GB2240689A

GB2240689A - Three speaker stereo system

Info

Publication number: GB2240689A
Application number: GB9024206A
Authority: GB
Inventors: Egils Ranga
Original assignee: Individual
Current assignee: Individual
Priority date: 1989-11-07
Filing date: 1990-11-07
Publication date: 1991-08-07
Anticipated expiration: 2010-11-07
Also published as: GB2240689B; GB8925156D0; GB9024206D0

Abstract

A stereo sound system is provided using three speakers A, B and C. The speakers can be powered by a specially designed three channel amplifier or by a specially adapted two channel amplifier. The invention also includes speakers for use with the system. The speakers may be different types, arranged in triangular configuration, and located at different levels. <IMAGE>

Description

STEREO SOUND SYSTENI The invention relates to stereo sound systems.

In sound reproduction systems, the illusion of sonic reality is improved with increasing numbers of perceived directions. Hence the advent of two channel stereophonic recording and reproduction was a great improvement on mono, as it could provide the illusion of a spread of sound sources between. and even slightly beyond, two loudspeakers, rather from just a single point. The word stereo comes from the Greek 'stereos' meaning solid or three dimensional. but is commonly misunderstood as meaning two channel.

As two channels have the potential for a circle of information in their relative phase and amplitude relationships, it should be possible to extend the soundfield round and behind the listener. So-called quadraphonic sound systems. launched in the early 1970's were relatively expensive. requiring the use of a matrix decoder to process specially produced recordings and feed four loudspeakers via a four channel amplifier.

These known commercial systems proposed horizontal loudspeaker layouts in which the loudspeakers surrounded the central listening area. the rear loudspeakers being elevated in some systems.

I have now deduced that a circle of sound can be obtained from just three loudspeakers in an appropriate layout, and accordingly my invention provides a sound reproduction system having three loudspeakers.

It can be shown that, compared to two speaker stereo, a three speaker system can produce a sound field that is wider and with the extra dimension of height, the effect being likened to a sound picture. This increase in potential sound directions enables more acoustic detail to be revealed from any stereo source, resulting in improved realism.

The three loudspeakers may be arranged in a triangular layout.

It is preferred that right hand and left hand speakers are arranged in an elevated position, facing a listening position, there being a third central speaker arranged lower down, also facing the listening position.

The speakers may be inclined to the vertical.

The upper speakers may be inclined downwardly, the lower speaker being inclined upwardly.

There may be right hand, left hand and central speakers, the central speaker being larger than the right and left hand speakers.

Alternatively, there may be right hand and left hand speakers, the central speaker incorporating a sub-woofer.

At least one speaker may have the woofer and/or tweeter circuit in which inductors and capacitors are each arranged in a star configuration.

There may be at least one speaker having a woofer and/or tweeter circuit in which inductors are arranged in a star configuration and capacitors are arranged in a delta configuration.

The system may include means for removing or reducing any volume imbalance.

The means for removing or reducing any volume imbalance may operate by reducing the stereo separation of an amplifier.

For example, electric resistance means may be connected across the channels of the amplifier.

Capacitance means may be connected in series with the resistance means to bring about the degree of phase compensation.

The reduction of the stereo separation of the amplifier may modify the amplifier output signal voltages to L'=O .789L+O .211R R/=o .7 89R+o . 211L where L and R represent the unmodified left and right channel signal voltages and L' and R' represent the modified voltages.

The three speakers may be electrically connected in a star arrangement.

The speaker signal voltages may have the following relative proportions: Va /=1.366 366L-O . 366R Vb'=1 . 366R-0 . 366L Ve /=L+R where Va', Vb' and Vc' represent the three speaker signal voltages and L and R represent the unmodified left and right channel signals.

A three channel amplifier may be used.

The three channel amplifier may have three amplifier stages, one for each speaker. signals to the three amplifier stages being derived from left and right hand signals via pre-amplifiers, inverters. bridge networks and attenuators.

However, I have further deduced that the three loudspeakers can be operated by a two channel amplifier suitably constructed or adapted for the purpose.

Accordingly, the invention also includes an amplifier constructed or adapted to operate three loudspeakers.

The invention includes a three channel amplifier for use in powering three loudspeakers of the sound reproduction system.

The three channel amplifier may have three amplifier stages, one for each speaker, signals to the amplifier stages being derived from right and left hand signals via pre-amplifiers, inverters, bridge networks and attenuators.

The invention includes a two channel amplifier constructed or adapted to provide output connections to three loudspeakers.

The two channel amplifier may be provided with means to reduce the stereo separation of the amplifier.

Impedance means may be connected across the channels of the amplifier.

An amplifier according to the invention may be provided with switching means operable to convert the amplifier into a conventional two channel amplifier.

The invention includes speakers constructed or adapted for use with a sound reproduction system or an amplifier according to the invention.

By way of example. specific embodiments of the invention will now be described with reference to the accompanying drawings, in which : Figure l shows a basic circuit connection of an embodiment of the invention; Figure 2 shows a basic physical layout of the system of Figure 1; Figure 3 is a view illustrating a drawback of the basic system: Figure 4 is a view similar to 3 but showing the effect of an improved arrangement.

