US3265984A

US3265984A - Symmetric peak amplitude equalizer of complex sound waves, without peak wave distortion

Info

Publication number: US3265984A
Application number: US341285A
Authority: US
Inventors: Meguer V Kalfaian
Original assignee: Individual
Current assignee: Individual
Priority date: 1964-01-30
Filing date: 1964-01-30
Publication date: 1966-08-09
Anticipated expiration: 1983-08-09

Description

Aug. 9, 1966 M. v. KALFAIAN 3,265,984

SYMMETRIC PEAK AMPLITUDE EQUALIZER OF COMPLEX SOUND WAVES, WITHOUT PEAK WAVE DISTORTION Filed Jan. 50, 1964 5 Sheets-Sheet 1 4 l k z 3 1 SPfC/I GAMFCONTROUED PHASE Sal/RC5 V AMPL/f/ER DELAY RC NETWORK 8 9 Q 6 7 SPE'ECI'I GA/N'CUl/TRUHED PHASE SOURCE AMPl/FIIR DElAY PEAK [IN/I IAS I WTWORK PITCH SET-RESET --1 PEAK-LIMIT SELECTOR FLIP-HOP Yaw BIAS 2 2 4 l5 J y ,6 5mm GAIN-(ONTROHED r PHASE 7 .srsAny-sm rs SOURCE AMPl lF/[R .D El Ay FEfD- BACK STORAGE 0 20 62L 90 PHASE nmy PIA K- a IPPER r012 5q.wAv i 3 INVENTOR 3,265,984 E EQUALIZER 0F COMPLEX SOUND 5 Sheets-Sheet 5 M. V. KALFAIAN WAVES. WITHOUT PEAK WAVE DISTORTION SYMMETRIC PEAK AMPLITUD PITCH SELECTOR Aug. 9, 1966 Filed Jan. 30, 1964 mbmmni .mbtkam SYMMETRIC PEAK AMPLITUDE EQUALIZER OF COMPLEX SOUND WAVES. WITHOUT PEAK WAVE DISTORTION 5 Sheets-Sheet 4 Filed Jan. 50, 1964 INVf/VIOR. q

5 sheets-Sheet 5 g- 9, 1966 M. v. KALFAIAN SYMMETRIC PEAK AMPLITUDE EQUALIZER OF COMPLEX SOUND WAVES, WITHOUT PEAK WAVE DISTORTION Filed Jan. 50, 1964 United States Patent SYMMETRIC PEAK AMPLITUDE EQUALIZER 0F COMPLEX SGUND WAVES, WITHOUT PEAK WAVE DESTGRTION Meguer V. Kalfaian, 962 Hyperion Ave, Los Angeles, Calif. Fiied Ian. 30, 1964, Ser. No. 341,285 4 Claims. (Cl. 330-423) This invention relates to automatic amplitude control systems, and more particularly to a system for normalizing the amplitude of propagated speech sound waves to a constant peak amplitude level. Its main object is to compress and expand the amplitude of propagated speech sound waves, so that wide variations of envelope peaks of ordinary speech sound waves may be reduced substantially to a constant peak amplitude without appreciable time delay. A further object is to provide envelope-peak equalization without any loss in voice quality, and further yet, to enhance the intelligibility of the average speaker. While the present invention is particularly contemplated for speech sound waves, however, its nature of amplitude normalization is adaptable for other complex waveforms where such normalization facilitates wave analysis.

In various forms of speech sound wave production and reproduction, it is often desired that the wide variations of envelope amplitudes in speech sound waves are compressed and expanded to a constant amplitude level. In one example, such amplitude equalization is desirable for transmitting speech sound waves through radio links where the signal-to-noise ratio is very low. In such noisy transmission, the phonetic sound a may be received intelligibly, because of its inherent high amplitude characteristics, but the phonetic sounds, such as t, k, s, may be completely lost in noise, because of their inherent low amplitudes. In another form of speech sound wave reproduction, envelope-amplitude standardization is desirable for the analysis of amplitude ratios between various resonances in different phonetic sounds of the speech, for example, in a system as disclosed in my Patent No. 2,921,133 issued January 12, 1960, or other systems utilized for synthetic recognition of phonetic sounds in spoken words. Since the term phonetic sound represents intelligibility, however, this sound may take other forms of intelligible characteristics, such as, for example, the sound echo from a detectable target, or the various sound patterns made by lower forms of life, for example, by dolphins in their particular form of communication one with another. Thus the principal object of the present invention is to provide amplitude equalization without any appreciable time delay, and without effecting distortion to the original characteristics of the sound wave, so that any final analysis can be standardized.

A wave-envelope in speech sound waves, as mentioned herein, is referred to a wave pattern having a major peaked wave and minor peaked waves. In a voiced vowel, these wave patterns are produced repeatedly by pufis of air from the glottis, which are set into vibrationsin the momentarily formed resonant cavities of the vocal system. As each pufli of air enters these cavities, an initial surge of pressure is formed, and accordingly, these vibrations are commenced by a high peaked wave representing the major peak of said envelope. The peak amplitudes of these wave-envelopes, however, vary widely, for example, an amplitude ratio of about thirty to one exists between the pronounced sound a and the sound s, even if the speaker tries to speak in monotone. Each one of these wave envelopes contains all the information necessary for conveying intelligence of the phonetic sound, as described by theory in my above mentioned patent. Accordingly, it is not necessary for these major peaks to diifer one from 3,265,984 Patented August 9, 1966 another for intelligible conveyance of phonetic sounds. In some previous literature it has been stated that these variations are necessary for smooth transition from one phonetic sound to another. But according to my analysis and tests, this smoothness at transition periods is effected by the gradual combination of resonances of both phonetic sounds (preceding and succeeding), rather than by the amplitude variations. Thus, a controlled constant level waveform of the original speech will have exactly the same quality of voice as the unmodified original; except of course, the former having higher intelligiblity than the latter. Such high quality of voice reproduction can be obtained, however, only by stepwise modification of each successive envelope between major peaks without introducing distortion to the original wave. For example, as each initial surge of air starts in forming a wave pattern in the vocal tract, the amplitude of the entire wave pattern is predetermined with a definite waveshape. In conventi'onal practice of amplitude equalization, the gain of an amplifier is reduced by negative feed-back through a resistance-capacitance network. In this case, when the time constant of this network is adjusted to be substantially long, distortion to the original waveform is negligible, and thereby eifecting little impairment to the original quality of the voice; but complete amplitude equalization is not obtained. Whereas, when the time constant of the network is adjusted to be too short, the original waveform of each successive wave pattern is distortetd, with consequential impairment of the voice quality. Two types of wave distortions are eifected: the first being peak clipping; and the second being destruction of the original amplitude ratios of the waves in a wave pattern one with respect to another. The first type of distortion adds unpleasant noise to the voice, besides changing the original characteristics of the voice. The second type just changes the characteristics of the voice, and intelligibility remains high. While in voice transmission the intelligibility is a primary concern, high quality reproducton is also highly desirable. Accordingly, the principal object of the present invention is to provide a system of amplitude equalization both for high quality reproduction, and high intelligibility of the original voice.

