CA1162484A

CA1162484A - High efficiency, light weight audio amplifier and power supply

Info

Publication number: CA1162484A
Application number: CA000342984A
Authority: CA
Inventors: Robert W. Carver
Original assignee: Individual
Current assignee: Individual
Priority date: 1979-04-05
Filing date: 1980-01-03
Publication date: 1984-02-21

Abstract

Abstract The present invention comprises an amplifier circuit and transformer based power supply wherein greater efficiency is achieved by using the input signal characteristics to control various aspects of the circuit operation. The transformer primary winding is energized by a pulsed power supply which is duty cycle modulated in response to the signal being amplified.
One embodiment of the amplifier employs output transistors connected to respective stepped voltage levels. Amplifier con-trol circuitry acts in relation to the input signal amplitude to more evenly distribute the voltage drop across the inter-connected transistors, thus reducing amplifier power requirements and minimizing distortion in the amplifier output. In another embodiment, a stereo amplifier constructed in accordance with the present invention contains a network for inverting the audio signals received by a first amplifier input channel. The inverted signals are thereafter processed in out-of-phase relationship to the signals in the second amplifier input channel to more efficiently utilize the power supply. In yet another embodiment, the power supply itself is constructed with a number of fault detecting circuits which sense fault conditions in either the power supply or the amplifier and shut the power off accordingly.

Description

Description High Efficiency, Light Weight Audio Amnlifier and Power Su~lv .. ~ . .

Technical Field This invention relates to methods for audio signal amplification and to audio amplifier circuitry and power supplies therefor.

Background Art Solid state circuit components have brought incredible reduction in the size, weight and cost of audio amplifier circuitry and have also achieved increased fidelity in sound reproduction as compared ~ with vacuum tube technology of a prior-generation.
In an attempt to exploit to the limit the potential of solid state circuitry, audio engineers have striven to provide the user with increased power ratings while simultaneously achieving decreased distortion levels. Their efforts have met with resounding success, but have produced some undesirable side effects primarily in the areas of increased weight, cost and power consumption. For example, a-commer-cially available state-of-the art 400 watt ampli-fier typically weighs anywhere from 16 kg to over 38 kg depending upon the particular design and choice of materials. Such amplifiers normally employ costly components necessitated by the peak loads which they must carry, and generate significant amounts of heat which must be dissipated to avoid component damage.
With regard to the transformer weight problem an obvious approach would be to reduce the number of windings and/or the gauge of the wire making up the transformer coils. However, reduction in the number of windings also reduces the inductance in the primary coil, thereby increasing idling currents through the coil and contributing to both heat gen-eration and increased power consumption. The conven-tional method for achieving low idling currents in the primary has been to use a large number of wind-ings. This approach also requires a large number of windings in the secondary to keep the voltage in the secondary at the proper level. The other obvious alternative for weight reduction, i.e. reduc-tion in the wire gauge, is not an acceptable solu-tion since the internal resistance of each coil would be increased, leading to excessive heat generation and power loss upon high power demands being placed on the transformer. ~onventional wisdom has thus taught the necessity of increasing the size and weight of the transformer whenever a transformer powered amplifier is redesigned for increased power rating.
An alternative approach for reducing ~he overall weight, size and cos~ of audio amplifiers has been to reduce the total input power requirement without decreasing output power capability~ Such increases in amplifier efficiency permit the use of less costly, lower weight power supplies, and can be achieved by reducing the power dissipation which normally attends the conventional use of output transistors in the output stage of the amplifier. When such power dissipation decreases are achieved, additional weight and cost savings are realized beyond those realized in the power supply since the weight~ size and cost of the heat sinks normally required b~ the output transistors in the amplifier ma~ also be 4~4 reduced.
U.S. Patent No. 3~426,290 to Jensen is represen-tative of one known approach for increasing amplifier efficiency by keeping the voltage supplied to the ..
output transistor of the amplifier very close to the output voltage level, thereby permit~ing opera-tion of the output transistor in a condition which is at all times only slightly out of saturation.
When operated in this condition, the actual voltage drop across the output transistor will be maintained quite low and the power dissipated by the transistor (equal to voltage across the transistor X current through the transistor) will be correspondingly reduced. A rather complex regulator is employed in the Jensen circuit to maintain the desired voltage supply to the output transistor wherein energy is stored in an inductive capacitive circuit by means of a switching transistor operated at high speed in response to a control signal derived from the audio input signal. By operating the switching transistor in full "on" or full "off" condition to maintain the desired voltage supply to the amplifier output transistor, energy consumption by the combined regulator and output transistor is reduced over that which would be consumed by an output transistor operated with a conventional fixed supply voltage~
While producing a decided advantage in amplifier efficiency, the Jensen circuit is only truly effec-tive if the switching transistor is operated at high frequencies, which can in turn cause transient in terference distortion in the amplifier output signal.
~.S. Patent No. 4,054,843 to Hanada discloses a i similar circuit to that disclosed in Jensen.
An alternative approach to achiev;ng improved 4~4~

amplifier efficiency is disclosed in the patent to Dryden (3,319,175) which discloses a stepped voltage supply operated in response to the voltage level of the amplifier output whereby the minlmum voltage from the available power supply voltages sufficient to achieve the desired amplification is applied across the power amplifying elementO While useful for the purposes disclosed, Dryden employs only a single transistor as the power amplifying element for each polarity of the output voltage and thus the entire difference between the load voltage and the connected supply voltage appears across the output transistor. Significant power losses will thus occur unless a large number of discrete supply voltages are provided by the power supply circuitry.
Each such discrete voltage requires a separate ampli-tude comparator and associated switching device thusadding significantly to the cost of the power supply.
Still another approach disclosed in the prior art is illustrated in U.S. Patent No. 3,622,899 to Elsenberg. In this patent a low power dissipation amplifier circuit is disclosed including plural transistors coupled in series to a load terminal wherein the transistors are energized by respective voltage sources having different magnitudes and wherein the transistors are biased to operate as amplifiers in sequence in response to an input signal of increasing magnitude. This type of circuit causes each output transistor to be driven into saturation as the next higher voltage output transistor is brought into operation, causing substantially the entire voltage drop in the amplifier output stage I (that is the difference between the supply voltage and the load or output voltage) to appear across only a single output transistor at any one time.
This arrangement of circuitry requires output tran-sistors having substantial power ratings unless a relatively ~arge number of output transistors and discrete supply voltages are provided. Either ap-proach will add to amplifier cost. The patents to Woehner (3,772,606) and to Sampei et al. (3,961,280) disclose circuit arrangements similar to that de-scribed above with reference to the Elsenberg patent.
~he patent to Schade, Jr. (3,887,878) discloses a transistor series amplifier wherein plural series connected transistors in the output stage are biased to share the total voltage drop in the output stage to permit use of lower cost components. ~owever, this patent fails to disclose a technique for re-ducing the total power dissipation in such transis-tors.
Still other techniques for reducing the cost of amplifier power supplies have been disclosed in the prior art. For example in the U.S. patent to Munch, Jr. (3,542,953) a technique is disclosed wherein a single power supply may serve two Class B amplifier circuits designed to amplify the same audio signal by phase inverting the input to-one amplifier to cause the amplifiers to draw peak cur~
rent from the power supply in alternation. Munch, Jr. does not, however, suggest how such a technique can be employed in a system employing dual amplifiers (such as in a sterophonic system) for amplifying two separate signals.
None of the prior art systems discussed above addre~sses directly the problem of reducing power I supply weight and costs by modifying the supply itself in a manner to employ less costly lighter, weight components while maintaining the power supply capabilities required by the amplifier circuit.
The patent to Chun (3,466,527) discloses a .~ circuit for reducing the cost and size of a trans-former based voltage supply circuit including a duty cycle controlled switch in the A.C~ power supply circuit of the transformer primary. The switch functions to regulate output voltage from the sec-ondary. However, the lower cost and weight capabil-ity achieved by the concepts disclosed in Chun arederived by operating the duty cycle controlled switch over only a quarter cycle volt-time integral and do not in any way suggest how such a circuit design could be employed in an audio amplifier circuit in lS a manner to obtain power supply weight and cost reductions based on the characteristics of the in-coming audio signals.

Disclosure of the Invention It is a general object of this invention to overcome the difficiencies of the prior art by pro-viding an amplifier circuit and power supply of signi-ficantly reduced weight and cost achieved by simul-taneously increasing the efficiency of the ampli-fier output stage and modulating the energization of the amplifier power supply in response to a charac-teristic of the signal being amplified.
Another object of this invention is to provide a transformer based power supply for an ampliier in which the transformer primary is energized by a pulsed supply which is duty cycle modulated par-tially in response to the signal being ampliied.
The duty cycle is controlled in a manner to insure that adequate power is available in the power supply output energized by the transformer secondary while at the same time minimizing the time during each cycle when idling currents are flowing through the primary winding of the transformer.
The above objects are achieved in one specific embodiment by providing means to cause current pulses to be transmitted to the primary winding of the trans-former and circuit means responsive to the signal being amplified in a manner to control the power of the pulses transmitted to the primary winding.
This is done to increase and decrease power of the pulses as the signal increases and decreases in amplitude, so that the power means provides to the amplifier means power of a magnitude related to power requirements of the amplifier means to provide an output corresponding to the signal. In one embodi-ment, the pulses may be formed by a pulse generator and the circuit means may include switch means op-eratively connected to the primary winding of the transformer to interrupt current to the primary winding. There are other means to make the switch means non-conductive at periodic intervals to cause the current pulses to flow through the transormer.
The circuit means further comprises modulating means to control the power of the current pulses to the transformer primary winding. This is desir-ably done by controlling the duration of open periods of the switch means in a manner that pulses of shorter duration pass through the primary winding when the amplitude of the signal is relatively small, and pulses of greater ~uration are delivered through the primary winding of the transformer when the - amplitude of the signal is relatively large.
The pulse generator based embodiment may also include comparator means responsive to a first value related to amplitude of the signal and to a second value related to power then ava;lable from the trans-former for the amplifier means. The comparator is arranged to provide a control signai of a value related to a difference ~etween the first and second values `to provide an output of a magnitude corre-sponding generally to the amplitude of the signal.
The first value is a voltage, the magnitude of which increases and decreases with the amplitude of the signal. The second value is a voltage rela~ed to the voltage at an output end of said transformer.
Desirably, there is an absolute value detector ar-ranged to receive the signal and to produce a direct current output having a magnitude related to the amplitude of the signal. The absolute value detector means is arranyed to transmit its direct current output to the comparator as said ~irst value. Prefer-ably, there is also a non linear peak detector ar-ranged to receive the direct current voltage fromthe absolute value detector and produce an output voltage for the comparator. The output voltage of the non-linear peak detector corresponds to the output from the absolute value detector. The non-linear peak detector is more responsive to relativelyrapid variations in amplitude in the signal and less responsive to relatively small variations in ampli tude of the signal. There is another comparing means operatively connected to two ends of the secondary winding of the transformer to produce a summing value related to a difference in absolute magnitude of the voltages at the ends of the transformer secondar~
i winding. This other comparing means is arranged to produce a second control voltage related to the summing value. The comparator means and the other comparing means are arranged to transmit a control output which is responsive to relative magnitude of the first control cueput signal from the compara-tor and the second control signal from the othercomparing means. The output is arranged to control power of current pulses in the primary winding of the transformer.
The pulse generator employed in the pulse gen-erator embodiment generates control pulses at a pulse frequency which is sufficient to cause each pulse cycle to have a duration no greater than the minimum response period required of the amplifier~ The pulses are transmitted to switch means to turn the lS switch means "on" and "off" at the pulse frequency to cause current pulses to flow through the primary winding of the transformer. In this manner, the power output of the transformer may be rendered more respon-sive to major amplitude variations of the signal~
In an alternative embodiment of the transformer based power supply of this invention, the primary of the transformer may be connected with a source of alternating current combined with a switch means arranged to cyclically interrupt the alternating current in response to a control signal produced by control means operatively connected to the switch means to cause the switch means to be conductive at selected portions of current cycles from the power input terminal means. The control means causes the switch means to be conductive for shorter periods of time for lower power requirements of the amplify-ing apparatus, and to be conductive for longer periods of time for higher power requirements of the ampli-Ey i ng appar at u s . Des i r ably, I he con tr ol me sns cau se s ~L6;~

the switch means to be conductive during a latter portion of each current pulse from said power input terminal. In one embodiment, there is a rectifying . means operatively connected with the power input terminal means and the primary winding to cause only positive current pulses to be directed to the primary winding. The switch means comprises voltage respon-sive switch means which becomes conductive at pre-determined voltage levels of the current pulses.
In the-preferred form of this embodiment, the switch means comprises a silicon controlled rectifier con-nected in series between the power input termina~
means and the primary winding of the transformer.
In another embodiment of the transformer based power supply, the power input terminal means is con-nected to the primary winding to cause alternating current to be delivered to the primary winding.
The switch means is a voltage responsive means to cause the switch means to be conductive at predeter-mined voltage levels during latter portions of thecurrent pulses. In the preferred form of this addi-tional embodiment of the power supply, the switch means comprises a first triac connected in series with the primary winding. Operation of the--first triac is governed by circuitry which responds to both the magnitude of the signal being amplified and the value of the voltage output from the trans-former secondary. The triac is caused to fire during precisely defined portions of the duty cycle, thereby regulating the amount of current flow through the transformer primary and the corresponding transfer i of energy to the transformer secondary.
I In another form of the transformer based power I supply, the triac switching means discussed above 4a~3~