Figure 5 illustrates sound positioning in two speaker stereo; Figure 6 illustrates the vectorial representation of encoded signals; Figure 7a and 7b relate to the encoding of two channel signals; Figures 8 and 9 show equivalent circuits which can be used to demonstrate the theory of the invention, Figure 8 relating to a mono signal and Figure 9 relating to an anti-phase signal; Figure 10 shows a resistive network for use in removing or reducing volume imbalance; Figure 11 is a view similar to Figure 10, but including more detail; Figure 12 is a view illustrating the effect of the circuitry of Figure 11; Figure 13 illustrates a further improvement to the three speaker system of the invention: Figure 14 shows a three channel power amplifier suitable for use with the invention: : Figures 15a to 16b demonstrate various speaker configurations which assist in an understanding of the invention; Figures 17a and 17b relate to a comparison of mono signal vectors; Figures 18a, 1 8b and 1 8c illustrate a preferred three speaker layout according to the invention; Figures 19a and 19b show a further speaker layout; Figure 20 shows the wiring of a typical high fidelity loudspeaker; Figures 2la and 21b illustrate bi-wiring arrangements of a typical high fidelity loudspeaker; Figures 22a and 22b illustrate arrangements for woofer and tweeter circuits of a further embodiment of the invention: Figures 23a and 23b are views similar to 22a and 22b, but showing a deltawired capacitor arrangement; and Figure 24 shows an embodiment of the invention incorporating a phase compensating network.

A simple basic arrangement according to the invention will firstly be considered.

Figure 1 shows three loudspeakers, A, B and C electrically connected to a two channel amplifier as shown in Figure 1. L represents the left channel output signal voltage and R represents the right channel output signal voltage.

Physically, the speakers are arranged in a substantially vertical plane, as illustrated in the elevational view of Figure 2, with left hand speaker A and right hand speaker B elevated, the centre speaker C being positioned lower down.

Assuming that the speakers are identical or have the same impedance, it can be shown that the speaker signal voltages can be given as :

Based on the power distribution of the speaker signals it can be shown that the corresponding sound locations should be contained within a circle bounded by the triangle formed by the three loudspeakers, as shown in Figure 2.

The horizontal position is determined by the amplitudes of L and R, whereas the vertical position within the available height increases with the phase angle between L and R. This is similar to the effect in an acoustic environment where sounds reflected off walls and ceilings are of random phase and incoherent whereas direct sounds are heard lower down.

As anti-phase signals would be mainly reproduced through speakers A and B, these are correctly positioned higher than speaker C.

This basic arrangement has a drawback in that it can be shown, for the speaker signal voltages given previously that the overall volume balance is nonlinear. The non-linearity can vary by a factor of up to 3, for anti-phase signals.

Furthermore, as a result, the directional effect is substantially incorrect with the lower sound locations being expanded and the upper sound locations being compressed. This is illustrated in Figure 3 where the single channel signals L only and R only are shown displaced upwards from their normal side positions.

I have appreciated that this drawback can be overcome or reduced by the rather surprising step of reducing the stereo separation of an amplifier forming part of the system. For normal two speaker stereo this would be a degradatory step, which would never normally be taken, but I have appreciated that for the proposed three speaker system, it produces improved effects.

The stereo separation reduction is easily implemented at a pre-amplifier stage, where signal levels are small, by cross feeding the left and right channels at a suitable stage through a resistor of an appropriate value.

It can be shown that in order to reduce the power level of anti-phase signals of equal amplitude to one third, the amplifier output signal voltages should be modified to : L/=0.789L+0.211R Rl=O . 789R+O .211L The corresponding speaker signal voltages can now be given as : Va '=0.456L-0.122R Vb'=o . 456R-O. 122L Vc'=O . 333L+0 . 333R It can be shown that three speakers responding to voltages having the above relative proportions will produce a linear overall volume balance and correct directional effect in a triangular layout.

Although it is preferred that a two channel amplifier be used, as described in more detail below, the invention includes an embodiment in which a three channel amplifier is used, for example as shown in Figure 14. In this arrangement each speaker A, B and C is driven by its own power amplifier 10, 11 and 12, each power amplifier being fed a signal proportional to one of the above values. These signals are derived from the pre-amplifiers 13 and 14 bv inverters 15 and 16, bridge networks 17, and attenuators 18 and 19. appropriately arranged.

Some of the effects of the modified system are indicated in Figure 4.

Although in practice the overall effect heard is dependent on the particular source that is being played, with some recordings it is possible to hear sounds which seem to emanate from areas outside the triangle formed by the speakers, particularly to the left and right, as shown in Figure 4.

The reason for these effects is not completely understood, but it is believed that it is related to recording studio techniques and speaker signal phase angles.

Sound waves and the human hearing mechanism are complex, whereas circuit and matrix theory are relatively simple.

To use a visual analogy, two speaker stereo is like looking at a circle, edge on. It has a width and depth, but no height. The three speaker system of the invention turns the circle to face the listener, to reveal a multitude of possible directions which make up a sound picture. As the sounds associated with depth are moved upwards, the result is a three dimensional sound stage in accordance with the true Greek definition of stereo.

