US2602860A - Loud-speaker structure - Google Patents

Description

July 8, 1952 L. s. DoulerrI LoUwsPEAKER STRUCTURE 2 SHEETS SHEET l *I Filed NOV. 18, 1947 III? III?

INVENTOR.

ATTORNEY Eo/v STEWART oouar m@ L m@ nl@ Patented July 8, 1952 UNITED STATES PATENT OFFICE LOUD-SPEAKER STRUCTURE Leon Stewart Doubt, Altadena, Calif.

Application November 18, 1947, Serial No. 786,563

4 Claims. 1

This invention is concerned with loud speakers and provides improvements in loud speaker arrangement to the end that improved low frequency response is obtained. More particularly, the invention provides a novel arrangement of loud speakers in a cluster to the end that the response of the cluster exceeds substantially the response of any individual loud speaker in the cluster.

The advantages of the invention are manifold, i'

including not only improved performance but also lower unit cost, lighter weight with unbraced cabinetry and non-reflex construction and smaller dimensions.

In its simplest form a loud speaker includes a cone of ilexible material such as paper, fastened at its edges rigidly to a baffle or the like and having a magnetizable diaphragm at its apex. A magnet is disposed behind the diaphragm and this is energized by a coil connected to the output of an audio-amplier or the like and either transformer or resistor coupled. The movement of the diaphragm due to iiuctuations of the field of the magnet vibrates the cone and thus produces sound. Dynamic speakers likewise employ a cone, but employ a voice coil instead of the diaphragm. The voice coil is wound on a small insulating tube that slides back and forth along the magnet and is energized by the amplifier. The magnet, in the case of a dynamic speaker, may be either a permanent magnet or an electromagnet that is steadily energized. The cone is attached to the tube by a spider of springy material. The fluctuating magnetic iieldv set up in the voice coil reacts with the steady eld of the magnet and moves the voice coil up and down the magnet. The spider permits the voice coil to move and forces it to come back once the pull ceases. The cone connected to the voice coil moves With it and vibrates to produce audible sound. The dynamic speaker is superior to the simpler type in that rattle due to the diaphragm striking the magnet is eliminated, but both types depend upon movement of the ilexible cone to vibrate and produce sound Waves.

The effectiveness of a vibrating device such as a speaker cone to produce an audible tone is determined by its ability to set in motion a mass of air. In the low frequency range of the audible frequency spectrum, the mass of air to be moved is large-so large in fact that many prior loud speakers and loud speaker systems are unable to respond fundamentally to tones in the low frequency range, i. e. below 100 C. P. S. One Way of increasing the mass of air is to increase the size of 2 the cone in an individual speaker. However, this approach has several disadvantages:

l. Paper, of which loud speaker cones generally are fabricated, is not suliciently strong to permit cone diameters much in excess of 18 inches.

2. As the size of the cone is increased, the ability of the speaker to respond to tones of a higher frequency, i. e. those of 4000 to 5000 C. P. S. and up, is severely reduced.

3. Large speakers and their attendant magnetic or field coils are unwieldy and the baiiles upon which such oversize speakers are mounted are too large for many uses.

.As a result of my investigations, I have developed a loud speaker system in which a plurality or cluster of speakers is mounted on a single bale in such fashion that the acoustic result in terms oi base frequency is closely analogous to that obtainable from a single large speaker having a cone area equal to the areas of the cones of the cluster. This result is achieved, however, without sacrifice of high frequency response. Thus the low frequency response of the cluster approaches the theoretical maximum while the high frequency response is not attenuated, as would be the case with a single speaker of equivalent cone area. In fact, because of variations in the mechanical resonance frequency of the individual speakers mounted in clusters in accordance with my invention, the high frequency response of the cluster generally exceeds in accuracy the response of any single component speaker in the cluster. Moreover, the low frequency response of the cluster is much lower than that of any individual speaker of the cluster.

In accordance with my invention the individual cones of a plurality of speakers are arranged in a cluster on a common baffle, with the edges of the cones very close together. This mounting of the speaker cones close together, i. e. with the distance between cone rims less than the cone radius and preferably less than 3 inches, is hereinafter referred to as tangential mounting.