further includes cut off circuitry for shutting off current flow through the primary of the transformer before the normal waveform of the alternating current supply has returned to zero. Such cut off circuitry may include a second triac which operates to com-mutate the first triac into a non-conductive state~
In the method of the present invention/ a series of current pulses are directed through a primary winding of a transformer to cause voltage pulses to be imposed on a secondary winding of the trans-former. These voltage pulses are in turn transmitted to power input terminals of an amplifier means.
The method further comprises controlling power of the pulses transmitted to the primary winding to increase and decrease power of the pulses as the signal increases and decreases in amplitude. In this manner, power delivered to the amplifier means is matched to the power requirements of the ampli-fier means, thus providing an output corresponding to the signals being amplified. Preferably, this is done by controlling the duration of the pulses delivered to the primary winding. A control signal is generated by comparing a first value related to amplitude of the signal and a second value r-elated to power then available from the transformer.
With respect to one embodiment of the method, power is supplied to the amplifying apparatus at a relatively constant voltage. In one form of this method, there is transmitted a series of first direct current pulses through a primary winding of a trans-former. The primary winding of the transformer is caused to be conductive at pre-selected time periods ! during latter portions of each pulse, so as to cause i current pulses to flow through the primary winding and cause a ma~netic field to build up around the transformer. While the magnetic field is building up, substantial current flow in the secondary wind-ing of the transformer is prevented. The secondary winding of the transformer is caused to bë conduc-tive after each first current pulse in the primary winding, so as to cause second current pulses to flow in the secondary winding. These second current pulses are directed to power output terminal means.
The first current pulses are controlled in response to power requirements at the output terminal means in a manner to cause first current pulses of greater power to flow during periods of higher power require-ments at the output terminal means, and to cause first current pulses of lower power to flow during periods of lower power requirements of the output terminal means. Power of the first current pulses is controlled by utilizing a silicon controlled recti-fier in series with the primary winding. The silicon controlled rectifier is made conductive at higher voltage levels when greater power is needed, and is made conductive at lower voltage levels when less power is needed.
In accordance with a second form of the-method, power at a substantially constant voltage is supplied by means of alternating current pulses transmitted to the primary winding of the transformer. The pri-mary winding of the transformer is caused to be con-ductive during latter portions of each half cycle of the pulses, so that positive and negative current pulses flow through the primary winding to cause corresponding positive and negative current pulses ! to flow through the secondary winding of the trans-i former. This form of the method further comprises ~6~34 rectifying the positive and negative current pulses from the transformer to transmit positive current to a positive output terminal and to transmit nega-tive current pu~ses to a negative output terminal.
Further, the periods of conductivity for the primary winding are controlled in a manner such that the primary winding becomes conductive at higher and lower voltage levels of the alternating current pulses delivered to the transformer, thereby causing current pulses of greater or lesser power to be transmitted throuyh the primary winding, in response to a power requirement of the output terminals.
A triac may be utilized by connecting it in series with the pr;mary winding. When used, the triac is caused to be conductive at higher voltage levels in response to greater power requirements at the output terminals, and to become conductive at lower voltage levels during periods of lower power require-ments at the output terminals.
Still another object of this invention is to provide a high efficiency audio amplifier including series arranged transistors connected to respective stepped voltage levels wherein the transistors are controlled in a manner to more evenly distr-ibute the voltage drop over the interconnected transistors, thus reducing power rating requirements and resulting in less distortion in the amplifier output.
In one embodiment of the audio amplifier, a plurality of series connected output transistors are connected in emitter follower configuration to the output of the amplifîer. The transistor closest ; to the output is connected through its base to signal input means. The remaining transistors are controlled through their base electrodes by transistor control ~6~4 means designed to cause the series connected transis-tors to be non-conauctive under conditions where said amplifying apparatus is amplifying a signal having an amplitude below a predetermined magnitude.
` 5 The transistor control means causes the second tran-sistor to become conductive under circumstances where the input signal is of a higher magnitude so that current at a higher voltage is delivered to the output terminal, with the result that current flows from the higher voltage point to the second and first transistors to the output terminal. In the preferred form, the transistor control means is responsive to output voltage from the first transistor to the load terminal. Further, the transistor control means is operatively connected to a base electrode of the second transistor in a manner that control current to the base electrode of the second transistor begins to flow~when the output voltage reaches a predeter-mined voltage level, so as to cause the s~econd tran-sistor to be conductive and cause current to flowfrom the higher voltage point through said first and second transistors to the output terminal.
Desirably, the transistor control means is further characterized in that the~first control means sup- .
plies base current to the base electrode of the second transistor according to a functional relationship of the output voltage at the output terminal. This is done in a manner that voltage of the current sup-plied to the base electrode of the second transistor varies as a function of magnitude of the output volt-age, with the voltage of the base current to the second transistor being a voltage level intermediate j the voltage at the higher voJtage point and the out-put voltage, with the result that voltage drop across 4~3~

~15-the second and first transistors is shared between the second and first transistors.
Specifically, the transistor control means com-prises a control transistor having a first main cur-rent carrying electrode connected to the base elec-trode of the second transistor, and a second main current carrying electrode connected to voltage divid-ing means connected between said output terminal and a related higher voltage source. The base elec-trode of the control transistor is connected to ajunction point between a pair of voltage dividing resistors, which in turn are connected between higher and lower voltage sources.
There is diode means inter-connecting the lower voltage point with the second electrode of the first transistor, so that when the signal voltage is at a level higher than a voltage level of the lower voltage point, the lower voltage point is blocked off from the higher voltage point.
In the preferred configuration, there are two sets of transistors arranged in push pull relation-ship, with at least two transistors being in each set. One set of transistors conducts during positive portions of the signal voltage, and the other set conducts during negative cycle portions of the signal voltage. The first set is connected to greater and lesser positive voltage points on the secondary wind-ing of the transformer, while the other set of tran-sistors is connected to greater and lesser negative voltage points on the secondary winding of the tran-sistor. The first and second control means functions in substantially the same manner as indicated pre-¦ viously herein.
; In the method of the present invention, a siynal 4~

is amplified by directing the signal to a base elec-trode of a first transistor, having a first main current carrying electrode connected to a load ter-minal, and a second main current carrying electrode connected to a lower voltage source~ Current is caused to flow from the lower voltage source through the first transistor during periods when amplitude of the signal is within a lower predetermined range.
During periods when the signal is within a higher predetermined range, a control current is directed to a base electrode of a second transistor. This second transistor has a ~irst main current carrying electrode connected to the second main current carry-ing electrode of the first transistor, and a second main current carrying electrode connected to a higher voltage source.
Said method is further characterized in that the base current directed to the base electrode of the second transistor is at a voltage level inter-mediate a voltage at the higher voltage terminal - and voltage at the load terminal. Thus, voltage drop across the second and first transistors is shared by the second and first transistors.
In still another embodiment of the subject inven-~5 tion, the audio amplifier includes a primary outputtransistor the base of which is responsive to the audio signal and the emitter collector circuit of which is connected between a first supply voltage and the amplifier output. Second and third tran-sistors have their emitters connected through diodesto the collector of the first transistor and have their colle~ctors connected, respectively, to second I and third supply voltages which are hiyher than the , first supply voltage. In this embodiment control ~6~t34 circuit means are provided responsive to the ampli-fier output to cause the second and third transistors to sequentially commence conducting in response to the output voltage from the amplifier exceeding the first and second suppiy voltages, respectively.
By this arrangement the voltage drop within the output stage of the amplifier occurs in one of the three ways as follows:
(a) solely across the first transistor, (b) solely across the first and second transistors, or (c) solely across the first and third transistors.
An additional feature of this invention is to provide a transformer based power supply circuit for, an audio amplifier in which the transformer primary coil is energized by an A.C. current duty cycle modu-lated by a solid state switch which is, in turn, controlled by a phase shift network wherein the amount.
of phase shift is a function of the power supply output voltage and of the audio signal being ampli-fied~ The phase shift network may be connected by means of a light photon coupled communication link to control circuitry responsive to the output voltage of the power supply and a signal which tracks the audio signal being amplified. By this arrangement, the duty cycle of the A.C. signal applied to the transformer primary may he modulated to cause.the amount of current flowing through the primary to be adjusted to be just sufficien~ to satisfy the power needs of the amplifier, thereby substantially reducing primary coil idling currents. In a pre-ferred embodiment, the power supply circuit may be j equipped with an automatic shut down capability in I . response to any one of the following conditions:

over current or over voltage in the arnplifier output or a direct current fault in the amplifier circuitry~
In yet another embodiment of this invention, the duty cycle controlled power supply circuit is desirably equipped with a transformer having a sec-ondary to primary turns ratio below 1.0 with a primary induction above 30 millihenries and a coil wire gauye diameter above no. 18 when the transformer i5 used to produce approximately 1000~ watts at a maximum + 75 volt D.C. output from conventional 117-125 volt, 60 cycle alternating currentO
Still other and more specific objects of this invention will become apparent from a consideration of the following Brief Descr.iption of the Drawings and Best Mode for Carrying Out the Invention.

Brief Description of the Drawinqs Figure 1 is a schematic diagram illustrating the basic concepts of the~present invention;
Figure 2 is a block diagram of one embodiment of a power supply and amplifier designed in accor-dance with the present invention;
Figure 3 is a circuit diagra~. of the power supply and amplifier illustrated in Figure -2; -Figure 4 is a modification of the switching circuitry for controlling the operation of the power supply illustrated in Figure 3;
Figure 5 is a circuit diagram of a stepped voltage power supply constructed in accordance with the present invention, Figure 6A, 6B and 6C illustrate various current cycle waveforms flowing through the primary winding ¦ of the power supply of Fiyure 5;
i Figure 7A is an alternative embodiment of a g~

stepped voltage power supply constructed in accor-dance with the present invention;
Figure 7B is a~first modification of the switch-ing circuitry for controlling the operation of the stepped voltage power supply of Figure 7A;
Figure 7C is a second modification of the switch-ing circuitry for controlling the operation of the stepped voltage power supply of Figure 7A;
Figure 8A through 8C illustrate various current cyle waveforms flowing through the primary winding of the power supply of Figure 7A;
Figure 9A through 9C illustrate various current cycle waveforms flowing through the primary winding of a power supply switched in accordance with the circuitry of Figures 7B or 7C;
Figure lOA is a circuit diagram of a conven-tional power supply and audio amplifier;
Figure lOB is a diagram of typical conduction periods of a conventional transformer secondary of the type illustrated in Figure lOA;
Figure lOC is a diagram of typical conduction periods of a transformer secondary constructed in accordance with the present invention;
. Figure lOD is a diagram of the peak load con- -duction periods of a transformer secondary constructed in accordance with the present invention;
Figure 11 is an alternative embodiment of-a push-pull amplifier constructed in accordance with the present invention and adapted to receive stepped voltages from the power supply of the present in-vention;
Figures 12A, 12B and 12C are diagrams o the voltage drops across the push-pull transistors em-ployed in the amplifier of Figure 11;

Figure 13 is another embodiment of an amplifier constructed in accordance with the present invention;
Figure 14 is a diagram of the voltage output of the amplifier illustrated in Figure 13;
Figures 15A and 15B are circuit diagrams of a preferred embodiment of the left channel of an amplifier constructed in accordance with the present invention;
Figure 16 illustrates a portion of the right channel of an amplifier constructed in accordance with the present invention; and Figure 17 is a circuit diagram of a power supply for the preferred embodiment of the present invention, wherein various safety control features have been incorporated into the circuit.

Best Mode For Carrying Out the Invention Referring to Figure 1, a highly schematic diagram of an amplifier circuit 2 and power supply circuit 4 designed in accordance with the subject invention is illustrated. Electrical energy is supplied to the system by means of an oscillating power supply 6 which may take the form of a pulse generator or a source of commercially available current-such as conventional 117-125 volt, 60 cycle alternating current. The primary coil 8 of a specially designed, light weight transformer 10 is connected to the oscillating power supply 6 by means of a duty cycle control circuit 12 designed to modulate the amount of energy suppiied to primary coil 8 in at least partial dependence upon a characteristic of the audio signal which is to be amplified by the system for I conversion into sound by speaker 14. The audio ~ signal characteristic may be supplied directly over 4~

the audio input signal conductor 16 or may be pro-vided through a feedback line 18 connected with the audio amplifier output 20. As will be explained in greater aetail hereinbelow, the provision of a duty cycle control 12 in the supply circuit to the primary coil 8 results in the possibility of employ-ing a much lighter weight transformer than has beenthought to be necessary heretofor.
- The transformer secondary coil 22 is connected to a rectifier and amplifier power supply 24 designed to provide a direct current supply voltage to ampli-~ fier 2 which may take the form of a voltage which varies widely in amplitude in dep~ndence upon ampli-tude changes in the input audio signal or the form of a relatively constant output voltage. If the later type of amplifier power supply is employed, a voltage-level feed back line 26 maybe employed to cause the duty cycle control 12 to adjust the amount of power supplied to the primary coil 8 in order to assist in maintaining constant the output voltage from amplifier power supply circuit 24.
As will also be described in greater detail below, amplifier 2 has been specially designed to permit the use of low cost, low power rated-trans istor components. When designed in accordance with this invention, the amplifier may be operated in a manner to minimize the amount of heat which must be dissipated by the heat sinks in the output stage of the amplifier. This operation permits further reduction in the weight of the amplifier and power system by permitting the use of smaller, lighter weight heat sinks than have previously been required by amplifiers providing comparable power output.
In particular the amplifier, which may be of the ,,. ., . ~

Class B push-pull type, includes at least a pair of output transistors 30 and 32 for amplifying the audio signal supplied on line 16 to the respective base electrodes thereof. In one embodiment of this invention amplifier 2 includes additional transistors 34 and 36 connected in series with transistors 30 and 32, respectively, to provide higher absolute supply voltage levels to the amplifier output 20 when necessary~ When these higher voltages are unnecessary, transistors 30 and 32 remain non-con-ductive as determined by supply voltage control means, 38 and 40, respectively~ When transistors 34 and 36 are non-conductive, voltage is supplied to the collectors of transistors 30 and 32 through diode5 42 and 44, respectively, which are in turn connected to lower voltage level taps 46 and 48, respectively, of amplifier power supply 24. To further reduce the power rating requirements for transistors 30 through 36~ control means 38 and 40 are designed to cause the voltage drop across the transistor pairs 30~ 34 and 32, 36 to be equally shared wherever transistors 34 or 36 are conductive, respectively. The manner by which this is achieved and the beneficial results that flow therefrom will be discussed in greater detail below.
~ eference is now made to Figure 2 which is a block diagram illustrating one embodiment of the ~present invention. There is a speaker 100 which is driven by an amplifier 102, which in turn derives its power from a transformer 104, having primary and secondary windings 104a and 104b, respectively.
The speaker 100 and amplifier 102 are or may be o~
I conventional design. As shown herein, the amplifier , 102 comprises a pair of transistors 106 and 108 .

having signal input terminals 110 and 112, respec-tively, through which the audio signal is fed into the amplifier 102. Positive portions of the audio signal cause the transistor 106 to be conductive ` 5 while negative portions of the audio signal cause the transistor 108 to be conductive in a manner such that an output current corresponding to the audio signal is supplied to the speaker 100. In the par-ticular embodiment shown herein, the amplifier 102 is arranged to operate at maximum power output when plus 80 volts is applied to the transistor 106 and minus 80 volts is applied to the transistor 108.
For an appreciation of the significance of the present invnetion, attention is now directed to the transformer 104. In a conventional power amplifier (e.g. a 400 watt amplifier), the power transformer would weigh at least 9 kg. The reason for this is as follows. Current flow in the primary winding is equal to the voltage input multiplied by the time the voltage is applied, divided by the inductance of the transformer. For a time varying input volt-age, such as a 60Hz, 117 volt house current, analysis reveals that in order to keep the current in the primary winding (i.e. the magnetizing current-) within -a proper value, the inductance must be made rather large. This requires that a large, heavy transformer be used in a conventional high power amplifier.
In the embodiment of Figure 2, a power trans-former 104 can be made of a size which is a very small fraction of the size of a transformer of a conventional amplifier of comparable power output.
In this particular embodiment, the transformer 104 weighs only about ~2 kg, or 1/40th the weight of the transformer of typical prior art amplifiers of the same power ratiny. The reason for this phen-omenal reduction in transformer size is that the present invention permits the inductance of the trans-former 104 of the present invention to be made rela-tively small. When a voltage i5 applied across theprimary winding 104a of the transformer, the mag-nitizing current climbs rapidly. Within several microseconds the current will have reached a high value of approximately twenty amps or so, and at this time an electronically controlled switch 116 is opened. At this point in time, an amo~nt of energy is stored in the magnetic field surrounding the primary winding. This stored energy can be considered to be analogous to the energy stored in the electric field of a capacitor. The opening of the switch 116 causes the field to begin collapsing, - which causes the energy to be transEerred to the secondary winding to deliver the energy to the amp-lifier 102. By turning the switch 116 on and off at a relatively high frequency (i.e. 20 KHz), 20,000 voltage pulses are delivered each second to the amplifier 102.
The power delivered by the transformex 104 is controlled by controlling the time within which the switch 116 is open for each current pulse. This is accomplished by tracking the audio signal which is to be amplified by the amplifier 102 and then comparing this tracking signal to the voltage imposed across the amplifier 102. This produces a control signal which is u~ilized to control the duration of each current pulse delivered to the primary wind-I ing of the transformer 104. In other words, on the ¦ assumption that the switch 116 is being opened and ! closed at a frequency of 20 KHz, the duration of ~6~