Peak clipping in amplitude equalization is usually present in previously suggested systems, either due to the particular system utilized, or 'due to instrumentation inefliciency. This difficulty is obviated in the presently proposed system by delaying the gain control feed-back, so that the original major peak of each successive wave pattern is produced in the output circuit before compression or expansion starts taking place. This delay can be degrees or less, as long as the original peak has been transmitted before control of the gain feed-back has started. I have found in my tests that when the major peak wave is in its downward travel towards zero crossing, any sudden change in gain control of the amplifier will have no ill effect in the output voice quality, whatsoever. Thus by controlling the gain of the amplifier stepwise from major peak to major peak in 90 degree delay periods, the original sound wave can be amplitude equalized without peak clipping or any type of distortion imposed on it. In such an arrangement, the first and last major peaks in the directions of increasing and decreasing amplitudes will pass through the amplifier output without compression or expansion. In normal speech sound waves, how ever, the rising and falling amplitudes are gradual, and these rare first major peaks can be further clipped without being noticeable in the output sound. The main feature of the presently proposed system is, accordingly, a delayed feed back from the original sound wave, for controlling the gain of the amplifier.

In speech communication over a carrier, it is desirable that the original amplitude variations of the sound Waves are first normalized to a constant level, so that the maximum swing of carrier power may be modulated by the sound Waves for maximum transmitter efliciency, and high signal-to-noise ratio at the receiving end. The characteristic waveform of speech sound waves, however, is not symmetric in the positive and negative peaks, and gross amplitude equalization, such as practiced in the previous art, will not satisfy the ultimate purpose. According to my tests, both negative and positive peaks of the sound wave can be equalized to a constant amplitude level without impairing the original quality of the speaking voice. Accordingly, the present invention also provides a balanced system for normalizing both the positive and negative peaks of the speech sound waves without introducing sound distortion at the modified end. This will be more clearly explained in the following specification when considered in conjunction with the accompanying drawings, in Which:

FIGURES 1 through 4 are block diagrams in various arrangements for peak amplitude equalization of the complex waves without causing peak wave distortion, in accordance with the invention.

FIG. 5 illustrates phase relations between various waves describing the function of the invention.

FIG. 6 is a schematic arrangement of a balanced amplitude equalizer utilizing vacuum tubes in accordance with the invention.

FIG. 7 is a modified arrangement of FIG. 6.

FIG. 8 is a modified arrangement of FIGS. 6 and 7, utilizing photoconductor elements in balanced bridge circuits, for varying the impedances in one of their legs in accordance with projected light intensities, as compression elements of the original sound level in steady states, in accordance with the invention.

FIG. 9 is similar to the arrangement in FIG. 8, except that FIG. 9 does not utilize steady state amplitude equalization, for simplicity of apparatus.

From the foregoing brief it is apparent that various methods may be employed for amplitude equalization of the speech sound waves. The simplest method by which this can be accomplished, in accordance with the present invention, is a direct feed-back after phase delay of the original signal. This is shown by a block diagram in FIG. 1, wherein speech signals in block 1 are applied to the gain controllable amplifier in block 2. The output signals from block 2 are phase delayed in block 3, and fed back in degenerative direction to reduce the gain of amplifier 2, in series with a resistance-capacitance network in block 4, and a limiting bias in bock 5. The limiting bias in block 5 determines the amplitude level above which signal level suppression occurs. The preadjusted time constant of the RC network then determines the speed in which the amplitude variations of the speech sound wave is normalized. As described in the foregoing, the phase delay in block 3 prevents peak clipping, and therefore avoids harsh distortion of the original quality of the speaking voice. One objection of such a simple arrangement is that, the time constant of the network in block 4 must be short enough (for example, about one-fiftieth second) to establish complete normalization. This short time constant, however, causes unnatural distortion of the minor peaked Waves between major peaked waves, and thereby destroy part of the original quality of the voice. Another objection is that, it will not provide symmetric amplitude equalization.

The block diagram in FIG. 2 is arranged for symmetric normalization of the speech sound waves, wherein, the output of speech source in block 6 drives the balanced gain controllable amplifier in block 7. The output of block 7 is applied to a balanced phase delay circuit in block 8, from which the signals are further applied to separate RC networks in

block

9 and 10 for completing balanced feed-back loop to the amplifier 7. The limiting bias in block 11 can be common to both sides of the feed-back loop. The output signals may be taken either from the output of amplifier in block. 7, or from the output of phase delay circuit in block 8.