The ability of the cone cluster of the invention to reproduce fundamentally lower tones than any which can be produced by any of the individual speakers of the cluster is apparently due to an acoustic effect rather than to an electronic or magnetic effect. I have determined experimentally that the acoustic effect is at a maximum when the speakers are mounted so that the rims of the speaker frames are actually in contact, i. e. tangential to each other; that the eifect is decreased sharply when the distance between the acoaseo 3 speakers is increased; and that the effect is completely absent as soon as the distance between the rims of the speakers exceeds the radius of the speakers involved.

In short, my invention contemplates a combination in a loud speaker structure which comprises a bale, at least two speakers mounted rigidly in the baiile with their edges close together, the distance between the edges of the speakers being less than the radius of any of the speakers, with common means for energizing the speakers. As indicated above, it is preferable that the speakers be mounted tangential with each other.

Some improvement in low frequency response is obtained with only two speakers mounted tangentially. If two speakers which will respond individually to no frequency lower than 200 cycles are mountedv tangentially in a common bafe and connected to a common energy source, the two together will respond fundamentally to 100 cycles. If a third speaker is added to the cluster of two no additional improvement results'. However, if four such speakers are combined, preferably in a symmetrical two-row group, these will respond to 50 cycles. Eight such speakers, mounted tangentially and symmetrically in a common baffle will respond to cycles.

The addition of one, two or three speakers to a cluster of four symmetrical and tangentially mounted speakers does not bring any improvement in low frequency response. However, when a fourth speaker is added to give a total of eight the response is improved by one octave. In sum, each time the number of speakers is doubled, the lowest frequency to which the cluster will respond fundamentally is lowered by one octave.

While the foregoing relationship actually exists, there is a practical limit beyond which a gain in fundamental frequency response is impracticable regardless of the number of speakers. This limit seems to be a function of the individual speakers. However, a grouping of eight or nine speakers is sufficient for all practical purposes, because even employing cheap individual speaker cones which do not have a frequency response below 200 C. P. S., the combination of eight such speakers in accordance with the invention permits a lowering of frequency response to 1/8 of 200 or 25 C. P. S. No musical instrument, except the single pipe of the largest organ, is capable of producing tones below a frequency of 25v C. P. S. andno human voice is capable of producing tones below this frequency. So, from a practical standpoint thereis no need for increasing the number of speakers above eight or nine.

e The preferable number of speakers for the practice of vmy invention is nine. As indicated above, optimum acoustic result is obtained with eight speakers and no further improvement in frequency response is obtained by adding the ninth. There are advantages however, since nine speakers -permit symmetry to be more easily obtained and impedance matching more easily accomplished. Thus I prefer to employ nine identical speakers arranged in three rows of three each. These speakers are wired half series half parallel, so that the voice coil impedance of the total cluster is the same as that of the speakers taken singly. With s uch an arrangement, standard amplifiers designed to match such single voice coil impedance may be used to energize the 4 cluster and no special transformer coupling or' other device is required between the amplifier and the cluster.

In addition to the foregoing considerations, the following observations should be made:

l. If the speakers are mounted tangentially, but not symmetrically with respect to two axes perpendicular to each other in the plane of the baffle, the improvement in lowering of frequency response tends to be diminished and in some cases lost.

2. The type of speakers which compose the cluster is not critical. Either permanent magnetic or field type speakers may be employed. Improvement of base response occurs regardless of the diameter of the individual speakers, but for matching purposes it is preferable that all of the speakers of the cluster be substantially identical.

3. Since the low frequency response can be brought to almost any desired level in accordance with the invention it is possible to employ sqeakers having hard or even metallic cones. Such speakers generally have a better middle and high frequency range response than is the case with softer paper cones. The base response resulting from the clustering of the speakers in accordance with the invention may thus be employed to compensate for the usual lack of low frequency response in this type of speaker while retaining or even improving response at middle and high frequency.

4. The only baling requirements are that the cluster of speakers be mounted on a single bafe of a dimension large enough to extend past the outside edge of the cluster. In other words, the space between the speakers of the cluster (which should be as small as possible) should be baifled. No noticeable effect is obtained by further baffling and there is no gain in true response by reflexing the speaker cluster.

