) is done via the NAND (22) for (T1) and the NAND (23) for (T2) which decode the information provided by the sign bit (21) and by the output bus bit of the subtractor (20) associated with the loudspeaker (4),

local system (6) controls.

  The bit associated with the loudspeaker must be in the high state for the command to be possible, therefore, if the sign bit (21) is in the low state, (T1) will be blocked because the NAND (22) will go low, otherwise if the sign bit (21) is high, it will be (T2) which will be blocked because the

  NAND (23) will then go low.

  The inverter (25) is used to adjust the output level of the

  NAND (22) for controlling the transistor (T1).

  The selected control voltage ((V1) (-V1) or 0V)

  is transmitted to the transistors (T3) and (T4) mounted in push-

  pull, the threshold compensation of which has not been detailed, which serves as a power amplifier and which is connected to the helmholtz coils (36) and (37) of the loudspeaker (4)

  a description of which is given in FIGS. 7 to 9

  The loudspeakers (4) are all built on the same principle: They are composed of a more or less long rigid tube (30) made of insulating material; on the diagrams this tube is

  of rectangular section, but it is not an obligation.

  The front and the bottom of the tube are open, and two opposite faces have slots (31) to allow

  air from the surrounding box circulates.

  Two metal plates (32) and (33) are arranged inside the tube (30), along the faces which comprise the slots (31). Between these plates is established a constant electric field sufficient to ionize the air

contained in the tube (30).

  These plates (32) and (33) can carry metal fins (34) in order to increase the cross-section of the tube and the volume of air contained in it while maintaining a reduced gap between the two plates (32) and (33) from the point of view of the electric field that is created between them. The number of fins given on the diagram is indicative. Just like the tube (30), the plates (32) and (33) are pierced with long openings (35) in order to also allow the air circulation of the box up to 'in the center of the tube where the electric field prevails. On the two opposite faces which are not drilled, there are arranged, outside the tube (30), two - 9 flat coils (36) and (37), electrically connected in

  series or parallel to the output push-pull (T3 and T4) of the associated local control system (6), in order to create a uniform magnetic field between them, i.e. inside the tube (30). field, which is established at art-

  the tube when the reference voltage (V1) or its opposite is transmitted to the coils (36) and (37), is directed, due to the arrangement of the coils described above, perpendicular both to the direction of the established electric field between the plates (32) and (33) and at

the main axis of the tube (30).

  Therefore, the ionized air, sensitive to the magnetic field because charged, will move in a direction perpendicular to both that of the electric field and that of the magnetic field, so by construction it will move in the direction of the axis of the The direction of this displacement will depend on the direction of the magnetic field established in the tube (30), therefore indeed on the sign of the tension

  transmitted to the coils (36) and (37) by the communication system

  local demand (6) ((V1) positive, (-V1) negative.).

  Likewise, the intensity of the displacement of ionized air operated will depend on the intensity of this magnetic field, therefore

  of the absolute value of the reference voltage (V1).

  There are many parameters, such as the spacing of the plates (32) and (33), the size of the tube (30) or the coils (36) and (37) for example, which make it possible to vary the efficiency of the loudspeakers. loudspeakers (4) in order to obtain the characteristics specific to N loudspeakers

required by this device.

  However, you can limit the number of speakers (4)

  different to produce by creating high-

  speakers as shown in Figure 10. On this one, four speakers were assembled in line, each using a flat coil (36) or (37) from its neighbor, and by interconnecting the five coils in series or parallel, a loudspeaker is obtained which, for a given voltage command, performs an air displacement four times greater than that effected, for this same control voltage, the

  basic speaker model that makes up the whole.

  In figure 6 we can see the electrical diagram

- 10 -

  of the high voltage generator (9): The transformer (38) is used in order to galvanically isolate the rest of the circuit from the mains, and the negative pole of the high voltage generator output will be connected to the midpoint of the 'low voltage power supply (14). The voltage raising device itself is a conventional assembly derived from the Schenkel voltage doubler with diodes and capacitors, usable insofar as the current passing between all of the plates, due to the ionized air, remains low. current must be higher, we can provide each speaker with its own Schenrikel voltage booster. A possible value of the voltage delivered by the high voltage generator required

  to ionize the air could be of the order of 2 to 8kV.

  Symmetrical low voltage power supply (14) :( pole

  positive + E, mass, negative pole -E) does not require an explanation

  specific cations; it is conventionally formed by a transformer, a diode bridge and voltage filtering and stabilization circuits. The operational amplifiers will all be supplied between the positive pole and the negative pole, unlike logic circuits ,

  powered only between the positive pole and ground.

