GB2488534A

GB2488534A - A system for addressing loudspeaker units in an array by multiplexing

Info

Publication number: GB2488534A
Application number: GB201102921A
Authority: GB
Inventors: David Marston; Graham Thomas
Original assignee: British Broadcasting Corp
Current assignee: British Broadcasting Corp
Priority date: 2011-02-18
Filing date: 2011-02-18
Publication date: 2012-09-05
Also published as: GB201102921D0

Abstract

The system has an input for an audio signal, a multiplexer 38 connected to the input and arranged to produce a multiplexed version of the audio signal for connection to each of a first set of connections of the loudspeaker array 41, and a processor for providing modulation signals for connection to each of a second set of connections of the loudspeaker array. The modulation signals and multiplexed signals are such that, when combined at each loudspeaker unit, they reproduce the audio signal addressed to that loudspeaker unit. The multiplexer could be a time-division multiplexer, and the modulation signals could be clock signals. Alternatively, the multiplexer could be a frequency-division multiplexer.

Description

METHOD AND APPARATUS FOR ADDRESSING AUDIO DEVICES

BACKGROUND OF THE INVENTION

This invention relates to addressing of audio visual components and in particular to addressing signals to loudspeaker units arranged in an array.

Multiple channel loudspeaker arrangements are known, for example the standard home cinema arrangement of two front, two rear and one front centre channel, with each speaker being separately wired from a surround sound processor. Loudspeaker design has allowed individual loudspeakers to become ever smaller and thereby to increase the number of loudspeakers used in a given environment.

SUMMARY OF THE INVENTION

We have appreciated the possibility of providing driving signals to multiple loudspeakers by addressing individual loudspeaker units as a logical array using frequency division or time division multiplexing techniques. In broad terms1 a loudspeaker addressing system comprises an input for receiving an audio signal, a processor arranged to process the received audio signal to produce multiplexed signals for connection to a first set of connections for loudspeaker elements and a second output for providing signals to a second set of connections for the loudspeaker elements, wherein the signals provided to the first and second sets of connections are such that when combined at each loudspeaker element allow individual addressing of the loudspeaker elements.

A first aspect of the invention provides addressing of loudspeaker elements in an array by time division multiplexing. An input is provided for an audio signal connected to a processor and multiplexer arranged so as to produce time division multiplexed versions of the audio signal which are coupled to separate amplifiers for connection to each column of a loudspeaker array. Clock signals are provided for connection to each row of a loudspeaker array. The clock signals and time division multiplexed signals when combined at each loudspeaker unit reproduce the audio signal addressed to that loudspeaker unit.

In a second aspect, a processor and OFDM unit are provided for producing signals for loudspeakers arranged in an array with an input for receiving an audio signal and arranged to produce multiple outputs each comprising the sum of signals for each loudspeaker unit multiplied by a frequency to be addressed to a first set of connections and to produce frequency signals for connection to a second set of connections of loudspeaker units in the array, wherein the signals are arranged such that when they are multiplied together at each loudspeaker unit they reproduce the audio signal addressed to each loudspeaker unit.

A third aspect of the invention provides an arrangement as in the second aspect but with each frequency signal replaced by the square wave representation of the frequency signal. In this aspect, a separate multiplexer is not needed at each loudspeaker unit simplifying the hardware arrangement.

A fourth aspect of the invention provides a new hardware arrangement for an addressable loudspeaker comprising a layer of loudspeaker material having a first set of electrical connections arranged as rows and a second set of electrical connections arranged as columns with loudspeaker units connected at intersections of the rows and columns to thereby provide a matrix of addressable loudspeaker units. The addressable loudspeaker units may be addressed using the signals according to the first, second or third aspects of the invention.