5. Since a large mass of air is set in motion by the tangential speaker cluster of the invention, no restriction of air movement should be imposed by the cabinet. This is true of the back of the unit as well as of the front. Thus a solid backed cabinet with no space for expansion and contraction of the body of air contained within the cabinet will have a damping effect, especially with respect to the low frequency response of the cluster.

6. The amplifier which energizes the cluster of speakers of the invention cooperates with it and the maximum theoretical range of frequency response of the clustered speakers is not attained in practice unless the amplifier is capable of delivering an equivalent range to the speakers. The amplifier, therefore, should have a at frequency response over the frequency range of the speaker cluster. It should also have a fiat power output over this range. For many good commercial ampliers, frequency response is said to be iiat or linear from 20 or 30 to 10,000 cycles and such afrequency response actually is available, in terms of frequency. However, if available power in the amplier is plotted against frequency there will usually be sharp deviations from linearity. An average commercial 15 watt output amplier may produce l5 watts at some frequency or even over a certain frequency range, say 400 to 2000 C. P. S., but at frequencies above ,and below this range the power output will decrease until the outputof 20. cyclesnmay be aslow aspone-half watt and at 10,000 cycles, one or two watts. The

full response of the speaker structure of the invention is not available if such an amplifier 1s employed, for linearity of power output as wellV as frequency is required for optimum results. Accordingly, I prefer to employ an amplifier which has a nat power output or response over the audible range. In short, the available power should be linear over the audible range, say from to 15,000 cycles.

7. Because of the extremely low frequency response obtainable with the loud speaker cluster of the invention, faults of amplifier design which would not be apparent with conventional loud speakers become noticeable. If a conventional amplifier is employed with the clustered speakers of the invention there is a tendency for increased hum. This hum or unfiltered noise from the average conventional ainplier is not objectionable with conventional speakers, but with the speaker structure of the invention what was previously considered as a quiet amplifier is frequently shown to have considerable low frequency noise. Therefore, in order to obtain optimum results the amplier employed with the speaker structure of the invention should have complete filtering of low frequency noise and otherwise be as free from noise as possible.

8. Coupling of the loud speaker structure of the invention to the amplifier circuit is important. Transformer coupling is preferable and the construction of the transformer is important since it must meet the requirements of linearity in terms of both frequency and power output.

These and other aspects of the invention will be more thoroughly understood in the light of the Fig. 3 is a diagram illustrating tangential.

mounting of two speakers;

Fig. 4 is a diagram showing tangential and symmetrical mounting of four speakers in a preferred arrangement;

Fig. 5 is a diagram showing tangential and sym- -metrical mounting of eight speakers in accordance with the invention; and

Fig. 6 is a wiring diagram showing a preferred form of amplifier connected to the input of the nine speakers of the cluster of Figs. 1 and 2.

As shown by Figs. l and 2, nine conical speakers I0, II, I2, I3, lli, I5, It, Il, d, are arranged symmetrically in three rows of three in a single baille I9. This baiile may be at throughout or, as shown in Fig. 2, may be composed of rigidly joined tilted segments for increasing the broadcast angle, which is sometimes desirable, for example, in an auditorium. As shown in Fig. 1, the speakers are disposed with their cone edges or rims close together and substantially tangential to each other.

The half series, half parallel electrical connection of the voice coils (which individually have impedances of 6-3 ohms) of the speakers is seen in Fig. 6. There are three series of three speaker each, the three series being connected in parallel with each other. In such an arrangement, the impedance of the total cluster of nine speakers is the same as that of any of the individual speakers, assuming them all to be equal.

A variety of amplifier circuits which meet the requirements of the speaker cluster may be employed. Thus, the amplifier should have response which is substantially linear over the audible range both in terms of frequency and in terms of power. By way of example, if the amplifier has a rating of 15 watts it should be capable of producing 15 watts over the audible range of say, from 20 or 30 to 15,000 cycles. Otherwise the available frequency response of the clustered speaker unit is not employed to full advantage.

The amplier of Fig. 6 fulfills the foregoing requirements. It consists of three audio- amplification stages 30, 3l, 32 and a power supply 33.