  (If it is a ttl circuit, a separate + 5v supply will be added). A possible value of the voltage delivered by this supply could be + 15v, O0v, -15v. (Filtered for the control of the coils, but regulated for the supply of the logic circuits and operational amplifiers, regulated for example in each local circuit

(6).

  A variant of the local control circuit (6) shown in Figure 12 allows the sound to be smoothed in order to eliminate the traces of the original sampling. This circuit requires the presence of the clock signal (17 ) which must then be transported by bus

main control (7).

  The frequency of the clock signal (17) is divided by two by the flip-flop T (39) .On the output of this flip-flop there are obtained two signals in phase opposition of double period to that of the sampling. These two signals ( Q) and

- 11 -

  (Q) each control and synchronize a positive triangular signal generator (40) and (41) whose operation is described below, which supply the triangular signals (Si) and (S2), of amplitude equal to the reference voltage (Vl), which will therefore be in phase opposition, because phase shifted by a sampling period and a period double that of the sampling. The inverters (24) and (42) allow to obtain the opposites (-SI ) and (-S2) of these signals (Si) and (S2). The signals (Si) and (-Si) will serve as control voltages for every second input sample, (S2) and (-S2) serving for the others, since the triangular control voltage has a period twice that of the samples and therefore monopolizes a control circuit over two periods of the clock

  (17), the circuits therefore all having to be doubled.

  The flip-flops (43) and (44) controlled by the output signals in phase opposition of the flip-flop (39) serve

  to memorize, over a period double of the sampling period

  the value of the bit associated with the loudspeaker (4) of the

  local system in question which changes with each sample.

  It is the same for the flip-flops (45) and (46) which memorize the values of the sign bit (21). The values thus memorized allow the whole of a period of the control signal (Si) and (-Si ) or (S2) and (-S2) to be available at the input of the adder (26), which in addition to its previous role of analog switch,

  is also the sum of two successive orders.

  As shown in the voltage diagrams of FIG. 14, showing the signals (Si) and (S2), the signals of the outputs (Q) and (Q) of the flip-flop T (39) and the clock signal (17 ), the triangular signal generator (40) is synchronized so that its output (Si) goes to 0 during a rising edge of (Q) which corresponds to the transmission of a new sample to this system of

ordered.

  The same is true for the triangular signal generator.

  area (41) during a rising edge of (Q).

  Figure 13 shows the diagram of the triangular signal generator (40). The reference voltage (Vl) and its opposite (-Vl), obtained thanks to the inverter (47), are

- 12 -

  applied to the input of an operational amplifier mounted as an inverting adder (48). The ratio of resistors that makes up the latter is such that, when the integrated transistor which forms the 'Discharge' output of the circuit (50) (Timer 555) is conducting putting the input of the adder (48) connected to the reference voltage (V1) to ground, the inverting adder (48) presents at output a voltage equal (Vl), while when this same transistor is blocked ( integrated transistor of the 'Discharge' output in the blocked state), the adder (48) has a voltage at output (-Vi). This voltage is applied to an integrator (49) whose capacitor Cl will be chosen as slightly lower than the result of its definition formula: Cl = T1 / R4 o R4 is the resistance of the integrator (49) and Ti half the period of the triangular signal to obtain. Under these conditions, when, by the output of l adder (48), 1a voltage (-V1) (integrated transistor of the 'Discharge' output of the circuit (50) in the blocked state) is applied to the integrator (also inverter) (49), the capacitor C1, initially discharged, charges at the reference voltage (V1) on a duration equal to the quasi-half-period of the triangular signal to be obtained, i.e. Tl. The integrator output is connected to the 'threshold' input of a Timer 555 circuit (50); like the pin 'controls Voltage 'of this same circuit is connected to the voltage of

  reference (V1), The 'Discharge' output of this circuit, previously

  in the blocked state (which made it possible to obtain (-Vi) at the output of the adder (48)), turns on and the output of the adder (48) then applies the reference voltage (V1) at the input of the integrator (49) .C1 then discharges up to 0V over a period of almost

  half period remaining of the triangular signal to be obtained, T1.