S

BRIEF DESCRIPTION OF THE DRAWINGS

The invention will now be described in more detail by way of example with reference to the drawings, in which: Figure 1: is a representation of a flat panel loudspeaker comprising an array of loudspeaker units according to an embodiment of the invention; Figure 2: shows two rows and columns of a loudspeaker arrangement according to a first aspect of the invention; Figure 3: shows a circuit diagram of the circuitry for addressing the arrangement of Figure 2; Figure 4: shows two rows and columns of a loudspeaker arrangement according to a second aspect of the invention; Figure 5: shows circuitry for producing signals for addressing the arrangement of Figure 4; Figure 6: shows a multiplier and low pass filter forming a demodulator at a loudspeaker unit; Figure 7: shows time domain and frequency domain modulation signals; and Figure 8: shows a class D amplifier arrangement as may be used in a time multiplexed system.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

The embodiments of the present invention comprise a loudspeaker arrangement comprising loudspeaker units arranged in an addressable array, and methods and circuitry for providing signals for addressing row and column connections of the array.

In conventional loudspeaker arrangements, if a large number of loudspeakers are to be driven by different signals, then this would require twice the number of wires as loudspeakers (each loudspeaker requiring two wires). In contrast, the embodiments of the present invention use matrix addressing so that the number of wires required may be twice the square root of the number of loudspeakers.

For example, if 100 loudspeakers are provided, instead of requiring 200 wires they can be driven by 20 wires (this being 2 x the square root of 100).

The reduction in the number of connections or wires required is achieved by either time or frequency multiplexing signals presented on a row of loudspeaker connections which when combined with either a clock or frequency signal on column connections reproduce the desired audio signal at each loudspeaker unit.

In the example of time multiplexing, the multiplexing is achieved by switching the output of the amplifiers used to provide signals to the loudspeaker matrix at a high frequency (above that of the highest audio frequency) so that when an amplifier's output coincides with the corresponding switched column, the corresponding loudspeaker unit receives current. By switching the rows in sequence, the individual loudspeaker units are then individually driven. The loudspeakers may contain low pass filters to remove the high frequency switching which may be additional low pass filters or may be the inherent characteristics of the loudspeaker units.

In order to understand the circuits and method for driving a loudspeaker array, a loudspeaker array embodying the invention will first be described with respect to Figure 1. The preferred embodiment is a flat loudspeaker panel which may be considered to act as "electronic wallpaper in which rows and columns of electrical connections are used instead of separate wires.

The loudspeaker array 10 shown in Figure 1 comprises row electrical connections 2 and column electrical connections 4 arranged in a regular array.

External connections 5 connect to corresponding column connections 4 and external connections 3 connect to corresponding row connections 2. The row and column connections within the array may comprise wires or tracks within corresponding non-conductive layers. At the intersection of row and column connections, individual loudspeaker units may be formed. A loudspeaker unit is used herein to describe a part of the overall loudspeaker arrangement capable of independently emitting an acoustic signal. One such loudspeaker unit 9 is shown at the intersection of column labelled "B" and row labelled "1". Preferably, each loudspeaker unit is formed of an arrangement of layers of the larger panel 10 and is separately moveable by electrostatic, electromagnetic or piezo electric force caused by electrical current between the column electrical connections and row electrical connections. The loudspeaker may in fact comprise continuous deformable layers such that each loudspeaker unit is a logical part of the continuous layers and not physically separate from adjacent loudspeaker units but provides separate acoustic response due to the separate electrical connections. In the example shown, each loudspeaker unit may be defined by a square comprising the overlap of the rectangular electrical row connections and column connections.

As can be seen from the arrangement of Figure 1, each row receives a corresponding signal as does each column. The signal resulting at each loudspeaker element therefore depends upon the manner in which the row and column signals combine as will now be described. For the avoidance of doubt, the embodiment is described in terms of "row" and "column", but this does not imply any particular physical circulation of the device and these terms could equally be "first dimension" and "second dimension" but are used for ease of understanding.

The embodiment shown in Figures 2 and 3 uses time division multiplexing to address each speaker unit. Figure 2 shows two rows of connections 2 and two columns of connections 4 providing signals to four loudspeaker units 19.