The first stage is resistance-coupled to the signal input through a condenser CII] and a potentiometer RI the control grid of a 6SJ7 pentode of the first stage being connected to the slider of the potentiometer for volume control purposes.

The first and second stages are resistancecoupled through a condenser C3 and a resistance R5, the tube of the second stage being a 6J 5 triode.

The third stage contains two 6L6G tetrodes connected in push-pull in order to obtain requisite power output, and the second stage is resistance-coupled to it through a condenser C4 and resistances R9, Rl 0, with the cathodes of the two tubes connected to the grounded midpoint between these resistors through the resistancecondenser combination RI I, CS. The output of the third stage is transformer-coupled to the Voice coils of the speakers through a transformer TI, the primary of which is center-tapped because of the push-pull arrangement of the third stage.

The power supply consists of a transformer T2, the primary of which is energized by line current through a fuse and switch. The transformer has a low voltage center-tapped secondary winding which supplies heating current at about 6.3 volts to the filaments (not shown) of all tubes. The transformer also has a center-tapped high voltage secondary winding connected to a conventional full-wave rectifier tube VI, having a filament energized at 5 volts by a third secondary winding. The output of the rectifier is the high voltage supply for the amplifier at about 350 volts and is sent to the three stages through a resistance RM provided with grounded condensers C8, C9 at its ends.

To consider the first stage in greater detail, .the cathode and suppressor grid of its tube are connected for feed back purposes to the output side of the transformer TI through a resistance RI3 in series with the condenser CI. 'I'he cathode is grounded through a potentiometer R2, the slider of which is connected between the resistance R13 and the condenser CI. Bias for'the screen grid of the tube is obtained by connecting this grid to the power supply through the resistor R3, with a condenser C2 connected to ground between resistor and screen grid. The plate is energized through a plate resistor R4.

The cathode of the tube of the second stage is coupled to the third stage'through a resistance RB and a condenser C5 and connected to ground through another resistance RI. The plate of this stage is connected to the power supply through a plate resistor R8, and an interstage resistor RI2, through which power is also supplied to the rst stage. A condenser C1 is connected to ground between resistors R8 and RIZ.

In order to form a clear idea of the values of .condensers, resistors, etc., employed in the circuit, attention should be directed to the following table:

Table Condensers Resistors 01;'25 mid. 25 v. (J2-.2 mrd. 400 v.

Rl 500 K pot.

R2 1000 ohm pot. WW.

(D3-.(7)5 mdf. 000 v. R3 1 mcg. t w. (14e-.1 mfd. 600 V. R4 270 K t W. (l5-.1 mid. 600 v. R5 .5 meg. t w. C6--50 mid. 25-50 V. R6 2400 ohm t 'W. C7--16 mid. 450 v. R7 47 K t w. @8G40 mid. 450 V. R8 47 K 5c W. Cil-40 mid. 450 v. R9 270 K l w. C10-.05 mid. 600 V. R10 270 K l W.

R11 250,0hm 10 W R12 l5 K 1 W. R13 15 K la W.

The amplifier just described has a gain of 42 db with an output of not less than 15 watts within 1 db through a frequency range of 20 to 12,000 cycles. There is less than one percent (1%) total harmonic distortion. The frequency response of this amplifier is within 1/4 db from 18 to 22,000 cycles and within 2 db from 15 to 52,000 cycles. The noise level, as determined experimentally, is more than 60 db below one volt, measured with 10 ohms of pure resistive load across the output of the amplifier.

At some frequency, determined by the mechanical construction of the speaker and the character of the amplifier which energizes it a single speaker of the prior art tends to blow, i. e. to resonate. Thus many speakers tend to fblow at about 80 to 110 or 120 C. P. `S. I have discovered that even employing a speaker which tends to peak, i. e. increases its output at some frequency in the audible range, the speaker structure of the invention is desirable, Vbecause the several speakers of the combination tend to damp each other mechanically, even though no damping action is provided by the amplifier.