The balanced arrangement in FIG. 2 will provide symmetric amplitude equalization, but as stated in the foregoing, the changing gain of the amplifier by the RC network will distort the original waveform of the speech sound wave. Such a distortion can be avoided by an arrangement of stepwise gain control of the amplifier, as shown in FIG. 3. In this arrangement, the output of speech source in block 12 is applied simultaneously to the pitch selector in block 13 and the gain controllable amplifier in block 14. The output of amplitfier 14 is further applied to a phase delay circuit in block 15, which charges a storage device in block 16 when the voltage peaks in block 15 rise above the limiting bias voltage in block 17. The stored charge in block 16 is then fed back to the gain control element of the amplifier in block 14, as a gain suppression loop. The stored voltage in block 16 remains in steady state, so as to prevent distortion of the amplified sound wave. The discharge of the storage device in block 16 takes place during a short time period immediately after the original sound peak has been amplified without clipping. This is accomplished, as in the following:

The pitch selector in block 13 produces at its output pulse signals at the major peaks of the input sound wave, and applies them to a set-reset flip-flop circuit in block 18 for driving it into reset position. The set and reset operating states of the flip-flop in block 18 produce sharply defined square waves which may be differentiated into pulse waves by a small coupling capacitor C1. The coupling polarity from the output of block 18 to the input of a normally inoperative discharger in block 19 is so poled that the discharger 19 receives a forward operating pulse only when the flip-flop in block 18 is driven into set position, and therefore, the discharger 19 remains idle when the flip-flop in block 18 is driven into the said reset position. The phase delayed sound wave from block 15 is further phase delayed by degrees in block 20, and applied to the peak clipper in block 21 for changing it into sharply defined square waves. These square waves are further differentiated into sharp pulses by a small coupling capacitor C2, and applied to the flip-flop in block 18 for driving it into set position. The coupling from block 21 to block 18 is so poled that the pulse transmittal occurs at the wave peak from block 15 when the storage device in block 16 is already charged to said peak. The set position of the trigger in block 18 now transmits a forward pulse to the discharger in block 19, so that it discharges the storage in block 16 by a quantity necessary when the amplitude of the major peaks of the sound wave is diminishing, instead of increasing.

The timing relations between major peaks of the original sound wave; the phase delayed wave; and the pulses for discharging the feed-back storage, may be shown graphically in the illustration of FIG. 5. The original sound Wave is shown at A in asymmetric form with its positive peak being larger than the negative peak from reference zero level. The pulse wave at B represents the major peak selection (of the Wave A) at the output of pitch selector in block 13. The sine wave at C represents the phase delayed sound wave at the output of phase delay circuit in block 15. This Wave C is further ninety degree phase delayed, as shown by the wave at D, which represents the output wave of phase shifter in block 20'. The wave at D is peak clipped to an extent as to form square wave, as shown by the wave at E, which represents the output of peak clipper in block 21. The square wave at E is further dilferentiated by the coupling capacitor C2, so as to form the pulse wave at F, for operating the trigger in block 18. With the phase relations of different waves just mentioned, it appears that the phase shifter in block can be eliminated, and the square wave at E obtained from the original sound wave at A. However, asymmetry of the wave at A will delay generation of the pulse Wave at F due to the delayed zero crossing, and the stored peak voltage in block 16 will be discharged to a lower value than had been charged to the peak value of the major peak. Thus the inclusion of phase shifter in block 20 is found necessary for greater precision of amplitude equalization, although it may be dispensed with if so desired.

The block arrangement of FIG. 3 has been given to describe the basic principles of amplitude equalization without distortion to the original waveform of the sound for high quality reproduction. As explained in the foregoing, the same quality of reproduction is also obtained by symmetric amplitude equalization, which may be achieved by an arrangement, as shown in block diagram of FIG. 4. In this arrangement, the output of speech source in block 22 is simultaneously applied to the gain controllable push-pull amplifier in blocks 23, and two separate pitch selectors in

blocks

24 and 25; the sound waves in positive polarity being applied to the block 24, and the waves in negative polarity being applied to the block 25. The output of amplifier 23 is phase delayed by a balanced circuit in block 26, and further applied to the storage devices in

blocks

27 and 28. These storage devices are charged in alternate time periods by the positive and negative major peaks of the sound from the output of phase delay circuit in block 26, when their peak amplitudes exceed that of the limiting bias in block 29. The phase delayed output wave of block 26 is further delayed by ninety degrees in block 30, and from thereon peak-clipped in block 31 for changing it into sharply defined square waves. These square waves are then differentiated into pulses by the small coupling capacitors C3 and C4 and further applied to the set-reset flip-flops in

blocks

32 and 33, for driving them into set positions. Assuming then, that the flip-flop in block 32 is operated into its set position, its output is differentiated into a pulse wave by the small coupling capacitor C5, and applied to the normally inoperative discharger in block 34 for discharging the storage device in block 27, as a new adjustment of gain suppression feed-back to the amplifier in block 23. Similarly, when the flip-flop in block 33 is driven into its set position, its output is differentiated into a pulse wave by the small coupling capacitor C6, and applied to the normally inoperative discharger in block 35 for discharging the storage device in block 28,

as a new gain suppression adjustment to the amplifier in block 23. In order to establish the set operating positions of the flip-flops in

blocks

32 and 33, these are first operated into reset positions by the output pulses of respective pitch selectors in

blocks

24 and 25. The sym metric amplitude equalized wave of the speech sound may then be taken from either the output of amplifier in block 23, or from the output of phase delay circuit in block 26.

Up to this point the basic principles of stepwise amplitude equalization of sound waves has been described by the block diagrams. For further details of a practical system, however, reference is now being made to schematic arrangements, one embodiment of which is shown in FIG. 6. In this arrangement, the sound wave is generated in a balanced circuit, as represented by the block 36. The output of block 36 is applied in push-pull mode to the first control grids of pentagrid vacuum tubes V1 and V2, and is amplified in the output transformer T1. The two end terminals of primary coil L1 of transformer T1 are connected to the respective anode elements of tubes V1 and V2, with its center tap connected to the supply voltage B1, so that the voltage waves applied to the first grid elements of V1 and V2 are changed to current waves through the primary coil L1. Due to the reactive characteristics of coil L1, the voltage wave leads the current wave by 90 degrees, and since audio transformers are not tuned within the range of audio spectrum, the voltage wave in the secondary L2 will be in phase with the primary coil L1. While available audio transformers will not yield this degree phase relation within the entire audio spectrum in the arrangement given, there will exist sufficient phase delay at all frequencies to satisfy the required performance. The voltage waves developed across the secondary coil L2 are then rectified by the diodes D1 and D2, and charged .across the capacitors C7 and C8, the common return connections of which are terminated to the center tap of coil L2 in series with the limiting bias supply B2. The voltage value of bias D1 is adjusted to the minimum level at which the sound amplitude is desired to pass the transformer T1. Any sound wave above this adjusted level causes conduction of the diodes D1 or D2, and charge its associate capacitor C7 or C8 to charge. The capacitors C7 and C8 are charged in negative polarity, and directly connected to the second control grids of tubes V1 and V2, respectively, so that any negative voltage across these capacitors causes proportional gain reduction of the respective tubes. Thus with sufficient gain provided by the amplifier, a wide range of peak amplitude variations of the original sound waves in block 36 may be compressed to a constant level across the secondary coil L3 of transformer T1. As described by way of the block diagrams, the discharge of capacitors C7 and C8 is established in short periods of time by normally idle dischargers, the operation of which is as follows:

The positive poled waves of the original sound from block 36 are applied to the pitch selector in block 37, and the negative poled waves of this sound are applied to the pitch selector in block 38. The output pulses of pitch selector 37 are applied to the set-reset flip-rflop in block 39 for driving it into reset position, and the output pulses of pitch selector in block 38 are applied to the flipfiop in block 40 for driving it into reset position. The output voltage Wave across secondary coil L2 is further phase delayed by 90 degrees in block 41, and further peak clipped for translation into sharply defined square waves in block 42, as shown by the wave at E in FIG. 5. These square waves are then differentiated :by small coupling capacitors C9 and 010, such as at F, and applied to the flip-flop in

blocks

39 and 40, respectively, for driving them into set operating positions. As described by -way of the block diagrams, the square voltage waves developed at the outputs of flip-fiops in

blocks

39 and 40, in their set operating positions, are further differentiated into negative pulses and applied to the base elements of normally inoperative P'NP transistors Q1 and Q2 for becoming conductive and discharging the capacitors C7 and O8 in respective pulse time periods. The transistors Q1 and Q2 are normally biased in backward direction by the positive voltage across B l in series with the load resistors R1 and R2, respectively. The series connected resistors R3 and R4 may be included to avoid momentary shorting of the capacitors because transistors in general are low impedance devices in their conductive states; although these resistors may be dispensed with, if so desired.

The schematic arrangement in FIG. 6 is shown with one stage of amplitudeequalization. When the amplitude of input sound wave from block 36 experiences extreme variation, however, the dynamic characteristics of vacuum tubes V1 and V2 may not satisfy the required normalization with a single stage, and more than one stage of equalization may be necessary. Thus the first stage may be repeated, as required. With multiple such stages it is preferable that each succeeding stage is driven with in-phase sound waves. This may be done with a slight modification of the schematic, as shown in FIG. 7. In this arrangement, the speech sound wave in block 43 is applied in push-pull mode to the first control grids of pentagrid tubes V3 and V4, and is amplified in the output transformer T2. The primary coils L4 and L5 of this transformer are split wound, so that they may be separately connected to the anode electrodes of V3 and V4, and to the plate supply voltage B3 in series with re sistors R5 and R6, respectively. With such an arrangement, the voltage waves developed across resistors R5 and R6 will be 180 degrees out of phase with the applied input voltages to the first control grids of tubes V3 and V4, and the voltage waves developed across primary coils L4 and L5 will be 90 degrees out of phase with said voltage waves. Since the voltage across secondary coil L6 will be in phase with the voltage in the primary coils L4- and L5, the lagging voltage wave is then obtained across the coil L6 for gain suppression feed back and the in-phase voltage wave is obtained from the resistors R5 and R6 for application to a succeeding stage of amplitude equalization from the secondary coil L8 of output transformer T3. This output is taken from R5 and R6 by the coupling capacitors C13 and C14 connected to the end terminals of primary coil L7 of transformer T3.

As described by way of the arrangement in FIG. 6, the positive and negative voltage waves across secondary coil L6 are separately rectified by the diodes D3 and D4, and charged in negative polarity across capacitors C15 and C16. The limiting bias voltage of B4 will normally prevent charge of the capacitors G15 and 016 until the voltage level across secondary coil L6 exceeds the voltage of B4. The capacitors 015 and C16 are directly connected to the second control grids of V3 and V4, respectively, so that any amplitude of'the original sound wave above a predetermined level is suppressed to a constant level by the negative fee-d back from 015 and 016. The discharge of these capacitors is established in short time periods, as in the following:

The positive poled waves of the original sound from block 43 are applied to the pitch selector in block 44, and the negative poled waves of this sound are applied to the pitch selector in block 4 5. The output pulses of pitch selector 44 are applied to the set-reset flip-flop in block 46 for driving it into set operating position, and the output pulses from pitch selector 45 are applied to the set-reset flip-flop in block 47 for driving it into set position. The output voltage wave across secondary coil L6 is further phase delayed by 90 degrees in block 48, and still further peak clipped for translation into sharply defined square waves in block 49 as shown by the Wave at E in FIG. 5. These square waves are then differentiated by small coupling capacitors 017 and 018, such as at F in FIG. 5, and applied to the flip-flops in blocks 46 and 47, respectively, for driving them into reset operating positions. The set operating positions of flip-flops in blocks 46 and 47 are polarized to apply forward biases upon the base electrodes of transistors Q3 and Q4, respectively, while the reset positions of said [flip-flops apply backward biases upon said transistors. Thus when major peaks appear at the outputs of pitch selectors in

blocks

44 and 45, the transistors Q3 and Q4 become conductive in their respective order and remain conductive until pulses from the block 49 operate the [flip-flops in blocks 46 and 47 into their reset positions. In such an arrangement, the capacitors 015 and 016 are discharged during quarter wave periods of the sound wave, instead of during pulse periods, as had been the case in FIG. 6. Since the transistors Q3 and Q4 are generally of low impedance devices, the values of series connected resistors R9 and R10 are chosen to effect slow discharge of the capacitors C15 and C16 (this may be averaged for different frequencies of the sound for discharge), so that during the said quarter cycle period the capacitor voltages are adjusted to the peak values without causing short circuit of the secondary coil L6, or possible injection of pulse signals in the amplifier circuit. Either mode of operation, however, may be utilized as found suitable.