Columns are labelled "A", "B" and rows are labelled "1", "2". Signals are provided to the columns by amplifiers 6 and to the rows by switches 7. The switches allow one row at a time to provide a return path for current from the amplifiers, for example by being connected to the ground (zero volts) part of the amplifier. Each loudspeaker element 19 is connected on one side to the column electrical connection 4 and on the other side to the row electrical connection 2. The instantaneous voltage at each loudspeaker element 19 will depend upon the signals provided at the rows and columns. More generally, there may be N amplifiers arranged to drive signals to N columns creating an N x M matrix. The signals provided to the amplifiers are time sliced using time division multiplexing at a rate such that each row may be addressed in turn with a portion of the signal on each column. Accordingly, each loudspeaker unit 19, in a given column, receives a portion of the signal provided to that column every M period of time such that each loudspeaker is effectively driven one at a time. Due to the slower responsiveness of the loudspeakers and inability of the user to perceive the time difference, the loudspeaker units may all appear to be playing at the same time.

The rate of switching is arranged to be sufficiently high that the time switching itself does not provide audible ar-tefacts. For example, the switching may be 2 x the highest audio frequency x the number of rows, so that all rows may be addressed in turn without the switching being audible to a user. In the example of a 10 x 10 matrix of the loudspeaker units, the switching frequency will be 2 x 20,0010=400KHz.

Time division multiplexing is known to the skilled person but for ease of understanding, consider column A of the arrangement of Figure 2. The signal provided to amplifier 6 of column A comprises a time multiplexed version of the audio signal to be addressed to loudspeaker unit at column A row 1, and at column A row 2. For a first switching period the signal provided to the column comprises the signal for Al, for a second switching period the signal for loudspeaker unit A2 and more generally an Nth period the signal comprises the signal for loudspeaker unit An.

The demultiplexing occurs by switching the switches 7 at a corresponding switching frequency. For a first period, switch for row 1 is closed and row 2 open so that the portion of the signal is asserted to the loudspeaker unit Al. For a second period of time, switch for row 1 is open and for row 2 is closed so that the signal is asserted to loudspeaker unit A2 and so on. More generally, for the Nth period the Nth switch is closed and all other switches are open so that the loudspeaker unit on row N receives the corresponding portion of the time division multiplex signal on column A. Provided that the switching to all rows is completed before the first loudspeaker unit has changed physical position, each loudspeaker unit receives an approximation to the original audio wave form,

S

which may be smoothly reproduced due to the inherent characteristics of the loudspeaker unit itself.

The arrangement using switches on each row as shown in Fig. 2 will work so long as each loudspeaker unit can only pass current in one direction: if current can flow in either direction then when the switch on row I is closed and the switch on row 2 (and any other rows) is open, then speakers on the row(s) with open switches will still receive some drive signal, as a current could pass from amplifier A, through the speaker at A2, then through the speaker at B2 and back into amplifier B. A first approach to avoiding the leakage' of signals to non-selected rows is to place a diode in series with each loudspeaker so that each one can only pass current in the same direction, e.g. from the column wire to the row. With traditional loudspeaker structures, this would halve the peak-to-peak motion range of the moving element, but by taking account of this in the speaker design, for example by biasing the moving element to sit near one end of its travel when no voltage was applied, this problem can be reduced.

A second approach to avoiding the leakage' problem is to place an electronic switch, such as a field-effect transistor (FET) between each loudspeaker and the signal ground, which is controlled by the row selection signal. Thus each speaker not on the selected row is individually isolated and no leakage will occur. The disadvantage of this is the need for an active component with each speaker.

This approach relies on the loudspeaker units having a low-pass filter characteristic, either by virtue of the inertia of the drive unit itself, or by the incorporation of circuitry to implement a low-pass filter, similar to that which might be found in a cross over' filter in a conventional loudspeaker having several drive units that handle different frequency ranges. The signal received by each loudspeaker, after the row switching has removed the signal during the times that other rows are being driven, consists of a series of impulses, which at simplest could resemble samples of the continuous audio waveform with periodic gaps corresponding to the times that other rows are being driven. The effect on the spectrum of the signal that these gaps cause is to create additional frequency bands at multiples of the switching frequency, leaving the low frequency components of the signal reduced in amplitude but otherwise unchanged. The low-pass characteristic of the loudspeaker unit removes these higher frequency spectra, so that the speaker only reproduces the required low frequencies.