~5 each period would be 50 microseconds. During thosetime periods where the power requirements o~ ampli-fier 102 are low, during each 50 microsecond periodthe switch 116 would be open for a relatively large., ' 5 fraction of that time (e.g. 25 to 35 microseconds).
When the power requirements of the amplifier 102 are relatively high, the switch 116 would be open in each period for a much shorter duration.
The audio signal which is to be amplified is directed into an absolute value detector 118. This signal can have both positive and negative portions.
The absolute value detector 118 provides an output when the negative portions of the audio signal become positive while maintaining these negative portions at the same magnitude relative to a zero line. The output of the absolute value detector 118 causes the negative portions of the audio signal to be in~
verted.
The output of the absolute value detector 118 is then directed to a non-linear peak detector 120 which is characterized in that it has a rapid re-sponse time for rapidly varying large signals, is less responsive to more slowly varying signals, and is virtually unresponsive to small signals~varying 2S about any arbitrary average level.
The output of the non-linear peak detector 120 is fed into a comparator 122. There is also a power output feedback 124 which is responsive to the volt-age impressed across the power input terminals of the amplifier 102. This power output feed back 124 ,transmits to the comparator 122 a voltage generally proportional to the voltage at the power input ter minals of the amplifier 102. The comparator 122 then "compares" the input from the non-linear peak detector 120 and the input from the power output feedback 124 to produce a control signal generally proportional to the difference between the two in-puts.
This control signal generally corresponds to the increment of increase or decrease in the dif-ference between the two input signals and is used to control the duration of the regularly timed cur-rent pulses in the primary winding 104a of the trans-former 104.
There is a pulse generator 126 which functions to generate a pulsed wave of a constant voltage,where the gaps between the pulses are of approxi-mately the same duration as the pulses themselves.
The pulses are of a same frequency as the desired current pulses for the transformer 104. In the particular embodiment described herein, where the frequency of the current pulses in the transformer are 20 KHz, the output from the pulse generator 126 would be of the same frequency.
The output from the pulse generator 126 is di-rected to a square wave to triangular wave converter 128. This converts the square wave form from pulse generator 126 to a wave form where each pulsé has the configuration of an isosceles triangle, where during the duration of each pulse, the voltage climbs at a substantially constant rate to a peak at the middle of the pulse, and then declines at a constant rate through the latter half of the pulse.
The output from the square wave to triangular wave converter 128 is transmitted to a ramp time modulator 130, ana this ramp time modulator 130 also - receives the control signal from the comparator 122.
The modulator 130 in effect "chops off n the upper L6f~

portion of the triangular wave form produced by the square wave to triangular wave converter 128.
The output of the ramp time modulator 130 is a constant voltage pulse signal having the same fre-quency as that of the pulse generator 126. The dura-tion of each pulse is directly proportional to the duration of the "unchopped" bottom portion of the triangular wave form from converter 128. Thus~ it can be appreciated that the duration of the pulses from modulator 130 are proportional to the magnitude of the control signal from comparator 122.
The pulses from ramp time modulator 130 are used to open and close the switch 116 in a manner that the switch 116 is closed during the duration of each of the puls~s from ramp time modulator 130.
Thus a pulse of relatively short duration produces a corresponding current pulse of a relatively small magnitude, since the current has such a very short time period to build up or l'ramp up". It can be appreciated that as the voltage pulses from modula-tor 130 increase in duration, the magnitude of the current pulses in the transformer primary winding 104a increase, correspondingly, in a manner that a pulse of longest duration from modulator 13-O pro- -duces a current pulse of the largest magnitude in the transformer primary winding.
Power may be suppIied to the primary by means of a conventional plug 132 and bridge rectifier 134 for converting regular 110-120 volt, 60 cycle current to direct current. The bridge rectifier 134 is con-nected to the upper terminal 136 of the transformer primary winding 104a. The lower terminal 138 of the transformer primary winding 104a is connected to the aforementioned switch 116, which in turn is ~6~

connected to ground. When ~he switch 116 is conduc-tive, direct current flows through the primary 104a.
The secondary winding 104b of the transformer 104 is center tapped at 140 to ground. The upper terminal 142 of the secondary winding 104b is connected to an upper positive output terminal 144 through a diode 146 which permits only positive current to be directed to the output terminal 144. In a similar manner, the lower terminal 148 of the secondary wind-ing 104b is connected to a lower negative output terminal 150, through a second diode 152 which permits only negative current pulses to be transmitted to the output terminal 150. There are a pair of shunt connected capacitors 154 and 156, one of which is connected to the positive output terminal 144 at a location between that terminal 144 and diode 146, and the other capacitor i56 being connected to the negative output terminal 150 at a location between that terminal 150 and diode 152. The other plates of the two capacitors 154 and 156 are both connected to ground.
In view of the foregoing description, the op-eration of the present invention may be understood in relation to the amplification of a typica-l audio signal, such as the audio signal of a musical com-position. This signal will be made up of some lower frequency oscillations (fundamental tones) on which are imposed any number of higher frequency oscilla-tions (overtones), with the amplitude of these os-cillations varying over a wide range (e.g. from the sound generated by full orchestra to the quiet sound ; of a single woodwind instrument playing a melodic theme). With regard to the amplitude variations in the signal, although these amplitude variations might seem to the listener to be in some cases very abrupt, in actuality the very sharp amplitude changes of any great magnitude would generally take place in a time period no less than about one thousandth .
of a second. For example, the rise time associated with the noise generated by a very sharp percussion, such as that generated by slapping two wood blocks together is generally greater than a thousandth of a second.
Assume that an audio signal is to be amplified by the circuit of Figure ~. This signal would be directed into the two signal input terminals 110 and 112 of the amplifier 102, and would also be di-rected to the absolute value detector 118. As indi-cated previously, this audio signal is converted by the absolute value detector 118 to a direct cur-rent wave form which in turn is transmitted to the non-linear peak detector 120 to provide a ~smoothed"
signal to the comparator 122.
Since upon initial start up, there is no voltage generated at the output terminals 144 and 150, the feedback signal provided by the power output feed-back 124 would be zero or substantially zero. Accord-ingly, the comparator 122 would generate a-rather strong output signal to the ramp time modulator 130.
The ramp time modulator 130 would in turn transmit pulses of the desired ~requency to the electronic switch 116~ with these pulses being of the maximum duration. In other words, the switch 116 would con-tinue to turn "on" and "off" at the same frequency, but the duration of the "on" periods would be at a maximum. Accordingly, the current pulses passing ! through the primary winding 104a would ramp up to maximum amperage and thus deliver full power to the ~6~4!t84 output terminals 144 and 1500 Within a very short period of time, (i.e. about two hundred microseconds) the voltages applied to the power input terminals 158 and 160 of amplifier.l02 would build up to the proper operating level.
At this time, the power output feedback 124, would transmit to the comparator 122 an output signal related to the voltage level at the power input ter-minals of the amplifier 102. ThereaEter, the compara~
lQ tor 1~2 would continue to provide to the ramp time modulator 130 a control signal related to the power requirements of the amplifier 102. In other words, when the comparator 122 receives an input signal from the non-linear peak detector 120 which indicates that the amplitude of the audio signal is increasing, there will be a greater disparity between this audio-related signal and the existing signal from the power output feedback 124 so that the voltage of the control signal to the ramp time modulator 130 increases.
This will in turn cause the current pulses through the primary winding 104a to increase in duration to deliver more power to the amplifier 102 and thus raise the voltages supplied at the output terminals 144 and 1500 On the other hand when the amplitude of the audio signal declines, the comparator 122 would detect that the difference in the signal from the non-linear peak detector 120 and that from the power output feedback 124 is smaller, so the control signal transmitted by the comparator 122 to the modu-lator 130 would be of a lower voltage. This wouldin turn shorten the duration of the current pulses through the primary winding 104a, thus delivering I less power to the amplifier 102. From the above I description it can readily be appreciated that com-parator 122 will in effect "track" the audio signal to maintain the voltage impressed upon the power input terminal 158 and 160 of the amplifier 102 so that these voltage levels are varied in such a manner that they remain only moderately above the power requirements of the amplifier 102. In actual prac-tice, there would generally be a voltage drop across each of the transistors 106 and 108 of approximately five volts. In the particular embodiment shown herein, the operating components of the invention are selected so that the maximum voltage which would be applied acxoss the terminal 158 and 160 of the amplifier 102 would be plus 80 volts and minus 80 volts.
As a further advantageous feature of the present invention, attention i5 directed to the two shunt connected capacitors 154 and 156. Inherent in the operation of the present invention is the feature that at any particular time, only a relatively small amount of power need be stored in-the two capacitors 154 and 156 to respond to rapid increases in the power requirements of the amplifier 102. The reason for this is that the power pulses through the primary winding 104a are of such a high frequency, and the response time to increase the duration (and--thus increase the power) of these current pulses in the transformer ]4 can occur in such a very short time, that the transformer 14 can respond in a matter of a fraction of a millisecond to begin delivering full power to the amplifier 102. Thus, these capacitors 154 and 156 can be made with reasonably small capa-city, thus creating a further savings in both weigh and expense.
! Another desirable feature of the present inven-tion is the efficiency achieved. Let it be assumed that the circuit constants have been arranged to cause a constant voltage of five volts across each of the amplifier transistors 106 and 108~ Let it further be assumed that the res;stance of the load (i.e. the speaker 100) is eight ohms. Then let us examine three situations:
1. Where the voltage applied to the amplifier 102 is plus 25 and minus 25 volts,