In the schematic arrangements of FIGS. 6 and 7, the gain controlled amplifiers have been shown utilizing variable-mu vacuum tubes. Transistors (for example, tetrodes) may be substituted to obtain equivalent performance of sound normalization. However, these devices, especially transistors, are susceptible to distortion when operated within wide range of amplitude variations, and therefore, require more than one stage of amplitude equalization for the desired degree of sound compression. Instead of utilizing these variable-mu devices, however, it is possible to vary the resistance in one leg of a bridge circuit against any range of signal amplitude variation without causing distortion of any kind, whatsoever. For fast control of resistance value, the resistive element in this leg may be substituted by a photoconductor, so that its resistance may be changed at will by an optical light, such as produced by a filamentary bulb, or by a solid state device, for example, the electroluminescence emitted from the gallium-arsenide mesa diode. Since photoconductors possess extremely high resistance (for example 800 megohms) in dark states, and low resistance in lighted states (for example, 4,000 ohms), a single stage of amplitude compression may be found satisfactory for any purpose without encountering undesired nonlinear characteristics. The frequency responses of presently available photoconductors, however, are to slow for short pulse responses, but experimental photoconductors opera-ting at high speeds are announced in periodicals, and may soon appear in the market. Similarly, while filamentary light bulbs are too sluggish to produce light variation at the required speed, the solid state devices operate at high speeds, and high illumination devices may also be available soon commercially. A schematic arrangement utilizing light modulated bridge circuit, is accordingly, shown in FIG. 8.

In FIG. 8, the input sound signal is applied to the primary coil L9 of transformer T4. The secondary is split into two separate coils L10 and L11. Across the coil L10 are series-connected two resistors R11 and R12 of equal values, and their junction point connected common to ground. From the upper terminal of coil L10, there is connected a resistor R13, and in series with it connected a photoconductor device PC-1. From the lower terminal of coil L11), there is connected a resistor R14, and thereon to the photoconductor PC-l. Thus, when the total resistance of R13 in series with the resistance of PC-l is equal to the resistance of R14, the voltage between the junction point of resistors R11, R12, and the junction point between R14 and PC-l becomes zero at all times because of the balanced bridge arrangement. Whereas, when the resistance in one leg, for example, the resistance of R13 and PC-1 is varied from zero to infinity, the signal across L10 appears across said two junction points from zero to maximum in reversed polarities. Thus assuming that the resistances of resistors R13 and R14 are of equal values, and assuming that the resistance of photoconductor PC-l in maximum lighted value is much lower than the value of R13, the output signal at junction point between R14 and photoconductor PC1 will have a predetermined minimum amplitude during maximum lighted state of the photoconductor, and maximum signal amplitude during its dark state, since as stated, photoconductors usually have extremely high dark resistance values. Accordingly, when the input signal amplitude across coil L10 is at minimum, the photoconductor PC1 may be left in dark state so that the signal is transmitted to the output terminal without suppression whereas, when the input signal amplitude is at maximum, the photoconductor may be optically lighted at maximum, so as to reduce the signal at the output terminal. By varying the light upon the pohtoconductor according to the signal strength, the signal amplitude at the output terminal may then be held constant continually without the use of gain controlled amplifiers. For symmetric amplitude equalization, however, it is necessary that two such bridge modulating arrangements are used for both positive and negative peaks of the original sound wave. This second bridge may be obtained across the split secondary coil L11, the two end terminals of which are coupled two series-connected resistors R15 and R16 of equal values, and the junction point between the two resistors connected to ground. To the lower terminal of coil L11 is connected a resistor R17, and in series with it connected a photoconductor PC-2. To the upper end terminal of coil L11 is connected a resistor R18 having equal resistance value as of R17, and the resistor R18 is connected to the PC-2 as a junction point for the output circuit from ground.

The output circuit terminal between the junction point of R14 and photoconductor PC-l is coupled to the base electrode of amplifying transistor Q5 by way of coupling capacitor C19, and the output circuit terminal between the junction point of R18 and photoconductor PC-2 is coupled to the .base electrode of amplifying transistor Q6 by way of coupling capacitor C20. A normal operating bias is applied upon the base electrode of transistor Q5 from across voltage divider resistors R19 and R20 connected between ground and the voltage source B5. Similarly, a normal operating bias is applied upon the base electrode of transistor Q6 from across voltage dividing resistors R21 and R22 connected between ground and supply voltage B5. Thus the input signals from the two bridge modulating circuits are amplified separately by the amplifying transistors Q5 and Q6 in their collector circuits which are connected to the center tapped primary coil L12 of transformer T5. The balanced signal voltages developed across the secondary coil L13 of transformer T5 are separately rectified through diodes D5 and D6, for charging these voltages across capacitors C21 and C22 when said voltages exceed the limiting bias voltage of B6. The capacitors C21 and C22 are directly coupled to the base electrodes of emitter follower transistors Q7 and Q8, respectively. The emitter circuit resistors R23 and R214 of Q7 and Q8, respectively, are coupled to the base electrodes of transistors Q9 and Q10 in series with current limitingresistors R25 and R26, respectively. The collector electrode of transistor Q9 is connected to the supply voltage B5 in series with the filament of an optical light bulb LB-l. Similarly, the collector electrode of transistor Q10 is connected to the voltage supply B5 in series with the filament of an optical light bulb LB-2. The function of the arrangement just mentioned is that, when the signal voltage developed at either end terminal of the center tapped secondary coil L13 of transformer T5 is higher than the bias voltage of B6, current passes through either diode D5 or D6, and accordingly the capacitor C21 or C22 is charged proportionately. Thus assuming that the capacitor C21 is charged with some voltage, current flows through the emitter circuit resistor R23 of Q7, which further drives the transistor Q9 con- \ductive for energizing the light bulb LB-1, so that its light emission can reduce the resistance value of photoconductor PC-l. As this resistance value is lowered, the bridge circuit becomes more and more balanced, with the result that the signal amplitude applied to the base electrode of amplifying transistor Q5 is reduced; thus efiecting the required amplitude compression. Similarly, assuming that the capacitor C22 is charged with some voltage, current flows through the emitter circuit resistor R24 of transistor Q8, which further drives the transistor Q10 conductive for energizing the light bulb LB-2, so that its light emission can reduce the resistance of photoconductor PC-2. As this resistance is reduced, the bridge circuit becomes more and more balanced with the result that the signal amplitude to the base electrode of amplifying transistor Q6 is reduced; thus effecting the required sound amplitude compression.