The method and circuitry for generating the time division multiplexed signals will now be described with reference to Figure 3. An input line 21 receives an audio signal which may be a single channel audio signal, or any known multiple channel audio signal. A processor 20 processes the received audio signal(s) to produce the separate signals Al, A2, B2, B2 to be addressed to the separate loudspeaker units. These separate audio signals may, in fact, be identical versions of the audio signal or may separate left and right stereo channels or separate different frequency ranges. The precise nature of the processing will depend upon the desired addressing of the loudspeaker arrangement, but the key point is that separate individually addressable signals may be provided. The separate addressable signals are provided to a multiplexer 22 on lines 23. The multiplexer 22 is itself driven by a clock signal 27 which is also used to drive the switches 7 (Figure 2) so that the clock source for multiplexing the signals and for demultiplexing the signals is essentially the same, avoiding any phase or frequency shift problems, although it may be advantageous to make the row switching signals activate slightly after the multiplexer switches, to allow time for the amplifier outputs to settle at their new values. The multiplexer 22 takes the signals destined for each column of the array and time multiplexes them to produce output signals on lines 25, for example output signals shown as Al, A2 being the time multiplex signal from combing signals Al and A2. Amplifiers 30 may be provided as part of the driver electronics instead of as part of the loudspeaker arrangement of Figure 2 or may be omitted in favour of the amplifiers in the loudspeaker arrangement, or any combination of some amplification within the driver electronics and some within the loudspeaker arrangement.

The amplifiers used in this arrangement may be any conventional audio amplifier.

However, the design is particularly well-suited to Class 0 amplifiers (whose output switches between high and low, to produce a series of pulses that when filtered with a low-pass filter produce the desired audio signal). Each speaker

I

already needs to include (either explicitly or implicitly) a low-pass filter, as explained above so this can be taken advantage of to allow the use of a Class D amplifier without needing additional filter components. Class D amplifiers are generally smaller and more power-efficient than other types. However, it can be advantageous if the operation of the Class D amplifier takes account of the frequency and phasing of the row selection signal, so that the output signal, when chopped into time slices for each row and low-pass filtered, gives the correct output for each speaker. For example, if the Class D amplifier uses a pulse-width modulation approach that involves comparing the analogue audio signal to a triangle wave, there should be a whole number of triangle wave periods in the row switching period. An arrangement using class D amplifiers showing how the triangle wave used to produce pulse-width modulated output would be synchronous with the row selection signals is shown in Figure 8.

A second embodiment using frequency division multiplexing is shown in Figures 4 and 5. The arrangement is similar to the first embodiment with loudspeaker units 31 arranged in a matrix of rows and columns, each unit having a connection to one of the column connections 4 and a connection to one of the row connections 2. Each loudspeaker unit 31 will comprise some additional circuitry in comparison to the first embodiment due to the different nature of frequency division multiplexing. In this embodiment, local demodulator circuits for each loudspeaker unit 31 are used as well as local amplifiers. However, the amplifiers may be low powered as they only need to drive a small loudspeaker unit, in comparison to the amplifiers needed in the first embodiments which drive multiple loudspeaker units. The use of local amplifiers require additional power connections.

The OFOM addressing of the loudspeaker units operates as follows. Each row 4 (two rows are shown A, B) is provided with a continuous frequency, Fl, F2... FN.

The frequency spacing is arranged such that frequencies are orthogonal. The columns are supplied with a multiplexed version of the signals intended for each loudspeaker unit multiplexed with the corresponding frequencies Fl, F2. As an example, column 1 is provided with a signal being the audio signal for Al multiplexed with frequency Fl and the signal for loudspeaker unit A2 multiplexed with frequency F2. At each loudspeaker unit, multipliers multiply the row signal by the corresponding column frequency signal. As a result, each loudspeaker unit will extract the corresponding audio signal intended for that unit only, and all other units will not reproduce the signal for that unit due to the orthogonality of the multiplexing.

Consider element Al, for example, which will receive frequency signal Fl on row A and the signal shown by equation I below.

Al signal = Al *sin(Fl)+iot.2*sin(F2) where Alsignal (as shown as 40 in Figure 6) is the signal seen at element Al before the local demodulation.

The element Al includes a multiplier (38 in Figure 6) so that the resulting signal provided to the acoustic part of loudspeaker unit Al is as shown in equation 2.