To further assist in an understanding of the invention there will now follow a discussion of some of the considerations involved in signal encoding using two channels.

I shall firstly consider a conventional two channel sound reproduction system in which a sound apparently moves between two loudspeakers.

If the sound becomes neither louder nor quieter. the combined power output from the two loudspeakers is constant. The apparent position of the sound is determined by the ratio in which the power is shared between the two loudspeakers.

Because power is proportional to the square of voltage. it follows that L2 + R2 is constant. where L and R are the left and right channel amplifier output signal voltages.

Figure 5 shows the apparent position of the sound 9 related to the amplitudes of the left (L) and right (R) signal voltages, which in this case will be assumed to be in phase.

L and R are the amplified versions of encoded signals 1 and r. In additional to the amplitude variation described, the phase angle between I and r may also vary.

as shown vectorially in Figure 6. These phase and amplitude variations can be plotted on the surface of a sphere as shown in Figures 7a and 7b.

Due the symmetry and anomaly of leading or lagging phase angles the encoded relationships can be considered in the two dimensions shown in the elevational view, Figure 7a. In a three speaker soundfield, the horizontal position is therefore determined by the amplitudes of 1 and r whereas the vertical position within the available height increases with the phase angle between 1 and r.

It is also of assistance to consider equivalent circuits of a three speaker system.

If one considers the circuit diagram of Figure 1, responding to a mono signal. from which L and R are therefore equal and in phase, such that L = R = V. say. if the impedance of each speaker is p, the centre speaker can be considered as two impedances of value 2p in parallel, as shown in Figure 8. The equivalent load impedance for each channel is therefore 3p.

If the phase angle between the two output voltages is increased from 0 to 180 . the signals will then be in anti-phase such that L = V and R = -V. The voltage at the common junction of the three speakers is therefore zero and the central speaker C produces no output. As shown in Figure 9, voltages of magnitude V appear across speaker A and speaker B in anti-phase. The equivalent load impedance for each channel is therefore p.

It follows that the volume of the mono signal is therefore one third that of the anti-phase condition or the two speaker connection. To achieve a linear volume balance, the level of anti-phase signals should be correspondingly reduced and this can be achieved by reducing the stereo separation of the amplifier, as already mentioned.

One method of utilising a two channel amplifier. and reducing the stereo separation. is shown schematically in Figure 10. A proportion x of the single channel signal is fed to the other channel at low level via a resistor ns.

The modified amplifier output signal voltages can now be given as /=(1 -x)L+xR where 0 s x s R /=(1 -x)R+xL and the corresponding speaker signal voltages as

where L and R are the amplified versions of encoded signals 1 and r.

The mono signal reproduced by the centre speaker is of course unaffected.

It is required to determine the value of x which will reduce the volume of anti-phase signals to one third. Having found x, the relative value of the crossfeeding resistor can also then be found.

Consider the anti-phase signals L = V and R = -V.

Then

Because power is proportional to the square of voltage, the volume of antiphase signals will be reduced to one third when

The corresponding signal voltages are

Proof of linear overall volume balance of the above speaker signal voltages is confirmed by :

which is proportional to L2+R2 The resistive network of Figure 10 determines the value of x in accordance with Ohm's Law such that 4ns+s)=(1 -x)s

Therefore

Substituting

gives

Therefore a linear overall volume balance is achieved when n = 2.732.

The required relative values of network resistors are shown in Figure 11, which shows the complete schematic diagram.

The amplifier can easily be reverted to two speaker operation by short circuiting the speaker C amplifier terminals with a link or switch, and open circuiting the cross-feeding resistor with a series connected switch.

It is of further assistance to compare the direction of sound locations which correspond to modified and unmodified signals.

Consider the circuit of Figure 1 responding to the unmodified amplifier signal voltages L and R. As previously stated, the corresponding speaker signal voltages can be given as :

Now consider the amplifier signal relationships for which there is no output from speaker B i.e. Vb = 0 when L = 2R and Va = Vc As the same signal passes through speaker A and speaker C, the sound is located midway between them at the 'heard' position marked h in Figure 12.

However, the signal relationship L = 2R corresponds to 'intended' position i. The lower soundstage is therefore expanded if the amplifier signals are unmodified.

The directional effect would be correct if the condition Vb = 0 corresponded to the signal relationship:

This in fact is the case for the modified speaker signals Va'= 0 . 4 56L-O.122R Vb/=O . 456R-O . 122L Vc'-O 333L+O.3331? as

It can be shown that the above speaker signal voltages will also produce a correct directional effect for all other encoded positions.

A linear overall volume balance with correct directional effect in a circular soundfield will be produced by the circuit and resistive network shown in Figure 11.

Auditioned results convey the impression of a wider and three dimensional sound field with stable front stage images.