The base to emitter resistance paths of transistors are usually low, impedance characteristics, and it is desired in the arrangement of FIG. 8 that, the capacitors C21 and C22 retain their stored charges after they are initially charged. For this reason, these capacitors are first coupled to base electrodes of emitter follower transistors Q7 and Q8, so as to offer high impedance to the capacitors. By the proper choice of these transistors, diode action 10 can be obtained. For example, the type 2N-404 transistor has an emitter to collector current cut-off characteristics when the base voltage is close to zero and lower toward positive direction with respect to the emitter, and starts emitter to collector conduction with the base voltage starting close to zero toward negative direction with respect to the emitter. The emitter circuit resistor of this transistor may then drive a higher power transistor Q9 or Q10, for providing the necessary current to the light bulbs LB-l or LB-Z. It must be remembered here that,

the higher powered transistors, especially silicone transistors, do not start conduction, except of course by leakage, from zero bias upon their base electrodes. Similarly, the diodes, for example D5 and D6, have some voltage gap from zero level before they start conduction. All these voltage gaps add to the limiting bias B6, so that the voltage of B6 is chosen accordingly. In fact, the bias voltage B6 may be eliminated completely, if the said voltage gaps are sufficiently high as a substitute for the bias voltage of B6.

The discharge of capacitors C21 and C22 is effected in the same manner as described by way of the arrangements in FIGS. 6 and 7. For example, the original sound Wave in the secondary coils L10 and L11 of transformer T4 is applied in opposite poles to the pitch selectors in

blocks

51 and 52. As explained in the foregoing these pitch selectors produce at their outputs pulse signals at major peaks of the sound wave. These pulses are applied to the set-reset flip-flops in

blocks

53 and 54, respectively, for operating them into set positions. At the same time, the output sound wave from across secondary coil L13 (or in any convenient mode) is degree phase shifted in block 55, and changed into square waves by clipping in the block 56. The output square waves from block 56 are then further differentiated into pulses by the small coupling capacitors C23, C24, and applied to the flip-flops in

blocks

53 and 54 for operating them into reset positions. As described in the foregoing, the set operating positions of the flip-

flops

53 and 54 apply forward biases in series with the current limiting resistors R27 and R28 upon the base electrodes of transistors Q11 and Q12, respectively, while their reset operating positions apply backward biases upon said transistors. Thus assuming that the flip-flop in block 53 is in set position, the transistor Q11 becomes conductive and discharges the capacitor C21 in series with the resistor R29. The value of resistor R29 is so chosen that conductance through the transistor Q9 is just sufficient to discharge the capacitor C21 while at the same time effecting its charge to the peak of the applied voltage; at which point the flip-flop in block 53 is driven into reset position for isolating the capacitor from further charge or discharge. This action is exactly similar regarding the discharging transistor Q12 which discharges the capacitor C22 in series with the resistor R30 when the flip-flop in block 54 is driven into set operating position, and stops said discharging action when the flipflop is driven into reset position.

In reference to the arrangement in FIG. 8, the output sound wave may be taken from the transformer T5, for example, from the secondary coil L14, or in any mode as desired. As described in the foregoing, the positive and negative amplitudes of the speech sound wave are not always symmetric. Thus in each branch of the bridge circuits across secondary coils L10 and L11 of transformer T4 the original sound wave is amplitude equalized separately. This does not mean, however, that rectified action takes place, since variation of the resistance of either photoconductor PC1 or PC-2 results in amplitude variation of the sound wave in both positive and negative polarities simultaneously. The sound wave in both bridges are combined in the output transformer T5, and due to the mutual inductance between both legs of the balanced transformer, the equalized amplitudes of both positive and negative polarities ofv the sound wave are averaged out, so that disproportionate amplitude equal- I 1 ization of appreciable magnitude does not appear in the final output wave. As described in the foregoing, however, some disproportionate amplitude equalization in the positive and negative polarities of the sound wave does not affect the original quality of the voice, as long as the waveform itself is not distorted.

As described in the foregoing, quality degradation of the original sound is caused by peak clipping and wave distortion. Peak clipping is the most serious for unpleasantness of the listener. Whereas, when wave distortion is present without peak clipping, the amount of quality degradation is not sufficiently high to cause unpleasantness to the listener. For some purposes where simplicity of apparatus is of primary importance, and degradation of the original sound can be tolerated, the arrangement given in FIG. 9 may be preferable. In this arrangement, the required amplitude equalization is established by suppression feed-back at a time rateas determined by the time constant of a resistance capacitance network. With such an arrangement, the original sound wave is compressed at each high peaked wave, and expanded at a constant time rate, so that each succeeding peak will be compressed to the minimum predetermined amplitude level. The wave distortion of the original sound is caused by the constant expansion of the sound amplitude, and the faster this expansionoccurs the greater the distortion becomes. However, peak clipping is eliminated in this arrangement, because the suppression feed-back peaks are delayed in phase from the original, as had been described in the foregoing. This arrangement is shown in FIG. 9.