  Cl cannot charge negatively due to the low threshold Schottky diode Dl. This gives a single triangular slot. Only a short negative synchronization run, of sufficient amplitude, applied to

  the 'trigger' input of the circuit (50) via

  diary of the differentiator R5, C2 allows this input to reach its tilting value, equal to (Vl) / 2, (at rest, the resistor R5 maintains this input at a

- 13 -

  potential close to + E), which changes the discharge output of the circuit (50) to the blocked state, and reapplies a voltage equal to (-V1) to the input of the integrator (49) whose capacitor is The triangular signal generator5 (40) then emits a new triangular slot and is interrupted until the next synchronization signal. This signal comes from the output (Q) of the flip-flop T (39) which is reverse thanks to the inverter (51), signal which can be shaped by a transistor. As this signal has a period being precisely worth 2 * T1 and that the triangular signal generator (40) must be ready (Ci again discharged ) during each rising edge of (Q),

  (which at the output of the inverter (51) gives a falling edge

  dant), we are obliged to take Cl slightly lower than

  its theoretical value, as indicated above, by choosing

  sant the value which corresponds to the minimum distortion of the triangular output signal (Si). The diode D2 is used to protect the 'Trigger' input during throws

  positive that occur when the sync signal

  nisation (Q) returns to low state. On power-up, the initial state does not matter and the operation stabilizes after two out of three synchronization signals, as long as Cl remains below its value theoretical, also taking into account the dispersion of

component characteristics.

  The use of a triangular control signal with a period double that of the sampling allows

  thus a better fusion of the sounds emitted by the diff-

  different loudspeakers (4), due to the gradual rise

  of the command signal; and, in addition, two triangular commands

  successive steps of the same sign and in opposite phase add up to form a voltage plateau (of value (Vi) or (- Vi)) and we then find the same

  result only for a rectangular order.

  The device according to the invention is mainly

  intended for the field of high fidelity.

- 14 -

Claims (7)

  1) Audio device that directly reproduces the sound to   from a digital signal obtained by coding the amplifier   sound study on N bits with fixed sampling frequency, characterized in that it comprises a general control system (8) which calculates the variation of the digital input signal by subtracting thanks to an N bit subtractor (18) each new sample with the one which preceded it and then performs the pressure variation associated with this variation of the input signal coded in N bits by a set of N speakers (4) each being fixedly associated with a single bit of this variation by through a local control system (6) and performing a pressure variation implemented in the loudspeaker (4) by ionizing at rest the air located between two metal plates (32) and (33) carried to a potential difference raised by a voltage source (9), then, when the loudspeaker (4) is effectively controlled, operated by the transition to the active state of the bit associated with it, in   establishing a magnetic field of perpendicular direction   cular to the previous electric field thanks to two helmholtz coils (36) and (37) which for this purpose receive from the local control system (6) a voltage equal to either a reference voltage (Vl) regulated by a potentiometer (19) , or at its opposite, according to the logical state of a sign bit (21) from the subtractor (18) indicating the sign of the variation of the input signal; the main axis of the loudspeaker (4) being arranged along the direction perpendicular to the two fields where the combination of the latter causes the ionized air to flow, and where the opening of the loudspeaker opens towards the outside by a vent (3) in the facade (2), the rest being immersed in a box (1); the efficiency of the N loudspeakers is adjusted so that for the same voltage supplied to the coils (36) and ( 37), each loudspeaker operates a pressure variation which is the product of an identical variable for all by the power of two associated with the bit which controls it, in - 15 -   the number N bits of the variation of the input signal.   2) Device according to claim 1 characterized in that the plates (32) and (33) of the loudspeakers carry slots (35) and that the insulating tube (30) which surrounds these plates (32) and (33) carries also slots (31). 3) Device according to claim 1 or claim 2 characterized in that each speaker (4) can be replaced by an assembly of several speakers (4) controlled simultaneously by the same   local center (6), the assembly having an identi   than that of the speaker it replaces, for the same control voltage.   4) Device according to any one of claims   previous characterized in that less than N speakers are used to reproduce an input signal coded in N bits, using only the least significant outputs of the N bit subtractor (18) corresponding to the number of speakers given.   ) Device according to any one of the claims   previous characterized in that the local control system (6) uses, for the control of the loudspeaker associated with it, and this for each sample o this activation is necessary, a period of a triangular signal, of zero value at the start and end of the command, of positive or negative sign depending on the sign bit (21), of amplitude in absolute value equal to the reference voltage (Vl) and of period double to that of reception of the samples of the digital signal input,   two successive commands added together.   6) Device according to any one of claims   previous characterized in that each speaker   has an independent box (5).   7) Device according to any one of claims   previous characterized in that the front face (2)   of the device is curved in an arc.   8) Device according to any one of claims   previous characterized in that the metal plates (32) and (33) of the speakers (4) carry fins of metal (34).   9) Device according to any one of claims - 16 -   previous characterized in that each speaker (4) can be controlled by a voltage proportional to the reference voltage (V1).