2. Where the applied voltage to the ampliier is plus 35 and minus 35 volts, and

3. Where the applied voltage is plus 45 and minus 45 volts.
The power output is equal to the voltage squared divided by the resistance. For the first situation (where the applied voltage îs plus 25 and minus 25 volts), there would be a drop of five volts across each transistor 106 and 108 when each is conducting, and a drop of twenty volts across the eight ohm load.
The actual power output would then be twenty squared divided by eight, which gives fifty watts. The loss at the transistors 106 and 108 would be 12 1/2 watts.
Therefore out of the total of 62 1/2 watts used (50 watts and 12 1/2 watts) fifty watts are actually utilized in the speaker, for an efficiency of 80%.
In the second situation (where the voltage ap-plied across the amplifier 102 is between plus 35 and minus 35 volts), the power delivered to the speaker would be equal to thirty squared divided by eight for 112 watts. The power dissipated in the transis-tors 106 and 108 would be 18.75 watts, which gives an efficiency of 85.7%.
In the third situation (where the voltage applied to the amplifier 102 i~ plus 45 and minus 45 volts), i 200 watts are actually utilized in the speaker 100 and only 25 volts dissipated in the transistors 106 and 108, for an efficiency of 88.8%.
In a conventional amplifier where the f~ll vol-tage of plus 80 volts and minus 80 volts is impressed across the amplifier power input termianls àt all times, the efficiencies for the three situations outlined above would be 25%, 37% and 50% respectively.
Since the efficiency of the apparatus of the present invention is quite high, a relatively small amount of energy is dissipated in the transistors 106 and 108. For this reason the heat sinks for these transis-tors 106 and 108 can be made relatively small. It has been found that with these various weight savings, a power amplifier can be built according to the present invention with a power rating of four hundred watts, and with a total apparatus weight of only 5.5 kg.
With reference to Figure 3, the circuitry o the embodiment of the present invention illustrated in Figure 2 will now be described in more detail.
The plug 132 is adapted to be connected to a conven-tional wall socket which produces 110 to 120 volts at 60 Hz. The plug 132 connects to the bridge recti-fier made up of four diodes D101, D103, D105-, and D107 arranged to transmit positive D~Co current to the primary winding 104a. A small capacitor 162 (fifty to two thousand microfarads) is connected between the output of the bridge rectifier 134 and ground to prevent the D.C. output from the bridge rectifier 134 from going to zero.
The absolute value detector 118 has two input terminals 164 and 166 which are arranged to receive a signal input from a stereo player unit. The signals ! transmitted to terminal 164 and 166 are in turn trans-mitted to an operational amplifier 168 through a set of four diodes D109, Dlll, D113 and D115. These diodes D109-D115 are so arranged that whenever there is a relatively s~rong signal at one input 164 and 166 and a relatively weak signal at the other input 164 and 166, the highest value will be transmitted to the operational amplifier 168. The highest nega-tive value will pass through the diodes D109 and D113 while the highest positive value will pass through the diodes Dlll and D115. As indicated previously, the output from the operational amplifier 168 is an output signal where the negative portions of the audio signal have been made positive.
This signal passes through a diode D117 of the non-linear peak detector 120. Signals of large value are within the operating envelope of the diodes Dll9 and D121 and thus are passed immediately with the capacitor 170 providlng a more constant output.
Smaller signal variations are largely blocked by resistor R101 and diodes Dll9 and D121.
The output from the non-linear peak detector passes through a resistor R103 to one input terminal of an operational amplifier 172 of the comparator 122. The input to the other terminal of the-opera-tional amplifier is from the power output feedback 124. Power output feedback 124 comprises two re-sistors R105 and R107 connected in series to the positive input terminal 158 of amplifier 102 and a thixd resistor R109 connected between ground and a location between the resistors R105 and R107.
These resistors R105-R109 step down the voltage from the level at the terminal lS8 to less than 15 volts, which is a voltage which the operational amplifier 172 is able to handle. The operational amplifier has a resistor Rlll to provide a feedback voltage.
- The output from the operational amplifier 172 is transmitted through a resistor R113 to the ramp time modulator 130. A aiode D123 is connected be-tween the location of resistor R113 and the modula-tor 130 to pass signals from the amplifier 172 above a certain value to ground.
The pulse generator 126 may be provided as one of a number of commercially available pulse genera torsr such as a Fairchild US 78540. As shown herein, this pulse generator comprises an operational ampli-fier 174 having a pair of Lesistors R115 and R117 provided as feedback loops. One input terminal 176 of the operational amplifier is connected to one plate of a capacitor 178, with the other plate of the capacitor 178 being connected to ground. The other terminal of the operational amplifier 176 is connected through a resistor Rll9 to ground.
The square wave to triangular wave converter 128 comprises a resistor R121j which receives the voltage pulses from pulse generator 126. The resistor R121 is connected to one plate of a capacitor 180, the other plate of which is connectea to ground.
Capacitor 180 converts the square-wave pulsed signal from generator 126 to a triangular shaped waveO
The triangular shaped wave is transmitted through one input terminal of an amplifier 182 which simply amplifies the triangular wave and transmits it to the ramp time modulator 130. The ampli~ier 182 has a feedback resistor R123 connected to the other input terminal thereof, and also a resistor R125 connected between the other input terminal and ground.
The ramp time modulator 130 may be a differen-tial amplifier of conventional design such as the TL0074 from Texas Instruments Corp. As shown herein, this modulator 130 comprises two transistors Q101 and Q103 which compare the two voltages directed to the two bases of the transistors Q101 and Q103;
the transistor having the lower input will conduct~
The emitters of the two transistors Q101 and Q103 are connected to a positive voltage source through the resistor R127. Thus, for purposes of illustra-tion, let it be assumed that the waveform from tri-angular wave converter 128 is imposed upon the transis-tor Q101 and the control signal from operational amplifier 172 in comparator 122 is imposed on tran~
sistor Q103. In this situation, those portions of the triangular shaped wave which are above the control signal voltage will cause transistor Q103 to conduct.
The collector of transistor Q103 is connected through a resistor R129 to a negative voltage terminal, with the resistance of R127 being substantially greater than the resistance of R129. Consequently, when transistor Q103 is non-conducting, the point 184 located between transistor Q103 and resistor R129 goes negative to cause a transistor Q105 to conduct, which in turn causes the transistor Q107 to be con-ductive and the transistor Q109 to be non-conductive.
Thus, the output from the ramp time modulator 130 is a series of pulses, with the duration of each pulse coinciding with the portion of the triangular shaped wave from converter 128 which is below the voltage control signal.
The switch 116 comprises a first transistor Qlll, the output of which is connected to the bases o~ two parallel connected transistors ~113 and Q115.
! Thus, when transistor Qlll becomes conductive, it causes the two transistors Q113 and Q115 to conduct, thus permitting a current pulse to pass through the primary winding 104a of the transormer 104. When transistor Qlll becomes non-conductive in response to the termination of the pulse from ramp time mod-ulator 130, transistors Q113 and Q115 also become non-conductive and switch 116 opens. The magnetic field created in primary winding 104a by the current pulse thereafter collapses, causing a transfer of energy between primary winding 104a and secondary winding 104b. In this manner, power is delivered to amplifier 102.
A modification of the circuitry of Figure 3 is shown in Figure 4. In Figure 4, there is shown only those portions of the circuitry of Figure 3 which are necessary for proper orientation o the components shown in Figure 4 to the other components of the present invention.
The reason for th~ modification of Figure 4 is that in some circumstances the voltage at the negative input terminal 160 of amplifier 102 may have an absolute magnitude that is too low relative to ~he voltage at the positive terminal 158. In this situation, it would be desirable to override the signal generated by the operational amplifier 172 of the comparator 122 and provide a correction signal of greater magnitude to the ramp time modu~
lator 130. This is accomplished in the circuitry of Figure 4 in the following manner. The power input terminals 158 and 160 o~ amplifier 102 are connected each through a related resistor R401 and R403 to a summing junction 400. The output from the summing unction 400 is a voltage which is the algebraic sum of the voltages at the amplifier input termina1s 158 and 160. Thus, if the absolute magnitude of the negative voltage is greater than the absolute magnitude of the positive voltage (minus 40 volts and plus 30 volts), the output at the junction 400 would be a negative value (e.g. minus ten volts).
On the contrary, if the positive voltage at the terminal 158 has an absolute magnitude greater than that of the negative voltage at the terminal 160, the output at junction 400 will be positive.
The summing junction 400 has a first connection through a diode D401 and resistor R405 to the nega-tive terminal of an operational amplifier 402~ The summing junction 400 has a second connection through another diode D403 to the positive terminal of opera-tional amplifier 402. The diode D401 permits only negative current to pass therethrough, while the diode D403 permits only posit;ve current to pass therethrough. At a location between the diode D403 and operational amplifier 402, there is a resistor R407 connected to ground. There is also a ~esistor R409 to provide feedback to the negative terminal of the operational amplifier. The output of the operational amplifier 402 leads through a diode D405 which will only pass positive current to the rPsistor R113. There is also provided a diode D407 which passes positive signals from the comparator opera~
tional amplifier 172 to the resistor R113O
When the voltages at the terminals 158 and 160 are equal, the output from the summing junction 400 is zero. In those circumstances where the negative voltage at the terminal 160 is greater than the positive voltage at terminal 158, there would be a negative output from the summing junction 400 which would be transmitted through the diode D401 to the ampli-fier 402, which would then produce an output signal 116'~4~34 through the diode D405. If the difference between the voltages at terminal 158 and 160 is sufficiently large, this signal will override the signal from the operational amplifier 172 (which is directed through diode D407) to cause ramp time modulator ` 130 to increase the power of the current pulses and thus correct the difference between the absolute-magnitudes of the voltages at terminals lS8 and 160.
In those circumstances where the positive voltage at the terminal 158 is higher than the negative voltage at the terminal 160, the output from the summing junction 400 will be positive, and thus a positive voltage will be transmitted through diode D403 to the amplifier 402, which in turn will produce an output signal. However, since the voltage at terminal 158 is already relatively high, in most circumstances the signal from the amplifier 172 would be suffi~
ciently large to overcome the signal from the opera-tional amplifier 402 and provide a control signal 2~ of sufficient strength to cause ramp time modula-tor 130 to increase the duration of the power pulses to the transformer 104 and thus correct the disparity between the absolute maynitude of the voltages at terminals 158 and 160.
As indicated previously herein, when the ampli-fier of the present invention is operating as an audio amplifier, the major amplitude variations in the signals to be amplified would occur in a time period no less than approximately 1/1000 of a second.
Therefore, the apparatus should be able to respond within this time period to change its power output from a low level tv a relatively high level. For this reason, the frequency of the control pulses, which in turn controls the frequency of the current pulses through the transformer primary winding 104a, should be at least a thousand cycles per second, and pxeferably at least two thousand cycles per second.
However, there are further advantages in operating the current pulses at a much higher frequency, in a range of fifteen to twenty-five thousand cycles per second (desirably about twenty thousand cycles per second). First, this permits the size of the transformer to be made quite small. Also, the fre-quency is at a sufficiently high level that it wouldnot produce any undesired sounds in the normal audio range, which is normally below twenty thousand cycles per second. Further, this frequency is n~t so high as to be beyond the capacity of the switching circuits used.
The basic concept of utilizing power output feedback in an audio amplifier transformer to control the energy transfer between the primary and secondary windings of the transformer in response to the magni-tude of the audio input signal can be effectively employed in a transformer which produces a fixed voltàge output across the secondary. Moreover, when a fixed output transformer is involved, the power output feedback can also be used to assist in main- -taining the voltage output constant. The Figure 5 embodiment of the present invention illustrates such a system. The main power transformer 500 (i.e.
magnetic field coil) has primary and secondary wind-ings 500a and 500b, respectively.
For reasons more fully explained in connection with Figure 11 below, the secondary 500b is tapped to provide plural positive and negative voltages i stepped in 25 vvlt increments from 25 to 75 volts.
I The positive terminals are respectively designated El through E3, while the negative terminals are respectively designated E4 and E6.
The opposite ends of primary winding 500a are connected through current rectifying diodes to the two terminals of a conventional power source, indi-cated at 502, which can be a wall socket providing current at 60 Hertz and 120 volts. Two leads 504 and 506 from power source 502 are respectively con-nected through related rectifying diodes D501 and D503, and then in series with a silicon controlled rectifier 508, to the upper end 510 of transformer ~ primary winding 500a. The leads 504 and 506 are also respectively connected through related second rectifying diodes D505 and D507 to the lower end 512 of primary winding 500a~ An examination of the diodes D501, D503, D505 and D507 makes it readily apparent that these four diodes are arranged in a rectifying bridge so that on each half cycle from the power source, a positive voltage is directed to the silicon controlled rectifier 508 and thence to the upper end 510 of the transformer primary wind-ing 500a.
Silicon controlled rectifier 508 is controlled by an audio input signal as will be explained-more fully hereinbelow. The audio input signal is directed into an input terminal 514, and thence to a junction point 516. The resistors R501 and R503 are connected between junction 516 and a suitable voltage source (e.g. a 75 volt source) to provide a base voltage level of, for example, 0.7volts. This voltage is developed by diode D509 connected from the junction of R503 and R501 to ground. The voltage is in turn directed to an operational amplif;er 518. Feedback to the opera~ional amplifier is obtained ~rom the junction 520 of a diode D515 and capacitor 522 in the transformer seconaary circuit. The point 520 is connected through two voltage dividing resistors R505 and RS07 to ground. At a junction point 526 between the two resistors R505 and R507 there is a feedback connection to the operational amplifier 518. The output of operational amplifier 518 is directed to a suitable control apparatus indicated at 528. This control apparatus 523 is connected to the control terminal of silicon controlled recti- ¦
fier 508 in a manner such that at higher outpu~ levels from operational amplifier 518, the rectifier 508 is caused to conduct at higher voltage levels on the latter portion of each half cycle. In like manner, when the output of the operational amplifier 518 is lower, the silicon controlled rectifier 508 is caused to fire at lower voltaye levels.
The center of secondary winding 500b is tapped at 530 to ground. The upper half of the winding 500b is tapped at two intermediate locations 532 and 534 to respectively provide the positive 25 volt and 50 volt output for power terminals El and E2, while the upper terminal 524 of winding 500b provides the positive 75 volt output for terminal E3~ In like manner, the bottom half of secondary winding 500b is tapped at intermediate locations 536 and 538 to provide the intermediate voltage levels of minus 25 and minus 50 volts at terminals E4 and E5, respectively, while the lower end 540 of the secondary winding provides the minus 75 volt output for terminal E6.
~ The three points 532, 534 and 524 are each con ! nected through respective blocking diodes D511, D513 I and D515 to their respective output terminals. A

~3~6~

first capacitor 542 i5 provided between ground and the lower voltage output terminal (i.e. positive 25 volt terminal). A second capacitor 544 is con-. nected between the positive 25 volt output terminal and the positive 50 volt output terminal, and thethird capacitor 522 is in like manner connected between the positive 50 volt output terminal and the positive 75 volt output terminal. Capacitors 542, 544 and 522 have sufficient capacitance to compensate for any abrupt power demand from the related output terminal so as to maintain the output voltage terminals at nearly constant voltage level.
The lower half of secondary winding 500b is connected to its negative output terminals E4, E5, E6 through three blocking diodes D517, D519 and D521.
Capacitors 546, 548 and 550 are respectively connected between the negative output terminals in substan-tially the same manner as the corresponding components for the upper half of the primary winding. However the blocking diodes D517 through D521 are reversed so as to permit only negative current to pass to the output terminals E4 through E6.
In the operation of the power supply of Figure 5, the output voltage in the winding 500b is-reg~
ulated entirely by the circuitry controlling the silicon controlled rectifier 508. The SCR control circuitry 508 is in turn governed by the audio input signal in a manner such that when the input signal is of a greater amplitude, the silicon controlled rectifier 508 is caused to conduct for greater portions of each power half cycle to transmit more current to primary winding 500a. This mode of operation ! can best be illustrated with reference to Figures 6A, 6B and 6C.

In Figure 6At there is a representation of the voltage delivered from power source 502 throuyh the two diodes D501 and D503 to the silicon controlled rectifier 508. It can be seen that because of the action of the diodes D501 through D507, a positive sine wave voltage pulse is delivered to silicon con-trolled rectifier 508 on each half cycle. Let it be assumed that the audio input signal to the amplifier - is of a relatively low amplitude, so that the power requirement of the amplifying circuit is quite low.
In this condition, the silicon controlled rectifier 508 is caused to conduct only at the very end of each half cycle. The point of conduction of each half cycle is illustrated at 600, and rectifier 508 remains conductive until the current has reached zero, at point 602. Thus, it can be seen that the current is being delivered to the primary winding 500a in rather short increments of time, and at a lower voltage level.
As the audio input signal reaches a greater amplitude, silicon controlled rectifier 508 is caused to conduct at a higher voltage level for the latter part of each half cycle, as illustrated in Figure 6B. The point of conduction occurs at 604, and each cut off point is indicated at 606. It can be seen that not only is the voltage higher, but also the time increment of each current pulse is longer, so that greater power is delivered to the primary wind-ing 500a.
Finally, in Figure 6C, there is shown the situa-tion where the audio input signal is at a maximum amplitude, thus Ieadiny to a demand for maximum power from the transformer. In this situation, silicon controlled rectifier 508 is caused to conduct near the peak voltage at the beginning of the latter half of each half cycle, as illustrated at 608 in Figure 6C, with the cut off point being indicated at 610.
Thus, it can be seen that current is being delivered at yet a higher voltage and for a longer duration during each half cycle.
As the current builds up in the primary winding 500a during the latter part of each half cycle of current, no current flows in the secondary winding 500b, because of the arrangement of the blocking diodes D511 through D521. However, at the end of each half cycle of current through the primary winding 500a, after the current is shut off, the field around 500a collapses so as to create a voltage drop across the secondary winding 500b and cause current to flow through the secondary winding to supply power to the six capacitors 522 and 542 through 550 and the six output terminals El through E6.
As indicated previously, when the amplitude of the audio input signal is at higher levels t the power requirements on the transformer 500 are greater.
During these periods, current will flow through the primary winding 500a for longer increments of time to store more energy in the magnetic field of primary winding 500a. When the current is shut off in the primary winding at the end of each half cycle, the magnetic field in the primary winding collapses to induce a flyback voltage across the secondary winc ng 50Qb. The diodes DSll through D521 are arranged so that current flows through the secondary winding to charge the capacitors 542, 544r 522, 546, 548 and 550 to maintain the voltage at the power terminals i El through E6 at the proper level.
Figure 7A illustrates another embodiment of 116Z4~4 a stepped voltage power supply constructed in accor-dance with the present invention. As in the Figure 5 embodiment, there is a power source such as wall plug 702 including two leads 704 and 706. Lead 704 is connected to a triac 708. The opposite side of the triac 708 connec~s to the ~pper end 710 of primary winding 700a in transformer 700. The other lead 706 of power source 702 connects to the lower end 712 of primary winding 700a.
The triac 708 serves a function similar to that of the silicon control rectifier 508, except that triac 708 conducts on both positive and negative half cycles of the current from power source 702.
Switching circuitry controls the triac 708 in a manner such that it conducts during the latter portion of each half cycle for longer or shorter periods of time, depending upon the power requirements of the amplifier. This switching circuitry is substantially the same as the circuitry which switches silicon controlled rectifier 508 in the Figure 5 embodiment of the~invention, and thus the circuit components ¦~
will not be described in further detail.
The secondary winding 700b of transformer 700 is tapped to ground at its center point 714._ The upper half of secondary winding 700b is tapped at two intermediate locations 716 and 718 to provide the positive 25 volt and 50 volt outputs for the power terminals El and E2. A connection 720 to the upper end of winding 700b provides the positive 75 volt output for power terminal E3. In like manner, the bottom half o~ the secondary winding is tapped j at three equally spaced locations, 722, 724 and 726 to respectively provide the negative 25, 50 and 75 volt outputs E4, E5 and E6.