In FIG. 9, the input sound wave is applied to the primary coil L15 of transformer T6, and induced inductively across the split secondary coils L16 and L17. The bridge circuit across the coil L16 is formed by first connecting the balanced resistors R31 and R32 to the end terminals of L16, and grounding the junction point of these resistors. From the upper terminal of L16, there is connected a resistor R33, and in series with it connected the photoconductor PS3. From the lower terminal of coil L16 there is connected a resistor R34, and thereon to the photoconductor PC3. The output is taken from the junction terminal between R34 and PC3, and coupled to the base electrode of amplifying transistor Q13 by the coupling capacitor C25. A normal bias upon the base of Q13 is obtained by the voltage dividing resistors connected across supply voltage B7. The bridge circuit across secondary coil L17 is formed by the balanced resistors R37 and R38 across said coil and the junction terminal between these resistors connected to ground. One terminal of the resistor R39 is connected to the lower terminal of L17, and to the other terminal of R39 is connected to the photoconductor PC-4. The resistor R40 is then connected between the upper terminal of coil L17 and the photoconductor PC-4, and the junction terminal between R40 and PC4 is coupled to the base electrode of amplifying transistor Q14 by the coupling capacitor C26. The normal bias voltage to the base of Q14 is obtained by the voltage dividing resistors R41 and R42 connected across the supply voltage B7. The arrangement thus far described indicates that the output terminals of the bridge across coil L16 are taken between the junction terminal of balanced resistors R31, R32, and the junction terminal of resistor R34 and the unbalanced total resistance of R33 and PC-3. Similarly, the output terminals of the bridge across coil L17 are taken from the junction terminal between balanced resistors R37, R38, and the junction terminal between resistor R40 and the total resistance of R39 and PC4. Sound suppression is now effected, as in the following:

The input signals from the bridge circuits are amplified by the transistors Q13 and Q14 in their respective collector electrodes, which are connected respectively to the end terminals of balanced primary coil L13 of transformer T7. The center tap of L18 is connected to the supply voltage B7, to complete the return circuits from collector to emitter of the transistors Q13 and Q14. The positive and negative voltage waves developed across the center tapped secondary coil L19 are separately stored across capacitors C27 and C28 through the respective rectifier diodes D7 and D8. These stored voltages are applied in forward direction to the base electrodes of transistors Q15 .and Q16, in series with the current limiting resistors R43 and R44, respectively, so that these transistors can vary the magnitude of currents flowing through the filaments of their respective light bulbs LB-3 and LB4. Accordingly to the current variations through the filaments of these light bulbs, their light intensities vary for varying the resistance of photoconductors PC3 and PC-4; thus establishing the required suppression of the original sound. The output may then be taken from the secondarycoil L20, or in any mode as found suitable. At this point it must be mentioned that photoconductors are not the only devices for displaying variable resistance characteristics, as thermistors and varistors, such as field effect resistors, also display variable resistance characteristics, and may be used instead of the former.

In reference to the arrangement of FIG. 9, it will be noted that the elements for discharging the capacitors C27 and C28 are omitted. In the previous arrangements, these capacitors were connected to emitter followers acting as high impedance loading elements, so that the inclusion of separate discharger elements were necessary for discharging them in short pulse time periods. However, the transistors Q15 and Q16 are used as straight amplifiers, and their base circuits act as low impedance loadings across the capacitors C27 and C28; thus simulating as resistance capacitance network. By choosing the proper values of capacitors C27 and C28, the rate of volume expansion of the original sound wave may then be predetermined. This constant rate of sound volume expansion will distort the original waveform of the sound, but as stated in the foregoing, this distortion will not alfect the sound quality as much as by clipping, .and since there will be no clipping action, the output sound in constant amplitude will have little deviation from the original quality. Thus with the presently disclosed arrangements and the specification thereof, it becomes apparent that various other modifications, substitution of parts, and adaptations may be made without departing from the true spirit and scope of the invention.

What I claim is: I

1. An arrangement for symmetric amplitude control of complex waves having asymmetric oppositely poled peaks, said control being in steady state steps between succeeding peaks, and without causing peak wave distortion, the arrangement comprising the following: a push-pull amplifier having first and second bridge circuit input and an out-put; each of said first and second bridge circuits comprising a center tapped impedance means having two end terminals connected to the two end terminals of a series-connected variably-controllable resistive element and a fixed-valve resistor, respectively, for forming said first and second bridge inputs by coupling the junction terminals between said fixed-value and variablycontrollable resistive elements of the said first and second impedance means to the said first and second inputs, respectively, from common parallel-connection of said center taps; means for generating said complex waves, and means for applying these waves in push-pull mode to said first and second impedance means, for amplifying said waves at the output of said amplifier; first and second storage means, and first and second rectifier means connected in series with said first and second storage means, respectively; first and second normally non-conductive discharger means in parallel with said first and second storage means, respectively; a phase shifting means con nected to said output, the last said means having first and second output-terminals; a reference amplitude level determining means for said phase shifting means; coupling means between said first and second output-terminals, and said first and second rectifiers, respectively, for unidirectional charging of said storage means, when the peak amplitudes of said phase shifted peaks are above said reference level; first and second peak selectors for deriving from said generated complex waves first and second signal-waves of said oppositely poled peaks at time periods substantially coincident with the respective pea/ks appearing at said output-terminals of the phase shifting means; means for applying the first and second signal waves to said first and second disch-arger means, respectively, for

rendering said discharger means conductive and re-adjusting the charges in said storage means to the instant levels at said output-terminals of the phase shifting means; means for translating the charges of said first and second storage means into controlling means; and means for varying said variably-controllable resistive elements of said first and second bridges by said first and second controlling means, respectively, for symmetrically compressing the admittance of the oppositely poled peaks of said generated complex waves to said output, thereby effecting said stepwise symmetric amplitude equalization at said output, and without peak wave distortion by virtue of gain readjustments at time shifted periods than the original peaks arriving at said output.

2. The arrangement as set forth in claim 1, wherein each of said variably-controllable resistive-elements of the said first and second bridges comprises an electrical sensitive resistive-means, the required electrical energy of each of which is controlled by the said charges across said first and second storage means, respectively.

3. The arrangement as set forth in claim 1, wherein each of said variably-controllable resistive-elements of said first and second bridges comprises a photoconductor and a light-emitting device projecting thereupon; and means for energizing each of these light-emitting devices independently and proportionally by the said charges of said first and second storage means, respectively.