It should be noted that throughout this description and analysis, the following symbols are used

SYMBOL DESCRIPTION UNIT A Left speaker in three speaker system B Right speaker in three speaker system C Centre speaker in three speaker system CT Capacitor in tweeter farads circuit Cw Capacitor in woofer farads circuit Cx Capacitor in phase farads compensating network h Heard position of a sound i Intended position of a sound j 90" operator 1 Left channel encoded volts signal voltage L Unmodified left channel volts output signal voltage or reference L' Modified left channel I volts output signal voltage LT Inductor in tweeter henries circuit L. Inductor in woofer henries circuit n Numerical multiplier operating on s P Speaker impedance ohms

r Right channel encoded volts signal voltage R Unmodified right volts channel output signal voltage or reference R' Modified right channel volts output signal voltage s Resistance in cross- ohms feeding circuit of two channel amplifier T Tweeter transducer V Voltage at output of two volts channel amplifier Va Unmodified left speaker volts signal voltage Va' Modified left speaker volts signal voltage Vb Unmodified right volts speaker signal voltage IVb' Modified right speaker volts signal voltage Vc Unmodified centre volts speaker signal voltage Vc' Modified centre speaker volts signal voltage (=Vc) w Angular frequency radians/second W Woofer transducer x Proportion of crossfed real or complex signal Angle subtended by two degrees stereo speakers I will now discuss in more detail some considerations of the practical application of my three speaker stereo sound system.

I will consider firstly speaker positioning.

A three speaker system introduces the extra dimension of height to the layout considerations, which cannot be contemplated with a two speaker system. In positioning three speakers, the essential requirements are that they are equidistant from and facing an optimum position and also that the angle there subtended between centre speaker C and each outer speaker A,B should be approximately 60".

These conditions ensure that for in-phase signals, the inter-speaker images are acceptably stable and relatively correctly positioned.

The subtended angle between speakers A and B is not so critical and in theory it may be varied between two extremes. The first is where the three speakers are in a horizontal layout as shown in Figures 16a and 16b. which is used to make comparisons with two speaker stereo. Here the lateral angular width between speakers A and B is maximised at 1200, whereas the vertical angle between speaker C and the line joining speakers A and B is zero.

The other extreme is where speakers A and B are adjacent in an elevated position above speaker C as shown in Figures 19a and 19b. Here the vertical angle is maximised at 60 but at the expense of the lateral angle which is now zero.

Consequently, the soundfield has no width and this arrangement is therefore not recommended, and is only shown for illustration.

An equilateral triangular speaker layout as shown in Figures 2, 3 and 4 provided a theoretical soundfield that is circular and thus simplifies directional considerations. However the locus of in-phase signals is too curved as the side locations are somewhat elevated compared to the centre image. Also the width of the soundfield is less than in two speaker stereo.

The layout of three speakers should therefore be biased towards the horizontal layout of Figures 16a and 16b, which can be considered as a starting point. It can be seen that if speakers A and B are now elevated slightly, a vertical angle is introduced which initially has little effect in reducing the lateral angle between speakers A and B. The sum of the vertical angle and the lateral angle can therefore exceed 1200, whilst maintaining the subtended speaker angles of 60".

The sum is maximised at 139.8 when the vertical angle equals 33.5 and the lateral angle equals 106.3 , as shown in Figures 18a, 18b and 18c. In the plan view, the lateral angle appears as 110 due to speakers A and B being elevated. A small reduction of horizontal angular width therefore enables a much larger vertical angle to be introduced.

It can be seen that when the optimum listening position, measured vertically, is mid-way between the lower and the upper speakers, the plane of the speakers is inclined at an angle of approximately 45". This would seem to maximise the combination of vertical and horizontal performance and confirm that the subtended speaker angle of 60 is a correct choice.

The arrangement of Figures 18a, 1 8b and 1 8c can therefore be considered as optimal. It has the advantages of the horizontal three speaker layout, plus height, which reveals a multitude of further potential sound directions which make up a sound picture. Furthermore, the locus of in-phase signal images is still perceived as level.

Other speaker layouts are also possible but the auditioned results will inevitably be biased towards particular aspects of the system performance, whilst compromising others. For example, a horizontal layout may be considered more practical, or the speakers may be positioned closer together.

In domestic situations the two upper speakers may appear obtrusive if they are large, whereas the lower centre speaker will tend to blend in with items of furniture. The use of smaller speakers may therefore be aesthetically preferred.

However, small speakers do not provide a good low bass frequency response.

One way of overcoming this limitation is to provide a larger centre speaker.

This achieves a good bass response whilst maintaining the use of small upper speakers. The larger lower speaker could be designed to match the characteristics of the other two speakers for their frequency range and also to have an extended response to lower frequencies.

However, an alternative solution involves the addition of a sub-woofer.

These units are commercially available in various forms according to different system design approaches. They may be of mono or stereo types and may be powered by their own amplifiers or driven by the same amplifier as the satellite speakers.

The sub-woofer is chosen to match the sensivity of the satellite speakers, so that low bass frequencies are neither exaggerated nor subdued. The crossover frequency of the sub-woofer may be determined by a tuned port and/or its electrical crossover network. This is chosen to match the natural roll-off frequency of the satellite speakers, which may be at around 80 Hz or less.