Al demothlow-pass-filter (Fl *Al signal)} low-pass-filter (Fl *(Al *sin(Fl)+A2*sin(F2)} tAl where Aldemod is the demodulated signal, which after low-pass filtering (39 in Figure 6), should recover the Al original signal (41 in Figure 6).

The method and circuit for producing the multiplexed signals will now be described with reference to Figure 5. An input line 31 receives an audio signal which, as with the first embodiment, may be a single audio signal or a multi channel signal of some sort. A processor 30 receives this and produces signals Al, A2, BI, 62 and so on for addressing to each corresponding loudspeaker unit.

These are provided on line 33 to an OFDM processor 32 which also receives a frequency signal F on line 37. The frequency signal F provided to the OFDM multiplexer is also provided to the frequency generator for the demultiplexing frequencies provided to the columns of the addressable array so as to avoid any frequency or phase shift problems. OFDM multiplexer provides the signals for each row on output lines 35.

A third embodiment provides a variation of the embodiment using frequency division multiplexing of Figure 4 and 5 using a square wave signal instead of a sine wave frequency signal for the multiplexing and demultiplexing.

The sine-waves Fl, F2,FN can be further modulated into digital signals using pulse density modulation. By oversampling these sine-waves it is possible to generate square-waves that still contain the sine-wave's fundamental frequency.

Instead of a multiplier being required at each speaker, the speaker can be driven directly (or via a passive low-pass filter) eliminating the need for local demodulation circuitry.

The square-wave that modulates the sine-waves must be a sufficiently high-frequency to be able to generate a clear sine-wave if it was low-pass filtered. For example if FN was a sine-wave of 384kHz then using 16x oversampling, the square-wave would be 6.144MHz.

Indeed, in general, the row control signals may be written R(1), R(2), ... R(N) where there are N rows These are treated as binary signals here, with values 0 and I, although a version using continuous signals (e.g. sinusoidal) would also be possible. These are time-varying signals, with possible waveforms including: * A short pulse on R(1), followed by a short pulse on R(2), etc * A square wave of frequency Fl on row 1, F2 on row 2 such that F2 = 2xFl, F23xF1, etc. The characteristics of the signals are such that the integral of R(a) x R(b) over a period I equal to the lowest frequency of any of the signals (which will be the time to pulse each row in the first example, or the modulation frequency of the lowest row in the second example) is zero for a not equal to b, and 1 for ab.

Another way of saying this is that the signals R are orthogonal to each other.

The amplifier signals driving the columns may be written C(1), C(2), The process of modulation for the amplifier driving column c can be written C(c) = Sum (over r=1..N) [R(r) x input(r,c)] where input(r,c) is the input signal that we wish to reproduce at the loudspeaker at (r,c).

I

The process of demodulation (or demultiplexing) at the intersection r,c works as follows: Output(r,c) = low...pass_filter(R(r) x C(c)) The multiplication operation becomes a simple switching process if R is a binary signal, and can thus be implemented by an arrangement such as a FET (Field Effect Transistor) in series with the loudspeaker, which is connected between the amplifier driving this column and the ground (zero volt) signal return path.

As long as the low-pass filter has the effect of integrating the signals over a period T (or multiples thereof, or a long period compared to T), and the input signals destined for each speaker do not contain any energy above frequencies that the low-pass filter will pass, then the output signal for each speaker will be equal to the input. This follows from the orthogonality condition, since only the signal that has been multiplied by the same signal as was used when modulating it will remain; all other signals will be zero because the product of two different row modulation functions when averaged over a period of time is zero.

The multiplication process is best thought of as demultiplexing (or separation in the time domain) when only one signal R(r) is nonzero at any one time. If the R(r) signals resemble periodic waveforms such as square waves, then the process can be thought of as demodulation (or separation in the frequency domain). If the signals R(r) were sine waves, then the spectrum of the signals driving a column would resemble a series of signals modulated around a set of carrier frequencies, with the individual signals non-overlapping in the frequency domain. If square waves are used, the signals targeted for each row are not cleanly distributed around the carrier frequencies (as each will effectively be sent on a lot of carriers whose spectrum is that of the square wave used in modulation), but the demodulation process (using the same square wave) will untangle these.