4. An arrangement for symmetric amplitude control of complex waves having asymmetric oppositely poled peaks, said control being in steady state steps between succeeding peaks, and without causing peak wave distortion, the arrangement comprising the following: a pushpull amplifier having first and second push-pull admittance inputs, and first and second gain-controlling inputs, respectively, and an output; means for generating said complex waves, and means for applying these waves to said first and second admittance inputs in push-pull mode, for amplifying said waves at said output; first and second storage means, and first and second rectifier means connected in series with said first and second storage means, respectively; first and second normally non-conductive discharger means connected in parallel with said first and second storage means, respectively; a phase shifting means coupledto said output, the last said means having first and second output-terminals; a reference amplitude level determining means for said phase shifting means; coupling means between said first and second output-terminals and said first and second rectifiers, respectively, for unidirectional charges of said storage means, when the peak amplitudes of said phase shifted peaks are above said reference level; first and second peak selectors for deriving from said generated complex waves first and second signal-waves of said oppositely poled peaks at time periods substantially coincident with the respective peaks appearing at said output-terminals of the phase shifting means; degenerative coupling means between said first and second storage means and said first and second gain-controlling inputs, respectively, for gain compression of said amplifier; and means for applying said first and second signalwaves to said first and second dicharger means, respectively, for rendering said disch-arger means conductive and readjusting the charges in said storage means to the instant levels at said output-terminals for the phase shifting means, thereby eifecting said stepwise symmetric amplitude equalization at said output, and without peak wave distortion by virtue of gain readjustments at time shifted periods than the original peaks arriving at said output.

References Cited by the Examiner UNITED STATES PATENTS 2,958,047 10/ 1960 Kalfaian 330144 ROY LAKE, Primary Examiner.

R. P. KANANEN, N. KAUFMAN, Assistant Examiners.

Claims

1. AN ARRANGEMENT FOR SYMMETRIC AMPLITUDE CONTROL OF COMPLEX WAVES HAVING ASYMMETRIC OPPOSITELY POLED PEAKS, SAID CONTROL BEING IN STEADY STATE STEPS BETWEEN SUCCEEDING PEAKS, AND WITHOUT CAUSING PEAK WAVE DISTORTION, THE ARRANGEMENT COMPRISING THE FOLLOWING: A PUSH-PULL AMPLIFIER HAVING FIRST AND SECOND BRIDGE CIRCUIT INPUT AND AN OUTPUT; EACH OF SAID FIRST AND SECOND BRIDGE CIRCUITS COMPRISING A CENTER TAPPED IMPEDANCE MEANS HAVING TWO END TERMINALS CONNECTED TO THE TWO END TERMINALS OF A SERIES-CONNECTED VARIABLY-CONTROLLABLE RESISTIVE ELEMENT AND A FIXED-VALVE RESISTOR, RESPECTIVELY, FOR FORMING SAID FIRST AND SECOND BRIDGE INPUTS BY COUPLING THE JUNCTION TERMINALS BETWEEN SAID FIXED-VALUE AND VARIABLYCONTROLLABLE RESISTIVE ELEMENTS OF THE SAID FIRST AND SECOND IMPEDANCE MEANS TO THE SAID FIRST AND SECOND INPUTS, RESPECTIVELY, FROM COMMON PARALLEL-CONNECTION OF SAID CENTER TAPS; MEANS FOR GENERATING SAID COMPLEX WAVES, AND MEANS FOR APPLYING THESE WAVES IN PUSH-PULL MODE TO SAID FIRST AND SECOND IMPEDANCE MEANS, FOR AMPLIFYING SAID WAVES AT THE OUTPUT OF SAID AMPLIFIER; FIRST AND SECOND STORAGE MEANS, AND FIRST AND SECOND RECTIFIER MEANS CONNECTED IN SERIES WITH SAID FIRST AND SECOND STORAGE MEANS, RESPECTIVELY; FIRST AND SECOND NORMALLY NON-CONDUCTIVE DISCHARGER MEANS IN PARALLEL WITH SAID FIRST AND SECOND STORAGE MEANS, RESPECTIVELY; A PHASE SHIFTING MEANS CONNECTED TO SAID OUTPUT, THE LAST SAID MEANS HAVING FIRST AND SECOND OUTPUT-TERMINALS; A REFERENCE AMPLITUDE LEVEL DETERMINING MEANS FOR SAID PHASE SHIFTING MEANS; COUPLING MEANS BETWEEN SAID FIRST AND SECOND OUTPUT-TERMINALS, AND SAID FIRST AND SECOND RECTIFIERS, RESPECTIVELY, FOR UNIDIRECTIONAL CHARGING OF SAID STORAGE MEANS, WHEN THE PEAK AMPLITUDES OF SAID PHASE SHIFTED PEAKS ARE ABOVE SAID REFERENCE LEVEL; FIRST AND SECOND PEAK SELECTORS FOR DERIVING FROM SAID GENERATED COMPLEX WAVES FIRST AND SECOND SIGNAL-WAVES OF SAID OPPOSITELY POLED PEAKS AT TIME PERIODS SUBSTANTIALLY COINCIDENT WITH THE RESPECTIVE PEAKS APPEARING AT SAID OUTPUT-TERMINALS OF THE PHASE SHIFTING MEANS; MEANS FOR APPLYING THE FIRST AND SECOND SIGNAL WAVES TO SAID FIRST AND SECOND DISCHARGER MEANS, RESPECTIVELY, FOR RENDERING SAID DISCHARGER MEANS CONDUCTIVE AND RE-ADJUSTING THE CHARGES IN SAID STORAGE MEANS TO THE INSTANT LEVELS AT SAID OUTPUT-TERMINALS OF THE PHASE SHIFTING MEANS; MEANS FOR TRANSLATING THE CHARGES OF SAID FIRST AND SECOND STORAGE MEANS INTO CONTROLLING MEANS; AND MEANS FOR VARYING SAID VARIABLY-CONTROLLABLE RESISTIVE ELEMENTS OF SAID FIRST AND SECOND BRIDGES BY SAID FIRST AND SECOND CONTROLLING MEANS, RESPECTIVELY, FOR SYMMETRICALLY COMPRESSING THE ADMITTANCE OF THE OPPOSITELY POLED PEAKS OF SAID GENERATED COMPLEX WAVES TO SAID OUTPUT, THEREBY EFFECTING SAID STEPWISE SYMMETRIC AMPLITUDE EQUALIZATION AT SAID OUTPUT, AND WITHOUT PEAK WAVE DISTORTION BY VIRTUE OF GAIN READJUSTMENTS AT TIME SHIFTED PERIODS THAN THE ORIGINAL PEAKS ARRIVING AT SAID OUTPUT.