As low frequencies have pressure characteristics of long wavelengths, compared with the distance between our ears, they carry little direction information and so the sub-woofer unit can be placed almost anywhere in the listening room, the whole of the sound frequency spectrum seeming to originate from the satellite speakers. However, it is preferred that the sub-woofer is placed centrally between the satellite speakers.

With a three speaker system, the sensitivity of the sub-woofer should be one third (-4.8dB) that of the three speakers, because the equivalent amplifier load impedance for a mono signal is three times that of each speaker. For example, if the small speakers have a sensitivity of 89dB for one watt at one metre, the subwoofer should match a sensitivity of 84.2 dB.

If the sub-woofer is powered by the same amplifier as the satellite speakers, it is connected to the left and right channel amplifier speaker terminals via low-pass filters (LP).

Figure 13 shows an arrangement applied to a three speaker system. In this case an additional high-pass filter element (HP) is required to complement the impedance with signal variation of the three speaker system.

The addition of a mono sub-woofer to the three speaker system lends itself to the combination with the centre speaker in one unit.

The advantages of the three speaker system according to the invention may be summarised as follows 1. Improved Realism (i) The extra dimension of height is added to the stereo sound field to produce the illusion of a three dimensional sound stage.

(ii) The resulting improved clarity and acoustic separation of individual sounds produces a greater sense of realism, the speakers seeming to disappear.

(iii) Artistic communication is improved resulting in greater enjoyment by the listener.

2. The use of a centre speaker produces stable front stage images.

3. The improved results can be achieved cheaply and cost effectively.

4. The system has total stereo and mono compatibility, left/right symmetry, a linear volume balance, and correct directional effect.

The advantages may be better understood by consideration of the limitations of two speaker stereo compared with the three speaker system of the invention.

Consider a two speaker stereo sound reproduction system as shown in Figure 15b in which the speakers are symmetrically positioned with respect to a listener, where they subtend an angle of .

If the speakers are fed with the same signal such that signal voltages L and R of equal amplitude and in phase, the apparent location of the sound heard by a forward facing listener would be mid-way between the two speakers. However, as the subtended angle + is increased towards 1800 as shown in Figure 15c, the central image becomes more unstable at low and mid-range frequencies, due to the opposing component of the speaker signals. High frequencies do not suffer to the same extent due to the attenuation properties of the human head.

Conversely, the central image becomes more stable and precise as the speakers are moved closer together until they are side by side as shown in Figure ISa.

The above considerations of inter-speaker signals exhibit different properties when the signals are in anti-phase. For example, if the speakers were positioned side by side as shown in Figure ISa, the asymmetric motion of the L and R speaker cones would have a cancelling effect, which is particularly noticeable at low frequencies, where the wavelengths are long, compared to the speaker dimensions.

This loss-of-bass phenomenum would also occur with in-phase signals if one of the speakers were incorrectly connected, and can be used as a check against this condition.

Assuming that the L and R speaker signals are in anti-phase and that the speakers are correctly connected but now positioned at each side of the listener as shown in Figure 15c, the asymmetric motion of the speaker cones would appear to have a sympathetic, rather than a cancelling effect. The behaviour is somewhat more complex however, and it is worthwhile to consider this case in more detail : The anti-phase signals leaving each speaker would reach each correspondingly nearest ear also in anti-phase, albeit at reduced level due to the distance travelled. At high frequencies the listeners head has a shielding effect which prevents interaction of the two sound waves in that area.At lower frequencies this shielding property is not so effective, but interestingly at a frequency of about lkHz the width of the head corresponds to half a wavelength so that the two sound waves are mutually in anti-phase.

At much lower frequencies, the shielding effect of the head is only slight and the two anti-phase sound waves, now with long wavelengths, would have a cancelling effect, although not to the same extent as when the two speakers are in front of the listener. The sympathetic motion of the speaker cones for anti-phase signals would only have a reinforcing effect when the half wavelength is much longer than the speaker spacing. As this would typically require the frequency to be less than 20 Hz and therefore below audibility, the reinforcing effect does not occur. In any case, low bass frequencies are usually associated with in-phase signals.

From all the foregoing, it would seem to follow that in order to minimise any sound wave cancellation effects, the angle subtended by the speakers should correspond to the phase angle between the L and R speaker signals, i.e., for inphase signals the subtended angle + should be zero, whilst for anti-phase signals the subtended angle + should be 1800.

The performance of a two speaker stereo sound reproduction system is clearly a compromise, as there is a three way trade-off, in order of importance, between: - (i) the stability and precision of in-phase inter-speaker images.

(ii) the width of the direct soundfield between the speakers.

(iii) the spacious quality of the reverberant or indirect sound field associated with near anti-phase signals.

A good compromise of the above requirements is achieved with speakers positioned as shown in Figure 15b, and it is generally accepted that the subtended angle + should not exceed 60".

It can be shown that the above three requirements can each be better achieved with a three speaker system. An initial comparison can be made with the three speakers in a horizontal layout, forward of and facing the listener, as shown in Figure 16. This shows a subtended angle of 60 between centre speaker C and speaker A, and 60 between speaker C and speaker B, giving 1200 between speakers A and B.