The positive and negative 75 volt leads 720 and 726 are attached at opposite ends to a first bridge rectifier, generally indicated at 728. The positive output of bridge rectifier 728 is connected by lead 730 to the positive 75 volt power terminal E3, and the negative output of the bridge rectifier 728 is connected to the negative 75 volt power output E6 through lead 732. The positive and negative 50 volt leads 718 and 724 are connected to the opposi~e ends of a second bridge rectifier 734. The positive output terminal of bridge rectifier 734 is connected by lead 736 to the positive 50 volt power terminal E2, while the negative output from the bridge recti-fier 734 is connected~through lead 738 to the nega-tive 50 volt power terminal E5. Finally, the posi-tive and negative 25 volt leads 716 and 722 are con-nected to opposite ends of a third bridge rectifier 740. The output leads 742 and 744 of bridge recti-fier 740 are respectively attached to the positive and negative 25 volt power terminals El and E4.
As in the Figure 5 embodiment, six~capacitors, designated 746 through 756, are provided between the power output terminals El through E6 to compen-sate for any abrupt power demand on the related output terminal so as to maintain the output voltage terminals at nearly constant voltage ~evel.
To describe the operation of the embodiment of Figure 7A, reference is made to Figures 8A, 8B
and 8C. The current through ~rimary winding 700a is not rectified and is thus an alternating current.
The triac 708 is caused to conduct in the latter half of each half cycle, whether it be a positive I or negative half cycle. When the power requirements ; of the amplifier are low, the control apparatus acts to cause triac 708 to conduct for only a very short time period at the end of each half cycleO This is illustrated in Figure 8A, where the point of con-duction of each half cycle is illustrated a~ 800.
Triac 708 remains conductive until the voltage has reached zero at point 802. Thus, it can be seen that current is being delivered to the primary wind-ing in rather short increments of time, and at a lower voltage level.
As the control signal reaches a greater ampli-tude, the triac 708 is caused to conduct at a higher voltage level for the latter part of each half cycle, as illustrated in Figure 8B, where the point of con-duction is indicated at 804, and each cutoff point is indicated at 8060 It can be seen that not only is the voltage higher, but also the ~ime increment of each current pulse is lGnger, so greater power is delivered to primary winding 700a.
Fina~ly, in Figure 8C, there is shown the situa-tion where the input signal is at a maximum ampli-tude, thus providing for maximum power requirements.
In this situation, triac 708 is caused to conduct near the peak voltage at the beginning of the latter half of each cycle, as illustrated at 808, with the cutoff point being ind;cated at 810.
Current flows in the secondary winding 700b simul-taneously with the flow of current in 700a, with the current in 700b also being an alternating cur-rent. With regard to the flow of current through the two 75 volt leads 720 and 726, since this cur-rent flows through the rectifying bridge 728, the output to the power terminal E3 is always positive, while the output to the terminal E6 is always nega-tive. In like manner, current is directed from the intermediate terminals 716, 718, 722 and 724 through the two bridge rectifiers 734 and 740 to provide positive current to the output terminals E2 and El at the positive 50 and 25 volt levels, and to provide negative current to the power output terminals E5 and E4 at the`negative 50 and 25 volt levels, re-spectively.
It has been found that by using the power supply circuitry of the present invention, the transformer can be made relatively small and still provide ade-quate power. For example, the transformer in the present invention can be made from 1/4 to 1/10 the size of the transformer in a conventional audio amp-lifier of comparable power rating, with one hundred and seventy five windings in the primary and two hundred in the secondary.
It is sometimes desirable to adjust the cutoff point of current flow through transformer primary 700a in order to more precisely control the charac-teristics of the energy transfer across the trans-former windings during each half cycle of current from the power source. For example, shutoff of the primary current prior to the zero crossover voltage point in the current waveform eliminates idling cur-rents in the primary winding during the remainingportion of the waveform half cycle. Accordingly, Figures 7B and 7C illustrate two modifications to the switching circuitry of Figure 7A, both of which modifications enable the transformer primary winding 700a to receive current during more narrowly defined portions of the power cycle~
In Figure 7B, a second Triac 758 is connected ¦ in parallel wi~h Triac 708. A control apparatus ~ as previously described in connection with Figure ~6~ 4 5 controls the operation of Triacs 708 and 758 re-spectively via leads 760 and 762. A capacitor 764 is connected in series with Triac 758 and acts to periodically shunt Triac 708. The circuit of Figure 7B operates as foiiows. In response to an audio input signal, output from the control apparatus causes Triac 708 to conduct at some point during each positive and negative half cycle of the current from power source 702. At a later predetermined time the output from the control apparatus causes Triac 758 to conduct, whereupon current is diverted from Triac 708 and begins to flow through capacitor 764. Triac 708 shuts off but current continues to pass through Triac 758 and capacitor 764 to trans-former primary 700a until the voltage buildup incapacitor 764 reaches a level sufficient to turn Triac 758 off, thus ending current flow through the transformer primary. Capacitor 764 is very small in order to limit the conductive state of Triac 758 to a short period of time.
Figure 7C illustrates a second modification o~ the embodiment of Figure 7A where the single triac 708 is replaced by a switching apparatus indicated generally at 766. A pair of GTO SCR's are connected in parallel with one another. One GTO, designated GTOb, conducts on positive half cycles and a second GTO, designated GTOa, conducts on negative half cycles. Blocking diodes are provided at D701 and D703. Each GTO becomes conductive at a predetermined voltage, as provided by the control mans, and becomes non-conductive, as provided by the control means, within a predetermined control time period, prefer--ably one millisecond.
Figures ~A, 9B and 9C illustrate the manner ~16~ 34 of switching common to both the Figure 7B and 7C
modifications. At low power requirements the cur-rent flow begins at 900 and ends at 902 near the . latter part of the last part of each half cycle.
At intermediate power requirements, the on-off switch-ing occurs earlier in the latter half of each half cycle, illustrated in Figure 9B at 904 and 906.
At peak power requirements, the switching occurs near the peak of each half cycle, illustrated in Figure 9C at 908 and 910. With this arrangement, the transformer can be made yet smaller.
For purposes of clearly demonstrating the advan-tages of the various duty cycle controlled power supply embodiments discussed above, reference is made to Figure lOA, which illustrates the basic con-figuration of a conventional amplifier power supply.
Conventional 117-125v 60 cycle AC is supplied from PS to the primary winding lOOOa of a transformer 1000. The secondary winding lOOOb of transformer 1000 has its upper end connected through a diode D1001 to the upper terminal of the amplifîer 1002.
The lower terminal of the secondary winding lOOOb is connected through a second diode D1003 to the lower terminal of amplifier 1002. Upper and--lower capacitors 1004 and 1006 respectively maintain the voltage imposed upon the amplifier 1002 at a substan-tially constant value. Normally, the supply voltage has a peak voltage input of approximately 169 volts.
Let it be assumed that the input voltage imposed at the upper terminal of the amplifier 1002 is de-signed to be plus 75 volts and the voltage at the lower terminal is minus 75 volts. The center of the secondary winding lOOOb is normally tapped to ground.

An audio sound typically has peak power require-ments of relat;vely short duration and average power requirements of longer d~ration which are possibly 1/20th of the peak power requirement. Thus, most of the time the amplifier is operating at only l/lOth to l/20th of full power. To understand the implica-tion of this fact, reference is made to Figure lOB
illustrating the sine wave o~ incoming voltage sup-plied to the primary of a conventional audio ampli-fier transformer connected to receive conventional117 to 125v alternating current. The turns ratio of the primary and secondary winding of the conven-tional transformer is such that with the primary conducting at least some current throughout the entire sine wave of the incoming voltage, the peak voltage generated in the secondary is just slightly larger than the plus 75 and minus 75 volt level required by a conventional audio amplifier. When the amplifying component of the amplifier is demand-ing only average power, current flows in the sec-ondary winding for only a very short period of time at the very peak of the sine wave of the input volt-age. This time period is indicated at 1008 in Figure lOB. When there are peak power requirements,--there is an immediate drain on the conventional storage capacitors 1004 and 1006 of the power supply to lower their voltage levels slightly, and the result is that the secondary winding is conducting for a longer time period~ so that the conducting portion of the sine wave of Figure lOB is broadened out to, for example, lines indicated at lOlOa and lOlOb. It should be noted that since the two lines lOlOa and lOlOb are spaced further apart, the voltage produced in the secondary winding lOOOa is moderately down from the peak voltage delivered at 1008.
In designing a transformer suitable for use in a conventional amplifier power system as described above, careful consideration must be given to acco-modating the idling current in the primaryn Idlingcurrent is the current which flows in the primary when no current is flowing in the secondary. In a transformer having a small number of windings in the primary and thus a small inductance, the primary idling current may become large enough to cause the transformer to heat up to an undesired extent. This fact dictates the use of a primary coil having a large number of windings.
A suitable audio amplifier transformer of con-ventional design must also be capable of accomodatinga relatively large current flow through the primary and secondary coils in order to handle peak power demands. Thus the wire forming the coils must be of sufficient diameter to allow the transformer to deliver high current at peak loads, without too much internal resistance. The result is a very large, heavy transformer having a large number of windings - to keep the inductance in the primary sufficiently high, and relatively thick wire to keep the-resistance relative low in spite of the rather long length of wire in the transformer.
In contrast with a conventional power supply transformer, a transformer desi~ned for use in the duty cycle controlled power supply of the present invention will normally be ormed with a higher sec-ondary to primary turns ratio than is conventional in non-duty cycle controlled transformers employed ! in commercial amplifier power supply circuits. With ' such a turns ratio, the point on the sine wave input ~.~.6~

to the primary at which current would stop flowing in the secondary can be made to occur well down the back slope as illustrated at point 1012 of Figure lOC. Without duty cycle control the power supply equipped with such a transformer would be supplying voltage substantially above the desired plus 75 and minus 75 volt levels normally required by conven-tional audio amplifiers. With duty cycle control, no current will flow in the primary of ~he trans- --former when the duty cycle switching element is open except for very small leakage currents allowed by the solid state switching element when in the open condition. These leakage currents can be ignored for purposes of this discussion. The switching element remains in a non-conducting position until the voltage in the primary drops to a point 1014 just above the 1012 level. Then, current flows in the secondary between points 1014 and 10120 If the switching element is a self-commutating SCR, the switching ele~ent will remain in a conducting state down to point 1016 but no current will be flowi~g in the secondary from point 1012 t~ point 1016 since the diodes connecting the storage capacitor's in the power supply to the transformer secondar-y will become back biased.
Causing current flow in the secondary of a duty cycle controlled power supply during the time between points 1014 and 1012 of Figure lOC would appear at first observation to be a little more inefficient than causing secondary current flow in a conventional amplifier as represented by 1008 in Figure lOB.
This is because the capacitors 1004 and 1006 only ! want to accept current at the 75 volt level. Thus, there are some resistance losses which occur in the .

duty cycle controlled transformer, these being repre-sented by the shaded triangle between points 1012, 1014 and lOlB in Figure lOC. However, a duty cycle controlled transformer can get by with many fewer turns (only a small fraction of the turns in a conven-tional transformer), so that the length of wire in the transformer is reduced. This reduces the internal resistance of the transformer proportionately.
Let it now be assumed that the smaller duty cycle controlled transformer is operated at peak power requirements as is illustrated in Figur~ lOD.
In this situation the switching element moves the "switch on" point further up the sine wave, which would be at a maximum at approximately point 1020 near the peak of the sine wave. Let it further be assumed that the capacitors 1004 and 1006 are suffi-ciently large so that they are maintaining voltages of plus 75 and minus 75 volts pretty much at that level. The voltage generated in the secondary at point 1020 would be substantially higher than the plus 75 volt level (possibly 90 volts), ignoring losses in the transformer. Therefore, the voltage difference represents the losses in the transformer itself. These losses are represented in the shaded triangle between points 1012, 1020 and 1022 in Figure lOD. Due to the fact that peak power is rarely re-quired by audio amplifiers for more than a very short time, the somewhat greater losses represented by Figure lOD can be tolerated in order to obtain the compensat;ng gains of cutting off idling currents during the first part of each cycle of the conven-tional A.C. input sine wave. When the switching element in the primary coil includes means for turn-ing off the current flow in the primary coil prior ~6~B4 to return of the waveform to a zero voltage (such as is illustrated in Figures 9A through 9C), even greater reduction in idling current losses can be achieved.
In summary, most of the time an amplifier power supply is operating in the low power mode, as shown in Figure lOC~ Accordingly,~the smaller duty cycle controlled transformer of the present invention can operate with about the same efficiency as the much larger prior art transformer. This is due in part to the number of primary and secondary turns being cut down substantially, thereby permitting the trans-former wire to be much shorter and thus offering less resistance in the transformer itself. This lessened resistance makes the existence of the idling current in the latter portion of each half cycle more tolerable. When higher power levels are re-quired, there is potential for greater inefficiency.
However, this is offset by the low internal resist-ance of the transformer, and in any case it is pos~sible to live with these higher inefficiencies for a short period of time, since they will not be large enough to overheat the transformer.
The following table includes the resul~s-of ~ -testing several different transformer designs wherein the transformers were tested by connecting the 50 volt output terminals of the secondary to two 150 watt light bulbs. The output of the secondary wind-ing was maintained at 300 watts. Temperature was measured at the top center portion of the transformer.

TABLE I
Turns Turns Resist- Temp C
Trans~ in RatioPrimary ance after former Pri- Sec/ Wire Induct- Primary/ 21 Number marY Prim~ Size ance Sec. Min.
9 149 0.6~ #18~#17 58.6 485/733 31 8 131 0.82 #18&~17 52~7 400/78g 45 7 113 1.03 #18&#17 32.2 335/812 62 1 90 1.14 #18 33.7 330/88~ 54

4 113 1.14 #18 32.3 330/993 67 6 113 1.14 #1~&#17 33.7 330/88~ 71 3 141 1.14 #20 58.1 538/1560 75 2 113 1.14 ~19 ~3 452/1470 84 177 1.14 #20 97 800/2340 107 The results of these tests indicate that a preferable duty cycle transformer designed for opera-tion in a power supply built in accordance with this invention would be a transformer having a secondary to primary turns ratio below 1.0 with a primary in-ductance above 30 millihenries and a coil wire gauge diameter above no. 18 when the transformer is used to produce a maximum + 75 volt D.C. outpu~ from con-ventional 117-125 volt, 60 cycle alternatin~ current.
Figure 11 depicts an amplifying apparatus 1100 designed to utilize the stepped voltage power- supply illustrated in Figures 5 and 7A-7C. A signal voltage is provided at 1102, and the output of the amplifier is connected through a load (shown herein as a speaker 1104) to ground. The apparatus employs two sets of transistors arranged in push-pull relationship, with each set being connected in series. The first - set QllOl, Q1103 and Q1105 are NPN transistors, and these are used to amplify pos;tive portions of the input signal. The other set of transistors Q1107, i 35 QllO9 and Qllll are PNP transistors and are used ~6~

to amplify negative portions of~the input signal.
In the following description, the overall operaticn of the first set of transistors QllOl, Q1103, Q1105 will be described in detail, with the understanding that the same description would apply to the opera-tion of Q1107, QllO9 and Qllll with respect to the negative portions of the signal.
It will be noted that the emitter electrode llQ6 of QllOl is connected to a power output terminal 1108 of the load 1104, and the collector electrode 1110 of QllOl is connected through a diode DllOl to a D.C. voltage source El having a magnitude of plus 25 volts. The emitter electrode 1112 of the second transistor Q1103 is connected to the collector `15 electrode lllO of QllO1, and the collector electrode 1114 of the transistor Q1103 is connected through a second diode D1103 to an intermediate D.C. voltage source E2 having a magnitude of plus 50 voltsO
Finally, the third transistor Q1105 has its emitter electrode 1116 connected to the collector electrode 1114 of Q1103, and its collector electrode lllB is connected directly to a higher D.C. voltage source E3, which is shown as having a magnitude of plus 75 volts. --As discussed previously herein, under the topic heading "Background of the Invention'l, various ar-rangements of series coupled transistors, with the stepped voltage sources of increasing magnitude are shown in the prior art. It is believed that a better understanding of the operating features of the pre-sent invention will be achieved by preceding a de-tailed description of the present invention with i a general discussion of the general mode of operation of prior art devices using an arrangement of series connected transistors with stepped voltage sources.
To disc~ss generally the prior art modes of operation, when the signal voltage is relatlvely small (e.g. below 25 volts) only the first transistor QllOl would be conductive, and all of the power would be derived from the 25 volt power source E1. The obvious advantage is that there is less voltage drop across the transistor QllOl and thus an increase in efficiency.
When the signal voltage closely approaches the value of the first voltage level, in prior art de vices the signal voltage is then applied in some manner to the base of the transistor Q1103 to make it conductive, so that the power is derived from the 50 volt source E2, with the 25 volt source being blocked out by the diode DllOl. While the signal is fluctuating between the 25 volt and 50 volt level, substantially all or at least the major portion of the voltage drop is taken across the second transis-tor Q1103.
In like manner, when the signal voltage rises above the 50 volt level, the voltage signal is ap-plied to the base of the transistor Q1105 to make it conductive and thus derive power from the--75 volt power source E3. Also, with the signal voltage fluc-tuating between the 50 and 75 volt level, substan-tially all, or at least the major portion o~, the voltage drop is taken across the third transistor Q1105. Thus, with regard to the prior art devices, each of the transistors must be made with the capability of withstanding the voltage drop imposed across the transistor at the current levels existing at th~
i various voltage levels.
~ Reference is again made to Figure 11. To discuss .