I

Figure 6 shows a multiplier 38, low pass filter 39 and speaker element 41 demonstrating the demodulation arrangement, the process for which is described above. Figure 7 shows an example input audio signal and the process of modulation and demodulation.

The embodiments described thereby provide addressable arrays of loudspeaker units or elements. The arrays may be easily extended by simply adding extra columns and providing corresponding extra multiplexed signals to the extra columns. For a configurable device, therefore, the dimension likely to be extended should be the multiplexed signals rather than the row control signal for simplicity.

The processor and multiplexer circuitry may of course be combined into a single unit and have been described separately for ease of understanding only. The drive circuitry needs to have knowledge of the number of rows and columns being addressed to provide the appropriate signals and a configurable input may be provided for this. The Loudspeaker array being driven may provide such a configuration signal to allow a form of "plug and play" in which the drive circuitry automatically provides the appropriate output signals depending upon the array to which it is connected.

Various further modifications are possible including introducing various delays in the signals in the time divisional multiplex arrangement so as to ensure appropriate phasing of the clocked rows. In the frequency of division multiplexing arrangement, phase multiplexing may also be introduced so as to halve the number of frequencies needed and to introduce I and 0 multiplexing to effectively provide two columns for each frequency.

Claims

SCLAIMS1. A system for addressing of loudspeaker units in an array by multiplexing, comprising: an input for an audio signal; a multiplexer connected to the input and arranged to produce a multiplexed version of the audio signal for connection to each of a first set of connections of the loudspeaker array; a processor for providing modulation signals for connection to each of a second set of connections of the loudspeaker array; and wherein the modulation signals and multiplexed signals are such that when combined at each loudspeaker unit reproduce the audio signal addressed to that loudspeaker unit.
2. A system according to claim 1, wherein multiplexer is a time division multiplexer arranged to a produce time division multiplexed version of the audio signal.
3. A system according to claim 2, wherein the modulation signals are clock signals.
4. A system according to claim 2 or 3, wherein each time division multiplexed version of the audio signal comprises time multiplexed audio signals to be addressed to each speaker unit connected to a corresponding one of the first set of connections.
5. A system according to claim 4, wherein the frequency of time division multiplexing is greater than the number of second connections multiplied by the highest audible frequency of the loudspeaker units.
6. A system according to claim 1, wherein multiplexer is a frequency division multiplexer arranged to produce a frequency division multiplexed version of the audio signal.
7. A system according to claim 6, wherein the modulation signals are frequency signals.
8. A system according to claim 7, wherein each frequency division multiplexed version of the audio signal comprises frequency multiplexed audio signals to be addressed to each speaker unit connected to a corresponding one of the first set of connections.
9. A system according to any of claims I to 8, wherein the product of any two modulation signals, averaged over a period of time, is zero for modulation signals applied to different connections, and nonzero for any modulation signal multiplied by itself.
10. A system according to claim 1, wherein the modulation signals consist of a short pulse, with the pulse on different connections occurring at different times.
11. A system according to claim 1, wherein the modulation signals consist of square waves of different frequencies.
12. A system according to claim 1, wherein the modulation signals consist of sine waves of different frequencies.
13. A loudspeaker arrangement comprising the system of any preceding claim and an array of loudspeaker units, each loudspeaker unit having a connection to one of the first connections and a connection to one of the second connections of the array.
14. A system according to claim 13, comprising means for providing a demodulation process at each loudspeaker unit by switching the modulated audio signals on and off at each loudspeaker in accordance with one signal from the second set of modulation signals.
15. A loudspeaker arrangement according to claim 13, wherein each loudspeaker unit includes a multiplier arranged to combine by multiplying together the signals received on the first and second connections.
16. A loudspeaker arrangement according to claim 13, wherein the frequency signals comprise square wave signals and loudspeaker unit includes a switch arranged to combine the signals received on the first and second connections.
17. A loudspeaker arrangement comprising a layer of loudspeaker material having a first set of electrical connections arranged as rows and a second set of electrical connections arranged as columns with loudspeaker units connected at intersections of the rows and columns to thereby provide a matrix of addressable loudspeaker units.
18. A loudspeaker arrangement according to any of claims 9 to 17, the array of loudspeaker units being extendible by addition of further rows or columns.