The angle of 60 assures that the stability of inter-speaker images is comparable with that in two speaker stereo. For example, when there is no output from speaker B, the output from speakers A and C is of equal amplitude and inphase. so that the image is mid way between them. This condition occurs when the signal relationships are L=0.9659, R=0.2588, which with its right side mirror image equivalent R=0.9659, L=0.2588, occupy positions at + 30 of centre.

In two speaker stereo the + 30 positions correspond to the speaker directions when the signals are L only or R only, whereas in the three speaker system these signal channel images would occupy positions at + 36.2 of centre.

The three speaker system therefore displays more width of the direct soundfield.

In practice this is perceived as being even wider, due to the effects of speaker signal phase angles, which have been ignored during the deduction of the L and R image positions. For example, the L only signal produces a small output from speaker B which is in anti-phase to the A and C speaker signals. This has a cancellation effect which reduces the sound wave intensity at the listeners right ear.

The result of this is for the sound image to seemingly emanate from further to the left.

When there is no output from speaker C, the output from speakers A and B is of equal amplitude and in anti-phase. As these signals are intercepting at an angle of 1200, rather than 60 as in two speaker stereo, the signal cancellation effects in the low and mid frequency ranges do not occur to the same extent and therefore the spacious quality of the reverberant soundfield associated with these types of signals is better maintained.

When the source signals L and R are of equal amplitude and in phase, the condition is reached for maximising the output from speaker C, whilst minimising the combined output from speakers A and B, all speaker signals being in phase and having power levels in the ration of 4:1:1 respectively. This compares with power levels in the ratio of 3:3 for two speaker stereo. However, because the opposing components of two vectors of length 1 at 1200 is less than for two vectors of length 3 at 60 as shown in Figures 17a and 17b, it follows that the three speaker system has a more stable centre image.

From all the above comparisons it can be seen that the inherent and mutually exclusive limitations of two speaker stereo sound reproduction can be overcome by a three speaker system.

This is only possible because a stereo source has excellent channel separation whereas the ears are sensitive to just a few dBs. A third speaker can thus be beneficially employed, the inter-speaker signal separation being reduced to a useable 6dB. This compares with a poor 3dB for a four speaker system operating from two channels.

The improvements are achieved at the expense of the integrity of the L only and R only images, but these directions are not as important as the centre image, which in the three speaker system is more stable.

Additional embodiments of the invention deal with the compensation for phase cancellation.

Referring to Figures 10 and 11 the cross feeding resistor ns reduces the level of anti-phase signals in order to achieve a linear overall volume balance. However, the arithmetic sum of the three speaker outputs does not take into account the effects of speaker signal phase angles.

In the section detailing the limitations of two speaker stereo, it has been described how anti-phase sound waves can have a cancelling effect in the low frequency range. This is dependent on the subtended speaker angle and the speaker distance. The three speaker system, with its wider overall subtended angle, can reduce the extent of the effect. but further improvements can be made.

If a capacitor Cx is placed in series with the crossfeeding resistor, as shown in Figure 24, their impedance increases as the signal frequency is lowered. The volume of anti-phase signals is correspondingly increased by a factor approaching three. as the speaker signal voltages tend toward their unmodified values. High frequency signals remain virtually unaffected, however.

The capacitor also introduces a phase shift which presents an in-phase component which further assists to compensate for the effects of phase cancellation.

It should be noted that this modification of the difference signal does not effect the mono signal of the centre speaker. and has a gradual effect as the sound directions move towards the anti-phase signal conditions. The effect is also gradual with variation of frequency. Very low frequencies, however, are usually associated with in-phase signals.

The relative value of the series capacitor is theoretically dependent on the overall subtended speaker angle and the speaker distance. In practice a value which corresponds to a 3dB frequency of around 100 Hz is generally useful and offers a slight but noticeable improvement to the overall performance of a three speaker system. For example, small speakers would display an improved low frequency response with stereo source material.

If the crossfeeding resistor had a value of 3.3 kQ for example, the required value of capacitor would be 0.47yF.

With a three channel amplifier arrangement such as shown in Figure 14, the above modifications cannot be achieved so simply. It would initially be necessary to separate the difference signals (1- r) and (r - 1) and pass them through a filter which reduces their voltages at high frequencies by a factor of 1.732 (the square root of three). These can then be added to a proportion of the l and r channel signals respectively to obtain the A and B power amplifier input signals. The derivation of the mono signal is not affected by the modification to the difference signals.

The amplifier output voltages of Figure 24 can be given as : Ll=(l -x)L+xR Where

R '=(1 -x)R+xL and the corresponding speaker signal voltages as : Va' = 2/3L' - 1/3R' = (e/3-x)R- (l/3-x)R Vb' = 2/3R' - '/3L' = (2/3-x)R- (l/3-x)L Vc' = '/3L + '/3R The above voltages are in the form given previously, except that x is now complex and frequency dependent.

At high frequencies the capacitor Cx is effectively a short circuit, so that x = 0.211 as previously determined.

At very low frequencies the capacitor Cx approaches an open circuit. so that x tends to zero and the speaker signal voltages tend toward their unmodified values.