specifically the present invention, it will be noted that the signal input terminal 1102 is connected through an operational amplifier 1120 to a biasing transistor Q1113. The collector electrode of transistor Q1113 is connected through a resistor RllOl to a plus 75 volt source. The base electrode 1122 of the trans-istor QllOl is connected at a junction point between the resistor RllOl and transistor Q1113 to provide a forward bias to the transistor QllOl. The base electrode 1124 of the second transistor Q1103 is connected to a first switch and control means, indi-cated at 1126, and the base electrode 1128 of the third transistor Q1105 is connected to a second switch and control means, indicated generally at 1130.
lS The second set of transi5tors Q1107, QllO9 and Qllll are similarly connected. Thus, the biasing transistor ~1113 is connected in series~ with a re-sistor R1103 to a minus 75 volt source, with the base electrode 1132 of transistor Q1107 connected to a terminal between transistor Q1113 and resistor R1103. the respective base electrodes 1134 and 1136 of transistors QllO9 and Qllll are respectively connected to a third switch and control means 1138 and a fourth switch and control means 1140.--Nega-tively stepped voltage sources E4, E5 and E6 areprovided in the same manner as the sources El, E2 and E3.
As shown herein, the input from 1102 is through an operational amplifier 1120 to the base 1142 of biasing transistor Q1113~ There is a feedback from the output junction 1144 between the transistors QllOl and Q1107 back through the resistors R1105 ! - and R1107 to ground. From the junction 1146 inter-! mediate transistors RllOS and R1107, there is a ~6 feedback connection back to operational amplifier 1120. The resistors RllO9 and Rllll provide an initial biasing voltage to the transistors QllOl and Q1107, respectively.
The general function of each of the switch and control means 1126, 1130, 1138 and 1140 is to cause a related transistor to become conductive at the appropriate time and then to apportion the voltage drop across the related transistor so as to minimize the power which must be dissipated by any transistor at any particular time. The manner in which this is accomplished can best be described with reference to the graphs of Figures 12A, 12B and 12C.
In Figure 12~ the voltage drop across the first transistor QllOl is plotted against the output volt-age. Let it be assumed that the signal voltage has climbed to a low level of five volts. This voltage is applied to transistor QllOl to cause it to become conductive to transmit current from the postive 25 volt source El through transistor QllOl and to the output terminal 1108. Thus the voltage at output terminal 1108 will be approximately 5 volts, and the voltage drop across transistor QllOl will be approximately 20 volts. As the signal current increases -to a value closer to the 25 volt level, the voltage level at output terminal 1108 increases, while the voltage drop across transistor QllOl decreases.
When the signal voltage comes within one or two volts of the 25 volt level, the first switch and control means 1126 becomes operative and directs current to the base electrode 1124 of transistor Q1103 at a voltage level intermediate the level of the output voltage and the value of the positive fifty volt source E2. The graphs of Figure 12A and 12B illustrate this relationship in a somewhat idealized manner, where the first switch and control means 1126 functions to apply a voltage to base electrode 1124 consistently mid-way between the output voltage and the plus 50 volt level at E~, so that the voltage drop across the two transistors QllOl and Q1103 remain substantially equal for all output voltages between 25 and 50 volts. In the actual embodiment shown herein, the apportioning of the voltage drop across transistors QllOl and Q1103 would depart moderately from this idealized situation~
When the signal voltage comes very close to - ~
the 50 volt level, then the second switch and control means 1130 makes the third transistor Q1105 become conductive and also transmits base current to the base electrode 1128 of Q1105 at a sufficiently high voltage so that only a portion of the total voltage drop is across the transistor Q1105. In like manner, the first switch and control means 1126 continues to supply current to the base electrode 1124 of transis~
tor Q1103 so that the voltage drop across Q1103 is within its apportioned share of the total voltage drop across the three transistors Q1105, Q1103 and QllOl.
`25 Again, the somewhat idealized situation is shown in Figure 12A~ 12B and 12C, in that with the output voltage between 50 and 75 volts, the voltage drop is equally apportioned among all three transistors.
In actual practice, the apportioning would not be that precise.
The manner in which the four switch and control means 1126, 1130, 1138 and 1140 operate will now ! be described. Since each of the four switch and i control means are substantially identical,~only the first means 1126 will be described in detail.
In the first switch and control means 1126, there is a control transistor Q1115 having its col-.
lector electrode 1148 connected to the base electrode 1124 of the second power transistor Q1103. The base electrode 1150 of transistor Q1115 is attached to a junction point 1152 between two voltage dividing resistors R1113 and R1115. The other end of the resistor R1113 is connected to a positive 75 volt terminal, while the other end of resistor R1115 is connected to ground.
The emitter electrode 1154 of the transistor Q1115 is connected through a resistor R1117 to a juncture point 1156 between two voltage dividing re-sistors R1119 and R1121. The other end of resistor R1121 is connected to a positive 75 volt source, while the other end of resistor Rlll9 is connected to the main output line 1158 leading to the output terminal 1108. A capacitor 1160 is connected in parallel with the resistor Rlll9 to alleviate rapid voltage changes across the resistors Rll9 and R1121.
As discussed previously herein, it is aesirable to have transistor Q1103 become conductive when the signal voltage (and consequently the output._voltage which should be substantially identical to the signal voltage) reaches a level just below the 25 volt levelO
It is also desirable to have the current supplied to the base electrode 1124 of the transistor Q1103 at a voltage level approximately intermediate the output voltage and the next stepped voltage in the power source, which is the 50 volt power source E2.
Accordingly, when the output voltage reaches approxi--I mately the 25 volt level, it is desired that the ~ base electrode 1124 of transistor Q1103 have a cur--6~-rent delivered thereto at a voltage approximately midway between 25 and 50 volts (e.g. appro~imately 37 1/2 volts)~
The value of the resistances R1113 and R1115 are selected so that when little or no base current is flowing to the base electrode 1150 of the transis-tor QlllS the voltage at junction point 1152 is ap-proximately 37,5 volts. The values of the two resis-tors Rlll9 and R1112 are so selected that when the output voltage comes within one or two volts of the voltage of ~he lowest power terminal (i.e. 25 volts) the voltage at junction 1156 is approximately 3~.2 volts so as to apply a forward bias to the emitter electrode 1154 of the transistor Q1115 to cause tran-sistor Q1115 to conduct and transmit base currentto the base electrode 1124 of the transistor Q1103.
Since the collector electrode 1114 of transistor Q1103 tends to follow the voltage of base electrode 1124 within a fraction of a volt, the immediate ef-fect would be to bring the voltage at the emitterelectrode 1112 of Q1103 to approximately 37~5 volts.
Thus, with an output voltage of approximately 2~
volts, the voltage drop across transistor Q1103 would be approximately 12.5 volts, and the voltage-drop across transistor QllOl would be 12.5 volts, thereby causing the power dissipated to be shared equally by QllOl and Q1103.
As the signal voltage increases in the 25 volt to 50 volt range, the voltage at junction point 1156 likewise increases so as to tend to drive the voltage at the emitter electrode 1154 of transistor Q1115 upwardly. This causes an increase in current to the base electrode 1150 of Q1115, thus raising the voltage of junction point 1152 to a level closer to the voltage of emitter electrode 1154 and also causing transistor Q1115 to be more conductive so that greater current is supplied to the base electrode 1124 of transistor Q1103 at yet a higher voltage.
The effect of this is to raise the voltage at emitter electrode 1112 of transistor Q1103 even higher (i.e.
closer to the 50 volt level). Thus, as the output voltage increases from the 25 volt level toward the 50 volt level, the voltage drop across transistor Q1103 diminishes to apportion the voltage drop ~e-tween the transistors Q1103 and QllOl.
By the time the signal voltage reaches the second power source increment level (i.e. the 50 volt level), substantially all of the voltage drop is across the load, and there is very little voltage drop across the two transistors QllOl and Ql1030 At this time, the second switch and control means 1130 becomes operative to cause the third transistor-Q1105 to become conductive. Because this is accom-plished in substantially the same manner as in thefirst switch and control means 1126, the operation of the means 1130 will be summarized only briefly.
It can be seen that there is a control tran-sistor Q1117 having a collector electrode 1148a --connected to the base electrode 1128 of transistor Q1105. There are a pair of voltage dividing resis-tors R1113a and R1115a producing a voltage level of approximately 62.5 volts at juncture point 1152a.
Also, the two voltage dividing resistors R1119a and R1121a are arranged such that when the output voltage reaches a level just below the 50 volt level, the voltage at junction point 1156a is approximately ! - 63.2 volts.
i Thus, when the output voltage comes quite close ~.~..6~

to the 50 volt level, a forward bias is applied be-tween the emitter electrode 1154a of transistor Q1117 and base electrode llSOa to cause the transistor Q1117 to conduct, thçreby supplying base current to the base electrode ii28 of transistor Q1105 to cause transistor Q1105 to conduct. As soon as Q1105 becomes conductive, the voltage of the emitter elec-trode 1116 of Q1105 rises to a level close to that of base electrode 1128 of Q1105 (i.e. approximately 63.2 volts). This causes the diode D1103 to block off the 50 volt power source so that all of the power is derived from the plus 75 volt power source.
When the output voltage is slightly above 50 volts, the voltage at which current is being delivered through transistor Q1115 to the base electrode 1124 of transistor Q1103 is intermediate between the output voltage and the voltage of the current to the base electrode 1128 of transistor Q1105. Thus, the volt-age drop from the 75 volt source to the level just above the 50 volts being delivered to output terminal 1108 is apportioned between the three transistors QllOl, Q1103 and Q1105. As the output voltage in-creases further toward the 75 volt level, the voltages at the ]unction points 1156 and 1156a increase propor~. -tionately to raise the voltage o~ the currents respec-tivley delivered to the base electrodes 1124 and 1128 of transistors Q1103 and Q1105, thus increasing the voltage level at the respective emitter electrodes 1112 and 1116 of transistors Q1103 and Q1105. Conse-quently, the voltage drop across the three transis-tors QllOl, Q1103 and Q1105 continues to be appor-tioned between the three transistors. As indicated ! previously, the apportioning illustrated in the graphs of Figures 12A, 12B and 12C is somewhat idealized~

and the actual voltage drops will depart somewhat from the precisely equal apportionment.
The operation of the third and fourth switch and control means 1138 and 1140 is substantially the same as that of the first and second switch con-trol means 1126 and 1130 respectively, except that the means 1138 and 1140 operate on the negative por-tions of the input signal. Accordingly, the operation of the means 1138 and 1140 will not be described in detail.
It is sufficient to note that transistor of switch control means 1138 is designated Qlll9 while the control transistor of switch control means 1140 is designated Q1121. The control transistors Qlll9 and Q1121 operate in substantially the same manner as corresponding transistors Q1115 and Q1117 to make the power transistors QllO9 and Qllll, respectively, conductive at the proper negative voltage levels.
- Transistors Qlll9 and Q1121 also control the voltage level of the emitter electrodes of transistors QllO9 and Qllll to apportion the voltage drop across the three transistors Q1107, QllO9 and Qllll.
Referring now to Figure 13, an alternative ar-rangement of the output stage transistors and tran-sistor control means for use in an audio amplifierof the type illustrated in Figure 11 is disclosed.
In particular, those components in Figure 13 which are identical with the components of Figure 11 have been identified by the same reference numerals. The voltage dividing resistors R1115 and R1113 are se-lected so that the voltage at 1152 will be 37 1/2 volts when the voltage at 1156 reaches approximately 38 volts as described above, thereby causing tran-sistor Q1115 to conduct. This in turn causes the ~68-emitter of transistor Q1103 to jump up to the 37 1/2 volt level, raising the input voltage to transis-tor QllOl to 37 1/2 volts. Diode DllOl now operates to block off the 25 volt power source. As the output signal on line 1158 climbs toward 50 volts, the volt-age at 1156 also climbs upwardly toward the 50 volt level. By the time the audio signal reaches the 50 volt level, the voltage at 1156 will also have reached 50 volts, thereby increasing the voltage supplied to collector 1110 of Q1101 to 50 volts.
The operation of the Figure 13 circuit is iden-tical to that of Figure 11 up to this point. How-ever, as the input signal voltage rises still further above 50 volts, transistor control 1130 operates to switch transistor Q1105 on, thus causing the potential appearing at emitter 1116 of Q1105 to be applied directly to the collector of transistor QllOl through diode D1303. Due to the bias at 1156a at this time, the potential applied to the collector of QllOl will be 67 1/2 volts. This will have the effect of back biasing diode D1301 to cause voltage from source E3 to be applied directly to transistor Q1101 through transistor Q1105.
Figure 14 is a graph representing the operation of the circuit of Figure 13, wherein line 1401 repre-sents the output voltage on line 1158 of Figure 11 and line 1402 represents the voltage applied to the collector 1110 of transistor Q1101.
A preferred embodiment of the left and right channels`of a stereo amplifier constructed in accor-dance with the present invention is illustrated in Figures 15A, 15B and 16. Referring first to the I left channel circuitry 1500 illustrated in Figures 15A and 15B, a left channel input signal is received - ~9 -at terminal 1502 and preconditioned in a high fre-quency filter 1504 which rolls the audio signal off beyond 20KHz. This filter acts to prevent transient in~er-modulation distortion.. Upon leaving filter 1504, the input signal enters operational amplifier 1506 and thereafter passes to transistors Q1501 and Q1503, which split the input signal into positive and negative halves. The positive half of the signal is fed to the upper half of the left channel amplifier and the negative half of the signal is fed to the lower half of the left channel amplifier. Because the upper and lower halves of the left channel ampli-fier are symmetrical, only the upper half will be described in detail.
The output of transistor Q1501 is level shifted upwardly through the action of resistors R1512 and R1513 to the base of transistor Q1505. The signal appearing at the output collector of transistor Q1505 is transmitted to transistor Q1509. At very low output power requirements, the emitter current from transistor Q1509 flows through series arranged diodes, indicated at 1508, to the base of output transistor Q1513, whereupon transistor Q1513 begins conducting.
Current from 25 volt power source 1512 then-passes through output transistor Q1513 to an output inductor 1510 and on into the loudspeaker.
When output voltages above approximately 25 volts are required, the amplifier output current is derived from the 50 volt power source 1514 through output tran~sistor Q1517. Similarly, when output voltages above 50 volts are required, transistor Q1509 drives output transistor Q1521 to derive current ! from the 75 volt power soruce 1516. Switching circuit 1518, including transistors Q1525 and Q1527, acts ~3h6 2~34 to apportion the voltage drop across the power transis-tors Q1513, Q1517 and Q1521.
The left channel amplifier includes an over-current protection circuit 1520. In the event of a short circuit across the output transistors, heavy currents are drawn through the amplifier, thereby creating a voltage drop across the emitter xesistor R1571 of output transistor Q1513. This voltage drop in turn switches over-current protection transistor Q1533 on, and the current which normally flows through transistor Q1505 to the base of transistor Q1509 is instead diverted to flow through the collector of over-current protection transistor Q1533. Thus deprived of its drive current, transistor Q1509 will not turn on and the output transistors will not con-duct. Consequently, the high power dissipations otherwise occurring under short circuit conditions are prevented.
Cross-over distortion is minimized by the action of transistors Q1537 and Q1539 in cross-over preven-tion circuit 1522. Transistors Q1537 and Q1539~
together with the 1-2-3-4 series arranged diodes 1524, resistors R1520 and R1521, and capacitor C1513, form a bias network which develops a slight~forward voltage drop between the bases of transistor Q1509 in the upper half of the left channel amplifier and transistor Q1511 in the lower half of the amplifier.
This forward voltage drop places transistors Q1509 and Q1511 on the verge of conducting. When an audio signal is received by the ampliier, Q1509 and Q1511 will conduct immediately and without discontinuity in the amplifier output waveform, thereby resulting in very low distortion of the audio signal.
For purposes of convenience t the values of all the capacitors and resistors employed in the circuit of Figures 15A and 15B are listed in Table II below.
Figure 16 illustrates the input portion of the circuitry for the right channel ampli~ier. The right channel amplifier includes a network for shifting the phase of the incoming audio signal by 180 in order to better utilize the amplifier power supply.
In all other respects, the right channel amplifier circuitry is identical to that of the left channel amplifier illustrated in Figures 15A and 15B.
Statistical analysis of stereo broadcasts indi-cates that the vast majority of audio signals as-sociated with one channel of such broadcasts are in phase with the audio signals of the other channel.
Prior art high fidelity amplifiers generally process incoming stereo signals without any modifications of the phase between the channels~ operating in what is known as single end~d fashion~ The components of a stereo amplifier operating in single ended fashion, however, tend to drain additional energy from the power supply. When the amplifier output voltage is high, the positive side of the power supply furnishes power to both channels but the negative side of the power supply does no wo~k.
When the amplifier output voltage is low, the nega-tive side of the power supply furnishes power to the amplifier but the psoitive side is not working.
Greater efficiency can be obtained from the amplifier if both sides of the power supply work continuously. In such situations, the power supply is said to be operated in bridge. Power can be de-livered in the bridge mode to a two channel stereo I amplifier by inverting the incoming signals in one of the amplifier channels and thereafter processing both channels in an out-of-phase fashion. As a result of the change in phase relationship between the otherwise generally in-phaSe stereo signals, positive power will always be required by one of the two amplifier channels while the remaining channel during any given power cycle will require negative power. Thus~ regardless of the value of the ampli-fier output voltage, the posi$ive and negative excur-sions of the power supply during each power cycle will both be employed. The greater power available at the amplifier output due to the fact that the power supply is being utilized more efficiently can increase the output power by about 15-20%.
Again referring to Figure 16, the right channel ' amplifier is indicated generally at 1600. The audio input signal to the right channel is received at terminal 1601 and fed to inverting network 1602.
The inverting network, which consists of capacitors C1601 and C1603 in combination with resistors R1601 R1603 and R1605, drives the inverting terminal o operational amplifier 1604. The values of the net-work components are listed in Table III below. Driving the inverting terminals of operational ampli-fier 1604 produces an output signal which is 180 out of phase with the input signal. As p'reviously dis-cussed, the majority of audio signals in each channel of a stereo broadcast are in phase. Consequently, the use of the inverting network generally results in a 180 phase difference between the operation of the left channel and the operation of the right channel of the amplifier.
Figure 17 is a preferred embodiment of the power source for the left and right channel'amplifiers .6~