Nevertheless, when the speaker structure of the invention is to be 'run at something approachingk maximum output in terms of power, it is desirable to employ an amplifier which has as a substantially flat frequency and power response throughout the audible range, and this result can -bestbe yobtained by increasing nega-tive voltage feedback to a point considerably above the customary level, say to 15 db or more. Ar-1increase above 15 db, say to 601db is quite 'within the -frange of lpractice of the invention. However `it -must be recognized that such amounts of feed- 'back are only obtained through sacrice of gain vand require care in engineering 'of the amplifier -cir'cuit to .preserve performance and avoid stability, i. e. tendency to oscillate.

The high degree of `negative voltage feedback which is involved in the preferred practice of the invention requires an output transformer of excellent characteristics. Thus the transformer Tl of Fig. 6 at the 15 watt level has a fiat frequency response from .-10 to 42,000 C. P. S. within 1 db with power curve flat from 20 to 15,000 CfP. S. within 1 db, the percentage of harmonic distortion at the 15 Watt levelbe'ingless than 1%. With such a transformer, the 0feedback in the circuit of Fig. `6 may be from l5 to 30'db, and preferably is about db.

As indicated heretofore, some advantage accrues to the practice of -the invention with 'bnly two speaker 'cones mounted "on the 'baie 'in tangential arrangement. For, as illustrated in Fig. "3, if `the -base frequency Aof 'a `single speaker 8 is C. P. S. the mounting of two such speakers 40, 4l side by side on a single baffle 42 in tangential arrangement will reduce the base frequency response to 50 C. P. S.

Really outstanding results in the practice of the invention begin when four speakers 43, 44, 45, 46, are arranged tangentially and symmetrically in a baiiie -41 as shown in Fig. 4. Thus with four speakers having a base frequency response individually of 100 C. P. S., the base frequency of the cluster will be reduced to 25 C. P. S.

Further advantage accrues with still larger speaker clusters, one such arrangement with eight speakers, 50, 5l, 52, 53, 54, 55, 56, 51, being shown in Fig. 5. Here again it will be observed that the speakers are disposed tangentially on a common baffle 58 and arranged symmetrically about both a vertical and a horizontal axis. The base frequency response of these eight speakers is the same as that obtainable with nine speakers of the same type. However, impedance matching is somewhat more dinicult with the arrangement of Fig. 4 and so for practical purposes it is desirable to add an additional speaker to make a total of nine and to arrange these in three rows of three, as in Fig. 1.

The speaker cones need not be of the same diameter, but this arrangement is preferable. Moreover, it is desirable for purposes of impedance matching and ease of assembly and design, to employ speakers that are not only equivalent mechanically but also equivalent electrically.

The invention makes it possible to obtain frequency response over the entire audible range with speakers of small size which heretofore have given relatively poor reproduction. Thus, by combining lsay nine cheap speakers, in accordance with the practice of the invention, results may be obtained which are superior to those which are obtainable with the most expensive single speakers heretofore available. Naturally enough, the better the individual speaker the better the total result, since some speakers are individually better than others in terms of linearity of response throughout the audio spectrum.

l claim:

1. In a loud speaker structure of non-renex construction, the combination which comprises a single stationary baffle, nine substantially similar cone speakers mounted at their rims rigidly and symmetrically to vthe bale in three rows of three each and substantially tangential to each other, and common vmeans for energizing the speakers simultaneously and substantially equally, the individual 'speakers of the rows being connected electrically in series with the rows electrically connected in parallel with the energizing means.

l2, In a loud speaker structure of non-reflex construction, the combination which comprises a single stationary bale, a cluster of 'speakers arranged in parallel rows, with the same number of speakers in each row as there are rows, the speakers being rigidly fastened to the single baffle at their peripheries Withl their center portions movable, and with their peripheries substantially tangential to each other, and common means for energizing the speakers simultaneously 'and substantially equally, the speakers of each Vruw being electrically connected in Vparallel with the Venergizing means.

3. Apparatus according to claim 2 inwhich the energizing Vmeans is an amplifier having fa substantially flat power response over the frequency 'response range of the cluster.

9 4. Apparatus according to claim 2 in which the energizing means is an amplier having a substantially flat power response over the frequency response range of the cluster and having a nega tive voltage feedback of at least 15 db.

LEON S'IEWART DOUBT.

REFERENCES CITED The following references are of record in the Iile of this patent:

UNITED STATES PATENTS Name Date Williams June 24, 1930 Number Number Number

Images