There now follows some final comments about the rationalisation of speaker crossover networks.

In sound reproduction systems signal quality is better maintained if the transmission path losses, both resistive and reactive, are minimised. This can be achieved by keeping cable routes as short as possible, particularly at the large signal end i.e.. loudspeaker circuit. where the signal impedances are low.

The logical conclusion of this philosophy is for an amplifier to be positioned adjacent to, or integral with. the loudspeaker which it drives. These are known as active loudspeaker systems. The signals from the pre-amplifier to each power amplifier are usually via cables but could also be by limited range radio carrier frequency transmission, or fibre optic links.

Most domestic loudspeakers, however, are presently manufactured as separates, in matched pairs, to be connected to a two channel stereo amplifier via cables.

High fidelity loudspeakers typically incorporate two transducers, which are commonly called a woofer and a tweeter. The former handles the bass and midrange frequencies, whilst the latter handles the high frequencies. Their ranges are determined by filters in the form of crossover networks which comprise inductors capacitors and sometimes resistors.

Figure 20 shows a typical crossover network comprising two inductors and two capacitors. These components are housed within the speaker cabinet and internally connected to two speaker terminals which are connected to the amplifier output via a two-core cable.

One attempt at maintaining signal quality is known as bi-wiring, whereby the woofer and tweeter circuits are separated within the speaker cabinet, as shown in Figure 21a, and allocated their own pair of terminals. These are then separately connected to the amplifier output via two two-core cables. This approach can reduce cable losses and inter-modulation distortion between the low and high frequency ranges.

Further improvements could be made if the crossover components were removed from the speaker cabinet and positioned as close as possible to the amplifier, as shown in Figure 21b. With this arrangement the reactive currents of the crossover components do not flow along the cables to the transducers and therefore reactive voltage drop and associated distortion are reduced. As the cable currents are only those which actually produce sound output, its quality is thus better maintained. Furthermore the woofer and tweeter cables can now be individually selected to suit their respective frequency ranges.

The separation of crossover networks approach can also be applied to a three speaker system as shown in Figures 22a and 22b, where the woofer and tweeter circuits are drawn separately for clarity.

Additional improvements to sound quality can be made if the capacitors of each crossover circuit are repositioned so that, in each case, all three capacitors are adjacent to the star point of their respective crossover networks. A star/delta transformation can then be performed on the capacitors to enable their values to be reduced to one third as shown in Figures 23a and 23b. As capacitors are parallel devices. they can benefit by being smaller, and therefore cheaper, and having improved transient characteristics. Inductors and resistors, on the other hand, are series devices and would not benefit from a star/delta transformation.

The arrangement of Figures 23a and 23b therefore enables the sound quality of a three speaker system driven by a two channel amplifier to be optimised.

Admittedly, this is achieved at the expense of complicating the wiring arrangements.

Further improvements to sound quality would require a three channel amplifier so that the speaker signal paths are then only as long as their respective cables. The best sound quality would ultimately be obtained from three active loudspeakers, whereby each speaker incorporates its own single channel amplifier.

The reader's attention is directed to all papers and documents which are filed concurrently with or previous to this specification and which are open to public inspection with this specification, and the contents of all such papers and documents are incorporated herein by reference.

All of the features disclosed in this specification (including any accompanying claims, abstract and drawings), and/or all of the steps of any method or process so disclosed, may be combined in any combination, except combinations where at least some of such features and/or steps are mutually exclusive.

Each feature disclosed in this specification (including any accompanying claims, abstract and drawings), may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise. Thus, unless expresslv stated otherwise. each feature disclosed is one example only of a generic series of equivalent or similar features.

The invention is not restricted to the details of the foregoing embodiment(s).

The invention extends to any novel one, or any novel combination, of the features disclosed in this specification (including any accompanying claims, abstract and drawings), or to any novel one, or any novel combination, of the steps of any method or process so disclosed.

Claims

CLAINIS

1. A sound reproduction system having three loudspeakers.

2. A sound reproduction system as claimed in Claim l, in which the loudspeakers are arranged in a triangular layout.

3. A sound reproduction system as claimed in Claim 2, in which right hand and left hand speakers are arranged in an elevated position, facing a listening position, there being a third central speaker arranged lower down also facing the listening position.

4. A sound reproduction system as claimed in Claim 3, in which the speakers are inclined to the vertical.

A .

5 sound reproduction system as claimed in Claim 4, in which the upper speakers are inclined downwardly, the lower speaker being inclined upwardly.

6. A sound reproduction system as claimed in any one of the preceding Claims, in which there are right hand, left hand and central speakers, the central speaker being larger than the right and left hand speakers.

7. A sound reproduction system as claimed in any one of Claims 1 to 5, having right hand and left hand speakers, and a central speaker, the central speaker incorporating a sub-woofer.

8. A sound reproduction system as claimed in any one of the preceding Claims, in which at least one speaker has a woofer and/or tweeter circuit in which inductors and capacitors are each arranged in a star configuration.