illustrated respectively in Figures 15A, 15B and 16. Referring now to Figure 17, when switch 1700 is closed, current begins to flow from A.C. power line 1702 through a phase shift network 1704 to Diac 1706 and Triac 1708. Triac 1708 turns on, permitt-ing current flow through the primary 1710a of trans-former 1710. The magnetic field in primary 1710a builds up, transferring energy to transformer secondary 17]0b and then to electrolytic energy storage capa-lQ citor banks 1716, 1718 and 1720. Storage capacitor bank 1716 is designed to maintain a constant 25 volt output at the 25 volt power source, capacitor bank 1718 is designed to maintain a constant 50 volt output at the 50 volt power source, and capacitor bank 1720 is designed to maintain a constant 75 volt output at the 75 volt power source. The capacitors in the capacitor banks become fully charged within the first 100 milliseconds after the power supply is turned on.
When the voltages at the three power supplies respectively reach their preferred voltage levels of 25, 50 and 75 volts, control transistor Q1701 is forced into conduction and the emitter current from Q1701 flows through an LED diode 1712. --In response to the emitter current, LED 1712 emits a red light which shines on photoresistor 1714 and lowers the photo-resistance value thereof. The lowering of the resistance value of photoresistor 1714 in turn acts to shunt some of the current flow-ing through phase shift network 1704, subsequentlyshifting the phase of the A.C. line signal and caus-ing the Diac 1706 and Triac 1708 to fire at a later point on the incoming A.C. line sine wave.
Changes in the firing point of the Diac and Triac create variations in the conduction angles and corresponding variations in the amplifier out-put voltage. 5uch variations provide a means for tracking the audio signal whenever the audio signal

5 frequency is below the repetition rate of the power supply line, e.g., below a frequency of 120 Hertz (2x60 ~ertz). The incoming audio signal is summed at the junction of resistors R1765 and R1767, and is low passed by the time constant of the parallel combination of resistors R1765, R1767 and capacitor C1733. The resulting signal is t'nen rectified by diode D1709 to form a D.C. voltage which is propor-tional to the output of the power amp:Lifier. This proportional D.C. voltage is applied to capacitor C1735, from whence it is fed to control trans;stor Q1701. ~ontrol transistor Q1701 thereafter controls the operation o~ LED 1712 to vary the time constant of phase shift network 1704 as discussed above, generating greater amplifier output voltages under high signal conditions and lesser output voltages under lower signal conditions. The output of the power supply thus effectively tracks incoming audio signals having frequencies in the low audio range~
This tracking ability makes it possible to urther reduce the cost, size and weight of the amplifier unit.
Automatic shut down of the Figure 17 power sup-ply as a result of over-current conditions is achieved through the use of operational amplifier i722 and transistors Q1703 and Q1705. If a fault condition results in over-current delivery to the audio ampli-fier, an over-currerlt trip signal from the circuit of Figure 15A is fed to the base of transistor Q1707.
Transistor Q1707 switches on, causing the input port of operational amplifier 1722 to go high. The output of operational amplifier 1722 likewise goes high, switching transistors Q1703 and Q1705 on. The emitter of transistor Ql.705 is connected to the 25 volt source and the collector of Q1705 is connected to capacitors C1723 and C1725. When transistor Q1705 turns on, the charge from the 25 volt supply is transferred to capacitors C1723 and C1725. Current then flows through LED 1712, causing the LED to shine brightly on photoresistor 1714. The resistance of photoresistor 1714 is accordingly lowered to a value sufficient to shunt virtually all of the current from the phase shift network 1704~ thereby shutting the power supply down.
When the power supply is shut down, LED 1712 is maintained in a lighted condition by the charge stored ~n capacitors C1723 and C1725. After a short period of time ~i.e., somewhere between 1/2 and 1 minute) the charge on capacitors C1723 and C1725 dissipates through the LED and the LED begins to ~o dark again. The resistance of photoresistor 1714 consequently begins to rise and the power supply cuts back on. If the fault condition has been re-moved, the power supply remains on and the audio amplifier operates as before. If~ however, the fault still exists, the over-current trip line activates transistor Q1707 and the power shut-off sequence is repeated.
An over-voltage trip network is indicated at 1724. The audio signal from the output of the audio amplifier drives the network, including resistors R1751, R1753, R1755, R1757 and R1759, capacitor C1731, and diodes D1701 and D1703, to form a D.C. signal which represents the time-average half-way rectified audio signal. Note that diodes D1701 and D1703 serve as OR gates as well as rectifiers. The output at the jUnGtiOn of resistors R1751 and R1753 (e.g., the time-average audio voltage) charges capacitor C1731. Capacitor C1731 is chosen such that a value representative of a pre-selected over-voltage value will cause capacitor C1731 to trip operational ampli-fier 1722, thereafter turning on transistors Q1703 and Q1705 to shut down the power supply in a manner analogous to that which occurs during over-current conditions.
If for any reason ~i.e., amplifier failure or the dropping of a tone arm~ a D.C. component should appear at the amplifier output~ a D.C. voltage will appear at the junction-of resistors R1761 and Rl7630 This voltage is carried through D.C. fault trip line 1726 to operational amplifier 1722, causing the operational amplifier to trip. When the D.C~ component is posi tive, diode D1705 will conduct into the positive port of operational amplifier 1722 and the operational amplifier will go high. When the D.C. component is negative, diode D1707 will conduct into the nega-tive or inverting port of operational amplifier 1722 and the output oE the operational amplifier-w-ill also go high. In both cases, the power supply will be shut down following the switching of transistors Q1703 and Q1705 and the energization of LED 1712.
Tables listing the values of the various resis-tors and capacitors illustrated in Figures 15A, 15B, 16 and 17 follow. As previously discussed7 Table II contains listings for the circuitry of Figures 15A and 15B. Table III includes the components of I the inverting network of Figure 16, while Table IV
discloses preferred values for the resistors and capacitors of Figure 17.

~L~6~

TABLE II

VALUES OF THE RESISTOF<S AND CAPACITORS USED IN
THE LEFT CHANNEL AMPLIFIER OF E'IGURES 15A AND 15B
:
Resistors 5 R1501 - 15k52 R1526 - 39k Q R1551 -22kQ
R1502 - 2k~ R1527 - 100 n R1552 -18k n R1503 - 6.2kQ R1528 - 100 Q R1553 -4.7k fl R1504 -390Q R1529 - lkQ R1554 - 39Q
R1505 - 2.7Q R1530 - 5.6k Q R1555 -27kSL
10 R1506 -- 9.lkQ R1531 -- 120 ~ R1556 -- 4.7Q
R15t)7 - 1.5kQ R1532 - lkS~. R1557 - 3.3kQ
R1508 - 1.5kQ R1533 - 5.6kQ R1558 -2.2kQ
R1509 - 9.1kQ R1534 - 1205L R1559 -3.3k Q
R1510 - 1.5kQ R1535 - 1.5kQ R1560 -33 n 15 R1511 - 1.5kQ R1536 - 1.5k Q R1551 -2.7k Q
R1512 - 4.7kQ R1537 - 2.4kQ R1562 -2.7kQ
R1513 -910SL R1538 - 22kn R1563 - lUk n R1514 - 47~ R1539 - 22k5L R1564 - 220 5L
R1515 - lkQ R1540 - 18k~ R1565 - 220Q
20 R1516 - 4.7kQ R1541 - 4.7k~L - R1566 - 6.2 S~
R1517 -910 Q R15~12 - 3!3Q - R 567 - 220 Q
R1518 - 47Q R1543 - 27kQ R1568 -220 Q
R1519 -- lkQ R1544 - 4.7kQ R1569 - 56Q
R1520 - 12kQ R1545 - 3.3kQ R1570 -6251 25 R1521 - 5.bkQ R1546 - 2.2kQ R1571 - .lQ
R1522 - 5kQ R1547 - 3.3kSL R1572 - .15 R1523 - 10Q R1548 - 33Q R1573 -2~7Q
R1524 - 10Q R1549 - 2.4kSL R1574 -2.7Q
R1525 - 12k Q R1550 - 22k Q

TABLE II (Continued) Capacitors C1501 - 200 pf C1517 - .01 f C1531 ~ pf '~ C1503 - .001 f ~1519 - ~01 f C1533 - .039 f 5 C1505 - 470 uf C1521 - 22 u~ C1535 - .039 f C1507 - 100 pf C1523 - .0033 f C1537 - .1 f C1509 - 100 pf C1525 - 22 uf C1539 - .1 f C1511 - 200 pf C1527 - .0033 f C1541 - .33 C1513 - 4.7 uf C1529 - 180 pf C1543 - .33 10 C1515 - 10 pf . _ . . _ TABLE III

THE INVERTING NETWORK OF THE RI&HT

_ 15 Resistors Capacitors R1601 - 3k Q C1601 - 200 pf R1603 - 12k ~ C1603 - 33 pf R1605 - 15k Q -TABLE IV

VALUES OF THE RESISTORS AMD CAPACITORS USED

Resistors -R1701 - 1.2kQR1725 - 9.1 megQ R1747 - 2 megQ
R1703 - 150kQ R1727 - 510k Q R1749 - 1.6 meg5 R1705 - 180kQ R1729 - 5.6k Q R1751 - l99k Q
R1707 -- 27kQ R1731 -- 6.8k52 R1753 -- 3~3kQ
R1709 - 27kQ R1733 - 22k~ R1755 - 3.3kQ
R1711 - 150k S~ R1735 - 1 Q R1757 - 4-7k~
R1713 - 1.5kQ R1737 - 3.6k Q R1759 - 4.7ksL
R1715 - 390 Q ~ R1739 - 20k ~L R1761 - 3.3kQ
R1717 680n R1741 - 100kQ R1763 - 6.8kQ
R1719 - 200Q R1743 - 200kQ R1765 - 30kQ
R1721 - 680Q R1745 - 20k Q R1767 - 60kQ
R1723 - 10k Q
, Capacitors C1701 - .lf C1713 - 2200u~ C1725 - 2200uf C1703 - .013f C1715 - 2200uf C1727 - 22u~
C1705 - .01~ C1717 - 2200uf C1729 - ~7uf C1707 - 2200uf C1719 - 3000uf C1731 - .033f C1709 - 2200u~ C1721 - 3000uf C1733 - 2.2uf C1711 - 2200uf C1723 - 2200u~ C1735 - 2.2uf ~6~

Industrial ApplicabilitY
The above disclosed amplifier circuitry and methods are of particular economic importance in the field of audio amplifiers. The concepts disclosed herein permit dramatic reduction in the weight and cost of providing suitable power supplies and power transistors for high ~idelity audio amplifiers and are particularly well adapted to stereophonic systems.
As an example of the dramatic weigh~ reductions possible through implementation of the concepts of this invention, a completed commercial embodiment of a power supply and amplifier rated at 400 watts weighs only 4 kg. In contrast, the lightest prior art commercial amplifier of comparlable power rating weighs approximately 16 kg.

Claims

1. An apparatus for amplifying an audio signal having a time varying characteristic, said apparatus comprising:
a) audio amplifier means for amplifying the time varying characteristic of the audio signal to produce an output signal corresponding to the audio signal, said amplifier means including a signal receiving means for receiving the audio signal and a power input means for receiving electrical energy for producing the output signal;
b) power means for supplying power to said amplifier means, said power means including a transformer having a primary winding and a second winding with a secondary to primary turn ratio of below 1.0, said primary winding having inductance above 30 microhenries and a coil wire gauge diameter above no. 18, said secondary winding being operatively connected to said power input means of said amplifier means, said primary winding being adapted to be connected to a power source;
c) power control means to cause variable electrical energy to be transmitted to the primary winding to the transformer in response to a control signal to induce corresponding variable electrical energy in said secondary winding; and d) control signal generating means responsive to the time varying characteristic of the audio signal to generate the control signal for said power control means to control a characteristic of the energy transmitted to said primary winding in correspondence with the time varying characteristic of the audio signal to cause said energy control means to transmit to said amplifier power of a magnitude which varies over time in a manner related to the time varying characteristics of the audio signal.

2. An apparatus for amplifying an audio signal having an amplitude which varies at an audio frequency, said apparatus comprising:

a) audio amplifier means for amplifying the audio frequency amplitude variations of the audio signal to produce an output signal corresponding to the audio signal, said amplifier means including an audio signal receiving means for receiving the audio signal and a power receiving means for receiving electrical energy for producing the output signal;
b) power supply means for supplying power to said amplifier means, said power supply means including a source of commercially available, sinusoidally varying supply voltage having a constant frequency within the audio range, a transformer having a primary winding and a secondary winding, said secondary winding being operatively connected to said power receiving means of said amplifier means, said primary winding being adapted to be connected to said source;
c) power control means to cause variable amounts of electrical energy to be transmitted to the primary winding of the transformer to induce corresponding variable electrical energy in said secondary winding, said power control means including switch means connected between said source and said primary winding for operating in a conductive state in response to a control signal for a selected portion of each cycle of said sinusoidally varying supply voltage to cause said selected portion of said sinusoidally varying supply voltage to be applied across said primary winding only when said switch means is in said conductive state; and d) control signal generating means responsive to the time varying amplitude of the audio signal to generate said control signal for said power control means to control the period of conduction of said switch means during each cycle of said sinusoidally varying supply voltage in correspondence with the time varying amplitude of the audio signal to minimize the time portion of each cycle of sinusoidally varying supply voltage during which idling currents pass through said primary winding.

3. Apparatus as defined in claim 2, wherein said switch means includes a solid state switch connected between said source and said primary winding, said solid state switch being responsive to said control signal to become conductive, said control signal generating means supplying variable control signals to said solid state switch at variable times during each cycle of said sinusoidally varying supply voltage to cause said solid state switch to begin conduction at varying times dependent upon the time varying amplitude of the audio signal.

4. Apparatus as defined in claim 3, wherein said power control means further includes means responsive to a cut off control signal for cutting off flow of current to said primary winding, said control signal generating means including means for producing a cut off control signal in a manner to permit sufficient current to flow through said primary winding to allow power of a magnitude which varies over time in relationship to the time varying characteristics of the audio signal while simultaneously minimizing the flow of idling currents through said primary winding.

5. Apparatus as defined in claim 3, wherein said control signal generating means includes a control link for transmitting an electrical signal representative of the voltage level produced by said secondary winding to said power control means for causing the control signal to respond to variations in the output voltage produced by said power supply means.

6. Apparatus as defined in claim 5, wherein said control link includes an electro-optical link for preventing transfer of substantial electrical energy between said primary and second windings over said control link.

7. Apparatus as defined in claim 3 further including fault condition sensing means for sensing faulty operation of said audio amplifier means to produce a fault signal, said control signal generating means including fault response means connected with said fault condition sensing means to cause said variable control signal to reduce the conductive duty cycle of said solid state switch.

8. Apparatus as defined in claim 2, further including audio signal filter means for generating an audio tracking signal representative of low frequency variations in the audio signal, said control signal generating means including audio signal responsive means to cause said power control means to modulate the power of pulses transferred to said primary winding in response to low frequency variations in the amplitude of the audio signal.

9. Apparatus as defined in claim 2, wherein said audio amplifier means includes first and second transistors having series connected emitter-collector circuits, said first transistor having a base electrode arranged to receive the audio signal and an emitter arranged to produce at least a portion of the output signal of said audio amplifier means, and wherein said power supply means includes first voltage means for providing a source of voltage at a first level to the connection between said first and second transistor and second voltage means for providing a source of voltage at a second level higher than said first level to the emitter-collector circuit of said second transistor; and transistor control means for holding said second transistor in a nonconductive state when the amplitude of said audio signal is below a first predetermined level and for causing said second transistor to conduct when the amplitude of said audio signal is above said first predetermined level.

10. Apparatus as defined in claim 9, wherein said transistor control means causes said first predetermined level to be equal to said first level and wherein said transistor control means further causes said first and second transistors to substantially equally bear the total voltage drop across said first and second transistors when the amplitude of said audio signal is above said predetermined level.

11. Apparatus as defined in claim 9, wherein said power supply means includes a third voltage means for providing a source of voltage at a third level higher than said second level and wherein said audio amplifier means includes a third transistor having an emitter connected with said first and second transistors, and a collector connected with said third voltage means and further including second transistor control means connected with the base of said third transistor for holding said third transistor in a nonconductive state when the amplitude of said audio signal is below a second predetermined level which is above said first predetermined level and for controlling the amplification of said third transistor to cause said first and third transistors to each bear a substantial portion of the total voltage drop across said first and third transistors when the amplitude of said audio signal is above said second predetermined level.

12. Apparatus as defined in claim 11, wherein the emitter of said third transistor is connected to the collector of said first transistor and wherein said audio amplifier means includes a diode connected between said first and second transistors to isolate said second transistor from said first transistor when said third transistor is conductive.

13. Apparatus as defined in claim 11, wherein the emitter of said third transistor is connected to the collector of said second transistor and said first and second transistor control means operates to cause said first, second and third transistors to equally share the voltage drop thereacross when said third transistor is conductive and wherein said audio amplifier means includes a pair of diodes between the collectors of said first and second transistors and said first and second voltage means, respectively, said diodes being arranged to isolate said first and second voltage sources when said third transistor is conductive.

14. An amplifier apparatus, comprising:
a) power input terminal means adapted to be connected to a source of sinusoidally varying supply voltage, b) a transformer with primary and secondary windings, said primary winding being connected to said power input means, c) switch means interconnected with said primary winding and said power input terminal means to interrupt current from said power input terminal means to said primary winding, said switch means operating in a conductive state in response to a control signal for a selected portion of each cycle of said sinusoidally varying supply voltage to cause said selected portion of said sinusoidally varying supply voltage to be applied across said primary winding only when said switch means is in said conductive state, d) voltage supply means connected with said secondary winding for supplying at least two voltage points, namely a higher voltage point and a lower voltage point, e) a load terminal adapted to be connected to a load, such as a speaker, f) a first transistor having a first main current carrying electrode connected to said load terminal and a second main current carrying electrode connected to the lower voltage point, g) a second transistor having a first main current carrying electrode connected to the second electrode of the first transistor, and a second main current carrying electrode connected to said higher voltage point, h) signal input means connected to a base electrode of a first transistor to make said first transistor conductive in response to an input signal;

i) first control means to cause said second transistor to be nonconductive under conditions where said amplifying apparatus is amplifying a signal of lower magnitude so that output current is at a lower voltage, with the result that power is derived from said second lower voltage point, and to cause said second transistor to become conductive under circumstances where the input signal is of a higher magnitude, with the result that current flow from said higher voltage point through said second and first transistors to said output terminal, j) second control means operatively connected to said switch means to cause said switch means to be conductive at selected portions of each voltage cycle from said power input terminal means, said second control means causing said switch means to be conductive for shorter periods of time for lower power requirements of said amplifying apparatus, and to be conductive for longer periods of time for higher power requirements of the amplifying apparatus.

15. The apparatus as recited in claim 14, wherein said second control means causes said switch means to be conductive during a latter portion of each voltage half cycle from said power input terminal.

16. The apparatus as recited in claim 15, further including rectifying means operatively connected with the power input terminal means and the primary winding to cause only positive current pulses to be directed to said primary winding, said switch means comprising voltage responsive switch means which becomes conductive at predetermined voltage level of said supply voltage.

17. The apparatus as recited in claim 16, wherein said switch means comprises a silicon controlled rectifier connected in series between said power input terminal means and the primary winding of the transformer.

18. The apparatus as recited in claim 15, wherein said power input terminal means is connected to the primary winding to cause alternating current to be delivered to said primary winding, said switch means being voltage responsive switch means to cause the switch means to be conductive at predetermined voltage levels during latter portions of each half cycle of said supply voltage.

19. The apparatus as recited in claim 18, wherein said switch means comprises a triac connected in series with said primary winding.

20. The apparatus as recited in claim 19, wherein there is first rectifying means connecting said lower voltage point to the second electrode of the first transistor, so that direct current is directed to said lower voltage point, and there is second rectifying means interconnecting the higher voltage point with the second electrode of the second transistor, so that a single polarity is applied to the second electrode of the second transistor.

21. The apparatus as recited in claim 14, wherein said first control means is responsive to output voltage from said first transistor to said load terminal and operatively connected to a base electrode of the second transistor in a manner that control current to the base electrode of the second transistor begins to flow when said output voltage reaches a predetermined voltage level, so as to cause said second transistor to be conductive and cause current to flow from said higher voltage point through said first and second transistors to the output terminal.

22. The apparatus as recited in claim 21, wherein said first control means is further characterized in that the first control means supplies base current to the base electrode of the second transistor according to a functional relationship of output voltage at the output terminal, in a manner that voltage of the current supplied to the base electrode of the second transistor varies as a function of magnitude of the output voltage, with the voltage of the base current to the second transistor being at a voltage level intermediate the voltage at the higher voltage point and the output voltage, whereby voltage drop across the second and first transistors is shared between the second and first transistors.

23. The apparatus as recited in claim 22, wherein said first control means comprises a control transistor having a first main current carrying electrode connected to the base electrode of the second transistor, and second main current carrying electrode connected to voltage dividing means operatively connected between said output terminal and a related higher voltage source.

24. The apparatus as recited in claim 23, wherein a base electrode of the control transistor is connected to a junction point between a pair of voltage dividing resistors, which in turn are connected between higher and lower voltage sources.

25. The apparatus as recited in claim 21, further including diode means interconnecting said lower voltage point with said second electrode, of the first transistor whereby when said signal voltage is at a level higher than a voltage level of the lower voltage point, said lower voltage point is blocked off from said higher voltage point.

26. An amplifier apparatus comprising:
a) a power input terminal means adapted to be connected to a source of sinusoidally varying supply voltage, b) a transformer with primary and secondary windings, said primary winding being connected to said power input terminal means, c) switch means interconnected with said primary winding and said power input terminal means to interrupt current from said power input terminal means to said primary winding, said switch means operating in a conductive state in response to a control signal for a selected portion of each cycle of said sinusoidally varying supply voltage to cause said selected portion of said sinusoidally varying supply voltage to be applied across said primary winding only when said switch means is in said conductive state, d) voltage supply means connected with said secondary winding for supplying plural voltage points, namely, a first greater positive voltage point, a second lesser positive voltage point, a third lesser negative voltage point, and a fourth greater negative voltage point, e) a load terminal adapted to be connected to a load, such as a speaker, f) a first transistor having a first main current carrying electrode connected to said load terminal and a second main current carrying electrode connected to the lesser positive voltage point, g) a second transistor having a first main current carrying electrode connected to the second electrode of the first transistor, and a second main current carrying electrode connected to the first greater positive voltage point, h) a third transistor having a first main current carrying electrode connected to said load terminal and a second main current carrying electrode connected to the lesser negative voltage point, i) a fourth transistor having a first main current carrying electrode connected to the second electrode of the third transistor, and a second main current carrying electrode connected to said greater negative voltage point, j) signal input means connected to base electrodes of the first and third transistors to make said first and third transistors conductive in response to positive and negative portions of the input signal, respectively, k) first control means to cause said second and fourth transistors to be nonconductive under conditions where said amplifying apparatus is amplifying a signal of lower magnitude so that output current is at a lower voltage, with the result that power is derived from said second and third lesser voltage points, and to cause said second and fourth transistors to become conductive under circumstances where the input signal is of a higher magnitude so that current at higher voltage is delivered to said output terminal, with the result that current flows from said first greater voltage point through said second and first transistors and from said fourth greater voltage point through said fourth and third transistors, l) second control means operatively connected to said switch means to cause said switch means to be conductive at selected portions of each voltage cycle from said power input terminal means, said control means causing said switch means to be conductive for shorter periods of time for lower power requirements of said amplifying apparatus, and to be conductive for longer periods of time for greater power requirements of the amplifying apparatus.

27. The apparatus as recited in claim 26, wherein said control means causes the switch means to be conductive during a latter portion of each voltage half cycle from said power input terminal means.

28. The apparatus as recited in claim 27, wherein there is rectifying means operatively connected with the power input terminal means and the primary winding to cause only positive current pulses to be directed to said primary winding, said switch means comprising voltage responsive switch means which becomes conductive at predetermined voltage levels of said voltage cycles.

29. The apparatus as recited in claim 28, wherein said switch means comprises a silicon controlled rectifier connected in series between said power input terminal means and the primary winding of the transformer.

30. The apparatus as recited in claim 27, wherein said power input terminal means is connected to the primary winding to cause alternating voltage to be delivered to said primary winding, said switch means being voltage responsive switch means to cause the switch means to be conductive at predetermined voltage levels during latter portions of the voltage cycles.

31. The apparatus as recited in claim 30, wherein said switch means comprises a triac connected in series with said primary winding.

32. The apparatus as recited in claim 31, wherein there is first rectifying means connecting said lower voltage point to the second electrode of the first transistor, so that said lower voltage point has a constant polarity, and there is second rectifying means interconnecting the higher voltage point with the second electrode of the second transistor, so that a single polarity is applied to the second electrode of the second transistor.

33. The apparatus as recited in claim 26, wherein said first control means is responsive to output voltage from said first transistor to said load terminal and operatively connected to a base electrode of the second and fourth transistors in a manner that control current to the base electrodes of the second and fourth transistors begins to flow when said output voltage reaches a predetermined voltage level, so as to cause said second and fourth transistors to be conductive and cause current to flow from said greater voltage points through said transistors to the output terminal.

34. The apparatus as recited in claim 26, wherein said first control means is further characterized in that the first control means supplies base current to the base electrodes of the second and fourth transistors according to a functional relationship of output voltage at the output terminal, in a manner that voltage of the current supplied to the base electrodes of the second and fourth transistors varies as a function of magnitude of the output voltage, with the voltage of the base current to the second and fourth transistors being at a voltage level intermediate the voltage at the higher voltage point and the output voltage, whereby the voltage drop across the second and first transistors is shared between the second and first transistors, and the voltage drop across the fourth and third transistors is shared by the fourth and third transistors.

35. The apparatus as recited in claim 34, wherein said first control means comprises for each of the second and fourth transistors a control transistor having a first main current carrying electrode connected to a related voltage dividing means operatively connected between said output terminal and a related higher voltage source.

36. The apparatus as recited in claim 35, wherein a base electrode of each control transistor is connected to a junction point between a pair of related voltage dividing resistors, which in turn are connected between a related higher and lower voltage source.

37. The apparatus as recited in claim 36, wherein there is a first diode means interconnecting said lesser positive voltage point with said second electrode of the first transistor, and second diode means interconnecting said lesser negative voltage point with said second electrode of the third transistor, whereby when said signal voltage is at a level greater than a voltage level of the lesser voltage points said lower voltage points are blocked of from said higher voltage points.

38. Apparatus as defined in claim 2, wherein said transformer has a secondary to primary turn ratio of below 1,0 with a primary inductance above 30 microhenries and a coil wire gauge diameter above no. 18.

39. Apparatus as defined in claim 2, wherein said audio signal includes first and second stereophonic signals in which the amplitude excursions tend to be in phase, and wherein said power supply means includes first voltage supply means connected with said secondary winding for supplying a voltage having one polarity and a second voltage supply means for supplying a voltage having a polarity opposite of the voltage supplied by said first voltage supply means, and wherein said audio amplifier means includes:
a) a first channel amplifying means for amplifying the first stereophonic signal, said first amplifying means including a Class B amplifier connected with said first and second voltage supply means for drawing power from said first voltage supply means when the amplitude excursions of the first stereophonic signal is of one polarity and for drawing power from said second voltage supply means when the amplitude excursions of the first stereophonic signal is of a second polarity;
b) a second channel amplifying means for amplifying the second stereophonic signal, said second amplifying means including a Class B amplifier connected with said first and second voltage supply means for drawing power from said first voltage supply means when the amplitude excursions of the second stereophonic signal is of one polarity and for drawing power from said second voltage supply means when the amplitude excursions of the second stereophonic signal is of a second polarity;

c) polarity inversion means for inverting the polarity of one of the stereophonic signals to decrease the amount of time during which said first and second amplifying means draw power simultaneously from said first and second voltage supply means when said stereophonic signals are being simultaneously amplified.

40. An audio amplifier system for amplifying an electrical signal having a time varying component within the audio range, comprising:
a) power supply means adapted to be connected with an electrical power supply for producing a plurality of different voltage levels having differing magnitudes, said power supply means including:
1) a transformer having inductively linked primary and secondary coils, 2) a source of commercially available sinusoidally varying supply voltage having a constant frequency, and 3) a power modulating means connected with said primary coil for modulating the flow of current through said primary coil, said power modulating means including switch means connected between said source and said primary coil for operating in a conductive state in response to a control signal for a selected portion of each cycle of said sinusoidally varying supply voltage to cause said selected portion of said sinusoidally varying supply voltage to be applied across said primary coil only when said switch means is in said conductive state;
b) audio amplifier means connected with said power supply means for amplifying the time varying component of the electrical signal by drawing electrical power from the said power supply means selectively from said plurality of different voltage levels; and c) system control means for minimizing the power losses within the system by controlling the operation of said power supply means and said audio amplifier means in response to the time varying component being amplified by said amplifier means, said system control means including:

1) first control means to cause said audio amplifier means to draw current from said power supply means at said voltage level having the lower magnitude when said time varying component is below a predetermined level and to cause said audio amplifier means to draw current from said power supply means at said voltage level having the higher magnitude when said time varying component is above a predetermined level, and 2) second control means operatively connected to said switch means for generating said control signal in response to the time varying characteristic of the audio signal to control the period of conduction of said switch means during each cycle of said sinusoidally varying supply voltage in correspondence with the time varying characteristic of the audio signal to cause said switch means to be conductive for selective time periods which are shorter for lower power requirements of said amplifying means, and longer for higher power requirements of said amplifying means.