9. A sound reproduction system as claimed in any one of Claims 1 to 7, in which at least one speaker has a woofer and/or tweeter circuit in which inductors are arranged in a configuration and capacitors are arranged in a delta configuration.

10. A sound reproduction system as claimed in any one of the preceding Claims, including means for removing or reducing any volume imbalance.

11. A sound reproduction system as claimed in Claim 10, in which the means for removing or reducing any volume imbalance operates by reducing the stereo separation of an amplifier.

12. A sound reproduction system as claimed in Claim 11, in which electrical resistance means is connected across the channels of the amplifier.

13. A sound reproduction system as claimed in Claim 12, in which capacitance means are connected in series with the resistance means to bring about a degree of phase compensation.

14. A sound reproduction system as claimed in any one of Claims 11 to 13, in which the reduction of the stereo separation of the amplifier modifies the amplifier output signal voltages to L' = 0.789L + 0.211R R' = 0.789R + 0.211L where L and R represent the unmodified left and right channel signal voltages and L' and R' represent the modified voltages.

15. A sound reproduction system as claimed in Claim 10 in which the three speakers are electrically connected to a star configuration.

16. A sound reproduction system as claimed in any one of Claims 11 to 15, in which the speaker signal voltages have the following relative proportions : Va' = l.366L - 0.366R Vb' = 1.366R - 0.366L Vc' = L + R where Va', Vb' and Vc' represent the three speaker signal voltages, and L and R represent the unmodified left and right channel signals.

17. A sound reproduction system as claimed in any one of the preceding Claims including a three channel amplifier.

18. A sound reproduction system as claimed in Claim 17, in which there are three amplifier stages, one for each speaker, signals to the three amplifier stages being derived from left and right channel signals via pre-amplifiers, invertors, bridge networks and attenuators.

19. A sound reproduction system as claimed in any one of Claims 1 to 16, including a two channel amplifier constructed or adapted to operate the three loudspeakers.

20. A sound reproduction system constructed and arranged substantially as herein described with reference to the accompanying drawings.

21. A three channel amplifier for use in powering three loudspeakers of a sound reproduction system.

22. A three channel amplifier as claimed in Claim 21, in which there are three amplifier stages, one for each speaker, signals to the amplifier stages being derived from right and left hand signals via pre-amplifiers, inverters bridge networks and attenuators.

23. A two channel amplifier constructed or adapted to provide output signals to three loudspeakers.

24. A two channel amplifier as claimed in Claim 23, provided with means to reduce the stereo separation of the amplifier.

25. A two channel amplifier as claimed in Claim 24 in which impedance means are connected across the channels of the amplifier.

26. An amplifier as claimed in any one of Claims 21 to 25, including switching means operable to convert the amplifier into a conventional two channel amplifier.

27. An amplifier constructed and arranged substantially as herein described, with reference to the accompanying drawings.

Amendments to the claims have been filed as follows 1. A sound reproduction system having three loudspeakers.

2. A sound reproduction system as claimed in Claim 1, in which the loudspeakers are arranged in a triangular layout.

3. A sound reproduction system as claimed in Claim 2, in which right hand and left hand speakers are arranged in an elevated position, facing a listening position, there being a third central speaker arranged lower down, also facing the listening position.

5. A sound reproduction system as claimed in Claim 4, in which the upper speakers are inclined downwardly, the lower speaker being inclined upwardly.

8. A sound reproduction system as claimed in any one of the preceding Claims, in which at least one speaker has a woofer and/or tweeter circuit in which inductors and capacitors are each electrically connected in a star configuration.

9. A sound reproduction system as claimed in any one of Claims 1 to 7, in which at least one speaker has a woofer and/or tweeter circuit in which inductors are electrically connected in a star configuration and capacitors are electrically connected in a delta configuration.

13. A sound reproduction system as claimed in Claim 12, in which capacitance means is connected in series with the resistance means to bring about a degree of phase compensation.

14. A sound reproduction system as claimed in any one of Claims 11 to 13, in which the reduction of the stereo separation of the amplifier modifies the amplifier output signal voltages to L'=O .7 89L+O.211R R'=O .7 89R+O.211L where L and R represent the unmodified left and right channel signal voltages and L' and R' represent the modified voltages.

15. A sound reproduction system as claimed in Claim 10 in which the three speakers are electrically connected in a star configuration.

16. A sound reproduction system as claimed in any one of Claims 11 to 15, in which the speaker signal voltages have the following relative proportions Va '=1 . 366L-O . 366R Vb/=l . 366R-0 . 366R Vc'=L+R where Va', Vb' and Vc' represent the three speaker signal voltages, and L and R represent the unmodified left and right channel signals.

18. A sound reproduction system as claimed in Claim 17, in which there are three amplifier stages, one for each speaker, signals to the three amplifier stages being derived from left and right channel signals via pre-amplifiers, inverters, bridge networks and attenuators.

22. A three channel amplifier as claimed in Claim 21, in which there are three amplifier stages, one for each speaker, signals to the amplifier stages being derived from right and left hand signals via pre-amplifiers, inverters, bridge networks and attenuators.

25. A two channel amplifier as claimed in Claim 24, in which impedance means are connected across the channels of the amplifier.